Supply Participation. 2.1 Prequalification applications. Rationale

Transcription

1 Supply Participation Rationale 2.1 Prequalification applications Prequalification of existing versus new capacity assets In order to participate in the Alberta capacity market, a new capacity asset must meet the requirements specified in this Section 2 of the CMD. Prequalification, among other purposes, is a mechanism to reduce the risk of non-delivery and is intended to provide the AESO with sufficient confidence that a new capacity asset will be in service in time to deliver its obligation volume during the obligation period, or inform the AESO that a capacity market participant plans to implement changes that could affect the future UCAP of an existing capacity asset. As described in subsections through below, a capacity asset undergoing retrofits may wish to prequalify in order to exempt all or a portion of the asset s UCAP from the capacity market power screen Existing generating assets, including seasonal assets that submit offers in the energy-only market today will automatically qualify to participate in capacity auctions provided the estimated UCAP for the asset is greater than or equal to 1 MW because these assets have demonstrated the ability to provide energy and follow instructions from the AESO System Controller. Automatic prequalification of existing generation will simplify the transition from the energy-only to capacity market. Existing load assets choosing to provide demand response will be required to prequalify as the AESO will require additional information to assess performance based on availability and delivery. Capacity assets under 5 MW of maximum capability do not have to offer in the energy market. However, if a small capacity asset chooses to offer into the energy market it may benefit from having UCAP calculated based on availability factor rather than capacity factor An external capacity asset will be required to prequalify to ensure the external asset can meet the eligibility requirements, such as demonstrating the possession of firm transmission to the Alberta border. Ineligible assets Resources selected for the Renewable Electricity Program (REP) Rounds 1, 2 and 3 are not eligible to participate in the Alberta capacity market as the REP program already provides compensation for the resource s capacity. Eligibility of future REP resources will need to be assessed subject to the contract terms for each round The AESO recognizes that energy efficiency is allowed to participate in capacity markets of other jurisdictions. However, the complexities other jurisdictions have faced with determining a capacity value and assessing performance for energy efficiency requires further study with respect to how this resource can be integrated into the Alberta capacity market. While energy efficiency will not be eligible to participate in the initial implementation of the Alberta capacity market it will be eligible for future participation. This is consistent with the AESO s design criteria for pursuing staged implementation where appropriate. General prequalification requirements A detailed project implementation plan is required as it will be utilized in assessing deliverability prior to commencement of the obligation period, as described in subsection of Section 8, Supply Obligations and Performance Assessments. Advancement through the AESO connection process was considered as a way to track project progress. While this would be feasible for transmission connected projects, it would not be appropriate to other capacity asset projects. Page 1 of 18

2 The commercial operation date of the capacity asset must be no later than the start of the obligation period for the capacity auction that the party is seeking prequalification for, regardless of whether the auction is a base auction or a rebalancing auction. The following list is provided as an example of what may be expected in the project plan submitted to the AESO for prequalification. Note that the list is not intended to be exhaustive. Key Milestones of Project Delivery (a) Development Activities associated with conceptualizing the project, engaging in stakeholder relations, and securing all required approvals and arrangements necessary for proceeding with construction of a facility including: Completion of Land Rights and Site Entitlement Attaining access rights; negotiating and executing lease/purchase agreements or options. Completion of Site Analysis Identifying critical site considerations, such as archeological and heritage sites; conducting geotechnical surveys and studies; evaluating site conditions for constructability. Attaining Project Connection (System Access) Initiating communication with relevant connecting authority; identifying and assessing suitable connection options; developing connection facility application and filing for approval to secure relevant connection agreements. Obtaining Environmental Approvals Identifying and conducting all necessary environmental studies to achieve required approvals. Obtaining Other Permits and Approvals Identifying, applying, and attaining all other necessary permits, and regulatory approvals. (b) Construction Activities associated with building, erecting, constructing, installing, testing, and commissioning of a facility, necessary for attaining commercial operation; including, without limitation: Site preparation and access complete. Facility equipment delivery, set-up, construction, and erection of facility components. Delivery, installation and commissioning of connection facilities and equipment for connecting the generating facility to the electrical system/network. Testing and commissioning of a facility in order to measure the ability of a facility can reach and maintain its capacity obligation. Compliance with permits, applicable law and notifying relevant stakeholders Achieving commercial operation. The AESO will not request information that is not project-related as part of the prequalification application, such as financial strength of the organization or the previous project development and delivery experience of the capacity market participant. The AESO will rely on a financial security requirement, and the monitoring of the achievement of project milestones to mitigate delivery risk. This approach is intended to facilitate the greatest amount of competition from new assets by reducing administrative burden, while still ensuring that the AESO can effectively vet new project developments. Please refer to the rationale for security requirements in subsection below. Asset-specific prequalification requirements Certain assets, such as aggregated assets, may be made of resources of different fuel-types or technologies. To ensure that these compound assets are properly prequalified, the prequalification requirements of all relevant asset-specific categories apply. Page 2 of 18

3 2.1.8 Demand response assets may consist of a large single load asset or an aggregated asset consisting of smaller load sites. Demand response will only be eligible to participate on the supply side of the market. Participation of a demand response asset on the demand side of the capacity market is not contemplated. Incorporating volumes into the demand curve can significantly change the demand curve shape during the capacity auction, adding significant complexity to the auction bidding process and clearing. Demand side participants are encouraged to develop demand response products and receive capacity payments for the ability to curtail load during periods of tight supply. A demand response asset must become a retail or self-retail asset belonging to a valid pool participant to ensure the appropriate metering data is captured and collected for the purposes of performance assessment and settlement. Demand response can take two forms: firm consumption level ( down to ) assets or guaranteed load reduction, ( down by ) assets. Allowing these two types of demand response assets to participate is intended to provide flexibility and incent load participation in the capacity market. Demand response assets must undergo a physical commission test to ensure that these assets can demonstrate: (a) that a relationship exists between the aggregator / provider of load reduction and the actual resources that will curtail load; and (b) ability to receive dispatches from the AESO System Controller Allowing external capacity assets to participate in the Alberta capacity market provides an additional source of supply, and increases market liquidity and competition. The prequalification requirements for external assets are intended to ensure that the capacity of an external capacity asset is deliverable to Alberta during EEA events and tight supply cushion. The AESO originally considered distinguishing external named assets from system external assets, which is the custom in US capacity markets. Given the differences in energy scheduling practices, treatment of transmission constraints and the additional complexities validating data submitted for a named external asset, the distinction was removed, but the prequalification requirements remain. The capacity market participant with an external asset will be an importer in to the energy market with a capacity obligation (i.e., an import with a capacity commitment). The importer will be responsible to acquire the capacity needed to meet their capacity obligation Allowing storage assets to participate in the Alberta capacity market increases overall market competition, provided that their reliability value is appropriately reflected. The prequalification requirement to maintain their energy production at the UCAP level for at least 4 hours is intended to ensure sufficient reliability value from the asset. The requirement for a storage asset to demonstrate the ability to sustain energy for at least 4 hours is derived from the historical observation of the average duration of system stress events, i.e. recent emergency energy alert declarations have lasted on average 4 hours. This does not mean that a capacity committed storage asset will not have an obligation beyond 4 hours; the asset will be required to provide its capacity obligation for the entire duration of the delivery event. Page 3 of 18

4 Table 1 Historical durations of system stress events in Alberta The AESO supports the participation of aggregated assets in the Alberta capacity market because it increases overall market competition and provides an opportunity for assets smaller than 1 MW to participate in capacity auctions. Stakeholders have suggested that the maximum UCAP size for aggregated capacity assets should be no greater in size than the single largest contingency (currently 466 MW), in order to prevent reliability issues, potential increases in the operating reserve requirement and any potential impact to capacity auction clearing. The AESO notes that the size of aggregated assets is not related to real time operational risk as each individual component resource of the aggregated asset is operated and dispatched independently (see below). However, the AESO supports initially limiting UCAP size for aggregated capacity assets to be no larger than the single largest existing generation capacity market asset UCAP to ensure that the introduction of aggregated capacity assets does not result in unanticipated capacity market distortions. The AESO will leverage the existing load settlement processes to ensure accurate and consistent measurement of aggregated demand response capacity assets. This restriction will limit aggregation of individual component resources to the pre-defined load settlement zones. At this time, it is anticipated that most aggregations will be demand response component resources. To be compatible with the AESO s load settlement process, sites must first be aggregated to an asset at the settlement zone level. Two or more zonal assets may be further aggregated into a single capacity asset with a single UCAP. To align with AUC Rule 021: Settlement System Code Rules, commercial and retail load participating in the capacity market must be aggregated to a retailer/self-retailer asset. For the purpose of aggregated demand response assets, all sites associated with that demand response asset will be considered individual component resource sites in the demand response program run by that retailer. The sum of the metering for all sites forms the basis of the UCAP, performance measurements (availability and delivery). The retailer is responsible for both energy market and capacity market settlement of the sites associated with the settlement zone asset in the aggregation. In other words, a demand response component site must have the same retailer for the capacity market and the energy market. Figure 1 provides an illustration of aggregating individual component assets across multiple settlement zones. This example only applies to load and small generation connected on the distribution system and subject to load settlement. Page 4 of 18

5 Figure 1 Aggregated capacity asset across load settlement zones If the individual components of an aggregated capacity asset meet the dispatch requirements in the energy market, the components of an aggregated asset may choose to be dispatched and operated as individual and separate energy market assets. However, the aggregation of generation assets in the energy and ancillary services markets will remain as described in the definition of aggregated generating facility in the Consolidated Authoritative Document Glossary that will be effective on September 1, 2018, and Section of the ISO rules, Transmission Loss Factors A refurbished capacity asset is considered a new capacity asset to incent asset life extension and potentially increase UCAPs. New capacity assets will not be assessed a default offer cap as described in Section 7, Capacity Market Monitoring and Mitigation if the capacity market participant fails the capacity market power screen. In order to determine if modifications to a capacity asset are significant enough for the asset to be considered a new capacity asset, a threshold test was developed using the same approach as North American and United Kingdom (UK) capacity markets. The thresholds levels for refurbished capacity assets in Alberta are based on a volumetric and cost analysis of recent or proposed refurbishments in Canada. Based on a range of different refurbishments across various technologies, the lower range of annualized refurbishment costs are approximately $45/kW-year. Assuming a useful capital life of at least 5 years, this equates to a capital cost of $200/kW. The $200/kW threshold is expected to be low enough to include some coal-to-gas conversions of the existing coal fleet as refurbished assets. The threshold is needed such that it represents a significant capital investment in the facility such that the refurbishment offer process is not abused as a way for units, which are not making material changes or whose cost structure is such that they can be effectively represented even under a standard mitigated offer, from being given an effectively free option to utilize a low risk economic withholding strategy in the market As described in the rationale for Section 7, Market Monitoring and Mitigation, market power mitigation will be implemented to limit the negative effects of economically withholding capacity volumes. Submitting a refurbishment plan should not facilitate the economic withholding of capacity volumes. A capacity market participant with market power that submits a refurbishment plan that is approved by the AESO is able to offer that capacity volume for the refurbished asset Page 5 of 18

6 at an unmitigated price. A capacity market participant will be required to indicate as part of the prequalification application, for each refurbished asset, whether the asset will: (a) permanently delist and retire if it fails to receive an obligation; or (b) not permanently delist and submit a mitigated offer for the existing asset to be used in the capacity auction in the event the unmitigated offer does not clear. The one-time use of option (b) is in place to prevent a free option that allows participants to submit a prequalification application for refurbishment each year and be permitted to continually have an opportunistic higher priced offer in the supply stack while mitigated. The introduction of this approach for refurbishments will require a multi-stage clearing process for the capacity auction whereby the unmitigated offers for refurbished assets which do not clear are added back into the supply curve at their mitigated price and the market clearing process is run again. This process will continue until all refurbishment assets have cleared the market or offered on a mitigated basis. To simplify the auction clearing process, unmitigated refurbishment offers will be required to be single, inflexible blocks. For example, in the illustration below three participant assets (A, B and C) submitted prequalification applications for refurbished assets and declared option (b). Figure 2 Multi-stage auction clearing process In this example the auction clearing engine iterates twice, each time determining the unmitigated offers that do not clear and replacing them with the mitigated offers. In this example a solution is found when the unmitigated offer from asset C clears in the auction If a firm subject to capacity market power mitigation enhances a capacity asset to add incremental capacity, the incremental volume will not be mitigated. However, the existing capacity will be subject to the mitigation cap. The rationale for the threshold values for incremental capacity assets are based on a threshold analysis. This analysis included a review of historical uprate volumes in Alberta, documented in the table below. The results indicated two uprate ranges: (1) in the 3 to 5% (12-19 MW) increase range; and (2) in the 12 to 15% (44-53 MW) increase range. Page 6 of 18

7 Table 2 - Summary of Alberta historical uprate maximum capability Since the smaller uprates require much lower capital additions relative to major uprates, the larger range of uprate volumes will be used to determine the volume addition threshold above which the entire asset will be considered a refurbished asset. The 15% and 40 MW incremental threshold described above is comparable to ISO-NE s New Capacity Eligibility requirements which state that: Uprate (increase above threshold) An existing resource can qualify as new if it is proposing an increase in output that is the greater of 20% or 40 MW above the existing qualified capacity. 1 Based on the AESO s analysis summarized above, the following thresholds will apply in Alberta: Assets whose maximum capability is increasing by at least 15 per cent or 40 MW (whichever is greater) or making a capital investment of at least $200/kW (across all installed MWs) may submit an unmitigated offer into the capacity market for the entire UCAP of the asset. Assets whose maximum capability is increasing by at least 1 MW and less than the 15 per cent or 40 MW threshold is classified as incremental. The incremental capacity will be unmitigated in the capacity auction. 1 Page 7 of 18

8 Figure 3 Qualification thresholds for refurbished and incremental capacity The AESO will consider a 1 MW increase in maximum capability as incremental capacity but it is possible that a 1 MW increase in maximum capability will not cause an increase in UCAP. Security requirement for new capacity assets The security requirement for a new capacity asset must balance the tradeoff between creating barriers to entry and ensuring that a capacity market participant is properly incentivize to deliver on its capacity obligation at the start of the obligation period. New capacity assets include assets that are not in service at the time they clear a capacity auction, assets planning to undergo refurbishment before the obligation period, and assets intending on adding incremental capacity before the obligation period. A number of factors may potentially interfere with the new capacity asset s ability to be in service at the beginning of the obligation period. These may include: delays in permitting, failure to secure financing, delays in equipment delivery, delays in construction and equipment installation, issues with installation that lead to a lower than expected capacity rating, and insolvency of the developer. The security requirement is intended to mitigate the risk of nondelivery by: (i) recovering the replacement capacity costs it incurs due to non-delivery through rebalancing; and (ii) creating an additional incentive for capacity assets to physically deliver. The security requirement for the Alberta capacity market is similar to those utilized in other situations where there is a risk of a supplier meeting a future obligation. For example, construction project performance bonds are normally a percentage of the invested project cost, and are a well-established mechanism for mitigating delivery risk. As outlined below, the security requirement for the Alberta capacity market will decline through time as project milestones are reached. The quantitative reduction in credit requirement with each completed milestone is intended to reflect the associated reduction in non-delivery risk. Reducing credit requirements as milestones are completed also provides incentive for resources to adhere to development timelines. Security requirement for brand new assets = (gross-cone * 1/CRF) * 5%. The capital recovery factor (CRF) delineates the number of years over which project investment can be recovered, and consequently how much of the project investment is recovered annually. For brand new assets, the CRF will use 20 years for n. For Page 8 of 18

9 refurbished and incremental assets, the threshold costs described in this subsection will be used. The 5% value is based on Surety Association of Canada s guideline for performance bond range from 4% to 10% of invested capital. 2 The following is intended to provide an example calculation of the security requirement for a brand new 100 MW asset using an estimated gross-cone of $148/kW: Figure 4 Declining security requirement i. CRF, assuming a 8.2% discount rate = {.082( )^20} / {[( )^20]-1} = or 10.3% ii. Security requirement = $148/kW *1/CRF = $1437/kw or $1.437m/MW * 5% = $72k/MW iii. Security requirement rate = $72k/MW / max number of auctions before the delivery year (6) = $12,000/MW iv. Declining security requirement = security requirement rate * Obligation * number of remaining auctions before the obligation period: prior to 2025 B (2022) = (.012 * 100 MW) * 6 = 7.2M prior to 2025 R1(2024) = (.012 * 100 MW) * 4 = 4.8M prior to 2025 R2 (2025) = (.012 * 100 MW) * 1 = 1.2M after commercial operation = (.012 * 100 MW) * 0 = 0M The declining security requirement is contingent on the project meeting its declared project milestones. In the event the project schedule is at risk of delivering in time for the obligation period the financial security will not be returned to the capacity market participant and will be held by the AESO until the project is back on schedule. If the capacity market participant is required to buy back its obligation in a rebalancing auction, the security will be held to guarantee the payment of the rebalancing costs during the obligation period. Refurbished assets will use the $200/kW cost threshold described in section above instead of 1/CRF as the life extension of a refurbished asset is not expected to last 20 years. Incremental capacity will use a $100/kW cost threshold derived from an AESO analysis of historical cost of capacity uprates in Alberta and jurisdictional comparisons. The AESO developed its security requirement proposal in consideration of Alberta s unique needs, and after reviewing the credit rules used in the PJM, ISO New England (ISO-NE), and UK capacity markets. Table 4 below summarizes the capacity market credit requirements in these other markets. Looking at other capacity markets, ISO-NE and the UK have comparable credit requirements. The PJM credit requirement is much higher, which may stem from a more conservative approach to credit risk, 2 Page 9 of 18

10 and the fact that PJM will increase a market participant s unsecured credit limit if they earn net revenues in PJM s markets. 3 Table 3 Capacity market credit requirements for new resources Component PJM 4 ISO-NE 5 UK 6 Applicability New capacity only New capacity only New capacity only Credit Requirement (After Clearing a Forward Auction) ~50% Annual net-cone 8.3% Annual Clearing Price 10,000/MW (~12% Annual net- CONE) Adjustment of Credit Requirement Over Time Increases with each forward auction cleared for separate delivery years Increases with each forward auction cleared for separate delivery years Increases with each forward auction cleared for separate delivery years Unsecured Credit Limit Increases with credit rating, net worth, and historical net revenues across all PJM markets Increases with credit rating, and net worth N/A Maximum: $50 million per market participant Maximum: $50 million per market participant Acceptable Forms of Secured Credit Cash Letter of Credit Cash Letter of Credit Cash Letter of Credit Mutual Fund Shares The AESO s proposed capacity market security requirement for new capacity assets aligns with its existing credit policy which specifies limits on unsecured credit, and the acceptable forms of secured credit for participants across the AESO s markets. At this time, the AESO does not believe that any change is required to its current guidelines on unsecured credit or acceptable forms of secured credit. Prequalification of a new capacity asset An application is required to properly assess a new capacity asset against the eligibility criteria for prequalification Unsecured credit limits in U.S. markets were tightened by the Federal Energy Regulatory Commission (FERC) following the 2008 financial crisis when U.S. RTOs faced severe credit stresses. In Order 741 and its subsequent modification, the FERC limited unsecured credit to $50 million per market participant. In both U.S. markets, unsecured credit limits increase with the credit rating of the market participant and its net worth up to the $50 million limit. PJM also allows a market participant s historical net revenues across PJM s markets to count toward its unsecured credit limit for the purposes of its capacity market credit requirement. These differences across markets may be explained by differences in volatility. Generally, a more volatile market will require higher credit requirements. Credit Overview and Supplement to the PJM Credit Policy, October 6, Exhibit IA, ISO New England Financial Assurance Policy, June 1, Applicant s Credit Cover Process, July 6, Government Response to the March 2016 consultation on further reforms to the Capacity Market, Page 10 of 18

11 Prequalification is intended to: (i) ensure that a new capacity asset meets the minimum standards for a capacity asset; and (ii) verify that a proposed project prequalification package contains all of the supporting information necessary to assess delivery progress, apply the appropriate security requirements, and determine the form of capacity asset in order to apply the correct UCAP, availability, and performance methodologies to the asset A prequalified asset will be assigned a UCAP for the upcoming auction. If the UCAP for that asset is less than 1 MW the asset is not qualified to participate in the upcoming auction. However, it will remain prequalified for future auctions given that improvement in performance can increase UCAP above the minimum sizing requirement. Prequalification for subsequent auctions Prequalification of an asset is a one-time step, unless circumstances surrounding the asset change. Allowing prequalified resources to remain eligible for future auctions until delisted, modified, or deemed ineligible by the AESO gives the AESO certainty on resources that will participate in the auction and in determining supply adequacy. This approach also reduces the administrative burden of prequalifying resources each year for every base auction and rebalancing auction. The modifications mentioned above include changes in self-supply status, refurbishment, and the addition of incremental capacity. A capacity market participant with incremental capacity must submit an incremental capacity prequalification application to the AESO prior to the auction in order for that incremental capacity to be included in the UCAP determination. 2.2 Self-supply designations The concept of self-supply, a best practice found in other capacity markets, was leveraged to accommodate existing cogeneration and other sites in Alberta where load is served by onsite generation. Such sites account for approximately 2,000 MW of generation. This also recognizes the unique nature of Alberta's system Sites must be able to physically deliver capacity to the rest of the grid in order to meet the criteria that capacity contributes to reliability and is a physical product. The rationale for requiring certain sites to self-supply is as follows: (a) The City of Medicine Hat is a site with onsite generation that is net metered at the connection to the Alberta Interconnected Electricity (AIES) system, and cannot physically flow their gross generation volumes due to system connection limitations. They must therefore self-supply. This includes generation not owned by the City of Medicine Hat located within the city limits. (b) Sites that do not have revenue quality metering at the generator terminus cannot be measured accurately for the purposes of capacity market settlement. (c) The Alberta capacity market is a physical market. The original criteria and assumptions for the design of the capacity market state that a capacity obligation is a forward obligation on capacity suppliers that requires the capacity sold in the market to be available to provide energy production or reduced consumption when needed. Based on this statement, sites with onsite generation that are net-metered and cannot physically flow their gross generation volumes to the grid due to system connection limitations must self-supply because they cannot physically deliver additional MWs to the system greater than that based on physical transmission limitations. Not all sites under this configuration are cogeneration sites and some manage their load with their own generation investments A site may choose to self-supply capacity provided they have a bi-directional net-interval meter at the connection point to the system. The bi-directional meter is necessary to accurately measure the net-to-grid energy in order to ensure delivery. Alberta's market does not have integrated utilities acting as load serving entities, as found in other capacity markets, but over 20% of the internal load is served by onsite generation. The capacity market design for Alberta must include consideration for this form of participant. Self-supply provides the market with a methodology to Page 11 of 18

12 deal with behind-the-fence (BTF) 7 locations with limited transmission capability. In addition, the ability to self-supply allows cogeneration sites that are tied to a host customer s load to be exempt from offering all of its capacity into the AESO-operated capacity market The 4 year requirement is intended to align with the proposed timing of the approval of the demand curve parameters to increase market certainty with respect to the demand curve. If the demand curve review cycle is shorter or longer than 4 years, the AESO will adjust this requirement accordingly. Self-supply volumes are not included in the procurement volume and therefore, the choice of whether or not to self-supply will impact the procurement volume. Stakeholders have suggested that the 4 year requirement is too restrictive and does not align with changes in operation and market conditions. In response to this feedback, the AESO will permit self-suppliers and capacity market participants to submit a change in self-supply status inside of the 4 years provided the self-supplier participant can demonstrate a physical change to the operation of the site. This allows reasonable certainty to be incorporated into the capacity market while ensuring that the capacity market structure does not negatively interfere with the business decisions of self-suppliers Stakeholder concerns related to self-supply An independent load and generator may pay, and are paid differently, from sites that are combined load and generation. Using a simple settlement example for the capacity market, it can be demonstrated that a site that is self-supplied will be allocated less of the reserve margin than a similar load without the ability to self-supply. To demonstrate the payment difference when comparing gross settlement to net settlement, Table 5 below provides a simple system with four cogeneration sites (I1 through I4), 1 pure load site (I5) and one new entrant pure generator (A1). This example assumes a reserve margin requirement of 15% as the additional amount to procure in the capacity market to ensure reliability. The internal load of this system is 44 MW, adding an additional 15% brings the procurement target for this system to 50.6 MW of capacity less 18 MW of self-supply equaling 32.6 MW. The volume of self-supply is calculated as the difference between the sites gross load and its net load. The size of the resource procured to serve this sample of load portfolio is calculated as difference of the necessary amount for the gross load minus the sum of the generators UCAP. Once the capacity market clears, the load will be allocated the cost of the capacity procured. The cost allocation formula used here is the total payment to all capacity assets multiplied by the load of the site divided by the total load of all sites. The illustrative example includes both a gross load and a net load calculation. The payments that generators receive in this illustrative example assume a capacity market price of $40/MW (over a particular obligation period). The capacity payment is simply the capacity obligation multiplied by capacity market price. The example includes both a gross generation and a net generation calculation. The results of the example show that by allowing netting of the generation out of the load: (i) the rest of the load on the system (pure load represented by I5) will pay more than it would if netting were not allowed; and (ii) the loads that have cogeneration sites would pay less if short of generation, or the generators would be paid more if long on generation. Currently in Alberta 20% of the gross load is selfsupplied. 7 The AESO 2017 Long-Term Outlook defines BTF as industrial load served in whole, or in part by onsite generation built on the host s site. Page 12 of 18

13 Table 4 - Gross vs net settlement for self-supply In this example, the difference between the gross and net calculation, an additional 2.08 MW, is allocated to the pure load at a cost of $83 ($1003 to $920). This is because the netted load is not carrying their reserve requirement under the same level of reliability criteria as the rest of the system. The rationale submitted by the cogeneration owners for this acceptable difference is that cogeneration provides a reliability benefit due to the fact that the load and generation are tightly coupled. When looking in aggregate at Alberta industrial systems there is a correlation between the load and generation. In Figure 4 below, it is apparent that as the generation at the site drops the load drops too. This correlation makes sense as, by definition, the electricity is a by-product of the steam used in the industrial process. If no steam is generated, then no generation output is provided and no industrial process is supported by the cogeneration. Historical analysis of industrial system designation sites from 2012 to 2017 showed the reduction in generation was roughly 500 MW greater than the decrease in load. This is partially due to the fact that some industrial system designation sites are not cogeneration sites. Determination of self-supply capacity Self-supply volume is the difference between a site s gross load and net load. Depending on when that difference is measured the value can change dramatically. No reliability risks exist if it can be assured that in the event of generation failure during a performance event the load will be at its net-load volume levels. However, examination of historical individual net site behavior has not demonstrated this in all cases. Loads are not always reduced when the generation is down and we find net loads increase to gross load levels in some instances. In the example graph below, when the generation at the site (blue line) goes to or approaches zero, the net load at the site (green line) increases. The gross load at the site (purple line) remains relatively constant. Further analysis indicates 7 of the 15 current industrial system designation sites demonstrate a high correlation of their load and generation. Figure 5 Net and gross measurements at a self-supply site Page 13 of 18

14 Due to this observation, the AESO proposed four options to industry for mitigating this risk in the form of the following question. How should the AESO determine how much capacity to procure for self-supplied load? Four options are listed below: 1. The AESO does not procure capacity for the netted-out load and requires the net load to be curtailed during delivery events if not meeting their delivery obligation. 2. The AESO does not procure capacity for the self-supplied load, but charges the load at the value of lost load plus the curtailed loads capacity payment (liquidated damages) if they rely on the system under shortage events. 3. The AESO procures some capacity based on a probabilistic assessment of each self-supplier s dependence on the system s capacity market. 4. Apply the cost allocation formula to net load only. If a self-supplier takes capacity in a prior year they pay for it in the future year. Option 1 is a true form of opting out of the market and would not compromise reliability. However, there are very few self-suppliers that could utilize this option, and the cost of mandating this on all sites would be prohibitive. Options 2 and 4, which are variations on a similar theme, provide a financial incentive for self-suppliers to make sure assets manage their consumption during delivery events. The most important difference is that Option 2 sets a maximum load obligation that is assessed during delivery assessment periods, whereas the Option 4 cost allocation method is tied to cost allocation periods. Option 3 is a combination of Option 4 plus an additional premium, equal to some fraction of the system reserve margin percentage, placed on the selfsupplied load to cover the risk of the load exceeding typical net levels during delivery events. This was seen by some stakeholders as incurring a double cost allocation and by the AESO as a highly administrative calculation requiring actuarial science to determine the right premium. Option 4 was seen by a majority of the working group as the simplest method to manage selfsupply as it is consistent with the current energy market treatment of generator station service load and net-measured sites. Some members felt this mechanism did not adequately address the reliability issue. The reliability concern comes from two places: (i) the method of cost allocation may not provide proper incentives for self-supplied load to not consume during system stress events if there is no alignment of delivery events and the times where costs are allocated; and (ii) the net load is highly variable, and most sites can incur non-coincident peaks in the hundreds of MW even though net loads are mostly in the tens of MW range. With the high variability of net loads combined with the fact that these loads are large, the treatment of self-supply must ensure that appropriate incentives are in place to discourage self-supply loads from consuming during the capacity delivery assessment periods. To not do so could present a reliability risk. Weighted energy cost allocation and self-supply Generation used for self-supply can participate in the capacity market only on a net-to-grid basis, while the load it supplies will be subject to capacity market cost allocation based only on net-togrid consumption. Concerns have been identified with the potential for self-supply loads to increase consumption due to onsite generation being off-line under the weighted energy methodology for cost allocation. The weighted energy methodology for cost allocation reasonably and fairly apportions capacity market cost to loads that operate in a predictable and consistent manner. The methodology can also be compatible with creating incentives to ensure that self-supplied loads have sufficient incentive to curtail during conditions when onsite generation is reduced. For example, to provide an incentive for self-curtailment, a rate could potentially be designed whereby additional costs are allocated to loads when net-to-grid consumption is significantly higher than average under defined conditions. The additional cost allocation would not impact loads that operate with a normal load profile that do not exhibit periods of intermittent high consumption. The additional cost allocation would be expected to account for a small percentage of the total cost of the Page 14 of 18

15 capacity market. A market participant could avoid incurring the additional cost allocation by avoiding intermittent periods of high consumption. This is the incentive the rate design is intended to provide. Additional details will be developed by the AESO and subject to further consultation. 2.3 Delisting For market transparency purposes, prequalified capacity assets that cannot participate in the Alberta capacity, energy and ancillary services markets for physical or economic reasons are required to temporarily or permanently delist from the Alberta capacity market. For clarity, participation refers to supply participation in the Alberta capacity, energy and ancillary services markets (i.e., providing energy production or demand response). A load that applies, prequalifies and obtains a capacity obligation to provide demand response is considered a supply of capacity The capacity market participant will have a must offer requirement in the energy market if the asset is committed to provide 5 MW or more in the capacity market in accordance with Section 10, Roadmap for Changes in the Energy and Ancillary Services Markets and a must offer in the capacity market in accordance with Section 5, Base Auction. As such, these demand response assets will be required to delist should they choose to no longer offer demand response. When a load delists it does not mean the load must no longer consume electricity or lose the ability to self-curtail their consumption. Self-suppliers that are net load and choose not to provide demand response are not considered to be participating capacity supply Capacity assets that are currently on extended mothball outages under Section of the ISO rules, Mothball Outage Reporting ( Section ) will be required to submit a temporary or permanent delist bid during prequalification for the first obligation period in order to remain offline. This will increase market information and transparency and will also facilitate the transition from Section 306.7, which will be amended to align with the delisting process. AESO review of impacts to the reliability of the interconnected electric system The AESO may review delisting submissions for reliability impacts and supply adequacy issues to ensure the safe, reliable and economic operation of the AlES. Temporary delist request for economic reasons Temporary delisting bids for economic reasons may be submitted during the prequalification period associated with the second rebalancing auction. The economic delist bid may be submitted after the asset has participated in both the base auction and the first rebalancing auction. It is only after participating in the base and first rebalancing auctions that a firm will be able to determine that the capacity asset has not earned sufficient revenue to remain economic. This notice period is in line with the notice period in Section The AESO acknowledges stakeholder feedback surrounding the requirement for a capacity market participant to offer into the base and first rebalancing auction even when it plans to temporarily delist its asset for economic reasons for the upcoming obligation period. As long as the offer reflects the net avoidable costs of temporary economic delist, the requirement that a temporary economic delisting request can only be submitted before the second rebalancing auction should not take away the opportunity for a capacity market participant to make arrangements to prepare, even in advance of the base auction, to temporarily delist the asset for the upcoming obligation period. In addition, the requirement to submit a temporary economic delist request during the pre-auction period for the second rebalancing auction should not have an undue impact on the market outcome. If the offer clears in the base auction or first rebalancing auction for the upcoming obligation period, there is no economic reason for the capacity asset to delist for the upcoming obligation period. If the offer does not clear, the temporary economic delist request for the upcoming obligation period may be submitted before the last rebalancing auction relevant to the subsequent obligation period. As such, allowing economic delist bids for the second rebalancing auction only does not restrict an asset from economically delisting for more than one obligation period. Page 15 of 18

16 2.3.6 In order to verify that the temporary economic delist request is not for the purpose of removing an asset from the market in order to increase capacity or energy prices to benefit the firm s portfolio, the AESO requires a firm to submit net avoidable cost data as described in subsection of Section 7, Capacity Market Monitoring and Mitigation and an attestation from a corporate officer of the firm. If the capacity asset will be available in the energy and ancillary services markets for a portion of the obligation period, the net avoidable cost submission associated with the temporary economic delist request must reflect that fact. For example; labor costs are not avoidable while participating in the energy market Allowing a temporarily economically delisted capacity asset to participate in the energy and ancillary services markets for a portion of the obligation period allows flexibility to capture energy market opportunities. However, a temporary economic delist request suggests a capacity asset is intended to be mothballed due to economic reasons. Participating in the energy and ancillary services market for a prolonged period would not be consistent with the intent of temporary economic delisting. Therefore, a 5 month limit on the duration of participation in the energy and ancillary services markets will be placed on such assets In the event of potential or eminent supply shock or sustained supply tightness caused by unplanned events, the AESO may permit a temporarily economically delisted capacity asset to delay the start of the outage or return to the energy market before the end of their outage term. Should these types of events occur, energy prices are expected to increase and the economic assumptions that went into the delisting decision may have changed to where the asset may not have delisted with foresight of these events From a marginal cost perspective, a capacity market participant would offer into the capacity auction at the net-avoidable costs of such asset. It is those costs the capacity market participant needs to recover in order to operate profitably in that obligation period. If the market clears above the asset s net-avoidable costs it will be economic for the capacity asset to remain active A capacity asset may not temporarily economically delist for more than two consecutive obligation periods in order to facilitate the optimal use of the existing transmission system and prevent price distortion caused by the uncertainty with respect to the duration and timing of a delisted capacity asset s return to operation. The requirement to submit a temporary economic delist request for each of the two periods separately is to prevent potential supply condition changes in the market from one obligation period to another and distortion of investment signals as a result of the uncertainty about the capacity delisting duration and the timing of its return to operation. Temporary delist request for physical reasons A capacity market participant is able to temporarily delist an asset if it is physically unable to operate. A 5 consecutive month period was chosen because this duration is greater than the average duration of a planned maintenance outage in Alberta, which is 2 to 3 months per year. Additionally, 5 months is short enough to enable seasonal assets to participate, and long enough to ensure the seasonal asset provides adequate value for reliability. A 5 consecutive month period is intended to preserve incentives to plan regular maintenance in periods of lower system risk while recognizing that longer than typical outages really mean a unit is unavailable and not expected to provide sufficient reliability. The obligations related to temporary physical delisting apply to all capacity asset types. For example, if an aggregate demand response asset loses individual component resources and those components cannot be replaced to maintain its capacity obligation, then the capacity market participant would be expected to delist due to a physical operational restriction thereby reducing performance risk Physical delisting presumes that an asset will not be physically operational for at least 5 continuous months of the obligation period. The AESO will approve physical delisting bids without the requirement for an economic review. If a capacity market participant wishes to delist a capacity asset for a period greater than 5 months but less than 12 months, the capacity asset is delisted from the capacity market for the entire obligation period. When the asset returns to Page 16 of 18

17 service it will be required to participate in the energy market. For example, a capacity asset that physically shuts down January 1 of a calendar year (the third month of the obligation period) and starts up again 5 months later, would be delisted from the capacity market for the entire obligation period but will be required to participate in the energy market from November 1 until December 31 and from June 1 to October 31 of that obligation period. In other words, a capacity asset that has received approval to temporarily delist due to physical reasons may participate in the energy and ancillary services markets during the period that the asset is physically capable of providing energy (able to function) assuming that period is less than or equal to 7 months of the obligation period, as shown below. Figure 6 Physical delisting participation in the energy and ancillary services markets Temporary delist from Obligation period 5 month outage Energy and AS participation Energy and AS participation The asset will not receive a capacity obligation for the obligation period but will be able to participate in the energy and ancillary services markets In the event of potential or eminent supply shock or sustained supply tightness caused by unplanned events, the AESO may permit a temporarily physical delisted capacity asset to delay the start of the outage or return to the energy market before the end of their outage term. Should these types of events occur, energy prices are expected to increase and the economic assumptions that went into the delisting decision may have changed to where the asset may not have delisted with foresight of these events. In addition, resources which may be able to enhance reliability should be made available The restriction of up to two consecutive obligation periods is applied to temporary physical delisting to facilitate the optimal use of the existing transmission system, as well as ensure that the capacity market price signal is an effective investment signal. Permanent delist notifications Because a permanent delisting decision is a long-term one and is likely not dependent on the price outcome of a single obligation period, permanent delisting notifications are allowed to be submitted during the base or first rebalancing auctions prequalification processes. Permanent delisting notifications may not be permitted during the last rebalancing auction in order to mitigate the potential risk of the market not having sufficient time to react to permanent change in supply conditions. The asset retirement date does not need to occur at the start of the obligation period. A capacity market participant that is currently participating in the energy market and is intending to permanently delist, if the asset has a must offer requirement in the energy or ancillary services markets, must participate in the energy market until to the physical retirement of the asset. If the retirement start date is greater than 7 months into the period, the asset may still receive a UCAP and be required to offer into the capacity market. Permanent asset delisting is an irreversible process and once the application is received by the AESO the asset will be required to retire on the declared retirement date. A capacity market Page 17 of 18

18 participant should consider temporary delisting if it would like to return the asset back to operation at a future period. This approach ensures the market has clear information in order to provide effective and accurate investment signals The same rationale for temporary physical delist in section applies to permanent delisting as well Demand response and external capacity assets may permanently delist. If the capacity market participant chooses to return to the capacity market they must prequalify as a new asset. This requirement is to recognize the physical difference between generating assets and external assets and loads. Generating assets in Alberta which permanently delist are physically dismantling while the other asset types may physically still exist (e.g. loads still consuming, wires still available to deliver energy from external sources) and may be able to return to the market at a future date The AESO will not review permanent delisting notifications for acceptability of financial data and cost information. The AESO, in response to multiple stakeholders, agrees that a legal owner of a capacity asset is entitled to make their own judgement about the economic viability of their assets and whether to retire them permanently. 2.4 Physical bilateral transactions Physical bilateral transactions will not be permitted in the Alberta capacity market. Physical bilateral transactions take place outside of the centralized capacity market. If permitted, buyers and sellers would find each other (i.e., self-matching) and report their matched commitments to the centralized market (i.e., the AESO) prior to a capacity auction. Contract prices are not reported, but remain private information between the buyer and seller. This practice negatively impacts the size of the centralized market by potentially reducing liquidity, thereby making the market less competitive. Prohibiting capacity assets and load to arrange for capacity outside of the market through physical bilateral arrangements promotes liquidity and competition in the market. The design requires the capacity market to achieve the desired reliability objectives through a real and measurable supply adequacy product that still respects the unique aspects of Alberta s electricity system. Page 18 of 18

19 Calculation of Unforced Capacity Ratings (UCAP) Rationale 3.1 Calculation of UCAP UCAP is intended to represent the reliability contribution of a capacity asset s physical reliability during tight supply market conditions. UCAP captures an asset s observed operational performance over a defined historical period, including its performance during system scarcity conditions. Adopting UCAP as the standard capacity product for the Alberta capacity market creates a consistent and measurable supply adequacy product that allows different technologies to compete on a level playing field (i.e., 1 MW of UCAP should deliver the same amount of reliability regardless of the underlying technology). The AESO considers that it is the entity best suited to perform UCAP calculations, given that the AESO is an independent entity that does not own or operate any power facilities, does not have a financial stake in the electricity industry, and currently collects the data needed to perform UCAP calculations. Furthermore, given that the AESO is responsible for maintaining system reliability, the AESO is well positioned to ensure a common approach for assessing the reliability value of each asset. The AESO will calculate an annual UCAP to align with the capacity obligation period. Feedback from stakeholders identified that a seasonal UCAP and seasonal obligation period could introduce the following complexities: (a) reduced investor certainty due to the difficulty in forecasting capacity market revenues; (b) difficulties associated with the need for a higher price cap in a seasonal auction, which is required for assets that might only clear one season but require a full year s worth of capacity revenue to remain in the market; and (c) the estimation of a seasonal capacity volume and seasonal UCAP becoming increasingly difficult as the period of estimation becomes more granular given the data available to the AESO. The intent of using a UCAP is to link the reliability of a capacity asset to the individual performance of that asset during tight supply conditions, creating an incentive for the capacity market participant to maintain high availability for the capacity asset when the system needs it the most. A capacity asset that performs, on average, better than or equal to its UCAP during periods of system stress may receive a higher UCAP for future auctions. The AESO will calculate UCAP by averaging the available capability or the metered volumes (capacity factor) of assets during hours with tight supply cushion over the previous five years. The availability capability or metered volumes determination for assets is described below. This approach has the following benefits: Using historical performance of assets during times when capacity was required provides an expectation of what the assets can be relied on to provide in the future. This approach implicitly captures the correlation of each capacity asset s capacity value and the drivers of tight supply cushions, such as: seasonal load, seasonal derates, seasonal output levels and planned outages. These historical observations provide the AESO confidence on each asset s contribution to supply adequacy without having to establish complex assumptions and modelling relationships that would be required in Page 1 of 20

20 alternate approaches such as Effective Load Carrying Capability (ELCC) or Equivalent Forced Outage Rate (EFORd). This methodology is simple and replicable allowing an asset owner to have clear signals on how to increase its asset s UCAP. This approach is connected to supply cushion hours and is directly aligned to Alberta capacity needs. Further the alignment with the performance assessment approach sends the right incentives to asset owners to maintain and increase the UCAP rating of their assets. Given Alberta s high load factor, planned outages can drive supply shortfall situations. By leaving all the drivers of availability, including planned outages, in the calculation the approach ensures that capacity market participants are measured on a full suite of resource adequacy contributions or limitations. The AESO recognizes that using historical data has limitations. For example, incentives were different in the past and history may not be a perfect indicator of the future. Tight supply periods may change in the future to be more heavily weighted in different periods than have occurred in the past. A UCAP determination approach that is based on five years of historical data will always have some amount of lag in reflecting changes to an asset s UCAP. The UCAP refinement process described in subsection 3.2 of CMD Final is intended to provide capacity market participants with an opportunity to request changes to the hours or data used to calculate an asset s UCAP due to specific circumstances listed in section 3.2, provided sufficient evidence is produced. The AESO explored methodologies for measuring reliability in other capacity markets, including approaches based on installed capacity, effective load carrying capability and equivalent forced outage rate: (a) Installed capacity (ICAP). ICAP reflects the nameplate capacity of an asset adjusted for temperature derates. Using the ICAP approach for determining the capacity value of assets may create an adverse selection problem for the AESO, where assets with lower performance and poorer reliability will clear the auction because they are able to bid at lower costs (by saving on plant maintenance) relative to the higher performing assets that are more expensive because they invest in their maintenance. ICAP may overstate an asset s ability to provide capacity during tight supply cushion hours since it does not account for forced or planned outages and other derates. (b) Effective Load Carrying Capability (ELCC). ELCC measures the capacity of an asset by simulating the asset s contribution to system reliability. This is accomplished by calculating the unserved energy expectation of two different scenarios, one with and one without the asset. The AESO will not be using an ELCC approach for the initial implementation of the Alberta capacity market because this approach is less transparent and far more complex to implement than the chosen UCAP methodology. Due to the large number of modelling inputs required to complete this analysis, capacity market participants would likely not be able replicate the UCAP that the AESO calculated for its asset. (c) Equivalent Forced Outage Rate (EFORd). EFORd measures the probability that an asset will not be available when required due to uncontrolled or unplanned outages or derates. The information market participants historically provided in an energy only market construct is not to the level of detail required for the AESO to accurately complete an EFORd statistic for all assets. The capacity factor and availability factor approaches, Page 2 of 20

21 which use energy market data in their determination, will provide an equivalent measure of unit reliability during periods of tight supply conditions. 1 Stakeholders that have advocated for the use of EFORd-based incentive schemes (and the use of 1,000+ hours) suggest that the AESO should assess performance in ways other than energy delivered during tight system conditions, such as rewarding an asset s long-run average availability. It is the AESO s view that this would not create the right signal for asset performance in the capacity market. An asset s UCAP that is based on the asset s long-run average availability over a significant number of hours (1,000 or more) will not reflect the likelihood that an asset will be available during shortage conditions. For example: Two assets run for 1,000 hours each. Many of these hours have adequate supply and reliability is not at risk. The system experiences 50 tight supply conditions during the year. Based on EFORd measures, an asset that delivered energy for every hour the system was in tight supply conditions and an asset that was offline for every hour of tight system conditions will receive substantially the same availability measure. Figure 1 below illustrates how EFORd and the use of 1,000 hours does not account for tight system conditions when evaluating the reliability potential of an asset. Figure 1 Evaluation of availability over 1000 using only force outage rates of assets Evaluation of availability over 1000 using only forced outage rates of assets Tight System Conditions Tight System Conditions Tight System Conditions Unit #1 Availability Unit #2 Availability UCAP based on forced outage rate (EFORd) Unit #1 Availability 85% Unit #2 Availability 85% Contribution to system reliability during tight system conditions Unit #1 Availability 25% Unit #2 Availability 100% The AESO also received feedback from stakeholders recommending the creation of an EFORd-based UCAP methodology. In the AESO s view, such an approach would misalign incentives by compensating assets for availability during periods when the system is not deficient in reserves, and when no additional reliability is necessary. 1 For clarity, the RAM described in subsection 4.2 of Section 3, Calculation of Demand Curve Parameters, makes assumptions regarding outage rates for thermal assets based on available capability data to determine conditions of tight supply in the Monte Carlo simulation, which assesses system reliability. Page 3 of 20

22 3.1.3 The AESO will not calculate UCAP for the asset types that will be ineligible for the capacity market. Please refer to subsection of the rationale document for Section 2, Supply Participation for assets that are ineligible for participation in the initial capacity market The AESO recognizes that calculating UCAP based on historical performance may not capture all of the transitional issues that arise from moving from an energy only market to a market structure with both capacity and energy markets, as well as be fully indicative of future performance in the capacity market. The purpose of the UCAP range is to recognize this transitional issue and provide capacity market participants with the flexibility to select a UCAP within a predetermined range that may better reflect the asset s potential performance in the capacity market. Each asset will be able to choose a UCAP by selecting the highest or lowest value determined from these three approaches: a UCAP range established in accordance with the elimination approach as in 3.1.6, a range of +/-2% multiplied by the asset s maximum capability. These values will be added and subtracted to the UCAP of the asset to calculate an alternate UCAP range. every capacity asset will receive a UCAP range of at least +/- 1 MW of UCAP The AESO believes that allowing participants to select the UCAP of their assets with these ranges reasonably addresses the concerns of market participants that assets would perform differently under the incentive scheme of the capacity market than they performed under the energy market only incentives The elimination approach involves eliminating 5% of the hours in which an asset was the lowest performing, effectively raising average performance to determine an upper range, and to eliminate 5% of the hours in which the asset was the highest performing, effectively lowering average performance to determine a lower range. The elimination approach will result in assetspecific UCAP ranges, which will reflect that the reliability value of each asset may vary based on historical performance. Below is an illustration of the AESO s elimination approach: Figure 2 Elimination approach for determining UCAP range Page 4 of 20

23 Table 1 below provides an example of each methodology for an asset that has a UCAP of 10 MW, and an MC of 12. The asset s availability factor is 83%. Table 1 Example UCAP ranges Elimination Approach* +/- 2% +/-1 MW Final Range UCAP (upper limit) 10.1 MW (12MW*0.0 2) MW 11 MW 11 MW UCAP 10 MW 10 MW 10 MW 10 MW UCAP (lower limit) 9.9 MW (12MW * -0.02) MW 9 MW 9 MW *Based on historical performance, the elimination method established a UCAP range at +/-1% of availability factor Stakeholders suggested using the minimum and maximum single-year UCAP values to establish the upper and lower bounds of the UCAP range (the minimum/maximum method ). The AESO has performed analysis on this approach and compared it with the elimination approach and has determined that the minimum/maximum method establishes very wide UCAP ranges for certain assets, capturing near zero to full maximum capability of an asset in some instances. Consequently, the minimum/maximum method does not align with the intent of UCAP, which is to represent the reliability value of an asset during the tightest supply cushion hours. There are a number of negative consequences if UCAP ranges are too wide: (a) Reliability concerns where market participants may use the ranges to overstate their UCAP because they assess the risk of performance events as low. This could introduce issues of adverse selection where less reliable assets select high upper values for their UCAP and potentially displace higher reliability assets (b) Market distortion concerns where market participants may use ranges to lower the UCAP in order to withhold capacity from the auction. (c) They may make the reliability value somewhat arbitrary and meaningless. Stakeholders also suggested using a standard deviation approach to establish a UCAP range. This approach would compare all of the data points used in determining UCAP, to the final UCAP value to determine the standard error. However, due to the large range established through this approach, particularly for variable resources, this would again deviate from the intent of UCAP. The UCAP range will not be applicable to demand response capacity assets or external capacity assets. The purpose of the UCAP range is primarily to handle transitional issues. Both of these asset types are considered new assets for the first capacity auction, and do not have any historical data to adjust to address transitional issues The minimum size requirement of 1 MW aligns with energy market minimum block size and the declaration of available capability. In order to perform an availability assessment, an available capability must be captured and maintained for the capacity asset Final asset level UCAPs will be shared publicly during the pre-auction period to support a fair, efficient and open market which involves minimizing information asymmetry. The data currently published on the AESO website may enable sophisticated market participants to estimate, with Page 5 of 20

24 some accuracy, the UCAP values for most of the generating units in the market. If sophisticated market participants can derive information from public data with a good degree of accuracy, then this information should be made available to all market participants. Otherwise, the market would not offer a level playing field for small and large participants. UCAP for capacity assets that meet the minimum threshold of hours for calculating UCAP per Section Due to the different operating characteristics of variable and dispatchable assets, the AESO will use two UCAP methodologies to appropriately represent the reliability value of each type of asset. Availability factor methodology The availability factor methodology relies on historical declarations of a capacity asset s available capability. The AESO is using the availability factor methodology for an asset that can respond to a dispatch and/or have metered volumes that align with its dispatch. For this asset the available capability declaration represents its full generating capability or load reduction in that period. Data in the Energy Trading System is presumed to be accurate given the must-offer must-comply obligation in subsection 3 of Section of the ISO rules, Offers and Bids for Energy. As a result, the available capability declared by dispatchable assets in the past provides a reasonable representation of a dispatchable asset s future ability to perform under similar conditions. Capacity factor methodology UCAP for all capacity assets whose generation capability is dependent on a fuel supply that is uncontrollable (i.e., wind, sunlight or water) and have no storage capability will be calculated using the capacity factor approach. While such an asset may be capable of producing energy, it may not be available to do so due to the variable nature of its fuel source. Therefore, the capacity contribution of these assets will be calculated using the capacity factor methodology. CMD 1 considered using maximum metered volumes as the denominator in the hourly capacity factor calculations. However, maximum metered volumes may capture outlying values where load goes offline and the generation remains active (i.e., outside normal operating conditions), which requires the AESO to identify outliers. The AESO has determined maximum capability as the denominator in the hourly capacity factor calculation provides a more stable value that only changes when the capacity market participant changes an asset s capability. Availability factor through linear regression The AESO will use a linear regression UCAP determination approach for self-supply assets that chose to be dispatched at a gross generation meter. This approach has been developed in response to stakeholder comments that the previously recommended capacity factor approach did not provide recognition for available but not dispatched generation levels. This approach will recognize the undispatched generation availability and ancillary services volumes, when applicable, while determining a UCAP level that also recognizes how gross dispatch generator volumes compare to net to grid energy volumes. The AESO will perform a linear regression of the capacity asset s net to grid metered output against dispatch level over the historical tight supply cushion hours. The linear regression provides a historical review of how gross generation dispatches have translated into net to grid energy volumes. This linear regression will create a formulaic representation of this relationship in the form of a line: Y = M(X)+B. This approach will rely on historical declarations of a capacity asset s available capability. The AESO will use these declarations to determine a gross to grid UCAP based on availability factor methodology described above and in section a. The AESO will use these gross to grid UCAP value as the (X) variable in the linear regression formula to calculate the self-supply site s UCAP for use in the capacity market. Page 6 of 20

25 For example, the linear regression determines the relationship between the site s gross generation dispatch (gross UCAP in the graphic below) to the site s net to grid energy volumes (net UCAP in the graphic below). This relationship can be expressed as a line formula. Figure 3 Example linear regression for self-supply dispatched gross-to-grid Net UCAP y = x Gross UCAP The site s gross to grid UCAP is calculated using the capacity asset s available capability (AC) and maximum capability over the historical tight supply cushion hours: UCAP AF gross = (AC/MC) * MC = (36/69)*69 = 36 MW Using the line formula the final UCAP determination, the availability factor through linear regression, would be determined as: UCAP AF net = (UCAP AF gross ) = (36) = MW This approach recognizes how gross asset dispatch translates into net to grid volumes as well as the capacity asset declared availability capabilities over the tight supply cushion hours. For this reason the AESO believes this is an improved approach to the previously suggested capacity factor approach for self-supply sites that choose to be dispatched on a gross basis. Five-year history Assessing an asset s capacity contribution over a 5-year period provides a reasonable estimate of future unit performance. This large sample over periods of low supply captures the variability in system conditions across different seasons. Tight supply cushion hours Supply cushion is a measure of real-time system resource adequacy risk. A large supply cushion indicates less real-time system resource adequacy risk because more energy remains available to the AESO to respond to unplanned market events. A low supply cushion indicates that the system has fewer assets available to react to unexpected outages or load increases, therefore, indicating a high real-time system resource adequacy risk. Evaluating the historical performance of a capacity asset during a subset of tight supply cushion hours captures the correlation of the asset s availability and capability with all other system factors that drive the tight supply cushion hours. This is expected to provide an indication of how the asset will perform in the future under similar conditions when capacity is needed. Page 7 of 20

26 CMD 1 and 2 considered using 100 tight supply cushion hours per year to calculate UCAP. The AESO has determined that calculating UCAP using 250 tight supply cushion hours continues to measure asset performance during periods that matter most for system reliability; namely, whether the asset delivers energy or reserves during stressed system conditions. The AESO has increased the tight supply cushion hours to 250 hours in response to stakeholder feedback indicating concern that UCAP calculations utilizing only 100 hours created excessive risk for market participants. Stakeholders requested the UCAP calculation to be based on 1,000 or more tight supply cushion hours. The AESO has determined increasing the number of hours used in the UCAP calculation beyond the 250 hours would not meet the intent of the UCAP methodology, as it would result in capacity value being measured under conditions with a supply cushion two or more assets away from a supply shortfall situation during a significant portion of hours. Figure 4 Evaluation of number of tight supply cushion hours Hours per Average Supply Cushion (MW) Capacity Interval 2012/ / / / /17 All , Planned and forced outages Unlike other jurisdictions, the AESO does not restrict the timing, duration and frequency of planned maintenance outages scheduled by an asset owner, as long as notification of the planned outage is provided to the AESO 24 months in advance. Firms have the flexibility and independence to schedule the outages of its assets without the need for AESO approval. Further, given the high load factor in Alberta, planned outages are a driver of tight supply hours. As such, Alberta s capacity market needs to ensure that planned outages are scheduled in a manner consistent with the assumptions used in developing the capacity obligation. This may result in planned outages occurring in, or leading to supply shortfall conditions. By reflecting the duration and timing of planned outages in an asset s UCAP, the incentive is placed on the asset owner to optimize outage frequency and duration in order to minimize supply adequacy risk, thereby maintaining system reliability. The probability of asset unavailability due to planned outages Page 8 of 20

27 should be reflected in an asset s UCAP values as they better reflect the realities of Alberta s outage planning framework. UCAP for capacity assets that do not meet the minimum threshold for calculating UCAP per section For an asset without 5 years of operating history in Alberta, the AESO will determine the UCAP using a class-average, a production or load estimate or through jurisdictional review. The class-average will be used for capacity assets without operational history that are of a similar design or have similar operational characteristics to other capacity assets in Alberta. This approach allows the AESO to approximate the reliability contribution of a new asset based on how similarly configured assets have performed in the past during tight market conditions. If a new asset does not have any similarly-designed or geographically located assets, the AESO will use production or load estimates based on engineering data provided by the new asset during prequalification. This will allow the AESO to approximate the reliability contribution of a new asset based on the best available information until historical data becomes available. In absence of class average or comparable class estimate the AESO will examine how similar assets or an asset class has performance in other capacity market jurisdictions during tight system conditions The AESO completed an analysis to determine a minimum number of hours that could be used to accurately reflect the UCAP of a new asset. This analysis used a discrete distribution to estimate the minimum number of hours required to estimate UCAPs for thermal and variable assets. A discrete distribution is characterized by a limited number of possible observations. The discrete probability distribution took different levels of asset performance into account; ranging from offline to full capacity. In theory, using the entire performance spectrum for units should lead to a minimum threshold that is aligned with the actual operational behaviour of the units. In order to determine the minimum number of hours required to accurately reflect the UCAP of an asset, the AESO applied the following methodology: Define the discrete distribution This step approximated the performance data for all 8760 hours in a year into the predefined number of bins listed in Figure 5. A five percent increment was used because it created a balance between a manageable number of categories and accurate indication of performance levels. For thermal assets performance levels were determined using availability factor data, and for variable assets performance levels were established using capacity factor data. The analysis assigned a weighting to each bin using historical data. The analysis examined historical data for a five year period and calculated a frequency count using the performance thresholds listed Figure 5. The frequency count and the total number of observations were used to assign a weighting to each of the bins in the distribution. Page 9 of 20

28 Figure 5 - Bins used to define the discrete distribution Availabilty Factor From (> than) To (<= than) 0% Calculate the Summary Statistics for the Distribution Discrete Value 0% 0% 5% 5% 5% 10% 10% 10% 15% 15% 15% 20% 20% 20% 25% 25% 25% 30% 30% 30% 35% 35% 35% 40% 40% 40% 45% 45% 45% 50% 50% 50% 55% 55% 55% 60% 60% 60% 65% 65% 65% 70% 70% 70% 75% 75% 75% 80% 80% 80% 85% 85% 85% 90% 90% 90% 95% 95% 100% 0% This analysis calculated the expected value, variance, and standard deviation of the discrete distribution using the formulas outlined below. A discrete random variable X follows the following probability distribution: The mean (expected value) of X is the sum of the products Xi* Pi: µx = X1*P1 + X2*P2 + + Xn*Pn = _(i=1)^n Xi*Pi With mean µx, then the variance of X is: σ2 X = (x1 µx ) 2 * p1+ (x2 µx ) 2 * p2+ + (xn µx ) 2 * pn = _(i=1)^n (Xi-µX)^2*Pi The standard deviation σx is the square root of the variance. Calculate the Sample Size The sample size was calculated using a targeted standard error of two percent. The standard error of the sample mean depends on both the standard deviation and the sample size; the standard error decreases as the sample size increases. The standard error indicates the uncertainty around the estimate of the mean measurement. The standard error is an indicator of the level of uncertainty that the AESO is willing to take around the expected value for performance factor calculations. The formulas for the standard error and the sample size are illustrated below. Page 10 of 20

29 Standard Error (se) =σ/ n Where σ is the standard deviation of the distribution and n is the sample size. Sample size was calculated as follows n= σ/ se ^2 ^2 Results for thermal assets Figure 6 shows the shape of discrete distribution for thermal assets. This illustrates the different levels of availability and the corresponding weightings used to assess the minimum number of hours required to calculate UCAP for thermal assets. Figure 6 Distribution of availability factor data for thermal units The summary statistics of the distribution are shown in Figure 7. The sample size calculated for the distribution is close to 250 observations. Having at least 250 hours of availability data for thermal assets should be sufficient to calculate a UCAP that is aligned with historical performance. Figure 7 Summary statistics for thermal assets Summary Statistics Expected Value 78.4% Variance 9.9% Std Dev 31.5% Std Error 2.0% Sample Size 248 Results for variable assets Figure 8 illustrates the shape of discrete distribution for wind assets. This shows the different capacity factor thresholds as well as the accompanying weightings that were used to assess the minimum number of hours required to calculate UCAP for variable assets. Page 11 of 20

30 Figure 8 Distribution of capacity factor data for variable units The summary statistics of the distribution are shown in Figure 9. The sample size calculated for the distribution is near to 300 observations. Having at least 300 hours of capacity factor data for variable assets should be sufficient to calculate a UCAP value that is aligned with historical performance. Asset-specific UCAP methodologies Figure 9 Summary statistics for variable assets The AESO s approach to the selection of a methodology to calculate asset-specific UCAPs is described in detail below: Wind & solar, & run of river hydro assets The AESO is using the capacity factor methodology to determine the UCAP of wind, solar and run of river hydro assets. Due to the variability of their fuel source, which is determined through environmental changes, these assets have limited ability to change generation levels relative to energy dispatches. Self-supply assets dispatched gross to grid Summary Statistics Expected Value 34.5% Variance 11.8% Std Dev 34.3% Std Error 2.0% Sample Size 294 Some self-supply sites (with load served by onsite generation) currently offer their energy into the energy market on a gross basis, meaning that the available capability for the asset is submitted without discounting the portion of its generation that is used to serve onsite load: Page 12 of 20

31 Figure 10 Gross metering of self-supply site Such self-supply sites are often dispatched to a level beyond the energy that they can deliver to the grid. Therefore, using an availability factor to calculate UCAP would risk overestimating the site s capability to deliver capacity. The AESO recognizes that using a capacity factor approach suggested in earlier versions of the CMD may not capture some of the undispatched energy these assets would be able to supply to the grid. For these sites, the AESO will now perform a linear regression of the net to grid metered output of the self-supply site relative to the weighted average energy market dispatches issued to the generating asset(s) on the self-supply site as observed in each of the 250 tightest supply cushion hours per year for the past five years. This methodology is described above in subsection Demand Response Demand Response (DR) assets can provide capacity by reducing energy consumption during system stress conditions. Comparable pre-qualification requirements for DR assets and generation assets will ensure all supply side capacity assets provide comparable reliability for consumers. Load customers in Alberta have never been subject to a must offer obligation in the energy market. Therefore, no availability factor, capacity factor or class performance can be established. In the absence of Alberta specific information, the AESO will establish an initial performance measure, the derating factor, for all new demand response assets based on the average demand response performance in system stress events in other jurisdictions. The AESO will apply this factor to all new demand response assets until an Alberta specific performance can be established. The AESO acknowledges that there is variation in how demand response resources are treated across other capacity markets. Treatment can vary significantly in how each of the markets take different approaches to calculating baselines, may or may not gross up demand response performance to account for transmission losses and reserve requirements and may have different procedures for notifying demand response resources of a performance event. Since no single market would provide an exactly analogous representation of general demand response performance during tight system conditions during the obligation period in Alberta, the AESO will average the historical performance of demand response assets in other capacity market jurisdictions where the assets had an obligation to provide capacity during a system stress event. The AESO considers this to be a reasonable approach, as it reflects the following: Demand response may not always provide or achieve its full nameplate capacity during a performance event. No data exists on how Alberta based demand response assets will perform in a capacity market. Page 13 of 20

32 The asset owner will still have a choice on the size of the reduction, the guaranteed load drop amount, or the firm consumption level. To determine the derating factor, four jurisdictions were reviewed including PJM, NYISO, ISO-NE and the United Kingdom. The markets have been chosen as they represent capacity market jurisdictions where loads curtail consumption in response to system operators rather than in voluntary response to price. The demand response resources had either to be certified prior to offering into the auction, or to satisfy credit requirements, similar to the Alberta capacity market. Event performance in PJM, the ISO-NE and ISO New York is measured as load reductions that were Y % of their capacity commitment. Figure 11 PJM demand response performance 2 Delivery year Event Performance No Events % % % % No Events No Events Average Event Performance 97% Figure 12 ISO-NE demand response performance 3 Data Source Date Performance Hours Type Response 2011 Report Jun-10 3 Event 91.8% 2012 Report Jul-11 6 Event 94.8% 2012 Report Dec-11 3 Event 87.7% 2013 Report Jan-13 3 Event 86.3% 2014 Report Jul-13 8 Event 115.5% 2014 Report Dec-13 4 Event 81.2% 2017 Report Aug-16 4 Event 102.2% Average Event Performance 94.2% Compiled from the ISO-NE Monthly Market Reports for March of : Page 14 of 20

33 Figure 13 NYISO demand response performance 4 Data Source Date Season Type Response 2017 Report Summer 2017 Test % 2017 Report Winter Test % 2016 Report Summer 2016 Test % 2016 Report Summer 2016 Event % 2016 Report Winter Test % 2015 Report Summer 2015 Test % 2015 Report Winter Test 95.90% 2013 Report Summer 2013 Event 75.20% 2013 Report Summer 2013 Event 67% 2013 Report Summer 2013 Event 71.80% 2013 Report Summer 2013 Event 63.70% 2012 Report Summer 2012 Event 97.50% 2012 Report Summer 2012 Event 72.60% 2012 Report Summer 2012 Event 79.60% 2011 Report Summer 2011 Event 92.80% 2011 Report Summer 2011 Event 90.10% Average Event Performance 81.16% C il d f NYISO A l R D d R P Notes: The table reports New York-wide average performance data for New York s ICAP/SCR resource type, which is capacity market DR. ICAP/SCR resources are occasionally activated for voluntary, rather than mandatory, events. Voluntary events are not included in the table. According to the 2014 Report, there were no mandatory events during Summer 2014 or Winter The 2011, 2012, 2013 and 2014 Reports did not report Test performance. The 2012 report listed an event that was mandatory in Zone J and voluntary elsewhere. This event was excluded from the above data. The 2011 Report listed two events without indicating whether these were mandatory or voluntary. The AESO included these in the table. UK De-rating methodology for demand side resources The demand resource de-rating factor is based on a combination of test and performance data for Demand Response providing operating reserves. The demand resource de-rating factor was 89.7% for 2014 and 86.8% for The AESO will not use demand response factors from the UK market as they represent performance in the ancillary markets (Short Term Operating Reserves (STOR) availability) and are not representative of capacity performance events. 4 Compiled from NYISO Annual Reports on Demand Response Programs: Page 15 of 20

34 As a result of this analysis, the initial derating factor the AESO will use for demand response will be 91% based on the average of past performance of demand response assets in PJM, the ISONE and the NYSIO. ( )/3= 91% Firm Consumption Level (FCL) This asset commits to reducing load consumption to a pre-defined level during an EEA event. The capacity contribution of an FCL asset is evaluated by comparing their consumption after a performance event to a pre-defined FCL. For FCL assets the qualified baseline is intended to capture the top range that an FCL asset may qualify to sell into the capacity auction. The methodology to determine the qualified baseline establishes the average, typical consumption of the load and provides the upper bound to the amount of capacity the asset may provide. The AESO will remove days that had tight supply cushion hours or performance periods from the qualified baseline to recognize that price responsive loads have historically reduced consumption as pool prices increase. Since loads have not historically participated in the energy market the AESO will not be able to adjust the qualified baseline by energy dispatches or load outages. Table 2 - Determination of qualified baseline for an FCL asset Date/Day 1-2 p.m 2-3 p.m 3-4 p.m 4-5 p.m 5-6 p.m 6-7 p.m 7-8 p.m 03-Apr Tuesday Day Apr Wednesday Day Apr Thursday Day Apr Friday Day Apr Saturday Weekend Apr Sunday Weekend Apr Tight Supply Day Apr Tuesday Day Apr Wednesday Day Apr Thursday Day Apr Friday Day Apr Saturday Weekend Apr Sunday Weekend Apr Monday Event Day Apr Tuesday Day Apr Wednesday Event Day Apr Thursday Day Apr Friday Day Apr Saturday Weekend Apr Sunday Weekend Apr Monday Day Apr Tuesday Day Apr Wednesday Day Apr Thursday Day Qualified Baseline hour An illustrative example of qualified baseline determination for tight supply cushion hour that occurred during 5-6p.m: A tight supply cushion hour occurred on April 27 between 5-6 p.m. of the previous year. To recognize that this load may have already reduced consumption as a response to high prices during the tight supply cushion hour the AESO will re-estimate the consumption by the following methodology: o The load will be averaged during the previous 15 non-holiday weeks days prior to the day with the tight supply cushion hour, using the same hour (5-6 p.m.) as the tight Page 16 of 20

35 o supply cushion hour. The tight supply cushion hours on April 9 th and the event day hours on April 16 and 18 will not be used in the averaging calculation. Using this approach, the qualified baseline value for April 27 between 5-6 p.m. will be determined to be 18 MW This methodology will be used to determine the qualified baseline hours for all 250 tight supply cushion hours of the previous year. Once an FCL asset has capacity market performance history, the capacity contribution will be calculated as the historical qualified baseline calculation minus the firm consumption level. This will reflect the performance of FCL asset s ability to reduce to its firm consumption level during delivery events. An FCL asset will be physically tested to ensure that the FCL, a pre-defined level claimed by the asset, can be realistically achieved by the asset during an EEA event. If the asset is unable to reduce to its FCL levels during a delivery event the UCAP of the asset will be reduced for the subsequent auctions to reflect that the asset is a less reliable resource. Guaranteed Load Reduction asset (GLR): A GLR asset will declare its capacity capability to the AESO. The capacity capability will represent the load the GLR asset will be able to reduce when required for capacity delivery purposes. The GLR value declared must be equal to or lesser than the maximum load the asset may be able to consume. All GLR assets will initially receive a 91% derating factor (see the discussion in the FCL asset, above) until an asset specific availability factors can be observed as the asset accumulates capacity market operational data Figure 14 - Illustrative example of GLR UCAP determination until operational data becomes available Typical on-peak consumption of a load selling the Demand Response product to the AESO The reduction 'capacity contribution' that DR asset declares it wants to sell 60 MW 20 MW Maximum UCAP that a DR asset may sell into the capacity market 18 MW The availability factor will be applied to the capacity contribution declared by the GLR asset: 20 MW x 91%* = 18.2 MW of capacity is the maximum UCAP the GLR asset may sell into the capacity auction. Once GLR asset performance data becomes available an availability factor will be established. This asset specific availability factor will be used to create a UCAP for the asset. Page 17 of 20

36 Aggregated assets The UCAP for aggregated assets will be based on the sum of the performance of each of the individual assets being aggregated. Performance will be measured using observed available capability declarations or metered volumes during the 250 tightest supply cushion hours of the previous 5 years. If the aggregated capacity asset contains both capacity factor and availability factor assets the capacity factor methodology will be used to determine UCAP. If the asset s UCAP is determined using an availability factor methodology the AESO would add the component available capabilities which would be zero and the aggregate asset would not meet the availability criteria described in Section 8 and would be subject to annual payment adjustments. Whereas, if the asset s UCAP is determined using a capacity factor methodology, the AESO will rely on meter data for determining availability. Determination of capacity limit of each Alberta intertie The capacity limit of each intertie limits the volume of capacity that can be provided from external assets and will be determined prior to each capacity auction. The capacity limit is intended to reflect the volume of capacity that could have flowed into Alberta during the 250 tightest supply cushion hours for each year during the previous 5 years. For the BC, Montana Alberta Tie Line (MATL) and Saskatchewan interties, the capacity limit is determined on an hourly basis by taking the minimum of firm transmission and available transfer capability (ATC). The BC/MATL combined path is additionally constrained due to the joint scheduling limit. The hourly capacity limit is determined using the minimum of firm transmission and ATC prior to load shed services for imports (LSSi) arming. LSSi is an ancillary service that is provided by loads that are armed and automatically trips following the loss of the intertie. The AESO has observed that the volume of load that is available to provide LSSi is lower when the Alberta system is in periods of tight supply. The AESO expects that these loads have reduced their consumption based on energy market price signals and also believes this practice may continue in the future. As a result, the AESO believes that determining the amount of capacity that could be delivered into Alberta over the intertie while including the LSSi volumes may overstate the intertie delivery ability during future tight supply cushion hours. As a result, the AESO will be determining intertie values without the increase flow volumes attributed with LSSi loads. The hourly capacity limits are then averaged to arrive at final capacity limits for the BC intertie, Saskatchewan intertie, BC/MATL path and MATL intertie. During a capacity auction the capacity procured from external capacity assets will not exceed the capacity limits identified for each intertie. External assets For the first capacity auction, a new external asset must declare its UCAP to the AESO, demonstrate that the external asset has firm transmission in the amount of the UCAP declared and confirm that the UCAP is being supplied from a source that is non-recallable. For new external capacity assets, in the absence of asset specific operating history, the AESO will apply the intertie derating factor to all external assets. This approach will be an interim measure until enough history is obtained in the capacity market, an existing external asset s UCAP will be established in the same manner as an internal capacity asset, as the AESO will have the data to determine UCAP based on an availability factor or capacity factor approach, as applicable. Mothballed or temporary delisted assets Section of the ISO rules, Mothball Outage Reporting (Section 306.7) enables market participants to exit the energy-only market by taking their generating units offline for a period of up to 24 months for non-operational reasons. Section is intended to help market participants manage fixed costs associated with generator maintenance. The available capability of a generating asset on a mothball outage is 0 MW. While this captures the real-time availability of a mothballed generator it may not accurately represent the unit s ability Page 18 of 20

37 to deliver capacity in future tight supply hours. The available capability of a mothballed generator is reflective of an economic decision by the asset owner and not reflective of an asset s reliability. Therefore, using available capability declarations while the asset was mothballed will not being considered in the calculation of the asset s UCAP. The AESO intends to use the historical observations of the asset s performance prior to the mothball outage to determine its UCAP. The methodology described in Section establishes the minimum number of hours that could be used to accurately reflect the UCAP of an asset that has been on a mothball outage. Long lead time assets, type 2 The AESO recognizes that per subsection 4.1 of Information Document # (R), Long Lead Time Energy, a long lead time asset is expected to restate its available capability in the Energy Trading System during economic shutdown to better reflect physical capability. As a result, observed available capability may not be an accurate representation of the asset s ability to deliver capacity in tight supply cushion hours and the use of availability factor may discount the true capacity contribution of the asset, especially if the tight supply cushion hour occurred in a relatively weaker economic period when the asset was offline. The expectation is that these assets would be able to return to service within a short amount of time if required for system reliability. The AESO has established the minimum number of tight supply cushion hours required to calculate a statistically significant UCAP for thermal units at 250 hours. Any hours where availability was reduced due to a long lead time configuration for economic purposes will be excluded from the sample set used to calculate a UCAP. If a long lead time asset has less than the 250 tight supply cushion hours the asset s availability will be supplemented with a class average for similarly designed assets. To confirm that availability was reduced due to a long lead time claim for economic purposes, the AESO will review the following: (a) The participant comment in ETS indicating that the unit was offline for a long lead time configuration. (b) The cost assessment for the asset in comparison to pool price during that period. The AESO will determine an approximate marginal operating cost for the asset based on past operational history. This marginal operating cost assumption will be compared to pool prices during the period for which UCAP is being determined. When pool price is less than the marginal operating costs of the asset those hours will be excluded from the UCAP calculation of the asset. 3.2 UCAP refinement process As part of the sequence of activities, leading up to a capacity auction, a capacity market participant will have the opportunity to submit a request a UCAP refinement based on a number of criteria The enumerated list in subsection of the proposal is intended to identify reasonable scenarios that would result in UCAP not being reflective of the reliability of the capacity asset. Acceptable refinement requests will be limited to the scenarios outlined in this section. The intent of this approach is to ensure that accurate asset reliability values are obtained for each asset while reducing the number cases taken to the formal appeal process. Reductions to available capability that occur as a result of a transmission constraint will be considered an acceptable reason to initiate a UCAP refinement. However assets whose availability was reduced due to distribution system constraints or transmission outages, where the asset was electrically disconnected from the transmission system will not be able to request a UCAP refinement. The AESO is obligated to plan and make arrangements for the expansion or enhancement of the transmission system. Therefore, transmission constraints resulting in violations of limits on the transmission system are an acceptable exemption from the calculation of UCAP. Unavailability due to a distribution constraint should be reflected in the UCAP Page 19 of 20

38 methodology, given it is a participant s choice of whether to connect to the distribution or transmission system If the AESO and the capacity market participant are unable to achieve resolution on a UCAP refinement issue, the capacity market participant has the ability to escalate by filing a formal dispute through the dispute resolution process. Page 20 of 20

39 Calculation of Demand Curve Parameters Rationale 4.1 Resource adequacy standard The resource adequacy standard announced by the Government of Alberta prescribes the minimum level of reliability to be achieved. The standard will be a normalized expected unserved energy metric set at a maximum of %. The Minister of Energy may make regulations establishing the resource adequacy standard, and AESO s duties now include procuring enough capacity to meet the established resource adequacy standard. The demand curve will be developed to meet this minimum standard. 4.2 Resource adequacy model A probabilistic approach is expected to provide greater information on the relationship between capacity and supply adequacy, as well as better capture the correlations between supply and demand variability. This results in a more informed and accurate estimation of the procurement volume. (a) Gross demand is currently the best measure of total provincial demand in Alberta, for which the AESO will need to procure capacity. Gross demand, as opposed to net-to-grid demand, is best suited to capture the overall behaviour between economic activity and load. Forecasting gross demand also aligns with the AESO's current planning and reliability mandate. The intent of the AESO s load forecast model is to minimize model error. Using multiple hourly weather and economic profiles introduces load-related uncertainty to the RAM, which provides a better reflection of the range of potential future conditions through which the reliability performance of differing capacity volumes can be tested. 1 Specifically, the RAM will consider probabilities of economic growth scenarios to capture uncertainty in the economic outlook underpinning the load forecast. These scenarios also capture uncertainty related to new sources of load growth and energy efficiency impacts. Each scenario is assigned an associated normal-curve-based probability totaling to 100%. (b) Currently, the AESO has visibility of generating units with a capabilities of 5 MW and greater and is able to reasonably determine outage rates and other key characteristics for these units. The AESO does not have sufficient visibility of assets with a maximum capability of less than 5 MW to include data in the RAM. In the future, the RAM will consider capacity assets less than 5 MW as data becomes available from their participation in the capacity, energy and ancillary services markets. Demand response capacity assets must be prequalified by the AESO in order to participate in the capacity market. Once these resources have been determined, they will be modelled in the RAM in subsequent auctions. 1 Additional details on the proposed capacity market load forecast model can be found here: Page 1 of 11

40 (c) The objective of the planned outage algorithm is to add maintenance events such that each event added impacts the lowest load days possible. (d) Sites with load served by onsite generation exhibit a wide range of generation, load and availability patterns. By aggregating the data across these sites, the AESO is able to capture the correlation between onsite generation and load. The individual unique characteristics of each site create assumption and modelling challenges which prevent the AESO from being able to model them like other generators. Daily gross peak loads of such sites are generally higher in the winter than in the summer. While it has been observed that, at times, these sites do have similar values in the different seasons, their facility daily peak availability does vary from winter to summer. Defining seasonal distributions takes these observed variations into account to better capture the variability in supply from cogeneration. (e) The RAM is set up to align with current system controller procedure for supply shortfall (EEA) events. The activation and utilization of contingency reserves is consistent with current EEA procedure and operating reserves will be used to meet energy requirements. As contingency reserves are used in real time for other reasons, those types of events are not captured in the RAM estimation of unserved energy unless the event leads to an overall supply shortfall. (f) It is necessary to develop renewable profiles to take into account the diversity of production from intermittent sources in Alberta. When evaluating resource adequacy it is important to use multiple hourly weather correlated profiles to represent uncertainty in renewable generation. There is insufficient historical data to cover 30 years of weather uncertainty for all sites. Simulated wind shapes were developed incorporating historical metered output from existing sites. Shapes were aggregated by geographic locations and correlations between wind output sites were maintained. Simulated solar shapes were developed using NREL National Solar Radiation Data. 2 Multiple profiles were created to represent diversity in location and technology (fixed and tracking solar PV). (g) The RAM considers weather correlated historic profiles to accurately assess the contribution of hydro to the system. Hourly, daily and monthly constraints need to be considered while allowing for flexibility inherent in the hydro system to meet load. (h) As part of the data review, transmission availability was identified as the binding import constraint rather than generation availability within adjacent jurisdictions during tight supply conditions. Therefore, imports within the RAM are a function of transmission availability with other jurisdictions There are two key principles underpinning the ICAP to UCAP translation: (a) The measure of capacity in the demand curve and supply curve need to align so the capacity that the AESO is buying aligns with what the capacity market is selling (i.e., UCAP). (b) The AESO is indifferent to the type of UCAP it procures. A MW of UCAP should deliver the same amount of reliability regardless of the underlying technology (e.g., 1 MW UCAP from wind equals 1 MW UCAP from simple cycle). 2 Page 2 of 11

41 4.3 Calculation of gross-cone and net-cone Reference technology Selection of a reference technology is meant to ensure the Alberta capacity market provides adequate revenue for required generation additions. The reference technology should represent a technology that can be developed to meet the capacity needs during the capacity auction timeframe at a low cost and, philosophically, be the unit most likely to be developed under predicted future market conditions. In all capacity market jurisdictions, the reference technology is based on a gas-fired power station. Some capacity markets refer to a combined-cycle plant, while other markets prefer a simple-cycle reference technology. The AESO will consider combined cycle and simple cycle capital and operating characteristics to determine the appropriate reference technology. Based on the AESO s assessment to date, simple-cycle technology may be the best fit to the criteria listed above. Additional details on the reference technology will be developed by the AESO and subject to further consultation. Gas-fired technologies including LM6000 s, frame turbines, reciprocating internal combustion engines, and LMS100 turbines all represent simple-cycle technologies with recent developments in the province. Fuel efficiency tends to favor LMS100 turbines and reciprocating internal combustion engines, while availability and maintenance costs may favour LM6000 or frame turbine power plants. 3 Approach to gross-cone estimate Gross-CONE and net-cone are significant inputs into the demand curve and are necessary for a functioning capacity market. Reasonable estimation ensures that new assets are attracted to enter the market when appropriate price signals are present. Working with experienced independent financing and engineering services firms to determine appropriate detailed cost estimates for the gross-cone will increase the objectivity and accuracy of estimates Using an approach that considers Alberta-specific conditions for financing generation projects will most accurately characterize on-the-ground conditions for developing supply to meet adequacy needs. The AESO will work with the external consultant to provide realistic financing assumptions in the gross-cone calculation, based on observable cost, and leverage data applicable to Alberta-based power projects. The ATWACC for individual firms is expected to vary greatly as different participants and projects will have asymmetric credit ratings, costs of debt and debt/equity ratios Since CONE values are not subject to significant changes over a several year timespan, CONE studies are updated in other capacity market jurisdictions every 3 to 5 years to reduce administrative burden. 4 Similarly, the gross-cone study will be updated with the demand curve review cycle to ensure that the reference technology is reflective of long term changes in the market and generation costs. In between review cycles, the gross-cone estimate will be indexed from year-to-year to reflect changes in the capital cost of the reference technology. The indices will reflect changes in labour, materials, and turbines (e.g., machinery and equipment) in order to track changes in the development cost of the reference technology. A composite index will be developed by weighting component indices by their relative contribution to installed costs initially. 4 See Charles River Associates Paper, Governance Institutions and Processes for Electric Capacity Markets: A Jurisdictional Review at PDF pp : Page 3 of 11

42 The specific indices used to index gross-cone will be developed by the AESO and subject to further consultation. Approach to energy and ancillary services offset To develop the energy and ancillary services offset (offset) approach to construct the net-cone, the AESO took into consideration that numerous historic and future market fundamentals will lead to significant challenges and uncertainty when modelling potential earnings derived from the offset. Some of the many examples of changes to market fundamentals that challenge analysis of the future power market include natural gas prices, the cost of carbon, renewables penetration, magnitude of coal to gas conversions, and energy and ancillary services market offer share ownership. To manage the uncertainty and to develop a transparent replicable methodology for creating the offset, the AESO is proposing a forward market methodology, which incorporates forward power and natural gas prices. To achieve certainty of revenues, the offset will not include revenues from ancillary services given that forward markets do not price ancillary service products. Approach to net-cone estimate The net-cone will not be a negative number because at a minimum, a plant operator could avoid generating to avoid any negative margin. Net-CONE reflects the missing money, that the capacity market is designed to compensate generators for. Thus, net-cone is the best index to reflect price parameters at points on the demand curve corresponding to the capacity price cap and the inflection point. 4.4 Shape of the Demand Curve The AESO s development of the demand curve was guided by the following principles: (a) Demand curve parameters should be set to ensure procurement of a sufficient amount of capacity for reliable operation of the electricity grid and to achieve the resource adequacy objectives, while avoiding significant over-procurement or under-procurement. (b) Demand curve parameters should be set to send an efficient price signal in the capacity market, avoid excessive capacity price volatility and reduce the opportunity for the exercise of market power. (c) Demand curve parameters should be set to balance between achieving resource adequacy and lowest possible long-term cost to consumers, and to sustain resource adequacy over time through a market-based outcome. (d) Demand curve parameters, including the relationship between these parameters and net- CONE, should be set to ensure that Alberta s market attracts investment in new capacity and maintains existing capacity in order to achieve the resource adequacy objectives. (e) Demand curve parameters should be compatible with, and robust to, reasonably foreseeable changes in supply, demand, energy prices and other factors in the electricity market. (f) To the extent applicable to the Alberta context, the demand curve analysis should incorporate experience and lessons learned from other jurisdictions. (g) Unique aspects of Alberta s electricity system (e.g., small size of the market, market transition) should be considered. Demand curve shape refined to meet the Government s resource adequacy objective The primary purpose of the demand curve is to express the Government s resource adequacy objectives as a quantity of capacity to be procured in the market, at prices high enough to attract entry as the Alberta system becomes short (and low enough to defer entry when the system is long). In 2017 and early 2018, preliminary demand curve analysis was conducted under the assumption that the Government would adopt a target resource adequacy level to be achieved on Page 4 of 11

43 average across many years. The AESO and stakeholders used a 400 MWh per year average expected unserved energy (EUE) target as a working assumption in developing its original candidate demand curves. Under that assumed reliability objective, candidate curves were designed to achieve 400 MWh EUE on average across many years, though reliability could drop to a lower level in any individual year. In March 2018, the Government of Alberta announced a minimum resource adequacy standard by specifying a minimum acceptable reliability of % EUE, which equates to approximately 964 MWh of EUE at the load level expected by Defining the reliability objective as a minimum acceptable (rather than a target average) introduces different implications for how the demand curve should be defined. Average reliability must now be, on average, greater than or equal to the % / 964 MWh EUE minimum. The minimum acceptable reliability level will also define the capacity level at which the AESO may need to intervene in the market to procure backstop capacity to avoid unacceptable reliability levels. The minimum acceptable reliability and the backstop intervention level are similar to the EUE intervention level in the energy only market in Section of the ISO rules, Adequacy of Supply. To represent this reliability standard as a demand curve for capacity, the AESO will aim to limit the frequency of a capacity auction clearing below the minimum acceptable level to no more than 5% of years. This will be achieved by developing a demand curve where the parameters are set such that procurement levels in the base auction are expected to exceed the minimum level 95% of the time. The demand curve necessary to achieve this objective is fairly wide and achieves average reliability considerably higher than the minimum. This reflects the reality that reliability deteriorates rapidly as the market becomes short, consistent with the exponential shape of the EUE versus a total UCAP curve. If the demand curve performs as expected, the demand curve would procure sufficient capacity to meet reliability needs with reliance on rebalancing auctions or backstop reliability interventions triggered only once every 20 years. The simulation performed by the Brattle Group, described in further detail below, showed that this 5% level will minimize the impact of out-of-market interventions on capacity and energy market performance. The AESO will not attempt to entirely eliminate the possibility of infrequent reliability interventions however, because doing so would require maintaining large excess quantities of capacity and imposing the associated costs. Given the ability to take action in the three years leading up to the obligation period to meet the resource adequacy standard, the AESO is of the view that this is a reasonable risk to take to prudently manage capacity costs. The AESO intends to meet the Government s minimum reliability standard by procuring all needed supply through the base auction. This approach will take advantage of the larger pool of resources available to commit on a three-year forward timeframe and the resulting higher elasticity of available supply. The AESO will leverage rebalancing auctions to correct imbalances arising after the base auction, but not to procure incremental supply on average through the demand curve shape. If the AESO consistently relied on the rebalancing auctions to true up its supply, it would risk entering the rebalancing auctions very short in years where net supply was under-forecasted in the base auction. The AESO would then have limited options to procure sufficient supply before the start of the delivery year, customers would be forced to pay potentially very high offer prices from the limited pool of available suppliers, and reliability could be negatively affected. Approach to evaluating potential demand curves In the fall of 2017, the AESO consulted on numerous candidate demand curves that would support the assumed reliability target of 400 MWh per year average EUE, perform well under a range of potential market conditions, and align with the other demand curve design principles. To examine potential performance of each of these curves, the Brattle Group developed a Monte Carlo model to simulate the likely price volatility, quantity procurement, cost, and reliability outcomes that could result from each curve. The Monte Carlo model evaluated capacity market outcomes probabilistically, given realistic variability in supply and demand under the long-run equilibrium assumption that merchant generation will enter or exit the market until average prices Page 5 of 11

44 equal net-cone. Results of the model include distributions of price, quantity, and reliability outcomes that might be realized over many years under AESO candidate demand curves. The AESO and the Brattle Group engaged with the Demand Curve Working Group to refine simulation assumptions and technically test a wide range of curves through an iterative process. This analysis revealed that there is a workable range of well-performing curves. The wellperforming curves met the desired outcomes of ensuring resource adequacy, providing a price signal of net-cone on average, and mitigating against net-cone error. 5 However, the simulation modeling identified trade-offs across these curves in terms of robustness to market conditions, price volatility, reliability outcomes and market power exposure. Curves outside the workable range of good performance parameters tended to have unacceptable performance in at least one area (such as unacceptably low reliability, high cost, or high price volatility). From March to June 2018, the AESO assessed additional curves consistent with the Government s % EUE minimum resource adequacy standard. In developing a new set of candidate curves, the AESO and the Brattle Group also implemented a refined simulation approach that accounted for stakeholder input. 6 All of the curves assessed in 2018 were downward-sloping and convex (i.e., the slope on the minimum-to-inflection point segment of the curve is steeper than the slope of the inflection point-to-foot segment). This downward-sloping, convex shape is generally consistent with the marginal reliability benefit of capacity. The AESO focused on curves with a price cap at 1.75x net-cone, which balances the trade-off between sending a strong signal for investment when the market is short and increasing price volatility under all market conditions. Rationale for selecting specific demand curve parameters Based on a qualitative and quantitative analysis of a large number of demand curves, the AESO has identified a workable range and preferred levels for specific demand curve parameters. Based on this analysis, the AESO and the Demand Curve Working Group developed a preference for a convex demand curve (rather than straight-line or concave) with the following: (i) a width comparable to or slightly wider than demand curves in other capacity market jurisdictions; and (ii) a higher price cap than other capacity market jurisdictions in the range of x net- CONE. The rationale for identifying the workable range for each demand curve parameter is as follows: Y-axis (or price) points. The Y-axis points for the demand curve will be set in reference to a multiple of the net-cone parameter. This approach to setting price points is consistent with the economic theory that capacity market prices must be able to rise to the long-run marginal cost of supply on average in the capacity market (i.e., to net- CONE). Setting demand curve prices around the expected net-cone will allow the demand curve to adjust to support higher (or lower) prices if market conditions change to increase (or decrease) the prices needed to attract new entry when needed. The price cap (zero quantity-to-minimum reliability segment) is set based on the maximum value of either a 1.75x times net-cone or a 0.5 times gross-cone. The 1.75x net-cone level is expected to set prices in times when energy prices and expected energy margins for the marginal new entrant are moderate or low. This price cap is high enough to attract supply offers and avoid shortage conditions, but not so high that it exposes market participants to excess price volatility or excess potential for the exercise of market power. The 0.5x gross-cone minimum on the price cap will prevail under conditions with high 5 Further material on the candidate curves examined earlier in the capacity market design process, along with the rationale for their selection, can be found here: 6 Further material can be found here: Page 6 of 11

45 energy margins and associated low net-cone. This minimum on the price cap will prevent reliability erosion, in cases of low net-cone or underestimation of net-cone. Inflection point. The inflection point is set at a level of 0.875x net-cone, which is within the workable range of at least 0.8 to 1.0x net-cone identified by the AESO. The exact placement of this point has a modest effect on demand curve performance in that placing it lower and to the left would make the curve more convex, while placing it higher and to the right would make it less convex. The more important performance issue related to the inflection point is that the inflection point (and the entire demand curve) needs to be right-shifted compared to the minimum acceptable reliability level. To avoid falling below the minimum acceptable level, the quantity points on the demand curve must be rightshifted sufficiently to mitigate year-to-year fluctuations in the supply-demand balance (caused by factors such as entry and exit from the market and load forecast changes). X-axis of demand curve. The X-axis points for the demand curve will be set in reference to quantity of UCAP MW of capacity, RAM metrics, and demand curve performance simulations. The minimum point is set at a value equal to the % EUE standard set by the Government. This minimum quantity is the point at which demand curve prices will reach the price cap. Setting the cap quantity at the minimum acceptable level will ensure that all in-market offered capacity will be procured before the AESO would potentially need to engage in backstop procurement. All other quantity points on the demand curve will be set at a higher level in order to ensure that average reliability exceeds the minimum acceptable, and that the capacity market is expected to fall below the minimum acceptable only 5% of the time. Foot of the demand curve. The foot of the curve is set at zero as negative pricing does not incentivize capacity additions. A price floor above zero is also not desirable because it would have the potential to attract and retain excess quantities of capacity resources, particularly if the cost of incremental supply is low. This was the experience in the early years after implementation of ISO-NE s capacity market with a price floor that attracted incremental low-cost supply into the already long market. By allowing capacity prices to drop to zero at higher quantities, the demand curve will ensure that customer costs are more aligned with reliability value and mitigate the potential for sustained periods with excess supply. The foot of the demand curve will be set at a quantity level such that the minimum acceptable reliability level will be violated only 5% of all years in the base auction procurement. Price outcomes can be expected to average at net-cone levels while also balancing capacity price volatility and maintaining the desired convexity of the curve. This combination is expected to best achieve the demand curve principles. Width of the demand curve. The AESO considered the following trade-offs in its evaluation of the width of the demand curve: (a) (b) (c) A steeper curve is more robust to a wide range of market conditions, has less reliability risk from underestimated net-cone, and less risk of excess capacity above the reliability requirement. A flatter curve has lower price volatility and less exposure to exercise of market power and need for strict mitigation. However, a flatter curve presents a risk of procuring more capacity than required to meet the resource adequacy standard. Assessments of Alberta's system indicate that a demand curve based on the marginal reliability value is too steep to achieve reliability. Due to Alberta s small market size, the AESO identified that the province is at greater risk than other markets of price volatility and resource adequacy concerns caused by the expected variability in supply and demand in the capacity market. To manage this volatility and variability, a relatively wider curve could be used. However, industry stakeholders raised concerns that a wider demand curve might have the undesirable effect of increasing average procurement levels and capacity market costs along with Page 7 of 11

46 dampening price signals in the energy market due to greater supply. In response to this feedback, the AESO has opted to choose a demand curve that is relatively steep compared to other candidate curves identified within the well-performing range. The steeper curve also requires a relatively higher price cap than that used in other capacity markets, to keep quantities above the minimum acceptable level and support revenues aligned with long-run marginal cost. Comparison of selected and alternative curves Figure 1 below illustrates two candidate curves based on the 5 % at the cap reliability objective and compares them with the curve that was previously proposed in CMD 1.0. Table 1 provides the performance result of each candidate curve displayed in Figure 1. The Less Convex curve (displayed in navy blue) has a price cap at 1.75x net-cone and a quantity at the cap corresponding to the minimum acceptable reliability (currently 964 MWh EUE). The kink is located at 0.875x net-cone and is a quantity corresponding to 40% of the overall curve width. The position of the foot is selected in order to ensure no more than 5% of draws occur at the price cap in the Brattle Group s sample Monte Carlo simulation. The More Convex curve (displayed in purple) has a slightly different shape, with the kink located at the intersection of net-cone and the quantity corresponding to 100 MWh EUE. The foot quantity is selected using the same approach as for the Less Convex curve in order to ensure no more than 5% of draws occur at the price cap. The horizontal portion of this curve is higher, which results in a steeper portion of the curve left of the kink, and a much wider foot compared to the Less Convex curve. The Less Convex curve is similar in shape to the proposed curve from CMD 2 and CMD 3, but slightly wider. The increase in width is the result of updated information and analysis since CMD 1, including: (a) updating RAM analysis resulted in a slightly steeper EUE curve (slightly widens the demand curve); (b) reducing the estimate of shocks to demand (slightly narrows the demand curve), and (c) revising the objective from a 400 MWh per year average EUE target to hitting the price cap no more than 5% of the time (slightly widens the demand curve). Figure 1 Revised candidate curves Page 8 of 11

47 As shown in Table 1, both candidate curves perform similarly in terms of average EUE and cleared quantity in the long-run Monte Carlo analysis. The only quantitative performance difference between the two curves is the approximately 15% higher price volatility associated with the Less Convex curve. Table 1 Revised candidate curve results 7 Demand Curve Average Price Price and Cost Standard Deviation of Price Frequency at Cap Average Cost Average EUE (Before Intervention) Reliability Average EUE (After Intervention) Average Cleared Quantity Average Uncleared Supply ($/kw-yr) ($/kw-yr) (%) ($mil/yr) (MWh) (MWh) (MW) (MW) More Convex $139 $46 5% $1, , Less Convex $139 $53 5% $1, , Figure 2 below shows the two revised candidate curves compared to other capacity market demand curves. The width of the Less Convex curve is 18%, which is wider than several of the U.S. capacity markets including PJM (8%) and ISO-NE (12%). The foot quantity of the Less Convex curve is consistent with the New York City zone, which is a similarly-sized market. 8 The More Convex curve is substantially wider than other markets with a width of 32%. 9 While the AESO s candidate curves are wider than curves in many other markets on a percentage basis, this is needed to manage entry and exit of lumpy supply in Alberta s smaller market. On a pure MW basis, the candidate curves are in fact steeper than in some other markets. For example, the Less Convex curve has a width of about 2,000 MW (UCAP) versus PJM s width of nearly 11,900 MW. The AESO s candidate curves also have relatively higher price caps than most other markets, which contributes to the curve s steepness. This higher price cap was chosen based on Demand Curve Working Group input and reflects a preference to avoid over-procurement in long conditions even though price volatility is somewhat higher. 7 Average price and average cost are outputs of the Monte Carlo simulation, not an approximation of an auction outcome. The Average EUE (After Intervention) represents the EUE estimate after adjustments to reduce EUE results from any simulation seed that exceeds the minimum resource adequacy standard to the minimum level. 8 The New York City demand curve is somewhat wider than the AESO s due to its lower quantity at the cap. 9 Note the percentage for the demand curve width refers to the percent of MW corresponding to the minimum acceptable reliability for AESO and percent of MW corresponding to the reliability requirement (1-in-10 loss of load expectation) for other capacity markets. Page 9 of 11

48 Figure 2 Cross-market demand curves comparison Notes: 100% of the Reliability Requirement corresponds to 400 MWh per year average EUE for the AESO and 1-in-10 loss of load expectation for the other capacity markets. NYISO and New York City curves are based on the 2018 summer period; the ISO-NE curve is the FCA11 Marginal Reliability Impact demand curve; the PJM curve is the 2021/22 BRA VRR curve; the Great Britain curve is based on the 2015/16 period; and the Western Australia curve is a proposed curve for their capacity auction. As shown in Table 1 above, both AESO candidate curves achieve reliability objectives and have similar performance on the quantitative measures in the long-run. However, the curves will likely perform differently in the short-term. If the market is even moderately long in the short-term, then the More Convex curve would likely cause significant over-procurement compared to the Less Convex curve. This could be a concern if existing supply is retained and incremental internal supply, imports, and demand response offer into the initial capacity auctions. While it is expected that excess supply would reduce over time through auction clearing even with the More Convex curve, it is reasonable to expect periodic long supply conditions, lasting a longer period of time, if the More Convex curve were selected. Table 2 below summarizes certain qualitative aspects of the candidate curves short-run performance as it relates to the capacity market design objectives. These factors suggest that the Less Convex curve is more consistent with the design objectives in the short run and support the proposal to adopt this curve in Alberta. Page 10 of 11

49 Table 2 Candidate curve comparison Consideration More Convex Less Convex Risk of Over- Procurement and Unnecessary Costs to Customers Sending an Efficient Price Signal Consistent with Incremental Reliability Value Exposure to the Exercise of Market Power Attracting and Retaining Investment Consistent with Resource Adequacy Objectives Consistency with Lessons Learned in Other Jurisdictions Has relatively lower risk when market is short and higher risk when market is long Over-procurement when long may be a particular concern if market is expected to be long in the early years Demand curve price substantially exceeds marginal reliability value at higher quantities Has relatively lower risk right of the quantity corresponding to 100 MWh EUE and higher risk to the left of that point Steep segment left of the kink will effectively attract supply when the market is short Substantially flatter segment right of the kink may be slow to incentivize retirements when the market is long Curve is wider than in other jurisdictions Has relatively lower risk when market is long and higher risk when market is short Over-procurement may be less of a concern when market is short Demand curve price may still exceed marginal reliability value at higher quantities, but by a much smaller margin Has relatively lower risk left of the quantity corresponding to 100 MWh EUE and higher risk to the right of that point Steep segment left of the kink will effectively attract supply when the market is short Somewhat flatter segment right of the kink will still incentivize retirements when the market is long Curve is somewhat wider than most full markets, but consistent with similarly-sized New York City zone 4.5 Demand curve for rebalancing auctions Using the same demand curve shape in the rebalancing auctions avoids the market distortions that would occur if the rebalancing auction demand curve were systematically different than the base auction demand curve. The AESO will update the procurement volume parameters of the demand curve using the updated resource adequacy assessment completed prior to the rebalancing auction, to reflect recent supply and demand information. These updates will allow the AESO to ensure reliability if it has under-forecasted procurement volume, reduce customer cost impacts if it has overforecasted procurement volume, and send an accurate updated price signal to suppliers about the tightness of capacity supply and demand in the market. The AESO proposes not to update the net-cone parameter in the rebalancing auctions. Net-CONE will likely be the subject of an extensive stakeholder process involving public release of draft parameter values. Since draft net-cone values may be available more than a year before they are used in a forward capacity base auction, use of an updated net-cone parameter in a rebalancing auction would introduce an opportunity for gaming. Since market participants would know with reasonable confidence whether net-cone is likely to increase or decrease in the rebalancing auction at the time they offer into the forward base auction, they would have incentives similar to those described above for systematic differences in demand curve shape. Page 11 of 11

50 Base Auction Rationale 5.1 Auction forward period and timeline The three-year forward period is long enough to achieve the benefits of a forward auction, namely the orderly entry and exit of capacity assets. At the same time, while supply and demand conditions are less certain three years forward, they can still be forecasted with reasonable accuracy. Many capacity markets have adopted similar forward periods, including PJM, and ISO- NE which have three-year forward periods; and the UK, and Ireland which have four year forward periods. The three-year forward period has received unanimous support from the Capacity Market Technical Design Working Industry Group. 1 As noted above, forward auctions support orderly entry and exit decisions by establishing market expectations well in advance of capacity commitment delivery. A capacity committed asset will be able to complete its interconnection and have additional time to complete construction prior to the start of the delivery period, allowing for competition between new and existing capacity assets. 2 Similarly, a capacity committed asset can signal to the market its intention to retire well in advance, or choose to reduce its obligation volumes in response to a reduction in forecasted load. Some larger capacity assets may require longer than three years of lead time to develop and build; therefore, there may be a preference for a longer forward period. These longer-term capacity assets may need to make significant investments before entering and potentially clearing the capacity market. While these capacity assets accept some additional risk by making investments prior to clearing a capacity auction, they are not excluded from the market. The capacity market s price signal allows these capacity assets to make investment decisions based on market fundamentals. While a longer forward period might benefit the subset of long lead-time capacity assets, these benefits may be offset by the costs of increased forecast error. The three year forward auction approach has certain drawbacks. In its report to Alberta s Market Surveillance Administrator, Potomac Economics drew a different conclusion about the most appropriate forward period and recommended a prompt auction, conducted only weeks or months before the start of the obligation period. 3 Potomac observed that forward auctions lead to greater uncertainty in load and supply availability relative to prompt auctions. While this view is acknowledged, it is important to allow new capacity assets to establish a capacity commitment and obtain some revenue certainty prior to the start of their construction period and equipment delivery period the time when capital expenditures increase dramatically for new assets. The AESO also acknowledges Potomac s observations that forward auctions may be less beneficial 1 See Capacity Market Technical Design Working Industry Group Recommendation, SAM 2.0, and 2 The AESO connection process shows that the target timeline between the initiation and the approval of energization of a connection project is 96 weeks. After the connection period, extra time and activities are also required before a project can begin commercial operation. 3 See Section III.2. of Potomac Economics, Report on Best Practices in Wholesale Electricity Market Design, November 2017, Prepared for the Alberta MSA, Available: Page 1 of 15

51 for capacity assets with longer construction lead times. 4 As discussed above, it is believed that such capacity assets may still be able to participate in forward auctions, and may benefit from the reduced price volatility of the three-year forward period relative to auctions that are settled more immediately prior to the obligation period. The three-year forward period strikes a balance between allowing enough lead time for capacity assets to complete construction after clearing the capacity market and managing uncertainty about future demand, and supply conditions. While a longer forward period would enable larger capacity assets more flexibility before making significant financial commitments, and a shorter forward period would reduce market uncertainty, a three-year forward period provides an appropriate balance of the aforementioned considerations Due to the short period between market design completion and the commencement of the capacity market, a transition period is being established in order to allow the AESO to procure capacity for obligation periods of 2021/22, 2022/23 and 2023/24 as shown in Table 1 of the Final CMD, Section 5, Base Auction. The AESO estimated that approximately eight months are required to complete all the activities necessary for conducting a capacity auction. However, the dates with asterisks will be determined at a future date when further details of the auction process are finalized. During this transition period, fewer rebalancing auctions will be held than once the transition period is completed. This is discussed in greater detail in the rationale for Section 6, Rebalancing Auctions. 5.2 Base auction guidelines and schedule The forward capacity auction will involve a series of activities that begin approximately eight months before the capacity auction. This amount of time is required for the AESO to complete the prequalification and qualification process for new and existing assets, as shown in Figure 1 in the Final CMD. This period of time should also allow capacity market participants to establish their auction participation strategy and to obtain internal approvals for participation. Figure 1 in the Final CMD provides a more detailed sequencing of key activities. 5.3 Dispute resolution The AESO will establish an appropriate framework with respect to the resolution of all capacity market related disputes so that disagreement between capacity market participant or a firm that has offer control of a capacity asset, and the AESO can be resolved in an efficient and timely manner. 5.4 Obligation period The one-year obligation period establishes a fair and competitive market for capacity assets. A one-year timeframe allows the capacity market to promptly reflect current supply and demand conditions and to respond to trends and changes as necessary. A longer obligation period may be more prone to inefficiencies due to forecast errors, may reduce the incentive for capacity suppliers to innovate and reduce costs, and may result in the AESO purchasing capacity that becomes inefficient relative to new technology prior to the end of the longer term obligation period. A longer obligation period also may result in inefficient retirement, mothball, and upgrade decisions. 4 Potomac also observed that a single year of capacity revenues is a small portion of the revenue requirement of a new resource. While this is of course true, this is a feature of capacity markets generally, and has no bearing on the choice of forward period. Page 2 of 15

52 In CMD 1, the AESO acknowledged that a potential downside of a shorter obligation period is that it could fail to provide enough certainty to attract investment in new capacity assets. Further assessment of whether various durations and term structures for an obligation period longer than one year would impact investment differently, and the pros and cons of different design alternatives that may allow up to a seven-year price lock-in has been conducted. The AESO relied upon an investment banker with knowledge of power industry financial matters and the Brattle Group's modeling expertise and its experience with other capacity markets for its assessment. These analyses are summarized below. Investment banker summary The investment banker completed a study of the financing arrangements that have been completed for generating assets built in PJM and the ISO-NE under their longer term capacity structures. The key findings of the study include: (a) (b) (c) (d) Project financing but with costs: The longer obligation period in the ISO-NE provided for project financing for three facilities. Even with the longer obligation period the risk characteristics of these projects have been rated high (below investment grade) and as such the financing costs have been high. Some of these projects had multiple tiers of financing that resulted in later recovery of cash for equity investors, increasing the return expectation for the equity investors. The debt financiers have not been typical US or Canadian lenders with offshore banks and non-banking lenders providing debt financing. Further, the recent low interest rate environment may have driven much of the financing for these projects. The investment banker points out that this environment may be changing and the durability of these lenders may be open to question. Incumbents still dominate: Many of the asset builds in the ISO-NE that qualified for the longer term obligation period have been completed by incumbents to the power sector that finance projects from their balance sheets such as: Dynegy, Exelon, NRG and PSEG. PJM relying on the energy market for financing: Very few new capacity assets qualified for the three-year obligation period in PJM and instead relied on bilateral hedges, often up to five years in length, to support their financing. The investment banker's considerations for a longer obligation period in Alberta include: i. Increased regulatory environment risk: The merchant power s sector risk in Alberta is likely higher in today s environment than in previous years with a recent change in the provincial government and the new policies that result from different long term environmental goals of this administration than the previous administration. ii. Deteriorating incumbent balance sheets: While the energy only market saw the development of a significant amount of merchant generation, much of that was balance sheet financed and supported by the contractedness provided by the PPA legislation. The benefits of that contractedness have largely disappeared and it s not expected the incumbents would be in a position to complete the same level of investment in the future as was completed in the past. Page 3 of 15

53 Brattle Group analysis The Brattle Group suggested a number of potential market design alternatives to the one-year term. The table below summarizes their thoughts as well as possible advantages and disadvantages of each alternative. Approach Description Advantages Disadvantages Have a longer term option when oneyear term does not attract needed investment Seek to secure needed capacity through a oneyear term; if that is not successful, allow all capacity assets to compete for a longer obligation period. The auction would clear at the price cap and shorter duration obligation period offers (i.e. two years) would clear prior to longer duration obligation period offers (i.e. seven years) Resource neutral Helps protect against reliability risk May reduce market distortions created by longer multi-year commitments Incents supply to offer at the shortest acceptable fixed price duration May provide an incentive for suppliers to hold out for longer term capacity commitments The long term average capacity market price is higher than oneyear, no lock-in approach, overall variability of capacity prices is greater and the number of auctions that settle at the price cap is greater Three-year term with auctions every three years Run a three-year auction every three years. The AESO would obtain all the capacity it needs for the next three years once every three years Some additional price certainty for sellers AESO runs risk of mismatch in needs vs. procurement quantity in many years and potential for additional costs to load Risk of large simultaneous retirement and new capacity asset entry every three years Laddered procurements Procure capacity needs through a variety of terms; purchase 20% of capacity in one-year, two-year, three-year, five-year and seven-year terms May be attractive to some market participants Provides significant pricing information on a number of different terms Untested not used in other jurisdictions No concrete theory on the most appropriate share of short and long term contracts May limit competition between new and existing capacity assets Potential for over-procurement Segments the market into smaller, less competitive slices, increasing need for monitoring and oversight Page 4 of 15

54 AESO's assessment of a one-year obligation period The AESO completed analysis regarding the one-year obligation period. The following table provides a summary of the advantages and disadvantages of a one-year obligation period for all types of capacity assets. Advantages Does not discriminate between capacity asset types. All capacity assets receive the same treatment, reducing the efficiency losses and costs due to early retirement of existing capacity assets Disadvantages Uncertainty regarding whether a one-year obligation period will attract sufficient new supply. The AESO recognizes a longer obligation period: reduces regulatory risk for new entrants; and may provide financing alternatives that are not available with a one year term Reduces the risk of over-procurement of capacity due to changing demand forecasts Has been successful in other markets - PJM attracted many thousands of MWs of new entrant capacity with one-year obligation periods Provide better liquidity in the capacity market all capacity assets are required to participate in each forward auction Provides better price fidelity for capacity - better represents marginal value and cost of capacity through time and provides valuable market information at more frequent intervals 5.5 Supply participation and offer format CMD Final specifies that a qualified capacity asset can offer up to seven blocks, as contemplated in the rationale for CMD 2. Allowing seven offer blocks is expected to be sufficient for firms to represent the cost structure of many different capacity asset types and configurations. This approach is also consistent with the number of offer blocks in other jurisdictions. A minimum block size of one MW allows for participation by nearly all assets. Capacity assets under this threshold can participate by aggregation. This size is also consistent with the energy market minimum resource size. Through the seven offer blocks, firms will be able to indicate whether blocks are flexible or inflexible. This option enables firms to prevent capacity assets that are under development from partial clearing and possibly requiring the firm to resize the asset. This option also allows firms to ensure that assets with a minimum stable generation level are able to ensure a minimum level of cleared capacity volume and the stable revenue associated with that volume. After a firm offers an inflexible block for an asset, all the higher priced offers for that asset have to be flexible, with the exception noted below for incremental capacity. This requirement is put in place in order to reduce computational complexity in the auction clearing algorithm. As incremental capacity may have a higher cost than the existing capacity from the same capacity asset, the incremental capacity may warrant an inflexible block that is priced higher than Page 5 of 15

55 the existing capacity from the same capacity asset. This gives a capacity market participant the ability to avoid its incremental capacity asset that is under development from being partially cleared and possibly having to be resized. Finally, the offer of each asset formed with price-quantity pairs is required to be monotonically increasing to reflect the increasing marginal cost of the capacity asset and to ensure the auction algorithm is solved efficiently. 5.6 Out-of-market capacity payments Assets that are the subject of a Renewable Electricity Support Agreement (RESA) under the first three rounds of the REP auction will be ineligible to participate in the capacity market. Compensation for the capacity value of these capacity assets is provided in the form of payment contained in the RESA. The capacity volumes for these assets will be accounted for in the target procurement volume identified in the demand curve. Subtracting this capacity volume from the target capacity volume avoids over procurement of capacity and reduces costs to load. Initially, any potential distortionary impact to the capacity market is expected to be minimal, given the expected magnitude of capacity value for the REP rounds one to three assets. If future RESA transactions are structured in a comparable manner, they too would be ineligible to participate in capacity market auctions. On a more general basis, there could be other forms of out-of-market payments made by the government to capacity market eligible capacity assets. The AESO expects that the value associated with the capacity payments and the rights to sell capacity from qualified capacity assets will form part of the negotiation between parties. With consideration to the future evolution of the capacity market, the AESO will establish a process to determine whether any alternative adjustments or incremental approaches should be implemented to incorporate future REP or similar programs. 5.7 Single-round uniform price auction The AESO is proposing to use a sealed-bid, single-round, uniform pricing auction for both the base and rebalancing capacity auctions. This is the most common auction format among existing capacity markets and is used in PJM, MISO and NYISO. It has a number of benefits relative to the other potential auction format: the descending-clock design used in New England and the UK s capacity markets. The sealed-bid, single-round design minimizes the opportunity for gaming, and encourages participants to offer at cost, a particularly important consideration given Alberta s small size and relatively concentrated market. The sealed-bid auction is also simpler to administer. Overall, a sealed-bid, single-round, uniform pricing auction should help facilitate a fair, efficient, and openly competitive capacity market in Alberta. The sealed-bid, single-round auction Sealed-bid, single-round auctions minimize the opportunity for gaming by limiting market participants access to information about competitors bids. Sealed bids ensure that market participants cannot directly observe their competitors offers. The single-round format allows auction participants to submit offers in only one clearing round. Unlike the descending clock auction format, the single round format does not provide further auction rounds that allow participants to revise their offers after seeing the result of previous rounds. While participants have some insight into how their competitors will offer based on the outcome of previous auctions and their knowledge of market conditions, this information may not be comprehensive. Without information about competitors offers, market participants are incentivized to offer at cost. This also allows the market to provide accurate price signals to capacity market participants entering or exiting the market. These considerations are particularly relevant given Alberta s relatively small electricity market. Page 6 of 15

56 The descending clock auction, an alternate approach During a descending clock auction, the auctioneer starts each round by issuing a price and asking firms to state the quantities they wish to sell at this price. If the quantity offered exceeds the target quantity to be procured, the auctioneer issues a lower price, and again asks firms the quantities they want to offer at the new price (hence, descending clock). This process continues until the quantity offered matches the quantity to be procured or until excess supply is negligible. The descending clock s multiple-round structure reveals information on supply offers after each round of bids (such as how many MW exited the auction), providing opportunities for some supply capacity assets to take advantage and coordinate offers or use market power to sway the auction results. Given the size and concentration of the Alberta market, this feature of the descending clock auction format introduces additional opportunities for gaming which could potentially offset the benefits from increased price discovery that this format might provide. In addition, the descending clock format favors incumbents relative to new entrants. Under the descending clock auction, established participants are better able to take advantage of the information revealed during the auction itself due to their better familiarity with the system. Given Alberta's unique characteristics of relatively small size and the concentration of incumbents, the AESO's view is that a sealed-bid, single-round auction is more appropriate. Uniform pricing Uniform pricing provides a single clearing price for every supply bid that clears the auction. This feature incentivizes market participants to submit cost-based offers to ensure they are cleared in the auction and make at least enough revenue to cover their net going forward costs. Uniform pricing is also fair in the sense that capacity assets supplying the same product receive the same price. In contrast, auctions with non-uniform pricing introduce incentives to offer above cost. For example, pay-as-bid auctions encourage low-cost capacity assets to offer above cost in order to capture a higher price for greater revenues. Sealed-bid, single-round and uniform pricing auctions are also simple and straightforward to implement. The operator builds the supply curve based on all of the bids received in the single round, and, along with the administratively determined demand curve, clears the market at a single price by maximizing social surplus between the two curves. By contrast, the descending clock auction is more challenging to implement as: (1) it requires additional parameters like step size (the reduction in volume between rounds), price band width, and infrastructure to enable communication between the ISO and market participants during the auction; (2) it creates challenges for the handling of scarce import capability; and (3) is intended for a single buyer auction which would introduce challenges during the rebalancing auction where market participants will be able to submit bids to buy out of their obligations. 5 The sealed-bid, single-round uniform pricing auction format supports a fair, efficient, and openly competitive capacity market by reducing gaming opportunities, limiting the possibility of tacit collusion, leveling the playing field between incumbent and new market participants, providing clear and accurate price signals, and incentivizing cost-based supply offers. 5.8 Auction clearing and price-setting The social surplus-maximizing clearing algorithm is the most commonly used clearing algorithm across all existing capacity markets, with the exception of the UK. 6 Maximizing social surplus will result in the most efficient long-term price signals which should provide the most efficient resource mix and lowest societal costs over time. This approach is also consistent with the clearing approach used in the current energy market. 5 See ISO-NE discussion, 6 In the UK, if the lump offer is marginal, it is only cleared if doing so economically benefits consumers. May result in lower shortrun consumers prices in some cases, but less efficient resource selection will increase prices over the long term. Page 7 of 15

57 Use of a different clearing algorithm may not have the same outcomes. For example, in the UK, if the inflexible block is marginal, it is only cleared if it is beneficial to the consumers. Figures 1 and 2 illustrate examples where this clearing algorithm does not maximize social surplus. Under the UK clearing algorithm, the auction clears at P 1 and Q 1 in Figures 1 and 2 as shown in the graphs below on the left. In these situations the clearing algorithm would have the AESO purchasing less capacity than its target purchase level. While this procurement level would still be above the level which would cause reliability concerns over time the AESO is concerned that it may systematically purchase less capacity than its target purchase levels and set auction price levels lower than what would be established under a maximization of social surplus approach. For example, in Figure 1 if the market had cleared at P 2 and Q 2, social surplus would be larger. In the graph on the right of Figure 1, the green triangle is larger than the red triangle, and thus there is additional social surplus by clearing at P 2 and Q 2 ; social surplus being the difference between the two triangles. In the graph on the left of both Figures 1 and 2, if Area A is bigger than the net social surplus gain, a net loss in consumer surplus may occur in the auction. Maximizing net consumer surplus instead of maximizing social surplus would clear the market at P 1 and Q 1 instead of P 2 and Q 2. Figure 1 Loss of social surplus: clearing at the block below the inflexible block Inflexible block Does Not Clear (Area A > Area B) Potential Addition Social Surplus Page 8 of 15

58 Figure 2 Lack of price signal for new entrants: clearing at the flexible block above the inflexible block The Flexible Block above the Inflexible Block Does Not Clear (Area A > Area B) Potential Addition Social Surplus By clearing at P 1, and Q 1 consumer surplus is maximized, but this reduces the effectiveness of the price signal by creating no market incentive for new capacity assets that could offer between P 1 and P 2. Instead, when social surplus is maximized in the clearing algorithm (auction clears at P 2 and Q 2 ), a more accurate price signal is provided compared to a clearing algorithm that maximizes consumer surplus only. Maximizing social surplus would attract new capacity assets to enter the market at price levels between P 1 and P 2, providing more capacity at a lower price The AESO expects that the supply curves created in the capacity auction will not be smooth, but will be built up by a number of independent supply offers resulting in a supply curve with a number of discrete steps. This will create scenarios where the market cannot clear at the intersection of the supply and demand curves, possibly due to the demand curve intersecting the supply curve between offer blocks, the marginal offer being inflexible or possibly due to the supply curve being below the demand curve. This section further describes the principles that will be used to clear the capacity market. Social surplus has two components: producer surplus and consumer surplus. Producer surplus represents the difference between total market revenues from the sale of the product and the total marginal costs of production. Consumer surplus represents the difference between a buyer s (in this case, the AESO s) willingness to pay for a product and the price of the product, summed over all units sold. When the market clears at the intersection of the supply and demand curves, as shown in Figure 3 below, the social surplus is maximized. Page 9 of 15

59 Figure 3 Consumer surplus, producer surplus, and social surplus In circumstances where the capacity market cannot clear at the intersection of the supply and demand curves due to the marginal capacity offer being an inflexible block, the market will then clear the capacity offer that maximizes social surplus. Figure 4 illustrates two scenarios: (1) on the left: a scenario where the entire inflexible block is cleared; and (2) on the right: a scenario where the inflexible block is skipped and the offer above the inflexible block is cleared. The social surplus resulting from clearing at P 1 and Q 1 is the same in both figures, depicted as the light green region. In the figure on the left, the additional social surplus from clearing the inflexible block (clearing at P 2 and Q 2 ) is indicated by the blue region (area A) minus the red region (area B). In the figure on the right, the additional social surplus from skipping the inflexible block and clearing the offer above the inflexible block (clearing at P 3 and Q 3 ) is indicated by the grey region (area C). In these illustrations, we see that the additional social surplus from clearing the inflexible block at P 2 and Q 2 (area A minus area B) is larger than the additional surplus if the inflexible block was skipped, and the next block was cleared at P 3 and Q 3 (area C). Therefore, in this scenario, selecting the entire inflexible block creates the greatest additional social surplus and the inflexible block would be cleared (auction clears at P 2 and Q 2 ). Staying at P 1 and Q 1 would result in smaller social surplus. The market-clearing engine used by the AESO to clear the capacity market will choose the higher quantity and price that maximizes the social surplus. Maximizing social surplus is also the approach used by the AESO in the energy market. Page 10 of 15

60 Inflexible block Figure 4 Maximizing social surplus Inflexible block Skipped When clearing the auction to maximize social surplus, the auction clearing price is set at the intersection between the supply and the demand curves, P p as shown in Figure 5 below. Figure 5 Auction price setting However, it is possible that the entire capacity supply curve or the portion of the capacity supply curve cleared in the auction lies below the demand curve as shown in Figure 6 and Figure 7 respectively. Page 11 of 15

61 In Figure 6 and Figure 7, if the procurement volume is Q p, the price value is not unique as the cost of the capacity at quantity Q p (represented by the supply curve SS at P s ) and willingness to pay at quantity Q p (represented by the demand curve DD at P d ) are not equal. In these situations, the AESO will set the capacity auction clearing price at the intersection between the vertical line drawn from the procured quantity Q p and the demand curve, i.e., the P d in the charts. Figure 6 The entire supply curve lies below the demand curve Figure 7 The supply curve of selected capacity assets lies below the demand curve The AESO will set the capacity market clearing price at the demand curve when the entire supply curve is below the demand curve, or when the entire procurement volume is below the demand curve. Setting the clearing price at the demand curve enables the price to reflect the market's value of additional capacity. Although it does not lead to the lowest procurement cost in one particular auction, it does provide price signals to support the efficient entry of additional, lower-cost capacity assets over time. When the clearing price is set at the demand curve, (P d ) in Figure 8, it provides a strong price signal for new or additional capacity assets to enter the market during the next auction. In the example below, the new or additional capacity asset enters at P n (indicated by the blue line in Figure 8), leading to additional social surplus (denoted by Area A) in the long run. Page 12 of 15

62 If the clearing price was set at P s instead of at the demand curve (P d ), this would result in a lower price but would not provide the price signal the new or additional capacity asset may need to enter the market. Over the long term this could lead to inefficient outcomes and reliability issues due to under procurement. Figure 8: The supply curve of selected capacity assets lies below the demand curve Setting the market price at the demand curve also prevents a situation where the entry of a new capacity asset in one auction can cause the market clearing price to collapse in the following auction (when it is the marginal unit, and there is no change in market supply and demand). This feature helps to ensure the overall market structure is attractive for new investment. Figure 9 illustrates a scenario assuming a new capacity asset enters the market and sets the clearing price at P n in its first capacity auction and as it transistions to being an existing capacity asset in subsequent auctions it reduces its offer price to P m due to lower going forward cost and, potentially market power mitigation. As illustrated in Figure 4, if the supply and demand remain the same, and the capacity asset offers at price P m in future auctions after recovering some of its fixed costs in earlier auctions, the market price would drop from P n to P m if the market price is set by the the marginal offer instead of the demand curve, even though there is no change in market supply and demand. This could discourage future capacity assets from entering the market if they can offer at a price between P n, and P m. However, when the price is set at the demand curve, the market price in the subsequent auction would stay at P n, correctly reflecting the fact that there is no change in market supply and demand and providing accurate price signals to other capacity assets. Page 13 of 15

63 Figure 9 Price set at the demand curve avoids new entry causing price to collapse Setting the capacity price at the demand curve would allow the market price to be set at the price cap if the scenario where there is insufficient capacity supply to meet the minimum procurement volume Subsection in Section 2, Supply Participation allows a firm that submits a refurbishment plan that is approved by the AESO to offer the refurbished capacity asset at an unmitigated price. For each refurbished asset, the firm will be required to indicate as part of the prequalification application whether the asset will: (a) retire if it fails to receive an obligation; or (b) not retire and submit an offer to be used for the existing asset s capacity. Subsection (b) outlines a multi-stage clearing process for the capacity auction whereby if offers associated with refurbishment capacity assets do not clear, the capacity assets are added back into the supply curve at non-refurbished offer prices as existing capacity assets, and the market clearing process will be run again. This process will continue until all refurbishment offers submitted under subsection (b) have either cleared the market or have been added back into the supply curve as existing capacity assets. To simplify the auction clearing process, refurbishment offers will be required to be single, inflexible blocks In the event transmission constraints are expected to cause capacity deliverability issues, the capacity price will be set at the level absent of transmission constraints to reflect the demand and supply balance of capacity assets. In this situation, offer blocks may be required that are priced above the unconstrained price. If this occurs, the offer blocks that are selected to meet the total capacity requirement would be paid an uplift payment in addition to the market clearing price. The uplift payment would be equal to the difference between the offer price and the unconstrained clearing price. This design feature is put in place because capacity market price, by itself, is not meant to be a trigger for transmission system development. In accordance with the provisions of the Electric Utilities Act and the regulations thereunder, the AESO takes into account a number of factors when determining whether an expansion or enhancement of the capability of the transmission system is or may be required. Page 14 of 15

64 5.9 Addressing intertie transmission constraints Section 2, Supply Participation discusses how capacity volumes are determined for individual external capacity assets. Alberta has limits on the amount of capacity that can be delivered through interties. Joint intertie scheduling limits will be determined and made available as part of the overall auction process. There may be auctions in which there are more qualified external capacity assets than there is available import capacity when joint scheduling limits across multiple interties are considered. For example, transmission delivery constraints may be observed on the Alberta BC Intertie and Montana Alberta Tie-line. The constraints will be a result of the combined flow limit on those two interties. The unforced capacity (UCAP) volumes of the external capacity assets will not be reduced to reflect the level of the joint scheduling constraint because this may result in an inefficient outcome where the higher cost capacity assets are cleared prior to fully utilizing the lower cost capacity assets. This treatment will not apply to intertie constraints due to issues outside Alberta interconnected system that are beyond the control of the AESO. Clearing lower-priced capacity assets first, subject to the principle of maximizing social surplus, results in a more efficient outcome and lower costs for consumers. Considering overall social surplus in situations where offers are priced the same also results in more efficient outcomes Addressing internal transmission constraints Alberta s transmission system is designed to support unconstrained operations under systemnormal conditions. It should be noted, however, that transmission development timelines can often extend beyond three years when considering regulatory approval and construction timelines. Development cycles of five to seven years are not uncommon. While constraints are not anticipated, any potential transmission constraints will need to be accounted for when clearing the capacity market so that the AESO does not procure volume that cannot be delivered. This would fail to provide value to consumers and would not meet supply adequacy requirements UCAP volumes of available capacity assets behind a transmission constraint will not be adjusted to reflect the limit of the transmission constraint. Doing so could result in the capacity market clearing some volume of the higher priced capacity asset prior to clearing all of the lower cost capacity asset. Clearing a lower priced capacity asset first, subject to the principle of maximizing social surplus, results in a more efficient outcome and lowers costs for consumers. Capacity assets compete for capacity based on their price structure. This competition promotes a fair and efficient market that treats all capacity assets equally, provided they meet the eligibility criteria. Considering overall social surplus in situations where offers are priced the same also results in more efficient outcomes. Capacity auction assessment against capacity market design criteria Adopting a sealed-bid, single-round auction with a three-year forward period and a 1-year obligation period for all participants promotes a capacity market that is fair, efficient and openly competitive, employs a market-based mechanism that incents competition in a transparent fashion and should result in a well-defined product and an effective and efficient capacity price signal. The one-year term for the capacity commitment is as short as possible and satisfies the design principle that investment risk should continue to be borne by investors. The auction design considers Alberta s unique approach to import and transmission constraint management by creating single price for capacity regardless of location. This is a simple and straightforward initial implementation. While other capacity market implementations differ, the selected design is one that best fits the unique needs of Alberta. Page 15 of 15

65 Rebalancing Auctions Rationale 6.1 Rebalancing auction timelines and procedures A rebalancing auction provides a market-based mechanism for the AESO and capacity market participants to adjust to changes in the load forecast, UCAP ratings, new capacity asset delivery expectations, and their portfolio of assets since the base auction. The updated demand curve reflects the value of capacity under updated system conditions. If the system is expected to be tight, rebalancing auction prices will be high. Capacity committed assets will be incentivized to deliver on their commitments as opposed to buying out at the high rebalancing price, and additional capacity assets will be incentivized to enter. If the system is expected to be oversupplied, prices in the rebalancing auction will be low. Capacity committed assets will be able to buy out of commitments relatively inexpensively, and new capacity assets may not wish to enter the capacity market. The rebalancing auctions are an important component of the AESO s effort to create an efficient capacity market that ensures the reliability of the Alberta electricity system. The rebalancing auctions support efficiency and resource adequacy by: Allowing the AESO to update demand for capacity based on the revised procurement quantity required to meet the supply adequacy requirement. Load forecast error is an unavoidable component of a forward capacity market. While the AESO aims to produce an accurate forecast, there will inevitably be some level of error. A key function of the rebalancing auction is to minimize the reliability and economic impacts of this error. If the AESO under-forecasted load in the base auction, the rebalancing auctions provides an opportunity to buy additional supply and ensure supply adequacy of the system. If the AESO over-forecasted load in the base auction, the rebalancing auctions provide opportunities to sell excess supply and recover costs for consumers. Allowing new capacity assets to enter with less lead time than the three-year forward period. The rebalancing auctions provide a mechanism for capacity assets that were unable to offer into the base auction to obtain a capacity commitment. Demand response providers may not have enough information about their underlying load three years ahead of the obligation period but may be willing to accept a capacity commitment a few months ahead. In addition, capacity assets that cleared the base auction for a given obligation period but came online in less than three years would be able to sell capacity early into a rebalancing auction for an earlier obligation period. Accessing this additional supply should reduce costs for customers. Allowing capacity committed assets with a capacity commitment to buy out if they are unable or unwilling to deliver. There may be capacity committed assets that cleared in the base auction that will be unable or unwilling to bring their capacity asset online in time for the start of the obligation period. There may also be capacity market participants that wish to increase or decrease the capacity obligations of their assets to optimize their capacity asset revenue. The rebalancing auctions provide capacity market participants the ability to complete these transactions while ensuring that the system has enough capacity online by the start of the obligation period. Page 1

66 Allowing up-rates or down-rates to capacity committed assets. Capacity committed assets that are able to increase the output of their plants through incremental investment may wish to make additional volume sales in the rebalancing auction in order to capture additional revenue. Capacity committed assets that must derate their plants to account for poorer than expected operating conditions or equipment problems will be provided an opportunity to purchase replacement supply. The AESO proposes that in the initial stages of Alberta s capacity market program, there will be a transition period in which capacity auctions are held using a compressed schedule whereby only one rebalancing auction will be held for each obligation period. Post transition period, there will be two rebalancing auctions for each obligation period. Rebalancing auctions will be conducted on a fixed schedule. During the transition period, the rebalancing auction will be held approximately three months prior to the obligation period. During the transition period, the reduced number of rebalancing auctions still provides an opportunity for capacity market participants to adjust capacity commitments and for the AESO to adjust procurement volume while avoiding the situation where it becomes impractical to administer too many capacity auctions within a compressed timeline. Post transition period, the first rebalancing auction will occur 18 months prior to the obligation period and the second will occur three months prior to the obligation period. Holding the final rebalancing auction approximately three months prior to the start of the obligation period allows for essentially final load forecasts and generator availability information to be used. The fixed schedule for a rebalancing auction will facilitate participation in the auction by reducing the uncertainty faced by capacity market participants. With a fixed schedule, capacity market participants offering new capacity into a rebalancing auction can ensure that their project development and implementation plan is sufficiently progressing to qualify by the time of the auction. Capacity committed assets at risk of being unable to meet their capacity commitment know exactly how much time is available to achieve their next milestone before the rebalancing auction offer and bid submission window opens. The AESO will establish a capacity auction schedule that allows sufficient time for capacity assets to qualify, to establish UCAP ratings for all capacity assets, to publish auction parameters, to determine auction results, and to evenly distribute the administrative requirements of running auctions over each calendar year. The alternative to fixed schedules running auctions only when certain criteria are met results in less predictability. The AESO s proposal to hold two rebalancing auctions post transition period strikes a reasonable balance between several competing factors. Holding more rebalancing auctions promotes transparency and rapid price discovery by making relevant information available to the market soon after it becomes available. For example, if a new capacity asset determines it will not be available in time for the obligation period and immediately buys out its capacity commitment in a rebalancing auction, the rest of the market will quickly become aware of the increased supply tightness through a higher rebalancing auction price. On the other hand, holding fewer rebalancing auctions increases liquidity in each individual auction; reduces transaction costs; and reduces the administrative burden of facilitating and participating in the auctions. Rebalancing auctions follow similar steps and timeline to those of the base auction, providing a consistent process for all capacity auctions. The AESO estimated that approximately eight months are required to complete all the activities necessary for conducting a capacity auction. Page 2

67 6.2 Bids and offers by capacity market participants The AESO s proposal allows capacity market participants to submit offers and bids into the rebalancing auctions under the following rebalancing auction objectives: Incremental sell offers. Enable capacity assets to enter the capacity market with less than the three-year forward period. These offers also ensure increased UCAP is offered into the capacity market. Repricing (buy-out) bids (incorporated in supply offers under gross clearing). Enables a capacity committed asset to buy out of its capacity commitment or to reduce its cleared capacity, contingent on market clearing prices. A capacity supplier looking to buy back who wishes to have a high degree of certainty to clear may submit a bid price at the price cap. UCAP reduction bids (incorporated in supply offers under gross clearing). Enable a capacity committed asset that is physically unable to deliver on its obligation, in part or in full, to buy out of its obligation regardless of the rebalancing auction price. A UCAP reduction bid price will be entered at a price in the final rebalancing auction marginally above the rebalancing auction price cap to ensure that it clears. Capacity suppliers choosing to buy back an obligation will not have the ability to submit a bid price above the price cap unless they are subject to a UCAP reduction in the final rebalancing auction or have not achieved development milestones. Since a capacity supplier is able to select a UCAP within an available range, UCAP reduction only occurs when the existing obligation volume of a capacity asset is greater than its final UCAP determined in accordance to Section 3, Calculation of Unforced Capacity (UCAP). Capacity suppliers which are not required to or do not wish to alter capacity commitments do not need to participate in a rebalancing auction. This type of capacity committed asset will be automatically entered as a price taker on the supply side of the auction, but will not incur any settlement as a result of the rebalancing auction. The majority of capacity suppliers who clear the base auction are expected to fall into this category Rebalancing auctions facilitate capacity suppliers buying out of their existing capacity commitments. Therefore, bids are required to be capacity asset specific and are not allowed to exceed a capacity asset's existing capacity commitment volume. The following requirements ensure that the auction clearing algorithm can be solved: sell offer quantities in each price-quantity pair shall be incremental quantities, such that the aggregate UCAP offered across all price-quantity pairs submitted increase monotonically with increasing price; buy bid quantities in each price-quantity pair shall be incremental quantities, such that the aggregate UCAP bid across all price-quantity pairs submitted decrease monotonically with increasing price Rebalancing auctions treat offer formats in the same manner as base auctions. The rationale for the proposed methodology is discussed in subsections 5.5 of Section 5, Base Auction. 6.3 AESO s bids and offers With a gross clearing methodology, a demand curve shift or firms bids and offers may cause the AESO to buy or sell capacity. All of the AESO s transactions will be facilitated through the demand curve, and the AESO will not submit offers through the supply curve. However, the AESO will submit bids on behalf of firms that have not submitted the required UCAP reduction bids in the second rebalancing auction. These bids will be submitted marginally above the price cap on behalf of the capacity assets with a UCAP reduction. The capacity supplier is responsible Page 3

68 for all costs associated with covering the obligation caused by a UCAP reduction. This measure is put in place to mitigate supply adequacy risks caused by capacity committed assets not having enough UCAP to fulfill their obligation volume. 6.4 Auction clearing, price setting, and settlement The AESO proposes to clear the rebalancing auction on a gross basis (i.e., including all supply, and demand in the market in the same way as the base auction), but to settle the auction on a net basis (i.e. only differences between the quantities cleared in the base auction and rebalancing auctions would be settled at the rebalancing price). Gross clearing in the rebalancing auctions increases transparency by allowing capacity market participants to easily see the effect of updated auction parameters on the AESO s demand curve and to see the volume cleared in the prior auctions. Clearing a rebalancing auction in the same way as the base auction reduces the likelihood of unanticipated outcomes due to idiosyncratic differences between forward and rebalancing auction mechanics. The gross clearing with net settlements approach is used by ISO- NE in its forward capacity market, and is also used in US real-time energy markets, which follow and rebalance day-ahead markets. 6.5 Anticipated transmission constraints Rebalancing auctions treat anticipated transmission constraints in the same manner as base auctions. The rationale for the proposed methodology is discussed in subsections 5.9 and 5.10 of Section 5, Base Auction. Rebalancing auction assessment against capacity market design criteria After the transition period, two rebalancing auctions occur before the obligation period. These rebalancing adjustments employ market-based mechanisms that should provide an effective balance between capacity cost and supply adequacy resulting in a reasonable capacity costs for consumers while still contributing to the reliable operation of the electricity grid. The use of rebalancing auctions are an effective best practice found in other capacity market implementations for dealing with forecast risk in the capacity procurement volume and availability risk for capacity assets. Inclusion of this design feature assists with satisfying the criteria of maintaining reliability objectives at lowest cost to consumers. Page 4

69 Capacity Market Monitoring and Mitigation Rationale Capacity market monitoring and mitigation background Due to the structure of capacity markets there may be an incentive and ability for a firm to influence market prices to enhance the value of its portfolio of capacity assets at the expense of other firms or ratepayers. Firms may attempt to influence market price in a number of ways. They may attempt to physically or economically withhold supply from the market to increase prices and augment the value of their remaining capacity assets, or those firms that have a large enough net-short capacity position may be incented to offer capacity at prices below cost to suppress market prices. The purpose of market power mitigation mechanisms is to limit behaviour that introduces inefficiently high or low market prices to the benefit of one firm, at a detriment to the market as a whole. The need for market power mitigation The AESO and the Brattle Group have conducted analysis that evaluated the level of market concentration and the likelihood that a firm may profitably influence the market clearing price in the capacity market. The AESO has determined that the capacity market in Alberta may provide an opportunity for firms to exercise market power under certain conditions; therefore, market power mitigation mechanisms are appropriate. 7.1 Mitigation of supply-side market power The need for supply-side mitigation The need for supply-side market power mitigation arises when a capacity market is concentrated and certain firms control enough capacity to effectively exercise market power. Supply-side market power refers to the ability of a firm or group of firms to withhold capacity from the capacity market to increase prices to benefit its remaining capacity assets. The level of market concentration in the Alberta capacity market can be assessed by calculating the percentage of the total unforced capacity (UCAP) controlled by the largest firms providing supply. At the time \CMD 1 was developed, the AESO calculated that the majority of the Alberta market fleet-wide unforced capacity available in the market is controlled by five firms. The five firms in Alberta control over 70% of the entire Alberta market fleet-wide unforced capacity in the market with the top two firms controlling almost 45% of the Alberta market fleet-wide unforced capacity. The latest unforced capacity data confirmed this observation. Table 1 indicates that the top five firms in Alberta control 79% of the entire Alberta market fleet-wide unforced capacity with the top two firms controlling almost 50% of the Alberta market fleet-wide unforced capacity. Page 1 of 9

70 Table 1 - Portion of the fleet-wide unforced capacity in the Alberta Market controlled by the top 5 firms Firm Offer controls based on fleet-wide unforced capacity (includes wind) Firm % Firm % Firm % Firm % Firm 5 3.2% Top 5 Total 79% Table 1 above also indicates that the Alberta capacity market will be sufficiently concentrated to raise concerns of market power. Nevertheless, not all firms that control large amounts of capacity have the incentive to exercise market power. The incentive to exercise market power depends on two factors: 1) how responsive the clearing price is to changes in supply due to withheld capacity; and 2) how much additional capacity a firm has left in the market to benefit from the increased price, after withholding a portion of its portfolio capacity. For example, consider a firm portfolio size of 500 MW of UCAP, and a competitive market clearing price of $75/kW-year. By withholding 200 MW of its portfolio s UCAP, the firm could increase the clearing price to $100-kW/year, thereby gaining $25/kW-year on the remaining 300 MW of UCAP in its portfolio. While this course of action would result in a gain of $7.5 million to the firm, it would lose $75/kW-year on the withheld 200 MW of UCAP, resulting in a loss of $15 million. In this example, the firm would not have an incentive to withhold capacity from the market. The AESO and the Brattle Group conducted analysis to determine at what size of a portfolio a firm begins to be incented to withhold capacity. The results of the analysis are dependent on the shape of the supply curve and demand curve utilized in the base auction. The shape of the supply curve and demand curve will determine how responsive the clearing price is to changes in supply due to withheld capacity. The analysis was conducted using six different demand curve shapes, and an estimated upward sloping supply curve developed by the Brattle Group. Table 2 shows, for each demand curve option and at three different quantities of withheld capacity, the minimum portfolio size at which a firm would have an incentive to withhold capacity. Table 2 illustrates that using a demand curve with a price cap at 1.75x net-cone, based on a resource adequacy target of 400 MWh of expected unserved energy (EUE), a firm with 1,290 MW of UCAP in its portfolio could profitably withhold 110 MW from the capacity auction. This would result in an increase in the clearing price by 10%, or by $13/kW-year. In general, the results of the analysis indicate that a firm with a portfolio size of 1,100 MW of UCAP or greater may have the incentive to withhold capacity from the Alberta capacity market. Page 2 of 9

71 Table 2 Market power incentive test results based on demand curves considered in CMD 1 Flattest Alberta Curve 400E 1.6x Net CONE Cap Middle Alberta Curve 400E 1.75x Net CONE Cap Steepest Alberta Curve 400E 1.9x Net CONE Cap Flattest Alberta Curve 100E 1.6x Net CONE Cap Middle Alberta Curve 100E 1.75x Net CONE Cap Steepest Alberta Curve 100E 1.9x Net CONE Cap 550 MW Withheld 2,090 MW $50/kW-yr 1,760 MW $63/kW-yr 1,550 MW $77/kW-yr 2,790 MW $34/kW-yr 2,310 MW $43/kW-yr 2,020 MW $52/kW-yr 225 MW Withheld 1,770 MW $20/kW-yr 1,420 MW $26/kW-yr 1,210 MW $32/kW-yr 2,440 MW $14/kW-yr 1,980 MW $18/kW-yr 1,690 MW $21/kW-yr 110 MW Withheld 1,630 MW $10/kW-yr 1,290 MW $13/kW-yr 1,080 MW $16/kW-yr 2,310 MW $7/kW-yr 1,840 MW $9/kW-yr 1,560 MW $11/kW-yr Since the publication of CMD 1, the AESO has conducted further analysis on the minimum portfolio size required for a firm to be able to profit from withholding capacity. The AESO examined the minimum portfolio of UCAP required to profitably raise the clearing price by 10% based on the demand curve which is expected to be used for capacity procurement. In this analysis, the AESO examined the price change along the demand curve and did not use an upward-sloping supply curve. In this approach, the AESO performed the price change assessment above and below the inflection point on the demand curve to account for the fact that withholding capacity at different segments of the demand curve will have different impacts on auction price. The AESO assessed the average amount of capacity to be withheld to introduce a 10% increase in clearing price on both demand curve segments around the inflection point. This analysis indicated that a firm with approximately 1070 MW of UCAP would be able to profitably raise the clearing price by 10% through withholding capacity assets. Additional sensitivity analysis illustrated that a small increase in the portfolio size would allow a firm to be able to profitably increase the clearing price by 15%. The above analysis demonstrates that the Alberta capacity market is structurally concentrated, such that there are several firms in the market that have the incentive and potential ability to exercise market power. To ensure that capacity market results are reflective of competitive outcomes, the AESO has therefore determined that ex ante supply-side market power mitigation measures, which consist of a must offer requirement, a market power screen and a default offer price cap (with a process for a firm to request an asset-specific offer price cap) are necessary. These market power mitigation components are consistent with those proposed for market power mitigation in the energy and ancillary service markets. The supply-side mitigation measures utilized in capacity markets in other jurisdictions, summarized in Table 3, provide context and comparison for the measures proposed by the AESO. All the jurisdictions described in Table 3 utilize the same components of the supply-side mitigation measures being proposed by the AESO. These are: (a) a must-offer requirement to mitigate physical withholding of capacity; (b) a market power screen to determine which firms could potentially exercise market power; (c) a default offer price cap that applies to all firms that fail the market power screen; and (d) an asset-specific offer price cap for a firm that has failed the market power screen but can demonstrate that its qualified capacity asset s costs are higher than the default offer price cap. Page 3 of 9

72 Table 3 - Supply-side mitigation measures in other jurisdictions Component PJM ISO-NE NYISO UK Ireland Must-offer requirement Yes Yes Yes Yes Yes Market power screen 3 Joint pivotal supplier Pivotal Supplier Pivotal Supplier All resources are mitigated All resources are mitigated Default offer price cap Net-CONE x previous three balancing ratios Dynamic Delist Bid is the cap; Estimated cost of supplying capacity Higher of projected auction price or net going forward costs 50% of net- CONE 50% of net- CONE Asset-specific offer price caps Yes, based on net going forward costs Yes, based on net going forward costs Yes, based on net going forward costs Yes, based on net going forward costs Yes, based on net going forward costs In CMD 3, the AESO also provided an alternative market mitigation approach. This alternative includes setting a default offer cap to 1x net-cone that would apply to all existing assets without market power screen or asset specific price caps. However, feedback from the capacity market Design Working Group on this alternative indicates both the concern that a default cap at 1x net-cone would expose consumers to higher capacity costs and the concern that capacity market participants are unable to request assetspecific offer price cap if their assets have net avoidable costs above 1 times net-cone. As such, the AESO is not suggesting the adoption of this alternative but rather implement the market power mitigation measures which consist of a must offer requirement, a market power screen and a default offer price cap (with a process for a capacity market participant to request an asset-specific offer price cap). The rationale for each mitigation measure proposed by the AESO in CMD Final is provided below. Must-offer requirement The must-offer requirement and the delisting process described in Section 2, Supply Participation, Section 5, Base Auction and Section 6, Rebalancing Auctions have been designed to prevent physical withholding in the capacity market. A must-offer requirement is employed by each jurisdiction in Table 3. Requiring all qualified capacity assets to offer into the capacity auction facilitates competitive prices for all firms and rate-payers. Market power screen The market power screen proposed by the AESO is a structural test designed to identify firms that have a capacity offer control of a UCAP portfolio large enough to profitably exercise market power. Those firms who pass the market power screen will not be mitigated. While any auction price increase caused by withholding capacity would lead to price distortion and an increase in consumer costs, setting a lower threshold percentage may result in overmitigation due to possible estimation errors of portfolio UCAPs. Therefore, the AESO proposes to set the threshold for the market power screen at the average portfolio UCAP size at which a firm is able to increase the clearing price by 10% above and below the inflection point on the demand curve by economically withholding capacity, without incurring any financial loss or gain (i.e., the firm would break even ). The portfolio size will only include the existing unforced capacity because new capacity may not be built if it does not clear in the base auction. Therefore, this new capacity should not be included in the firm s portfolio. Page 4 of 9

73 The AESO tested the sensitivities of setting the price impact level at 5% 20% and did not observe material changes to the size of portfolio that would fail the market power screen. The 10% price impact level considers the possibility that a firm contemplating a withholding strategy faces uncertainty as to the potential success of the strategy due to variability in market conditions and that it may employ a certain level of expected price change before employing an economic withholding strategy. However, a higher price impact level would increase the risk of consumers having to pay higher costs caused by firms exercising market power. The price impact level needs to balance the risks of over-mitigation with consumer exposure to higher prices due to firms exercising market power. The AESO is of the view that a 10% price increase around the inflection point strikes a balance between these two types of risk. By completing this analysis at points above and below the inflection point of the demand curve, the AESO is not predetermining where the auction will settle. While the net-cone value is above the inflection point, which might suggest establishing the portfolio test above the inflection point, the default offer price cap is at a level below the inflection point. The AESO considers that setting the market power screen using points both above and below the inflection point balances these two considerations. The AESO is not proposing the use of a fixed portfolio size or percentage of total market UCAP as the market power screen threshold. These static measures may not be appropriate if other market parameters, such as the demand curve shape or CONE levels change. Market power mitigation measures will not be applied to rebalancing auctions. The majority of capacity will be procured and cleared in the base auction; therefore, the capacity to be cleared in a rebalancing auction is expected to be minimal. Therefore, both the ability of a firm to profitably withhold capacity to raise the capacity price to the benefit of its portfolio and the potential for the clearing price in the rebalancing auction to have a material adverse impact to overall procurement cost, are limited. Not applying market power mitigation measures to a rebalancing auction will reduce the risk of over-mitigation and the administrative burden of the firms and the AESO. While the AESO will not apply market power mitigation measures to a rebalancing auction, a rebalancing auction will be included as part of the auction competitiveness assessment (see subsection 7.3.3). Should this assessment indicate that rebalancing auctions also require mitigation, the market power mitigation measures may be applied to future rebalancing auctions. Default offer price cap The use of a default-offer price cap is an administratively efficient mechanism (i.e., no subjective assessment a firm s behaviour), and focuses mitigation efforts on those firms that have the greatest incentive to exercise market power. In CMD 2, the AESO proposed a default offer price cap of 50% of net CONE. This was based on an assessment of net avoidable costs for different technology types conducted by the AESO. The charts in Figure 1 below illustrate the net avoidable costs of different technologies as a percentage of net-cone. These percentages indicate what percentage of net-cone a capacity asset needs to receive in order to recover its net avoidable cost. In the charts, the net avoidable costs of existing conventional coal, simple cycle and combined cycle gas technologies do not include return of or on capital because the capital costs are sunk. The net avoidable cost of coalto-gas conversion includes the return of capital because the conversion has not occurred and the capital costs have not incurred. Figure 1 shows that the net avoidable costs of existing combined-cycle and simple-cycle gas-fired generation are 0% of net-cone. This is because the expected energy and ancillary services revenues of these technologies are above their avoidable costs. The net avoidable costs of coalto-gas conversion units that need to be recovered in the capacity market are 20% to 40% of net- CONE (the higher number applying before conversion costs are expended and the smaller number applying after conversion costs become sunk costs). The net avoidable costs for conventional coal units that need to be recovered in the capacity market range from 60% to 80% of net-cone depending on the year and energy market mitigation scenario. This suggests that the default offer cap proposed by the AESO will be adequate to allow for the economic operation of the asset. Page 5 of 9

74 Figure 1 Estimate of net avoidable costs by technology type While the foregoing analysis provides the rationale for setting the default offer price cap at 50% of net-cone and demonstrated that 50% of net CONE would allow most technology types to recover their full net going-forward costs, some stakeholders suggested that the default offer price cap is too low, resulting in over-mitigation in the capacity market. There is also a concern that the possible overmitigation may discourage investments in capacity assets and eventually negatively impact supply adequacy. Therefore, the AESO proposes to increase the default offer price cap from 50% of net- CONE to 80% of net-cone. The 80% of net-cone is chosen based on the analysis that this level would allow existing conventional coal units to recover net avoidable costs and therefore represents a plausible range of the net avoidable costs of existing conventional coal units. Increasing the default offer cap to 80% of net-cone addresses the concern of over-mitigation and potentially reduces the number of asset-specific offer price cap requests, which will reduce administrative efforts for both capacity market participants and the AESO. There will be situations where the price cap is set at a gross-cone multiple. This will occur when the demand curve is expressed in gross-cone terms as described in subsection 4.4.2(c). For example, consider the demand curve price cap is set according to the formula max (1.75 net-cone, 0.5 gross- CONE). When net-cone is lower than 0.5/1.75x gross-cone, the default offer cap price would be set at 0.8 X 0.5/1.75x gross-cone, which is 0.23x gross-cone. When net-cone is very low (e.g. below 0.29x gross-cone) in above example, it is expected that some existing units may not be as efficient in the energy and ancillary services market as the reference technology and therefore still have positive net avoidable costs. Setting the default offer price cap at a gross-cone multiple would allow these capacity asset to offer at their avoidable costs without having to submit asset specific offer cap requests. An alternative is to set a default offer price cap at a fixed $/kw-year when net- CONE is very low. However, setting the default offer price cap at a fixed level may result in a default offer price cap that is higher than the costs of the reference technology. Asset-specific offer mitigation Asset-specific offer mitigation facilitates the participation of a qualified capacity asset that has net avoidable costs higher than the default offer price cap. This approach enables a firm that has capacity Page 6 of 9

75 offer control of a qualified capacity asset that has net avoidable costs higher than the default offer price cap to submit offers at levels reflective of net avoidable costs. In addition, this approach avoids over-mitigation, which may artificially drive capacity assets out of the market. Avoidable costs are the costs that can be avoided by a firm that has capacity market offer control of a capacity asset if the asset is delisted for a year. Net avoidable costs are avoidable cost less the expected margin from energy and ancillary service sales less the variable costs incurred to generate those sales. Costs such as major maintenance investment, depreciation, servicing of capital, depreciation, are not avoidable if a capacity asset is delisted for a year. These costs should not be included in the avoidable cost calculation because a firm owning only the single asset would not include these costs in their offers. Using net avoidable costs as the basis for asset-specific offer price caps is intended to reflect the price at which a firm that has an offer control of a qualified capacity asset but who does not hold market power, would be willing to offer that asset into the capacity auction. It is an estimate of the avoidable cost of making capacity available for the delivery period, taking into consideration expected margins from energy market operation. In all delist economic reviews, the expected net energy and ancillary services revenues will be deducted from the avoidable fixed costs of the asset. The guiding principles on how the AESO interprets avoidable cost are included in subsection Shared costs that are expected to be re-allocated, transferred or re-monetized, and sunk capital, including the return of, and on, capital, are not avoidable. Therefore, these costs should not be included in the avoidable cost calculation. Including only avoidable costs to assess the asset-specific offer price cap effectively restricts market power and allows the price signal to be closer to one that would be formed absent of market power. Expected payment adjustments costs are included in avoidable cost calculation because they are avoidable if the capacity asset does not take on a capacity obligation (e.g., if the capacity asset is delisted) and as such represent a cost of making capacity available to the market. Project investment for continuing operation and availability improvement, opportunity costs, major incremental capital expenditures and a return on incremental capital investments, net decommission costs that were previously included in the avoidable cost calculation in CMD 2 have been removed for the following reasons: Project investment for continuing operation and availability improvement can only be deferred but not avoided if a capacity asset is delisted for a year. Therefore, these costs should not be included in the avoidable costs submitted for asset-specific offer price cap requests. Opportunity costs are primarily applicable to situations of permanent delisting and there is no requirement for an economic test for permanent delisting. Therefore, opportunity costs are not relevant in the context of market power mitigation. Incremental and refurbished capacity assets, as defined in Section 2, Supply Participation, are not subject to market power mitigation. Therefore, major incremental capital expenditures and a return on incremental capital investments over a demonstrated capacity asset life are not relevant in the context of market power mitigation. An asset that has permanently delisted is not subject to market power mitigation. Therefore, net decommissioning costs are not relevant in the context of market power mitigation. The AESO also identified the need to establish the appropriate escalation rates that may be applied to calculate the costs measured in the dollar of the delivery year and the process of assessing firms asset-specific offer price requests. These additional details will be developed by the AESO and will be subject to further consultation. As an asset-specific offer price cap is based on net avoidable costs, a firm that requests an assetspecific offer price cap will be required to submit an asset s avoidable costs, including supporting evidence in relation to such costs. The AESO also requires a firm that requests for asset-specific offer price caps to provide variable operating costs in order to determine the energy and ancillary services market offset used to calculate net avoidable costs. The energy and ancillary services markets offset will be determined in the same manner as used for the reference unit net-cone calculation as described in Section 4, Calculation of Demand Curve Parameters. Requiring the Page 7 of 9

76 submission of both the cost data and the attestation from a corporate officer of the firm that requests the asset-specific offer price cap is to ensure the correct cost categories are included in the submission and the costs submitted are reflective of true avoidable costs. A firm may utilize the dispute resolution process described in Section 5, Base Auction if a disagreement between the firm and the AESO regarding to assets net avoidable costs arises. 7.2 Mitigation of suppliers with net-short positions The need for mitigation of net-short capacity positions A firm that has a large enough net-short position (i.e., the firm would benefit from a reduced capacity-auction clearing price due to a requirement to pay for capacity) may have the ability and incentive to offer capacity into the market below cost in order to reduce prices in the capacity market. This outcome would harm all other firms in the capacity market, and could potentially discourage future capacity investment. The AESO and the Brattle Group have conducted an assessment to estimate the minimum net-short capacity position needed to create the incentive to make uneconomic offers into the capacity market. Similar to the supply-side incentive test described above, the results of this assessment depend on the shape of the demand curve used in the capacity auction, the cost of the capacity to be offered below cost, and the overall size of a firm s net-short position. The analysis that was conducted when CMD 1 was developed tested the six different demand curves that were being considered for use in the Alberta capacity market and was based upon the assumption that capacity offered below cost was equal to 1.2x net-cone. Table 4 illustrates that a firm would need to have a net-short position of at least 370 MW to be incented to offer 110 MW of capacity into the market below cost. Under other demand curve assumptions, the net-short position needed to have this incentive increased. Table 4 Estimate of net-short capacity position incentive test based on demand curves considered in CMD 1 Flattest Alberta Curve 400E 1.6x Net CONE Cap 550 MW Net Short 1,200 MW $31/kW-yr 225 MW Net Short 770 MW $14/kW-yr 110 MW Net Short 640 MW $7/kW-yr Middle Alberta Curve 400E 1.75x Net CONE Cap 1,150 MW $33/kW-yr 640 MW $18/kW-yr 520 MW $9/kW-yr Steepest Alberta Curve 400E 1.9x Net CONE Cap 1,100 MW $35/kW-yr 650 MW $17/kW-yr 480 MW $9/kW-yr Flattest Alberta Curve 100E 1.6x Net CONE Cap 1,050 MW $38/kW-yr 570 MW $21/kW-yr 460 MW $10/kW-yr Middle Alberta Curve 100E 1.75x Net CONE Cap 990 MW $42/kW-yr 530 MW $24/kW-yr 380 MW $12/kW-yr Steepest Alberta Curve 100E 1.9x Net CONE Cap 950 MW $46/kW-yr 500 MW $25/kW-yr 370 MW $13/kW-yr Some firms in Alberta may a have net-short capacity position, such as a competitive retail entity or a selfsupplying load. However, under the capacity cost-allocation structure directed by the government, retailers are not expected to be exposed to capacity cost. Self-supply customers may have short capacity position but the AESO s review of self-supplying loads has indicated that none of these capacity market participants have a net-short position large enough to allow them to profitably exercise buyer-side market power. Based on this, the AESO has determined that no mechanism for mitigating net-short (or buyerside) market power is required at this time. This may need to be reviewed in the future in the event of changes to portfolio compositions, UCAP determination for self-suppliers, or capacity cost allocation structure. The AESO has also determined that implementation of a minimum-offer price requirement (MOPR) on identified net-short firms for the purpose of mitigating market power that may arise due to net- Page 8 of 9

77 short positions in the Alberta market is not required at this time. The AESO s analysis did not identify any firms with a large enough net-short capacity position needed to create the incentive and ability to gain from artificially suppressing capacity prices. The main difference between Alberta and other capacity markets, where net-short capacity positions do occur, is that there are no load serving entities (LSEs) in Alberta with customers for whom they are responsible for securing capacity for. Capacity is purchase by the AESO on behalf of the load and costs are flowed through to the end use customers. Alberta based load serving entities may create a short position through financial hedges meant to lock in the price of capacity for their customers. The AESO is not aware of any of these products being offered at this time. Currently, it is unlikely that these firms have a net-short position large enough to provide the incentive to suppress market prices: Regulated Rate Option ( RRO ) providers. The RRO providers in Alberta either do not own capacity assets or are prohibited to share information with an affiliated provider that owns capacity assets. Therefore, the RRO providers do not have the ability to exercise buyer-side market power in a capacity auction. Competitive retail entities. As a result of the capacity cost allocation methodology proposed by the Government of Alberta, these retail entities (despite competitive net-short position) are not exposed to the capacity price because the capacity cost is passed to consumers through the tariff. Self-supplying loads. The Brattle Group s analysis indicates that a net-short position of over 370 MW is necessary to create the incentive to suppress prices. Through analysis completed by the AESO, there are no self-supplying loads with a short position of this magnitude. 7.3 Reporting of auction statistics and market competitiveness Auction statistics and capacity market assessments may assist in ensuring the capacity market is competitive, efficient and supporting Alberta s reliability needs. This process is also intended to provide sufficient information to support business decisions, investor confidence, and allow for industry engagement on potential capacity market design flaws and possible solutions. The AESO intends to publish on its website, as soon as reasonably practicable following a capacity auction (likely within 1 week of the capacity auction), and concurrent with the notification of auction results, the following auction statistics reports: (a) clearing price ($/kw-year); (b) total cleared capacity (MW); (c) cleared capacity differentiating between existing and new resources as well as technology type; and (d) a list of assets with a capacity market obligation after the second rebalancing auction, including asset name, resource type, new/existing/uprates, submitting party, and awarded MWs. The AESO will publish (a), (b) and (c) in order to promptly provide transparency to the market of the capacity market auction outcome. This provides a starting point for market participants to assess their business decisions and for market observers to analyze the competitiveness of the market. The AESO intends to publish (d) to minimize information asymmetry and help firms identify possible parties that may able to facilitate capacity asset substitution. The AESO does not intend to release (d) prior to the second rebalancing auction because asset substitution occurs after the conclusion of the second rebalancing auction. The AESO will not publish the supply curve following a capacity market auction because in a concentrated market such as Alberta s, releasing offer prices may negatively impact the competitiveness of subsequent auctions. However, the AESO may consider publishing supply curve data when it determines that such data is not expected to have undue impacts on the competitiveness of the market. Page 9 of 9

78 Supply Obligations and Performance Assessments Rationale Overview of payment adjustments In exchange for capacity payments, capacity market participants take on an obligation to maintain their availability throughout the year, to perform when called upon by the AESO during shortage conditions, and to offer into the energy market. Payment adjustments are an asset-neutral approach developed to encourage capacity market participants to perform in accordance with their obligations. Capacity market participants are expected to reflect the expected cost of payment adjustments. In the long run, the payment adjustments will provide a financial signal to capacity market participants to maintain supply adequacy at lowest cost to consumers, as assets with lower performance risk will have a competitive advantage. Prior to the start of the obligation period, new capacity assets that are delayed in meeting their in-service date and existing capacity assets that anticipate not being available during the obligation period can participate in rebalancing auctions to reduce their obligation volumes in order to mitigate payment adjustment risk. During the obligation period, asset substitution (both ex ante and ex post) can be used by capacity market participants to manage delivery risk. Asset substitution is being allowed only after the final rebalancing auction with the intent of increasing liquidity in the rebalancing auctions. 8.1 Assessment prior to commencement of obligation period Failure to meet major milestones for new capacity committed assets Assessing a new capacity committed asset s project development and implementation plan allows the AESO to take action prior to the obligation period if it appears that supply will not be available during the obligation period. This approach helps to ensure required levels of supply adequacy and will apply to new assets at significant risk of failing to come online. Prior to rebalancing auctions, the AESO will identify capacity committed assets that are unlikely to be operational by the start of the obligation period. For new capacity assets, this assessment will be based on the completion of development milestones especially those on the critical path. New capacity assets identified by the AESO as being more than eight months behind their development schedule at the first rebalancing auction and more than five months behind their development schedule at the last rebalancing auction will be required to buy out their full capacity obligations. At the final rebalancing auction, completed three months prior to the obligation period, it s unlikely that a project delay of five months or more could be remedied prior to the commencement of the obligation period. As discussed in the subsection 2.3 of Section 2 Rationale, Supply Participation, regarding delisting, unavailability of five months or greater is not a long enough availability period to delivery adequate reliability value through an obligation period. At the first rebalancing auction, a delay of eight months recognizes the risk of a five month delay but also provides the capacity asset owner the opportunity to remedy some amount of the project delay. The development of demand response capacity assets will be reviewed at the final rebalancing auction. If a new demand response asset cannot demonstrate that it has obtained load sufficient to achieve a UCAP of at least 75% of its obligation volume it will be required to repurchase the difference between its obligation volume and its achieved UCAP in the final rebalancing auction. This approach recognizes that certain demand response assets may acquire customers up to the Page 1 of 13

79 commencement of the obligation period while for reliability purposes requires a substantial portion of the UCAP to be attained prior to the final rebalancing auction. Overall, the goal of this approach is to ensure that new capacity committed assets that may be experiencing development delays manage their obligation volume prior to the obligation period in order to ensure that the AESO is able to meet its reliability obligations. The AESO does not believe this approach pushes all new capacity to the month of October. Rather, this approach assesses the readiness of a new capacity asset based on the project development and implementation plan described in Section If a new capacity asset will not be available within five months of the original timeline the obligation volume for that obligation period will need to be repurchased. Conversely, this provision does not create the requirement that resources have a commercial operation date in October. The AESO anticipates that many resources may target commissioning prior to October to ensure availability during the obligation period. Further, asset substitutions are available for new resources to manage their payment adjustment risks. Updates to qualified UCAP ratings In addition to availability and delivery assessments during the obligation period, capacity market participants will have an incentive to maintain their asset s ability to perform because the UCAP for capacity assets will be updated annually in each auction qualification round. These updates will include recent operational performance. Over-availability and over-delivery in recent years translates into a higher UCAP, and therefore, greater potential capacity revenue in the future year. Payment adjustments during the obligation period create incentives for the capacity market participants to meet their forward capacity obligations before the obligation period by ensuring that new supply is on time, retaining existing capacity, or by securing a replacement capacity asset through the rebalancing auctions or asset substitution. 8.2 Assessment during obligation period The AESO considered phasing in the application of the performance assessment program but determined that it was not the appropriate approach. The performance framework is an integral part of the overall market design and is required to ensure that incentives for over-performance as well as consequences for under-performance are in place at the beginning of the obligation period. This approach also aligns with the payment of full capacity revenues to suppliers at the outset of the capacity market. The AESO had suggestions from stakeholders for alternate performance approaches. These suggestions were considered but not reflected in the final market design approach due to failure to meet one or some of these objectives: the approach should retain revenue neutrality for over and under performance capacity assets; the approach should ensure capacity asset performance is incented more highly during performance events rather than availability events; and performance assessment bonuses or penalties should be directly linked to obligation period revenues. Approaches that only assessed penalties against future UCAP determinations did not achieve this goal. The AESO also believes, based on design changes in the eastern US capacity markets, that performance penalties are required to incent compliance. The eastern markets have designed somewhat similar penalty structures in response to past non delivery on capacity obligations. Page 2 of 13

80 Unavailability payment adjustment Utilizing tight supply cushion hours for conducting availability assessments is intended to encourage availability when the system is at risk of reliability challenges. These hours will include emergency event hours where delivery payment adjustments are assessed. Availability will be assessed during the same number of hours as the UCAP assessments described in Section 3, Calculation of Unforced Capacity (UCAP) in order to align incentives and measurement to periods of greatest reliability risk to the system. The goal of this design element is to encourage readiness to be available and respond to dispatch instructions during the obligation period, particularly in times when the system is at risk. As the availability assessment is completed through the obligation period on a large number of hours, providers are able to use periods of higher availability to offset periods of lower availability. Additionally, in response to stakeholders feedback and to facilitate year by year unavailability payment adjustment risk management, the AESO will allow a capacity committed asset with availability volume greater than its obligation volume to be eligible to receive an over-availability payment adjustment. Availability assessment period Unavailability payment adjustments will be assessed by comparing each capacity asset s capacity obligation to its availability during a fixed number of hours during the obligation period. Availability assessment will be conducted during the obligation period over the 250 tightest supply cushion hours. These hours will include the hours in which EEA events occur. This means that if a delivery assessment and availability assessment hour overlap, both the availability and delivery of the capacity committed asset will be assessed. Availability will be assessed annually after the end of the obligation period. The AESO considered assessing availability over shorter hours, quarterly or semi-annually, but was concerned that the split would arbitrarily establish hours for assessment that did not correspond with system tightness. Additionally, if the split was uneven (e.g. 70 hours in the summer/ 30 hours in the winter) the outcome could be an unintended grouping of outages in the period with fewer assessment hours. Assessing availability during these hours is consistent with how capacity asset UCAP will be determined. The number of recommended hours for the availability assessment (250 hours annually) is based on the average number of hours historically between 2011 and 2017 in which supply cushion was approximately 500 MW; conditions which characterize system tightness (see Section 3, Calculation of Unforced Capacity (UCAP). Availability assessment volume definition During each year, capacity committed assets will be required to demonstrate that their actual availability was at least equal, on average, to their obligation volume (expected availability) during the availability assessment hours. Averaging the availability of assets throughout an entire availability assessment period allows capacity assets to compensate their unavailability in some hours with their over-availability in other hours, which also provides a way for assets to manage potential payment adjustment risk exposure. In measuring availability for firm consumption level assets, the calculated lookback baseline represents an ongoing baseline of consumption. This ensures that the characteristics of the capacity asset, including calculated capacity value assumed in the auction process are being delivered Page 3 of 13

81 Unavailability payment adjustment for negative availability assessment volume Tying the payment adjustment to the capacity asset-specific capacity payments (i.e. obligation price per MW) ensures that the payment adjustment level is consistent with each asset's maximum revenue from the capacity market. This approach most accurately reflects the amount of capacity revenues available for each capacity asset that cleared in any of the auctions corresponding to a particular obligation period. Therefore, setting a penalty based on assetspecific capacity payment will not lead to disproportionally high penalties in relation to total capacity revenues in the auction rounds when the rebalancing auction is cleared at a far lower price than the forward capacity auction. As the penalty is not based on the maximum of rebalancing and forward auction prices, this is not discriminatory against assets that have received their obligation in the auction which cleared at a lower price. Overall, this design change is expected to reduce risk exposure and provide more revenue certainty because the payment adjustment is directly linked to the amount of revenue received from the capacity market by each asset. The factor of 40% is an allocation factor representing the amount of the total payment adjustment to an asset that will occur through the unavailability payment adjustment. The 40% weighting is the difference between 100% and the AESO s choice of a 60% allocation factor to non-delivery payment adjustments. The 40% allocation of payment adjustments to availability reflects a higher weighting to delivery period assessments given that these periods represent hours when the system is at greater supply adequacy risk. The Working Groups through Q1 and Q reviewed a number of scenarios for these values 1. The 40% weighting for availability assisted the AESO in achieving a 1.3x maximum penalty structure for the poorest performing assets. The factor of 1.3 scales the total payment adjustment level up above the capacity auction price. A value greater than one ensures that capacity assets failing to deliver are exposed to a net payment adjustment, after accounting for capacity revenues they will receive. A value larger than one also discourages speculative capacity sales, by committing to a capacity obligation the capacity asset is at risk of losing more through poor availability and performance than through what might be earned through capacity payments. The value is believed to be of a magnitude that is sufficient enough for capacity assets to retain the incentive to deliver on capacity commitments, but will not be so large that new entrants will be discouraged from participating. Over-availability payment adjustment for positive availability assessment volume In alignment with multiple stakeholders feedback, the AESO agrees that capacity assets that have average availability greater than their obligation volume should be eligible to receive an over-availability payment adjustment. This design change would make the unavailability payment adjustment revenue-neutral as collected unavailability payment adjustments will be directed to capacity committed assets which are eligible for over-availability payment adjustments. This change is being implemented to help avoid an asymmetric risk exposure for capacity committed assets. In particular, in the years where capacity assets would have been unavailable, they would have assessed unavailability payment adjustments; while in the years where capacity assets would have been over-available they would not have been able to receive any additional payments. In the long run, this would result in only negative payments. While over-availability would be rewarded with higher UCAPs and higher capacity market revenues in future years, the timing of that over-availability payment doesn t provide as timely feedback to assets as within the Page 4 of 13

82 year availability bonuses. Overall, providing a possibility for assets to earn over-availability payments is seen as another way for capacity assets to manage their payment adjustment risk exposure and is expected to decrease the risk premium that would have otherwise been reflected in higher assets capacity offers. The AESO considered but did not include non-capacity committed assets as eligible for the overavailability incentive as these assets do not have a capacity obligation. The availability assessment is meant to ensure capacity committed assets are meeting their obligation during tight supply cushion hours. As described below, the maximum potential over-availability and over-delivery payment adjustments will be capped at a capacity asset s total annual obligation price per MW. Delivery payment adjustment Capacity assets failing to deliver during EEA events will be assessed a non-delivery payment adjustment based on the shortfall between their actual delivery and obligation volumes. Similarly, capacity assets that over-deliver on capacity obligations will receive an over-delivery payment adjustment. These payment adjustments are intended to create a strong marginal incentive to reduce consumption, deliver energy or operating reserves during periods when the system is most in need of supply. By applying a payment adjustment during EEA events, all capacity assets with capacity obligations effectively face a $/MWh incentive, incremental to the energy price, during these events. Delivery assessment period Delivery assessment periods will occur during EEA events, when the system is in need of all available capacity in order to maintain reliability, and operating reserve targets. Any time the AESO declares an EEA level 1 (i.e. all available capacity assets are in use) or higher (i.e. EEA level 2: load management procedure is in effect; EEA level 3: firm load interruption is imminent), the delivery assessment period will begin. A declaration of EEA 0 (i.e. a termination alert issued when energy supply is sufficient to meet AIES load and reserve requirements) will mark the end time of a delivery assessment period. These events are hard to predetermine, and as such, there will be no explicit prior notification before such periods occur. Likewise, there is no maximum duration of the EEA events that can be predicted or pre-defined ahead of time. The AESO will continue to provide the real-time supply adequacy report to market participants which may be a help in identifying periods of tight supply adequacy. Delivery volume definition The performance of a capacity asset is calculated as the capacity asset s actual performance minus the obligation volume, measured during performance assessment periods in MWh. The capacity asset s obligation volume is multiplied by the balancing ratio to determine the volume subject to an over-delivery or non-delivery payment adjustment. The balancing ratio is the ratio of energy and reserves produced by capacity assets during a delivery assessment period to the total committed capacity in that obligation period, and is a number less than or equal to one. The balancing ratio is intended to adjust required delivery volumes to reflect system conditions. The ratio is also meant to adjust an individual capacity asset s obligation volume in a delivery period to its pro rata share of the total capacity market obligation volume need during the delivery assessment period. Delivery volume definition for guaranteed load reduction assets Delivery of guaranteed load reduction capacity assets will be measured as the actual consumption of electricity during a delivery assessment period as compared to an hourly delivery Page 5 of 13

83 baseline consumption at a business as usual load level (i.e., what the asset would have been consuming had the delivery assessment period not occurred). The delivery baseline is calculated as a standard day baseline multiplied by an in-day adjustment factor, as described below. This standard day baseline is meant to capture ongoing asset consumption, as well as mitigate the incentive for loads to inflate consumption for a short period prior to delivery assessment in order to artificially increase potential load reductions and obtain over-delivery payments. The standard day baseline was selected because: it is accurate for a variety types of loads both above and below one MW of load reduction; it produces the narrowest distribution of errors; and this method produces very low load-impact error during the most common delivery events. The standard day baseline methodology is modelled after the 10 of 10 baseline analyzed and supported by the Federal Energy Regulatory Commission. 2 Under the AESO s standard day baseline, each corresponding hour during the past 10 similar days prior to a delivery assessment period is averaged to establish an hourly average baseline for those 10 days. It is generally accepted that a period of approximately 10 non-holiday weekdays represents normal operation consumption for a load. This baseline is short enough to capture any near-term consumption trends and long enough to limit opportunities for manipulation. For weekends and holidays the standard day baseline will be set based on consumption during the most recent 5 weekends or holidays, as applicable. The standard day baseline for weekends and holidays is shorter than for non-holiday weekdays because delivery assessment events generally occur during weekdays and therefore the AESO does not require as many samples to establish the normal weekend/holiday consumption level. The baseline is meant to reflect the normal consumption level or the consumption in the absence of a delivery event or a dispatch. For clarity, the standard day baseline calculation will exclude days: where the asset received dispatch instruction for an amount greater than 0 MW; a day with a delivery assessment period; days in which the load has undergone an outage, either forced or planned; and/or days where load was triggered and tripped for the provision of LSSi. Excluding these days helps estimate the normal consumption level of the load; averaging in hours when consumption has been reduced will unfairly penalize the delivery of a guaranteed load reduction asset. For example, in order to calculate the standard day baseline from the days in Table 1 below, the following days are excluded because they are not representative of the resources normal operations: April 14,15,21 and 22 (in gray) because they are weekend days; and April 16 and 18 (in orange) because they include a delivery event. 2 Measurement and Verification for Demand Response, February 2013, Page 6 of 13

84 Table 1 Consumption data for an example standard day baseline calculation Date/Day 1-2 p.m 2-3 p.m 3-4 p.m 4-5 p.m 5-6 p.m 6-7 p.m 7-8 p.m 11-Apr Wednesday Day Apr Thursday Day Apr Friday Day Apr Saturday Weekend Apr Sunday Weekend Apr Monday Event Day Apr Tuesday Day Apr Wednesday Event Day Apr Thursday Day Apr Friday Day Apr Saturday Weekend Apr Sunday Weekend Apr Monday Day Apr Tuesday Day Apr Wednesday Day Apr Thursday Day Resource Baseline Continuing the example of guaranteed load reduction delivery assessment based on consumption data from Table 1 above: An EEA event occurred on April 27 between 5 p.m. to 6 p.m. The AESO will measure the consumption of the load for the duration of the delivery assessment period (the assumption is that the load has reduced its consumption). The AESO will calculate the delta between the consumption during the actual delivery assessment period and the previous 10 day average load consumption for 5 p.m to 6 p.m. (April 11-13,17,19-20,23-26). The average consumption of the load during the last 10 weekdays (5 p.m. to 6 p.m.) is 18 MW. If the load consumed 10 MW during the hour when the EEA occurred, the AESO will credit the asset with eight MW of reduction volume. The AESO will then compare the eight MW reduction to the obligation volume (adjusted for the balancing ratio) of the load and determine if the load has met its delivery obligation. If the asset has an obligation of eight MW or lower it will be deemed fully delivering. If the asset has reduced consumption in the hours prior to the delivery assessment period, due to energy dispatches, and is already consuming at 10 MW at the time of the delivery assessment the load will be deemed fully delivering to its capacity obligation. In-day adjustments are included to align the standard day baseline calculated from recent nonevent days with the conditions of the delivery event day to establish a more accurate consumption level for the guaranteed load reduction asset. The adjustment shifts or scales the standard day baseline by a fixed amount so that it matches the actual load during a period before the event start. This adjustment can help correct for load changes due to weather, as well as for variable operations. Page 7 of 13

85 The adjustment factor will be limited to +/- 20% of the standard baseline. For greater clarity, the in-day adjustment factor will be rounded either up or down if calculated as being less than 0.8 or greater than 1.2 respectively. Analysis performed in other jurisdictions indicates that a same-day uncapped multiplicative adjustment increased accuracy of the baseline over the capped multiplicative adjustment (20% caps). However, an uncapped multiplicative adjustment does have somewhat greater susceptibility to gross inaccuracies under certain demand conditions. 3 In the example below in Table 2, the load may have been consuming at a higher rate, than it did in the previous 10 non-holiday weekdays, due to weather conditions just prior to the delivery event taking place. In this case, the in-day adjustment will shift the standard baseline up reflecting the higher consumption during the delivery event day. Table 2 In-day adjustment factor example First 3 of 4 hours prior to delivery assessment period start time Delivery assessment period 9-10 a.m a.m p.m Average Load 1-2 p.m 2-3 p.m 3-4 p.m 4-5 p.m Standard Day Baseline Day-of delivery event consumption X X X X Adjustment Factor Adjusted Baseline Adjustment Factor Avg load day-of /Avg load baseline 1-2 p.m 2-3 p.m 3-4 p.m 4-5 p.m Calculation 16.95/15 = *1.13 *1.13 *1.13 *1.13 adjusted standard day baseline *Load consumption in MW Delivery volume definition for firm consumption level assets For firm consumption level assets, the delivery volume during a delivery event will be measured as the load s consumption during the event which will need to be equal or less than the qualified baseline, as described in subsection of Section 3, Calculation of Unforced Capacity (UCAP), minus the capacity obligation multiplied by the balancing ratio. Measuring delivery using the qualified baseline ensures that the load has continued to operate at the level at which it qualified. A reduction in consumption of the obligation volume from this level ensures that the capacity has been delivered. Difference between guaranteed load reduction and firm consumption levels baseline determination The qualified baseline for firm consumption level assets is fundamentally different from the delivery baseline used to measure if the guaranteed load reduction asset has delivered to its capacity obligation. The delivery baseline for guaranteed load reduction is generally intended to track the daily and seasonal consumption patterns of the underlying load. The standard day baseline and the use of an in-day adjustment are used to capture the most recent, normal consumption pattern of the load. The use of the 10 non-holiday weekdays is best suited for such purpose as jurisdictional overview and load studies indicate that such a sample size (10 days) is appropriate to establish the normal consumption pattern of the load. The five weekend & holiday baseline is shorter as less delivery assessment periods generally fall during off-peak hours, while the five days still 3 Development of Demand Response Mechanism/Baseline Consumption Methodology Phase 2 Results Final Report/ Page 8 of 13

86 provide large enough sample size to establish the normal consumption level of the load in offpeak periods. The qualified baseline for firm consumption level assets is fixed and does not vary over time. It is meant to capture the top range that a firm consumption level asset may qualify to sell into the capacity auction and helps establish the upper range of the UCAP calculation for the asset. The qualified baseline is intended to capture the average annual consumption of the load providing the reduction. It is based on the asset s consumption during the 250 tight supply cushion hours of the previous year. Tight supply cushion hours may fall in various months and therefore the qualified baseline does not record the loads most recent consumption patterns. In order to recognize if reduction in consumption may have taken place during tight supply cushion hours the load pattern is re-estimated. For the re-estimation of the load consumption, the longer 15 weekday and 10 holiday & weekend timeframes are used to allow the AESO larger sample sets to select data from as a tight supply cushion hours may cluster in certain periods. The 45 day baseline cutoff for firm consumption level assets is meant to capture longer consumption patterns as days with tight supply cushion hours may cluster. The 35 day baseline cutoff for guaranteed load reduction assets is intended to prevent the phenomenon called a static baseline where an asset maintains a baseline which over-represents their consumption level due to higher consumption patterns in summer vs winter or vice versa, measuring performance from such baseline would be inaccurate as it would lead to phantom reductions. Demand response outage considerations A demand response asset that is on a forced or planned outage will not be exempt from availability or delivery payment adjustments. While a load outage reduces the risk of an EEA event to the system, exempting demand response asset outages is not aligned with the capacity product that the AESO has procured in the market. This approach is consistent with: The treatment of generation facilities that are not exempt from availability or delivery payment adjustments when they are undergoing forced or planned outage. How other jurisdictions treat demand response asset s performance obligations during the time when they are offline for planned or forced outages. Reducing the perverse incentive that may arise if a demand response asset is considered meeting its capacity obligation during assessment periods when the asset is offline for a planned or forced outage. In the long term this may incent load to be unavailable to provide the reduction to the AESO. Demand response LSSi Considerations A demand response asset may participate in the Load Shed Services for imports (LSSi) program. A demand response asset that is providing LSSi mirrors a generator that is providing operating reserves. A load that is providing LSSi and is continuing to consume is like a generator that has reduced production to provide operating reserves, and will be recognized as such during a delivery event. The provision of LSSi enables the increase of imports which adds to Alberta s supply adequacy. Non-delivery payment adjustment Non-delivery payment adjustments will be set based on the obligation price per MW, which would link the payment adjustment rate to the capacity asset s maximum available revenues from the capacity market. The obligation price per MW will be reset every obligation period, and the payment adjustment level will be adjusted accordingly. Tying the payment adjustment rate to the capacity asset-specific auction clearing prices - i.e., obligation price per MW - ensures that the payment adjustment level is consistent with the each capacity asset's maximum revenue from the capacity market. This approach most accurately Page 9 of 13

87 reflects the amount of capacity revenues available for each capacity asset that cleared in any of the auctions corresponding to a particular obligation period. Therefore, setting a penalty based on asset-specific clearing price will not lead to disproportionally high penalties in relation to total capacity revenues in the auction rounds when the rebalancing auction is cleared at a lower price than the forward capacity auction. As the penalty is not based on the maximum of rebalancing and forward auction prices, this is not discriminatory against assets that received their obligation in the auction which cleared at a lower price. Overall, this is expected to reduce risk exposure and provide more revenue certainty because the payment adjustment rate is directly linked to the amount of revenue received from the capacity market by each asset. The factor of 60% in the non-delivery payment adjustment rate formula is an allocation factor, representing the amount of the total expected payment adjustment a non-delivering unit will incur through the delivery payment adjustments. The AESO s choice of a 60% allocation factor reflects higher importance of delivery periods compared to availability periods. System reliability is most at risk during delivery periods and as such the payment assessments places a greater weighting on delivery events. The factor of 1.3 scales the total payment adjustment level up above the capacity auction price. A value greater than one ensures that capacity assets failing to deliver are exposed to a net payment adjustment, after accounting for capacity revenues they will receive. A value larger than one also discourages speculative capacity sales because by committing to a capacity obligation the capacity asset is at risk of losing more through poor availability and delivery than through what might be earned through capacity payments. The value is believed to be of a magnitude that is sufficient enough for capacity assets to retain the incentive to deliver on capacity commitments, but will not be so large that new entrants will be discouraged from participating. Normalizing by the expected EEA hours ensures that on average, the total non-delivery payment adjustment for a non-delivering asset will be 1.3 times the relevant capacity price. Due to variability in system conditions, the number of EEA hours during which delivery payment adjustments are assessed will vary from year to year. Since the payment adjustment rate is based on the expected number of hours, it will not vary as much from year to year as the actual number of EEA hours. The specific value of expected EEA hours will be revised for each obligation period based on reliability modelling. The resource adequacy model will define EEA1 and EEA2 events as the activation and utilization of contingency reserves. This is consistent with the current EEA2 procedure that operating reserves will be used to supply energy requirements. The model will then measure the average amount of hours that supplemental reserves and spinning reserves are dispatched over the number of iterations that are run to evaluate asset adequacy. The model will shed firm load once contingency reserves are depleted but regulating reserves will be maintained during load shed events. Ancillary services in the model are reported as a percent of gross load. Table 3 Ancillary services as a percentage of gross load Ancillary Service Type AESO Supplemental Reserves Requirement 2.5% Regulation Up Requirement 1.5% Spinning Reserves Requirement 2.5% The AESO will determine and communicate to market participants the specific value of expected EEA hours in advance of each base auction using the AESO's reliability modelling. The hour count will be determined at the demand curve inflection point, which is appropriate because it is Page 10 of 13

88 the point on the demand curve that is closest to the expected average outcome over time. This value will remain constant for that obligation period. This will inform market participants decisions in the auction bidding process. Additionally, if the expected EEA hours based on the reliability modelling is lower than 20, a floor of 20 hours will be used, which will add increased predictability to the non-delivery payment adjustment rate value from auction to auction. Over-delivery payment adjustment As described above, the over-delivering assets with capacity obligations will be eligible to receive payment adjustments funded from the collected non-delivery payment adjustments. Over-delivery payment adjustments are additive to the energy and ancillary services prices, creating incentives to deliver energy and capacity during shortage events. Over-delivery payment adjustments will allow assets to recover from non-delivery payment adjustments through over-delivery during future events. Over-delivery payment adjustments will be made for each MWh of over-delivery during EEA events, and will be paid at the $/MWh over-delivery payment adjustment rate. In the event when there are residual funds remaining or when there were no assets eligible for over-delivery payment adjustments, the collected funds will be directed to reduce total capacity charges to consumers. The rationale for doing so is that if capacity assets with capacity obligations do not deliver, the consumers pay less for the service that has been underprovided. Maximum amounts for unavailability and non-delivery payment adjustments Capacity assets will be subject to annual and monthly caps on payment adjustment exposure from the combination of availability and delivery assessments. The payment adjustment caps are necessary to protect participants from excessively high risk of participating in the capacity market by keeping payment adjustment exposure in line with revenues. This helps maintain the investment attractiveness of the Alberta market. Total payment adjustment exposure will be capped in two ways: 1. Annual cumulative unavailability and non-delivery payment adjustment cap at 130% of the annual capacity revenue based on the obligation price per MW. A poor delivering asset, or one that did not show up for the obligation year, would potentially have adjustments of up to 130% of annual revenue. This is also meant to dissuade speculative capacity market entrants that do not intend to materialize. Even with an annual cap on payment adjustments, the AESO anticipates that a profit maximizing firm will always be incented to avoid penalties and deliver energy during delivery and availability events. Even for firms that may own a portfolio of assets and may be able to easily engage in asset substitution, these firms will always be incented to maximize their revenues, which requires maximizing asset performance. Further, an asset that reached the maximum over-delivery payment in one year will still be eligible for energy revenues during delivery events. The energy and ancillary service markets will continue to provide incentive for these assets to provide energy during tight system conditions. By comparison, Great Britain and ISONE limit annual payment adjustments to 100% of a capacity market participant s annual capacity market payments; PJM caps payment adjustments to 1.5x net CONE. 2. Monthly non-delivery payment adjustment cap at 300% of the monthly capacity revenue based on the obligation price per MW. This monthly cap is not set to 100% of monthly revenue, because in a situation when a long-term delivery period or multiple delivery periods take place in a single month, a 100% monthly revenue cap could exempt a non-delivering capacity asset from the payment adjustment amounts, reducing incentives to deliver as expected. A 300% monthly cap limits monthly non-delivery penalties while still allowing the delivery assessment framework to achieve total annual payment adjustments of up to 1.3x annual capacity market revenue for the poorest performing assets. Multiple delivery and Page 11 of 13

89 availability payment structures were shared with the working groups through Q1 and Q that demonstrated the effectiveness of the payment adjustment structure. Maximum amounts for over-availability and over-delivery payment adjustments Maximum potential over-availability and over-delivery payment adjustments will be capped at a capacity asset s total annual capacity payment. This is implemented to mitigate potential excessive over-delivery and over-availability payments in the situations when the number of overdelivering assets is significantly smaller than the number of under-delivering assets (e.g., extreme case of one over-delivering asset and multiple under-delivering assets) would result in eligible payments potentially exceeding annual capacity revenue on small volumes of over-delivered capacity. 8.3 Ex ante asset substitution and volume reallocation The Comprehensive Market Design supports ex ante asset substitution and ex post volume reallocation. Ex ante asset substitution Asset substitution allows a capacity supplier to assign the delivery assessments to another qualified capacity asset as a tool to manage delivery risk while maintaining overall system reliability objectives. A capacity supplier may engage in asset substitution with a qualified but non-committed or partially committed capacity asset. The proposed ex ante asset substitution approach is modelled on the existing AESO approach found in the ancillary services market for operating reserve, as well as other capacity markets. Financial arrangements between counterparties will be outside the AESO s purview. The AESO will allocate the payment adjustments associated with under-delivery and over-delivery of the substitute asset to the original obligation holder and not the substitute asset. This will simplify settlement, should not impact credit requirements, and will allow counterparties to work out the terms of their agreement independently. Asset substitution will not transfer the obligation from one customer to another, but rather transfer the delivery assessment between qualified assets. Ex post volume reallocation Volume reallocation represents another way to mitigate the risk of non-delivery payment adjustment. The ex post volume reallocation transaction allows the buyer to meet its obligation via a combination of its own delivery and that acquired from other capacity providers. This provides an additional option for payment adjustment management and flexibility. In contrast to ex ante asset substitution, only capacity committed assets will be allowed to participate in volume reallocation. Primarily, this is because the AESO doesn t want to reduce the incentive for capacity assets to ensure they are able to deliver on capacity obligations during delivery assessment periods. Also, by allowing non-committed capacity assets into the ex post volume reallocation process there may be incentive for some capacity suppliers to withhold capacity from the obligation period with the expectation that they can receive a capacity payment through the ex post process. The AESO wants to ensure that all capacity assets participate competitively in the base and rebalancing auctions and that assets with capacity obligations have effective incentives to provide capacity when required by the system. Ex post volume reallocation is being implemented to provide an additional way for capacity committed assets to manage non-delivery risk exposure. Volume reallocation lowers financial risk Page 12 of 13

90 for capacity committed assets as it provides an additional way to manage cost incurred because of non-delivery to both participants with portfolios and smaller participants. Providing an additional way to balance the financial risk may lower the capacity assets' offers in the capacity auction, decreasing the cost of capacity to consumers. Supply obligations and performance assessment vis-a-vis the capacity market criteria The capacity market can achieve desired reliability objectives by creating a real and measurable supply adequacy product in which to assess whether capacity assets met their capacity market obligation and incent providers to live up to their obligation. The incentives are designed in such a way that a wide variety of technologies should be able to compete to provide capacity while ensuring a fair, efficient and openly competitive (FEOC) market. Costs to consumers are minimized by creating a product for which value can be demonstrated via delivery. The capacity market incentive mechanisms, outcomes and relevant data are also transparent. Leveraging best practices and lessons learned from other capacity market implementations to inform the payment adjustment framework is expected to maintain investor confidence and trigger sufficient private investment. Page 13 of 13

91 Settlements and Financial Security Requirements Rationale 9.1 Capacity market statements The AESO will issue monthly statements for settlement of capacity payments to capacity market participants based on the following rationale: (a) The capacity market should operate on a monthly billing cycle to align with the energy market. Aligning the settlement period of the two markets will reduce the administrative requirements by leveraging existing processes and will align the timing of common settlement activities across markets. (b) Statements across the capacity and energy markets will be kept separate to simplify the implementation. The AESO may consider consolidating statements in the future The information on the statement will roll up the settlement results into summary line items to better understand the statement. Details behind the line items will be available through settlement reports supporting the statement. 9.2 Settlements applicable to capacity assets Capacity payments A capacity market participant will receive capacity payments for their cleared obligation volume during the obligation period. This is consistent with other capacity markets and Alberta s current energy market. In addition to the capacity payment, a capacity market participant may receive a payment adjustment as described in Section 8, Supply Obligations and Performance Assessments. Until a new asset has reached commercial operation it may accrue payment adjustments based on performance assessments for availability and/or delivery. The AESO considered and rejected an alternative option to not pay or penalize a new asset for the months of the obligation period prior to reaching commercial operation. To be consistent with existing capacity assets, which are subject to performance assessments for the entire obligation period, new capacity assets should not get a performance exemption in the obligation period. Calculating capacity payments The capacity payment needs to include the change in obligation from the base auction through to the associated rebalancing auctions. The formula to calculate capacity payment provided in the Final CMD ensures that all changes in obligations are incorporated in the capacity payment. Below is a settlement example with payment adjustments. This example does not apply the payment adjustment caps. The example below is of a capacity asset that reduces its obligation prior to the obligation period by buying back a portion in each associated rebalancing auction: Capacity Payment = { [ O b * P b ] - [ (O b O r1 ) * P r1 ] - [ (O r1 O r2 ) * P r2 ] } / # months in obligation period Page 1 of 8

92 Table 1 Capacity payment example For a 12 Month Obligation Period Obligation (O) in MW Price (P) in $K Base Auction (b) st Rebalancing Auction (r1) nd Rebalancing Auction (r2) Annual Capacity Payment = {[80*200] [(80-30)*150] [(30-10)*400]} Annual Capacity Payment = $500,000 for a 10 MW final obligation 9.3 Calculating capacity payment adjustments The rationale for payment adjustments for unavailability, over-availability, non-delivery and overdelivery is provided in Section 8, Supply Obligations and Performance Assessments. The below is a continuation from the example above in subsection 9.2.2, where payment adjustments to the capacity payment are applied. Obligation price per MW The obligation price per MW represents the price that the capacity market participant will be paid for each MW. This price is calculated on a per asset basis Obligation Price per MW = Capacity Payment ($) / Obligation (MW) Obligation Price per MW = 500,000/10 = $50,000/MW Payment adjustment for availability Please note in the example below that division by the number of small supply cushion hours has been removed to simplify the example. However, the concept and ultimate outcome remain unchanged. Unavailability Payment Adjustment Rate ($/MW-Year) = 40% * 1.3 * Obligation Price per MW Unavailability Payment Adjustment Rate ($/MW-Year) = 0.4 * 1.3 * 50,000= $26,000/MW-Year Obligation Volume Table 2 Annual unavailability payment adjustment example Actual Availability Volume* Availability Assessment Volume* Unavailability Payment Adjustment Rate Annual Unavailability Payment Adjustment ,000 26,000*(-3) = -78,000 * determined at the end of the obligation period as described in Section 8, Supply Obligations and Performance Assessments. Capacity Payment = 500,000/12 = $41,666 (per month) Annual Unavailability Payment Adjustment = $78,000 assessed at the end of the obligation period as shown in Table 2. Page 2 of 8

93 In this example, the AESO will apply the unavailability payment adjustment amount to the last settlement period of the obligation period to reduce the capacity payment to zero ($41,666 $41,666, = $0). The $41,666 of the $78,000 payment adjustment will be made available for overperformers. The remaining $36,334 will be debited to the asset s payment adjustment balance (PAB) The over-availability payment adjustment is calculated after all asset settlements are completed in accordance with Section 8, Supply Obligations and Performance Assessments. The following is a simplified example of how the over-availability adjustment will be determined and applied: Over-availability Payment Adjustment Rate ($/MW-Year) = Total Unavailability Payment Adjustments Collected in an Obligation Period ($) / Total Over-availability Volume (MW) If we assume only 1 under-performer owed $78,000 and assume only 2 over-performers totaling 3 MW: Annual over-availability payment adjustment rate = $78,000 / 3 MW = $26,000/MW-Year. Asset Name Table 3 Annual over-availability payment adjustment example Obligation Volume Actual Availability Volume* Availability Assessment Volume* Over-availability Payment Adjustment Rate Annual Overavailability Payment Adjustment A ,000 52,000 B ,000 26,000 $78,000 * determined at the end of the obligation period as described in Section 8, Supply Obligations and Performance Assessments. For over performing asset A the annual over-availability payment adjustment entitled = 2 MW * $26,000/MW-Year = $52,000. Asset A is entitled to this $52,000, but may have to collect this over multiple months if the payment adjustments collected from other assets do not provide enough to cover the entire $52,000 in a single month. If the amount cannot be paid in one month, then the amount still owed to Asset A will reside in Asset A s PAB. The amount used to pay assets that have a positive PAB will come from the payment adjustments collected from assets that have a negative PAB. Payments will be distributed prorata to those with a positive PAB. Payment adjustment for delivery The following example illustrates how a capacity asset was assessed a delivery payment adjustment for an EEA (delivery) event that spanned 4 settlement intervals (22:23 to 01:05). In accordance with Section 8, Supply Obligations and Performance Assessments the following formula is used to calculate the delivery volume and non-delivery payment adjustment rate: Delivery Volume (MWh) = Actual Delivery - (Obligation Value * Balancing Ratio) The balancing ratio is the ratio of energy and reserves produced by capacity assets during a delivery assessment period to the total committed capacity in that obligation period. The balancing ratio assumed for this example is = 11,700 MW / 13,000 MW = 0.9 Non-delivery Payment Adjustment Rate ($/MWh) = 60% * 1.3 * Obligation Price per MW/max (Expected EEA hours, 20) Page 3 of 8

94 Expected EEA hours will be determined through the AESO s reliability modelling done in advance of the base auction (three years before the obligation period). For this example we will use 13 hours, and the obligation price per MW of 50,000 from the availability example above. Non-delivery payment adjustment rate (NDPAR) = 0.6 * 1.3 * 50,000/ 20 = $1,950/MWh In the event the EEA event straddles a settlement interval or occurs completely within a settlement interval, the determination of the obligation becomes more complex due to the fact that the actual delivery volumes are hourly and the EEA events may be less than an hour. To best approximate the amount of energy delivered during the EEA event, the following methodology will be used for capacity assets that are dispatched in the energy market and UCAP is based on availability factor. The figure below shows: an EEA event that ends 5 minutes into a settlement interval; dispatch level during the EEA period was 11 MW; after the EEA period ended the unit was dispatched down to 1 MW based on energy merit order dispatch; the obligation for this asset was 10 MW and its ramp rate is 5 MW/Min; and the actual delivery was recorded at 3 MWh for the entire settlement interval. Figure 1 Qualitative representation of dispatch instructed levels over a settlement interval Page 4 of 8

95 To calculate the energy provided during the EEA event the AESO will apply the following calculations: EEA obligated energy represents the amount of energy expected to be delivered during a settlement interval segment of an EEA event and is calculated as follows: EEA Obligated Energy = Obligation volume * minutes of the event in the settlement interval / minutes in the settlement interval EEA Obligated Energy = 10MW * 5 min/60 min = 50/60 MWh or MWh To calculate the energy delivered for an EEA event that occurs over part of a settlement interval (EEA MWh), subtract the actual delivery from the energy produced outside the EEA event. EEA MWh = Actual Delivery Outside Energy To calculate the outside energy, add the dispatch segments that occur outside the EEA event in the settlement interval, shown as the brown area and grey area in Figure 1. Energy outside of the EEA event is the sum of the area outside of the EEA event. Energy outside of the EEA event = 12/60MWh + 53/60MWh = 65/60MWh Therefore the EEA MWh is equal to 3 MWh 65/60MWh = 115/60MWh or 1.92MWh The table below shows the delivery calculation for the entire EEA event. Once each settlement interval is calculated the asset was found to under-deliver in hours ending 23 and 24. The asset was found to over-deliver in hours ending 1 and 2. The over-delivery adjustment is explained later in this section. Table 4 Example EEA event from 22:23 on March 8 to 01:05 on March 9 EEA Event Time Mins Actual Delivery EEA MWh Original Obligation EEA Obligated Energy BR Adjusted Obligation Delivery Adjustment Energy NDPAR Hourly Delivery Payment Adjustment 22:23 to 23: * 0.9 = = $ *1950 = :00 to 00: * 0.8 = = -3 $1950-3*1950 = :00 to 01: * 0.8 = = 3 $1950 0*1950 = 0 01:00 to 01: * 0.9 = = 1.18 $1950 0*1950 = 0 total $-13,533 Note: the balancing ratio can change from hour to hour and in the above table fluctuates from 0.8 to 0.9. The total non-delivery payment adjustment for the entire event equals ($13,533). This nondelivery payment adjustment will reduce the monthly capacity payment. If there is a remaining balance owing for the non-delivery payment adjustment; it will be added to the outstanding payment adjustment balance. Page 5 of 8

96 9.3.7 In the example shown in Table 5, hours ending 1 and 2 are eligible for an over-delivery payment adjustment. The over-delivery payment adjustment is calculated after all asset settlements are completed as described in Section 8, Supply Obligations and Performance Assessments. EEA Event Time Over-delivery payment adjustments are revenue neutral and will never exceed the under-delivery payment adjustments collected for a settlement period. Over-delivery Payment Adjustment Rate ($/MWh) = Total Non-delivery Payment Adjustments Collected $ / Total positive Delivery Volume MWh Assume for the example over-delivery payment adjustments of $600/MWh and $500/MWh as shown in Table 5. Mins Actual Delivery Table 5 Over-delivery payment adjustment example EEA MWh Original Obligation EEA Obligated Energy BR Adjusted Obligation Delivery Adjustment Energy ODPAR ODPA Owed 00:00 to 01: * 0.8 = = 3 $600 3 * $600 = $ :00 to 01: * 0.9 = = 1.18 $ * $500 = $590 total $2390 The overall settlement for the month for this example asset is a net of the capacity payment and the payment adjustments as shown in Figure 2 below. In that month the asset is entitled to a capacity payment of $41,667 but receives a $0 payment for capacity and debits the asset s PAB by $47,436. Figure 2 Illustrative example of capacity net settlement The over-delivery payment adjustments paid to over-performers will depend on the amount of monthly reduction of the overall capacity market payment adjustment balance and balances owing to over performers may be paid over multiple months. Page 6 of 8

97 For under and over-delivery for capacity factor resources: Dispatch levels and ramp rates are not applicable; As with availability factor resources, contingency reserve dispatch volumes and regulating rise range will be added to the actual delivery; and The AESO will prorate the difference between the actual delivery and balancing ratio adjusted obligation by the minutes of EEA event in the settlement period. 9.4 Capacity cost allocation settlements The AESO s application for approval of a cost allocation methodology will be filed with the Commission for review, approval, and implementation before the beginning of the first capacity market obligation period. Details of the rate design will be developed, including stakeholder consultation, prior to filing The timing of cost allocation settlement will align with capacity market settlement. Further details will evolve through the stakeholder consultation to be held in late Net settlement instructions (NSI) Net settlement instructions will not apply to the capacity market. Buying back obligation volumes in rebalancing auctions and asset substitution and volume reallocation are tools capacity market participants can utilize to facilitate the management of capacity resource obligation risk. NSI works in the energy market because the price paid for a MW of energy is equal to the price consumers will pay for a MW of energy in the same time period. Given the current thinking on cost allocation in the capacity market, this will not be the same for the capacity market. A volumebased NSI approach no longer works because the price paid for capacity no longer equals the price paid by consumers in that same time period. Facilitating NSIs will cause a discrepancy between the amount paid to capacity providers and the amount collected from capacity consumers. This does not eliminate the ability for counterparties to enter into independent financial hedges with each other; however, these will not be registered with the AESO and accounted for in capacity market settlement. 9.6 Financial security requirements for capacity assets The AESO will leverage the existing forms of allowable security established for the energy market for capacity market participants. The existing forms of security are appropriate, reliable and established within industry The assessment of availability is conducted at the end of the delivery period and looks back at the entire obligation period. To minimize the financial security risk, the AESO settlement will only claw back up to 100% of the capacity market payment on any one monthly statement until the balance of the (availability or delivery) payment adjustment is paid. If the capacity market participants asset does not continue to have an obligation in the next obligation period the AESO may require security against the estimated outstanding payment adjustment balance. In this case there is no offsetting revenue stream owed to the asset so default risk (consequence to consumers) increases. 9.7 Measurement, verification and tracking of capacity resources The capacity market will use metering data, ancillary service data, LSSI, transmission constraint data and available capability (if applicable), for the purposes of capacity settlement, Capacity will be measured based on historical observed availability factor or capacity factor in the obligation period depending on the type of capacity resource being settled In order to perform the settlement calculations and monitor rule compliance, the established metering and SCADA practices used in the energy market will be used in the capacity market. Page 7 of 8

98 Alignment with criteria The CMD should provide mechanisms for consumers to hedge the cost of capacity if and where appropriate. As described above, it was determined that facilitating capacity market NSI was not an appropriate tool for hedging the costs of capacity. Financial hedges may still be developed by market participants. Settlement design ensures the capacity market is compatible with other components of the existing electricity framework, such as load settlement and retail customer choice, and should be robust and adaptable to different government policy initiatives related to the electricity sector. Page 8 of 8

99 Roadmap for Changes in the Energy and Ancillary Services Markets Rationale Overview - Roadmap for changes in the energy and ancillary services markets Changes to the energy and ancillary services (EAS) markets are required to facilitate the implementation of the capacity market, monitor and mitigate market power in the energy market, respond to the changing Alberta generation fleet (i.e., increased variable generation, coal retirements, etc.) and improve market efficiency. The timelines for implementation of these changes are still being evaluated, but some of the changes have more defined implementation timelines driven by studied system needs. The majority of the proposed changes that will be implemented by 2021 or earlier are required for the capacity market. Any proposed changes that are required after 2021 have been outlined in subsection 10.8, Roadmap reforms in the EAS markets. Subsection 10.9, Out of scope reforms in the EAS markets, summarizes design elements related to the EAS markets that have been evaluated as part of the capacity market design but that have been taken out of scope. While some of these design elements may be linked to a future market or system operation trigger or a new business case, these elements are not required for either the implementation of the capacity market or the evolution of Alberta s generation fleet at this time Overview of the EAS markets After an evaluation a number of alternative approaches, many of the elements in the EAS markets were determined to continue to work in Alberta. Most alternatives were rejected as having limited value for Alberta at this time and are addressed in subsections 10.8 and The following aspects of the EAS markets continue to be appropriate for Alberta (a) Dispatching using the merit order. Energy market dispatches will continue to be based upon price-volume blocks comprised of offers and bids from available pool assets. Stacking competitively priced offers will achieve orderly and economic dispatch of available capability by: i. enabling dispatch of lower priced resources before utilizing higher priced resources; and ii. producing prices that reflect cost of production or opportunity cost and, in turn, producing efficient price signals for both generators and loads. The price signals in the energy and ancillary services markets will serve to provide incentives for flexible operational behavior (for dispatch and operational reliability needs), and incent an economically efficient level of investment in flexible resources. (b) Dispatching the ancillary services market as a separate market. Dispatching the ancillary services and energy markets separately continues to work in Alberta because the market clearly identifies separate products and pricing for capacity needed for system security. An evaluation on co-optimization of the markets indicated limited value at this time (discussed further in subsection (e)). (c) Self-commitment. The current self-commitment model continues to work in Alberta as participants are best able to manage their physical asset conditions and needs, and manage the risk and reward associated with their dispatch submissions. Additionally, directives due to Page 1 of 35

100 reliability reasons are rarely issued to assets, indicating that security management is not a pressing issue. This means that assets have been able to respond to market signals to selfposition their assets in the self-commitment model. The self-commitment model leaves the risks of operational decisions with the legal owners instead of with the AESO on behalf of loads. Additionally, the AESO s analysis of a centralized commitment model (discussed further in subsections (b) and (c)) indicated limited value compared to anticipated costs for moving to a centralized commitment model at this time. (d) Single part offers and bids. Single part offers can continue in the Alberta energy market as long as any mitigation proposal, discussed in subsection 10.7 below, recognizes that the single part offer must be able to include costs that would normally be included in a three part bid (i.e., start up, minimum run, cycling and other operating costs such as carbon costs). (e) Pricing methodology. Changes to the pricing methodology, including pricing at the margin, setting system marginal price, and averaging 60 system marginal prices to get pool price are not required to implement the capacity market at this time. The AESO considered a number of design aspects related to pricing, all which show limited value at this time. For further discussion, see subsections (a), (b) and (c), and (b) below Obligations in the EAS markets for a generating unit, aggregated generating facility, and energy storage facility Volume obligations A generation unit, aggregated generating facility, or energy storage facility, regardless of whether or not it has a capacity commitment, continues to have a must-offer obligation into the energy market. Pool assets with capacity commitments, by receiving payments for capacity, have taken on obligations to offer their assets into the energy or ancillary services markets. This approach is consistent with current energy market obligations that all available capacity must be offered into the market. The existing ISO rules contain must-offer requirements for the maximum capability (MC) of source assets (defined as generating units, aggregated generating facilities and imports). The AESO proposes to maintain those requirements for generating units, aggregated generating facilities and energy storage facilities with a capacity commitment. In contrast, other market jurisdictions typically use the concept of must-offer for the ICAP equivalent of the UCAP that a capacity asset supplies. However, the AESO has determined that it would be more efficient from an implementation perspective for Alberta to continue the must-offer requirement for the maximum capability of a source asset outlined in the existing ISO rules. Currently, the existing ISO rules require a generating unit or aggregated generating facility with a maximum capability of 5 MW or greater to offer into the energy market. In the AESO s view, it reasonable that this 5 MW threshold continues to be apply and those assets with a MC of 1 MW or greater and less than 5 MW will have the option to offer into the energy market, as this approach aligns with volumes identified in the existing dispatch tolerance requirements. For generating units, aggregated generating facilities and energy storage facilities that do not have a capacity commitment, the AESO considers that, in order to ensure system reliability, the AESO must continue to have visibility of the physical capability of these assets to assess supply adequacy in and near real time. It was determined that the existing ISO rules would continue whether an asset has a capacity commitment or not. All assets must offer in order to prevent physical withholding and to ensure that all capacity is in the merit order, which allows the AESO to properly respond to outages and issue directives. Page 2 of 35

101 Pricing obligations There is no proposed change from status quo pricing obligations. All offers must be between the offer floor ($0) and offer cap ($999.99) in the energy market. For the rationale for the continuation of these obligations, see subsection (a), Dispatching using the merit order, subsection 10.8, Roadmap reforms in the EAS markets, and subsection 10.9, Out of scope reforms in the EAS markets below. Any change to the offer floor, cap, shorter settlement or shortage pricing will be considered as part of the roadmap. The submission of offer control information is needed to calculate the market power screen. Cost data is needed to calculated asset-specific reference prices. Dispatch obligations The status quo dispatch obligations are proposed to remain the same. All units continue to selfcommit and are dispatched from the merit order. The AESO proposes to include an additional requirement to submit a rate table (or a ramp curve) that reflects the ramp rate at different MW levels of the asset. For the rationale for the continuation of these obligations, see subsection (a), Dispatching using the merit order, subsection 10.8, Roadmap reforms in the EAS markets, and subsection 10.9, Out of scope reforms in the EAS markets below. The use of a ramp table or ramp curve, whereby market participants would submit to the AESO a ramp rate table (or a ramp curve) that reflects the ramp rate at different MW levels of the asset, would represent ramp rate based on the operating state of the asset. The AESO considers that these operational ramp rates would provide better certainty to the AESO system controllers, and provide better information to system controllers than the single submitted ramp rate provided today. This better information does not require a change in dispatch behavior from market participants. Outage scheduling obligations There are no proposed changes to the current model for outage scheduling. A generating unit, aggregated generating facility or energy storage facility is required to submit outage information. While the AESO may cancel an outage, the AESO will not approve outages. The AESO anticipates that the current outage model, aligned with a capacity model, will continue to work appropriately. Given that the CMD proposal does not contemplate exemptions for performance measurement due to outage schedule, an examination into changes to the outage model is not required. The AESO already has existing ISO rules that pertain to outage scheduling of a generating unit, aggregated generating facility or energy storage facility. The AESO reviewed the options for outage scheduling for a generating unit, aggregated generating facility or energy storage facility and proposes to maintain the current outage model regardless of whether the asset has a capacity commitment or not. Further details on the alignment of the existing outage scheduling obligations with the timelines contemplated in the capacity market will be developed by the AESO and will be subject to further consultation. The existing ISO rules require outage reporting where a generating source asset with a maximum capability of 5 MW or greater changes its available capability by 5 MW of more. The AESO considers it reasonable that these requirements continue to apply to support the ongoing assessment of system adequacy. The AESO uses this outage information in order to assess and report Alberta s supply adequacy for both the short-term and the longer-term (next Page 3 of 35

102 two years). The AESO s assessment of supply adequacy is required to manage the reliability of the system. Supporting reasons for maintaining the current outage submission process includes: An asset has a financial incentive to plan outages to avoid stressed and higher priced times; An asset with a capacity commitment has a financial incentive to be available during system stressed periods in order to avoid performance based payment adjustments; The risks of taking a planned outage during system stressed period should reside with the market participant; therefore, comprehensive information is required in order to assess that risk by industry and the AESO; The capacity market design does not contemplate a performance exemption for an asset with a capacity commitment that takes an outage during a performance assessment period; Stakeholders expressed concerns in comments on SAM 2.0 that the requirement for AESO approving outages would pose challenges to the operation of assets and other processes; and The AESO will continue to have the ability to issue a directive to cancel a planned outage, including a mothball outage, and to direct the starting up of a long lead time asset through reliability unit commitment requirements Obligations in the EAS markets for a load or aggregated load asset Volume obligations A load or aggregated load with a capacity commitment has a must-offer obligation and must be able to be dispatched. A load with a capacity commitment can offer at the offer cap to be dispatched last in the event of an energy emergency alert (EEA). Based on the overarching principle that an asset with a capacity commitment has obligations to offer into the energy or ancillary services market, it is reasonable that a load asset with a capacity commitment must also offer into the energy or ancillary services market. This will enable the AESO to take these load assets into consideration when assessing supply adequacy. Other market jurisdictions take a similar approach. Additionally, it is reasonable that the same 5 MW threshold applied to a generating unit with a capacity commitment also apply to a load with a capacity commitment. A load or aggregated load facility with capacity commitment of 5 MW or greater must-offer into the energy or ancillary services markets. Those with capacity commitments 1 MW or greater and less than 5 MW will have the option to offer. A load asset that does not have a capacity commitment will not have the must-offer requirement, but will continue to have the option to offer into the energy or AS market if they are equal to or greater than 1 MW, in accordance with the existing ISO rules. For load or aggregated load with a capacity commitment, the AESO requires real time telemetry of the load consumption level. Pricing obligations A load or an aggregated load asset with a capacity commitment must offer its obligation volume between the market offer cap ($999.99) and offer floor ($0), unless it delists. Page 4 of 35

103 Offers from a load or an aggregated load asset with a capacity commitment must be incorporated into the energy market merit order (unless dispatched to provide ancillary services). As such, volumes offered must be associated with a price. The submission of offer control information is needed to calculate the market power screen for energy market power mitigation. Cost data from a load or an aggregated load asset is not required for energy market power mitigation (see subsections ). Dispatch obligations A load or an aggregated load asset with a capacity commitment must be dispatchable and its availability for dispatch must be accurately represented to the AESO in real-time. The current mechanisms that apply to generators for indicating availability (i.e., energy restatements for an acceptable operational reason) will be applied to a load or an aggregated load asset with a capacity commitment. This mechanism will ensure that the operating state and availability of dispatchable assets are accurately represented to the AESO in real-time. Outage scheduling obligations A load or aggregated load asset with a capacity commitment will be required to submit outage information similar to those requirements in Section of the ISO rules, Generating Outage Reporting and Coordination. While the AESO may cancel an outage, the AESO will not approve outages. The AESO anticipates that the current outage model aligned with a capacity model will continue to work appropriately. Given that the capacity market design does not contemplate exemptions for outages, an examination of changes to the outage model is not required. The AESO already has existing market rules that pertain to outage scheduling of load and generation. The current ISO rules require outage reporting where a generating unit or aggregated generating facility with a maximum capability of 5 MW or greater changes its available capability by 5 MW or more, and load reduces its available capability of 40 MW or more. A load or aggregated load asset with a capacity commitment greater than 5 MW will be required to follow outage reporting requirements similar to the generator outage reporting requirements. These additional requirements for a load or aggregated load asset with a capacity commitment are necessary to ensure that outage information is sent to the AESO to conduct assessments of short and long term supply adequacy, and to ensure that outages that impact the formation of the merit order are transparent by processing and publishing the outages in a timely manner Obligations in the EAS markets for an import asset Volume obligations Similar to other assets with capacity commitments, an import asset that has a capacity commitment must offer the obligation volume into the energy or ancillary services markets. Generators within the province can use their asset pricing strategy to ensure they are dispatched in the energy market when it is economic for them. Under the current requirements, an import can only offer into the energy market as a price taker and must be scheduled for an entire settlement period at minimum. This is not a level playing field given that an asset with a capacity commitment must offer its energy into the energy market. Providing the ability for imports to price their energy and dispatch intra-hour, similar to generators, levels the playing field between the two types of supply. The current ISO rules require imports to offer at $0 which would mean that all import offers are in-merit and will be dispatched and scheduled. As a result, it would mean that the obligation Page 5 of 35

104 volume will always be dispatched and scheduled even when it is uneconomical. Allowing imports to request and use priced assets addresses this issue and is aligned with all other energy market assets. An import priced asset may offer at a price between the market offer cap ($999.99) and offer floor ($0), similar to other pool assets, and are only dispatched and scheduled when they are in-merit. In order to implement this option, intra-hour scheduling of the intertie will be required. The AESO would accommodate the submission of an e-tag by an import pool participant following their receipt of a dispatch from the energy market and will approve e-tag submitted intra-hour corresponding to the dispatch. However, the import pool participant would be accountable for ensuring that balancing authorities and transmission providers along the transmission path into Alberta will approve the e-tag at any time during the hour. In addition, the AESO will continue to make available the option for import as price taker at $0 and be scheduled at an hourly interval. This option provides flexibility to import pool participants to either offer using priced assets or to continue to be price taker at $0. Pricing obligations To align obligations in the energy market across all assets with a capacity commitment, imports will be given the option to submit priced offers. The proposed must-offer requirement for an import with a capacity commitment is challenging to implement if an import asset is limited to offer as price-takers in the energy market. Under the current price-taker practice, a must offer requirement would effectively require import assets to schedule and flow energy at all times and receive the hourly pool price, at whatever value it clears at. Enabling an import with a capacity commitment to price its offers levels the playing field by treating imports in a similar manner as internal Alberta generators and loads with capacity commitments. This approach also supports economically efficient outcomes by enabling an import with a capacity commitment to use the offer pricing mechanism to reflect the cost-ofproduction from the import asset providing energy. The submission of offer control information is needed to calculate the market power screen for energy market power mitigation. Cost data from an import asset is not required for energy market power mitigation (see subsections ). Dispatch obligations ISO rules and dispatch processes will be revised to enable intra-hour dispatch of priced imports in the energy market in the same manner as priced offers in the energy market. An import asset is responsible for ensuring interchange transaction scheduling approvals from their source to Alberta will meet energy market dispatch requirements. To accommodate the priced offers for an import asset, new requirements are needed to dispatch imports at any time during the hour. The proposed dispatch protocol will be similar to the current dispatch of the energy market with the exception of specific details for scheduling, tagging and settlement. The intra hour scheduling practice will dispatch an intertie offer as part of the energy market merit order at any price instead of just $0. This provides a level playing field for all assets and an efficient dispatch protocol resulting in effective prices reflecting market conditions. The following scheduling example is based on BC Hydro Intra-hour scheduling business practice for worst case scenario which allows 5 minutes for transmission procurement and e-tag. Please note the example should be viewed as a worst case; it is preferable for the schedule ramp to start as close as possible to the dispatch time. Example: Pool participant receives a dispatch at XX:06. Pool Participant procures transmission (if required). Page 6 of 35

105 Pool Participant submits (or adjusts) e-tag prior to XX:25 for transaction start at XX:45 and a ramp duration of 10 minutes. Assuming the e-tag is approved by required entities, at XX:40 energy ramp starts. At XX:50 ramp is completed. Ramp starts 34 minutes after dispatch; schedule starts 39 minutes after dispatch. If, in the above example, an e-tag can be submitted by XX:10 (4 minutes after dispatch) for a XX:30; the ramp will start 19 minutes after dispatch and schedule starts 24 minutes after dispatch Obligations in the EAS markets for an export asset Volume obligations There is no proposed change from status quo for volume obligations for export assets; however, the option to offer using priced assets will be made available to an export asset. The AESO is proposing to apply the same pricing mechanisms to export assets in order to maintain a level playing field and fairness across imports and exports. Pricing obligations An export asset will have the option to continue with their current price taker asset or acquire a priced asset. Enabling the option for exports to price volumes supports a level playing field across different pool assets and is symmetrically consistent with the AESO s proposed approach for import assets. In general, this approach also enables efficient pricing by enabling export participants to reflect in their offers the cost of production. Dispatch obligations A price taker export asset continues to be dispatched using the existing protocols and requirements. A priced export asset will be dispatched intra-hour. A priced export asset is responsible for ensuring interchange transaction scheduling approvals from their source to Alberta will meet energy market dispatch requirements. To accommodate the priced offers for an export asset, new requirements are needed to dispatch exports at any time during the hour. The following scheduling example is based on BC Hydro Intra-hour scheduling business practice for worst case scenario which allows 5 minutes for transmission procurement and e-tag. Please note this is a worst case scenario; schedules should be implemented as soon as possible following a dispatch. Example: Pool participant receives a dispatch at XX:06. Pool Participant procures transmission (if required). Pool Participant submits (or adjusts) e-tag prior to XX:25 for transaction start at XX:45 and a ramp duration of 10 minutes. Assuming the e-tag is approved by required entities, at XX:40 energy ramp starts. At XX:50 ramp is completed. Ramp starts 34 minutes after dispatch; schedule starts 39 minutes after dispatch. Page 7 of 35

106 If, in the above example, an e-tag can be submitted by XX:10 (4 minutes after dispatch) for a XX:30; the ramp will start 19 minutes after dispatch and schedule starts 24 minutes after dispatch Obligations in the EAS markets for a long lead time asset As a subset of generation that requires more than an hour start time, a long lead time asset ( LLTA ) continues to have a must offer requirement. A LTTA continues to have similar obligations to generating units with the exception of requirements related to restatements and acceptable operational reasons when offline, as well as dispatches and directives. The existing requirements related to dispatching and directing a LLTA allow for a balance in allowing these assets some flexibility to manage their operations while still providing visibility of the units availability to the AESO. However, in the event of a forecast supply shortfall, the AESO may take action to direct a unit. The existing ISO rules related to compensation reflect the market conditions and the ability of the LLTA to choose to accept the dispatch or be directed for energy. Compensation will also reflect recovery of costs, as well as the forfeiture of capacity market compensation in circumstances where the LLTA did not meet its capacity commitment. This balance provides the correct incentives for LLTA to be online in anticipation of energy emergency events which usually align with higher price hours reflecting market conditions Monitoring and mitigation of market power in the energy market Market power mitigation framework The market power mitigation test is a three-part test of offer prices that will be assessed on an hourly, ex ante basis after T-2 in advance of each delivery hour: (a) Market power screen: Determine whether the firms that control energy offers have structural market power (net of obligations); and A firm will be considered to have market power if its value of the residual supplier index is less than 1. Firms may voluntarily submit supply obligations for inclusion in the residual supplier index calculation. (b) No-look scarcity test: If the market is very tight in a delivery hour, there will be no market power mitigation in that delivery hour irrespective of generator concentration or offer prices; The market will be deemed to be sufficiently tight in hours where the supply cushion is expected to be less than 250 MW. (c) Asset-specific reference price: Calculate the maximum price level that a generator would be expected to offer energy at if it had no market power based on the asset-specific short-run marginal costs adjusted through the use a market-wide marginal cost multiplier of 3x in order to account for cycling and start-up costs. When the market is scarce, as measured by a supply cushion of 1000 or less, a scarcity multiplier will be used, adjusting the market-wide marginal cost multiplier of 3x reflecting operating costs to a price multiplier of 6x reflecting scarcity conditions. The asset-specific reference price for non-thermal assets is based on opportunity costs. Page 8 of 35

107 An asset-specific reference price will not be less than $25/MWh. For each offer, if the following three conditions are satisfied: (a) The controlling firm has market power for the delivery hour (net of obligations); (b) The energy market is not tight; (i.e., the no-look level does not occur); and (c) The offer price is above the relevant asset-specific reference price. Then the price component of the offer will automatically be restated to the asset-specific reference price. In an energy-only market, the energy market provides the signal for both long-term decisions related to investment and retirement, and short-term decisions related to consumption and production, including those related to the operation of the electricity network, through the pool price. In most hours historically, offers in the Alberta electricity market have been made at approximately marginal cost. The effect is that accepted offers are paid at least their marginal cost from the energy market, with lower marginal cost assets earning more than marginal cost and typically covering all of their operating costs. To the extent that a firm expected to be on (or near) the margin and therefore set the pool price, it would make an offer to reflect its fulsome marginal cost in order to recover costs related to cycling and minimum run times, for example, from the market. That is to say, a single part bid equivalent of a cost model in the energy market must reflect more than standard operating and maintenance costs. The pool price may be quite high in some of these hours because of scarcity (i.e., little or no available supply in excess of demand). With the implementation of a capacity market, the purpose of the energy market will evolve. The capacity market will work in conjunction with the energy market to provide an efficient collective signal for dynamic, long-term decisions in investment and retirement. The energy market, through the pool price, must continue to provide the principal signal for short-term decisions related to consumption and production, including those related to the operation of the electricity network as well as price value for flexibility and ramping. In hours where available supply in excess of demand is scarce, the pool price may be relatively high. However, since the energy market is no longer required to provide sufficient revenue in excess of operating costs to cover the fixed costs of prudent investment in generation capacity, the AESO has proposed to put in place a framework to mitigate the exercise of market power in the energy market on an ex ante basis. Considering all of its elements together, the AESO s market power mitigation proposal is a practical solution that is consistent with relying on competition to the maximum extent possible to achieve efficient market outcomes for Albertans within the broader context of the existing market design, specifically one-part offers and self-commitment of generators. An alternative approach that relies less on competition and more on price regulation is also possible, but this would require additional changes to the broader market design in order to be workable. A high-level comparison of these approaches is set out in Table 1. Page 9 of 35

108 Table 1 - High-level comparison of mitigation approaches CMD 2 emphasized the importance of energy market price signals for future investment, retirement, and decisions related to consumption and production. The AESO recognizes that energy market scarcity events provide a signal for investment and lessens the magnitude of the missing money problem. The AESO realizes that real-time prices as a signal of energy scarcity at high prices provide important incentives to all participants. Allowing competitive outcomes and price formation expectations through market fundamentals is critical to deliver value for flexibility and ramping. The AESO proposes a new combined mitigation and graduated scarcity approach consistent with the goals and objectives from the CMD 2 proposal, which uses the market-wide marginal cost multiplier during normal market conditions, allows graduated levels based on the supply cushion, and lifts the screen in extremely tight conditions. The graduated scarcity approach considers factors relating to supply and demand, while recognizing that competitive prices are determined by competitive forces. The identification of firms that have market power assists in achieving competitive prices by ensuring that potential mitigation is focused only on those firms that may have an incentive to offer non-competitively, i.e., offer at substantial mark-ups above their marginal cost. Based on further analysis of historical mark-ups, the AESO is satisfied that the RSI threshold of 1 is appropriate. The RSI threshold of 1 meets market design objectives and balances mitigating egregious exercises of market power while providing incentives for self-mitigation through participation in the forward market. The AESO is of the view that when the market is sufficiently tight the pool price is expected to be very high. Options to achieve this include lifting the market power mitigation framework or raising the asset-specific reference price multiplier when the market is sufficiently tight, while ensuring that pool price offers allow for both the recovery of operating costs and signal scarcity but do not reflect the exercise of market power. As part of the market design, CMD 2 proposed that firms identified to have market power in hours where the market is not tight would have their offer prices limited to cost-based asset-specific reference prices that were set at a multiple of the asset s short-run marginal costs. CMD 2 also proposed a no-look scarcity test whereby there would be no mitigation in delivery hours when the market was sufficiently tight to ensure that the pool price can reflect these conditions by potentially being relatively high (a supply cushion less than 500 MW). Based on additional analysis and feedback from stakeholders, the AESO confirms the assetspecific reference price multiplier is 3 (the market-wide marginal cost multiplier ) when the supply cushion is above 1000 MW and proposes a graduated scarcity approach, whereby mitigation is Page 10 of 35

109 lifted when the supply cushion is lower than 250 MW, the asset-specific reference price multiplier is 6 (the scarcity multiplier ) when the supply cushion is between 250 and 1000 MW. Finally, the AESO is of the view that to the extent that the market power mitigation framework can support predictability (stability) in energy market outcomes, expectations about energy market outcomes will be formed with greater confidence. This will be reflected in better-informed offer prices in the capacity market and reduces the likelihood of the capacity market clearing on the basis of expectations of little exercise of market power in the energy market, only to have substantial market power be exercised in the energy market, with consumers having to pay twice for capacity (once in the capacity market and then again in the energy market). Identification of market power The purpose of the market power screen is to distinguish the firms that possess market power in a given delivery hour from those who do not. The market power screen will use the residual supplier index to measure the degree to which output from a firm is required in order for the energy market to serve load. In general terms, the residual supplier index for a specific firm expresses system supply less the firm s own supply as a fraction of market demand. The proposed formula recognizes the portion of a firm s supply that is forward contracted. Defined as such, it is a measure of the ability and incentive of a firm to profitably exercise control over the pool price, with a lower value implying that the firm has greater structural market power. The key issue with using the residual supplier index as a measure of structural market power is the determination of the threshold at which it is said to identify market power. Under the proposed framework, a firm will be determined to possess market power in a given delivery hour if its residual supply index (when measured after netting off supply obligations) for that hour is less than 1. Offers made for an asset at prices that are above the asset-specific reference price by firms with market power will have the price component automatically restated to the asset-specific reference price. Treatment of supply obligations in the implementation of the market power screen The proposed definition for the residual supplier index includes a term for the firm s fixed price physical and financial supply obligation volume that has the effect of reducing their offer control by the amount of the obligation. Firms may voluntarily submit data to the AESO regarding their supply obligations for inclusion in their calculation of the residual supplier index. The submitted obligation data may be subject to ex post auditing by the AESO. Fixed price supply obligations have the effect of reducing a firm s direct exposure to pool price outcomes and therefore reduce their incentive to exercise market power. The impact of the inclusion of an obligation value would be to raise the value of residual supplier index, resulting in an increased likelihood that a firm would pass the market power screen, and a reduced likelihood that the firm would be subject to offer price mitigation. The AESO is also of the view that the inclusion of an obligation term aligns the market power screen with strong incentives for market participants to engage forward contracting which is the most effective way of mitigating the exercise of market power. Interpretation and threshold of the residual supplier index The residual supplier index for firm i in delivery hour t is set out in the proposal document. When the residual supplier index for a particular firm is exactly equal to 1, the supply of all other firms plus its own supply obligations is exactly equal to total demand (including exports and reserves). When the residual supplier index is greater than 1, the supply of all other firms plus its own supply obligations is more than enough supply to meet total demand. However, when the residual Page 11 of 35

110 supplier index for a particular firm is less than 1, some amount of supply from the firm in excess of its supply obligations is necessary to meet total demand. The purpose of the mitigation screen is to identify firms with structural ability to exercise market power. A key issue addressed in CMD 2 regarding the RSI mitigation screen was the threshold at which it was set to identify pivotal firms. The AESO has considered historical market data over the period from 2013 to 2017 and is satisfied that the RSI threshold of 1 appropriately identifies the existence of structural market power. A RSI threshold at 1 balances mitigating egregious exercises of market power and provides incentives for firms to self-mitigate by entering in forward arrangements, allowing market outcomes guided by competition. This allows competition in most hours by focusing on key hours when structural ability to exercise market power is greatest. This determination is based on the following evidence and arguments, each of which is discussed in greater detail below: (a) A firm is, by definition, pivotal if its residual supplier index is 1 or less; (b) The mark-ups of price-setting offers are often high when the residual supplier index is less than 1 but are rarely high otherwise reflecting the competitiveness of the market; (c) High pool price hours tend to be identified by the residual supplier index beginning at less than 1; (d) Larger firms fail the market power screen based on the residual supplier index more often than smaller firms; and (e) The residual supplier index formula accounts for net supply obligations and as such creates an incentive for forward selling that mitigates the incentive to exercise market power within the market itself. (a) A firm is, by definition, pivotal if its residual supplier index is 1 or less. From its definition, when a firm s residual supplier index is 1 the supply of all other firms plus the firm s own physical supply obligations is exactly equal to market demand. Thus, the residual supplier index being less than or equal to one is a conceptually straightforward condition for the firm to possess structural market power. 1 (b) The mark-ups of price-setting offers are often high when the residual supplier index is less than 1 but are rarely high otherwise reflecting the competitiveness of the market. Figure 1 is a scatter plot of the mark-up 2 of the price-setting offer and the residual supplier index of the firm that controls that offer for each settlement interval of the years 2013 to 2017, if the firm was one of the three largest firms in the Alberta market. Each observation is also shaded to indicate the pool price in that settlement interval. 1 2 A combination of several firms may have joint market power even under circumstances in which none of them have market power on their own. Situations of this type will not be identified by a market power screen based on the residual supplier index. All else equal, this factor would suggest using a higher threshold for the residual supplier index. The mark-up of a specific offer is defined as: offer price less marginal cost, divided by marginal cost. Page 12 of 35

111 Figure to 2017 RSI versus mark-up with shaded pool prices As demonstrated in Figure 1, when applying an RSI threshold of 1, the results for 2014 and 2015 are similar to The analysis demonstrates the mitigation screen captures a significant number of hours where mark-ups were considerably above short-run marginal cost. Years 2016 and 2017 observed a substantial change in offer behaviour associated with the generators that were subject to Power Purchase Arrangements (PPA). Between December 2015 and May 2016 all of the then PPA Buyers terminated their PPAs and subsequently transferred offer control to the Balancing Pool, which appears to have generally followed a marginal cost offer strategy. As such, market outcomes in the years 2016 and 2017 cannot be considered to be indicative of competitive energy market outcomes. Specifically, at high levels of the residual supplier index (e.g., greater than 1.1) the markups of price-setting offers tend to be very close to zero with only rare exceptions. As the residual supplier index falls toward 1, larger mark-ups begin to be observed. As the residual supplier index declines below 1, much larger mark-ups (including some greater than 80 times marginal cost) are occasionally observed. While a residual supplier index that is less than 1 does not guarantee that the mark-up and pool price will be high, offer prices tend to be much closer to marginal cost when the residual supplier index is greater than 1. Moreover, there are few hours with very high mark-ups that would pass a market power screen based on the residual supplier index. This is consistent with the expectation that firms that lack market power will make offers very near to their marginal cost while those with market power will seek to exercise it by raising their offer prices above their marginal cost (i.e., that the effect of competition is to press mark-ups down and prices toward marginal cost). Market prices that are high during periods of scarcity do not reflect the exercise of market power but rather the fundamental and underlying supply-demand conditions. (c) High pool price hours tend to be identified by the residual supplier index beginning at less than 1. Figure 2 is a scatter plot of the same data as above (the mark-up of the price setting offer and the residual supplier index of the firm that controls that offer in 2013, if the firm was one of the three largest firms in the Alberta market) except that the data is organized into ten bins defined by pool price ranges (e.g., $0 to $99.99, $100 to , and so on). Page 13 of 35

112 Figure 2 RSI versus mark-up by pool price bin Figure 2 illustrates that: High pool prices are more likely when the residual supplier index of the firm that controls the price-setting offer is low and there are few high pool prices when it is high; At moderate pool prices there is a split of hours with residual supplier index values greater than 1 and less than 1, though most are less than 1; and At low pool prices hours are more are likely to have higher values of the residual supplier index but many of these hours still have low values of the residual supplier index (i.e., structural market power is present; some mark-ups exceed ten times marginal cost). As above, while a residual supplier index that is less than 1 does not guarantee that the mark-up and pool price will be high, offer prices tend to be much closer to marginal cost when the residual supplier index is greater than 1. Some of the high prices reflect scarcity conditions and some reflect market power. As discussed above, this was an expected outcome of the energy-only market. However, with the implementation of a capacity market and capacity payments, the purpose of the energy market will change. The market power mitigation framework is intended to balance a focus on identifying the subset of firms who can influence the pool price by exercising of market power with minimizing the risk of over-mitigation that causes prices to be very low during scarcity conditions. (d) Larger firms fail the market power screen based on the residual supplier index more often than smaller firms. Figure 3 illustrates residual supplier index duration curves for the five largest firms, using data from all hours of the forecast year 2021 that underlie the AESO s 2017 Long-term Outlook. The offer control for existing generation capacity is based on the Market Surveillance Administrator s 2017 Market Share Offer Control Report, 3 adjusted to unwind the effect of all power purchase arrangements. Entering generation capacity, including 1,200 MW of wind capacity, is assumed not to be part of any firm s portfolio. 3 Page 14 of 35

113 The shaded horizontal band covers the range of residual supplier index values from 0.85 to 1. Note that Figure 3 does not include any adjustment for the physical supply obligations of each firm. Figure 3 - Year 2021 forecast RSI duration curves for the five largest firms Figure 3 indicates the relative importance of supply from the various firms to meeting demand. Based on the residual supplier index threshold of 1, the largest firm was identified to have market power in approximately 65% of all hours, with the smaller firms being identified much less frequently. The fifth largest firm was not identified in any hour. At higher levels of the residual supplier index threshold than 1 (such as 1.1), the test would identify (i) larger firms more often; and (ii) additional market participants in at least some hours. At levels of the residual supplier index lower than 1, fewer firms would be identified by the test and those that did would be identified less often. For instance, with a residual supplier index threshold of 0.9, only the three largest firms would ever be identified by the test and the largest firm would be identified in approximately 25% of hours. Analysis using historical data yields qualitatively similar results. Figure 4 reports analysis for the years 2013 to 2017, assuming the maximum annual value of import available transfer capability and without adjustment for any net obligations. 4 With the exception of 2017, the largest firm is pivotal in between 8% and 25% of hours depending on the year, while smaller firms are identified as pivotal depending on the year. In 2017, the largest firm was pivotal in approximately half of the delivery hours, which was very different from other years as a direct result of the power purchase arrangement termination issue. 4 AESO. Energy market power mitigation. Presentation to the Energy and Ancillary Services Workgroup Meeting #4. June 13, Available online at: Page 15 of 35

114 Figure 4 - Duration curves for the largest firms for the years 2013 to 2017 At levels of the residual supplier index lower than 1, fewer firms would be identified by the test and those that did would be identified less often. At higher RSI, more firms would be identified as pivotal. (e) The residual supplier index formula accounts for supply obligations and as such creates an incentive for forward selling that mitigates the incentive to exercise market power within the market itself. Page 16 of 35

115 The inclusion of fixed price physical and financial supply obligations in the residual supplier index would reduce, perhaps substantially, the frequency with which a firm is identified as pivotal by the market power screen and potentially the number of firms who are identified by the market power screen. Figure 5 illustrates the impact on the residual supplier index duration curve of the largest firm in the market in the forecast year See the discussion of the duration curves in the previous subsection for additional details. Specifically, the no obligation duration curve (solid purple) for this firm is exactly the same curve as in the previous figure (also solid purple). The figure below illustrates the effect on the residual supplier index duration curve of this firm holding various levels of supply obligations. As the firm is expected to have slightly greater than 4000 MW of installed capacity, obligation values of 500 MW, 1000 MW (approximately one-quarter of installed capacity), and 2000 MW (approximately one-half of installed capacity) are considered. A number of specific points of interest are indicated in Figure 5, including: With a residual supplier index threshold of 0.9 and no supply obligations, the firm would be identified as having market power in approximately 25% of hours (point A); With a residual supplier index threshold of 1 and no supply obligations, the firm would be identified as having market power in approximately 65% of hours (point B); With a residual supplier index threshold of 1 and 1000 MW of supply obligations (approximately one-quarter of installed capacity), the firm would be identified as having market power in approximately 25% of hours (point C); With a residual supplier index threshold of 1 and 2000 MW of supply obligations (approximately one-half of installed capacity), the firm would be identified as having market power in approximately 1% of hours (point D); and With a residual supplier index threshold of 0.9 and 1000 MW of supply obligations (approximately one-half of installed capacity), the firm would be identified as having market power in approximately 1% of hours (point E). Figure 5 - Duration curve for firm 1 accounting for net obligations Thus, with a residual supplier index threshold of 1, the largest firm in the market would be identified as pivotal by the market power screen in 25% of hours if it forward sells onequarter of its installed capacity and only 1% of hours if it forward sells one-half of its Page 17 of 35

116 installed capacity. This illustrates the impact of the inclusion of supply obligations in the calculation of the residual supplier index. The inclusion of supply obligations accounts for some of the incentive to exercise market power, which makes it a more useful measure of market power for the mitigation framework. Identification of scarcity conditions: No-look scarcity test The pool price must reflect market conditions if the energy market is going to efficiently allocate resources. In circumstances when available supply in excess of demand, known as the supply cushion, is highly limited, it is both expected and desired that the pool price reflect this scarcity by being relatively high. However, as the supply cushion approaches zero, all firms become pivotal and will be identified by the market power screen based on the residual demand index threshold of 1. This raises the prospect of over-mitigating offer prices and preventing the energy market from price formation necessary for providing an efficient, market-based price signal during scarcity conditions, as a marginally pivotal firm may be identified as pivotal. Further, the AESO is of the view that scarcity conditions exist before the physical supply of electricity is exhausted. To prevent the realization of these concerns, the proposed market power mitigation framework includes a no-look scarcity test to identify highly scarce hours in which no mitigation will be applied as well as a scarcity multiplier (discussed below). The key issue with the no-look scarcity test is the determination of the specific metric used to distinguish scarce hours. Under the proposed framework, the test will be based on the supply cushion, with scarce hours defined by the supply cushion being less than 250 MW. As noted in CMD 2, the original scarcity screen of 500 was based on the measure of a contingency. The AESO recognizes that this threshold is somewhat arbitrary and is binary as it may limit competitive outcomes and, at the extreme when the supply cushion is zero, all firms would be identified in the screen. The proposed no-look level at 250 MW is part of the graduated scarcity approach for an assetspecific reference price that is adjusted by applying a scarcity multiplier (discussed further below). Adjusting the asset-specific reference price upwards when the supply cushion is below 1000 MW allows for price formation. The no-look level ensures perverse price levels as the market moves to supply shortfall do not occur. It is expected that volumes that are unmitigated are sufficient for price formation, but lifting the market power screen when the no-look level occurs ensures that the mitigation screen is not leading to administrative pricing when the market should determine the price. Further discussion on the graduated scarcity approach can be found in subsection below. The AESO will run the no-look scarcity test as close to real-time as reasonably practicable in order to account for contingencies potentially affecting marginally pivotal firms after offers are locked down at T-2. The determination that no mitigation will be applied in hours where the supply cushion is less than 250 MW is based on the following evidence and arguments, each of which are discussed in greater detail below: (a) The market will be approaching supply shortfall; and (b) Assessment of pool price-supply cushion outliers in historical market data suggests scarcity occurs when the supply cushion is sufficiently low. (a) The market will be approaching supply shortfall. A value of 500 MW is approximately equivalent to the most severe single contingency. Multiples of this number are used as part of the asset-specific reference price defined below. At 250 MW, the energy market would be near emergency conditions so the test identifies scarcity conditions. (b) Assessment of pool price-supply cushion outliers in historical market data suggests scarcity occurs when the supply cushion is sufficiently low. Page 18 of 35

117 Figure 6 is a scatter plot of the pool price (in logarithmic form) and supply cushion for the period from February 1, 2008 to June 30, As a result, these data relate to market outcomes from before the publication of the Market Surveillance Administrator s Offer Behaviour Enforcement Guidelines 6, which occurred in early Figure 6 illustrates that there is generally an inverse relationship between the pool price and supply cushion. Figure 6 Pool price and supply cushion correlation Each observation can be grouped into one of 13 supply cushion bins, each 250 MW wide (e.g., 0 MW to 250 MW, 250 MW to 500 MW and so on). For each of these bins, the mean (average) pool price and the standard deviation of the pool price was calculated. The mean for each bin is illustrated with a red horizontal line. Also illustrated is the mean plus and minus one standard deviation (blue horizontal line), the mean plus and minus two standard deviations (the green horizontal line), and the mean plus and minus three standard deviations (the orange horizontal line). Given the offer price cap of $999.99/MWh and the market price cap of $1000/MWh, when any of these values exceeds $1000/MWh the horizontal line is illustrated at $1000/MWh. Figure 6 can be used to identify pool price outliers. Specifically, pool prices that are either greater than the relevant mean plus three standard deviations or less than the relevant mean minus three standard deviations can be viewed as outliers from normal competitive market outcomes. Of particular interest is that the mean plus three standard deviations is greater than $1000/MWh for the supply cushion bins 0 MW to 250 MW and 250 MW to 500 MW. As a result, there are no pool price outliers when the supply cushion is in this range. While future market outcomes when the capacity market is implemented are invariably going to be different in nature from past market outcomes, the AESO is of the view that this 5 6 Market Surveillance Administrator (2012). Supply cushion methodology and detection of events of interest. %20Step%205/Offer%20Behaviour%20Enforcement%20Guidelines% pdf Page 19 of 35

118 historical analysis supports the conclusion that when the supply cushion is sufficiently low, supply is sufficiently scarce such that it is appropriate to expect the pool price to rise above marginal cost as a market-based signal of scarcity. As a result, when the supply cushion is in this range, no mitigation of offer prices will occur. Determination of asset-specific reference prices for thermal asset In the context of a market design with single-part offers and no uplift mechanism to recover startup or other similar costs, assets which need to recover such costs may not be able to if they are only paid their short-run marginal costs by the energy market. For thermal assets, the assetspecific reference price for asset j in delivery hour t is set out in the in the proposal document. Under the proposed market power mitigation framework, the asset-specific reference price for thermal assets will be set at a three times the asset s marginal cost, i.e., MM = 3 (discussed in subsection (a) below). The asset-specific reference price is intended to reflect marginal operating costs, including carbon. A market-wide marginal cost multiplier is applied to ensure that cycling and start-up costs can be priced into the energy market as part of a single-part offer or bid and selfcommitment energy market design, noting that marginal costs do not reflect these values. The asset-specific reference price is based on a formula for a reference asset type. A participant would be able to submit an exception request if the asset s actual costs are above the assetspecific reference price (see subsection ). The acceptable market-wide marginal cost multiplier on the asset-specific reference price is estimated based on historic run times and acceptable operating costs. Offers made for an asset at prices that are above the asset-specific reference price by firms with market power will have the price component automatically restated to the asset-specific reference price. CMD 2 presented data comparable to the industry standard. However, given the possibility that the costs are low because of an asset-specific cost or because of increased cycling in the future, a range of costs are considered. Determination of asset-specific reference prices for other assets, including imports and nonthermal, energy-limited assets For other assets, including an import or a non-thermal, energy-limited asset, the asset-specific reference price will be set based on a formula that captures the concept of opportunity cost. The AESO proposes a formulaic approach to the calculation of opportunity cost because it is (i) objective, (ii) transparent, (iii) forecastable, (iv) calculable by parties outside of the AESO, and (v) does not require the submission and verification of substantial amounts of information and modelling results. For an import asset, the opportunity cost of selling energy into the Alberta market is the value that they could obtain from selling that energy into another market, either contemporaneously or in the future. The value of energy in a neighbouring market is a reasonable proxy for this opportunity cost. The AESO proposes that the day-ahead, on-peak price of energy in the Mid-Columbia market provide a reasonable basis on which to make this calculation. A margin is added to this external price to obtain the asset-specific reference price in recognition that market conditions can change substantially from day-to-day and Alberta must ensure that importers are paid sufficiently such that they will schedule trade to Alberta when market conditions in Alberta are relatively tight. For non-thermal, energy-limited assets located in Alberta (including hydro and storage), the decision to produce energy in one period implies that the ability to produce output in future periods is reduced. For these types of assets, the value of foregone opportunities to use their limited energy in the future is the opportunity cost of current production. This is the relevant measure of short-run marginal cost for such an asset. For this type of asset, using its limited Page 20 of 35

119 ability to produce energy in the current period may not only require foregoing the opportunity to produce energy in the future, it may also result in the opportunity to produce ancillary services in a sequence of periods being foregone. Since there is significant uncertainty about future energy and ancillary services prices, and uncertainty about whether ancillary services are going to be used, the opportunity cost of energy-limited assets may be as high as the energy market price cap in some hours. Given the inability to forecast which hours these will be, the market power mitigation framework will not impose specific offer price mitigation on offers made by non-thermal, energy-limited facilities provided that offers from those facilities were made into the ancillary services market. If no offers for all available capacity from those assets were made into each of the ancillary services markets, then the relevant offer prices will be subject to mitigation as though they were made by imports. Given the inability to forecast which hours these will be, for non-thermal, energy-limited assets, there will be no offer price mitigation (in effect, the asset-specific reference price will be the energy market offer price cap) provided that offers for the maximum capability of these assets were made into the each of the markets for ancillary services. The link between the energy and ancillary services markets here is that if the energy limit is near binding, the high opportunity cost may come in significant part from the value of providing ancillary services and the potential inability to provide some of these services if the energy is immediately used. Thus, if an energylimited asset is going to effectively have its asset-specific reference price set at the energy market offer price cap, then it will be expected to have at least made offers to provide ancillary services. Said another way, the higher energy price is used to limit flow of water, which could similarly be achieved by an offer into the ancillary services market, meaning the water will only flow if required for system emergency. If no such offers are made, the argument that an element of opportunity cost is related to the ability to provide ancillary services is less convincing. Thus, if such ancillary services offers are not made, then the relevant assets will be assigned asset-specific reference prices based on the rolling average pool price as set out in the proposal document. The AESO recognizes the importance of water management to the operation of hydro assets and that opportunity cost is a critical element of determining when the available water is best used to produce energy. The AESO continues to be of the view that in periods when water is scarce and opportunity cost is relatively high it is appropriate for there to be no energy market mitigation for such an asset if that asset makes offers to provide ancillary services, especially active spinning and supplemental reserves and standby reserves. The AESO will determine the relevant products and volume obligations, which will be subject to further consultation. Determination of the price multiplier The asset-specific reference price is proposed to be a multiple of the asset s short-run marginal cost that represents the maximum price level at which the asset would be expected to be offered if the firm that controlled it had no market power. The proposal includes both a market-wide marginal cost multiplier and a scarcity multiplier to reflect market fundamentals, in addition a no-look scarcity level (discussed above) when the mitigation framework is entirely lifted. As such, the proposal allows offer prices to reflect cycling and start-up costs and energy scarcity. During periods when the supply cushion is relatively high (i.e., over 1000 MW), the (market-wide marginal cost) multiplier will be 3. This value was adopted following extensive consideration of market data and discussions with market participants. The AESO is satisfied that it ensures that mitigated firms are able to recover their operating costs through the energy market while still mitigating market power by placing an upper bound on the offer prices of firms with market power (see subsection (a) below). Page 21 of 35

120 During periods when the supply cushion is relatively low but not very near to zero (i.e., between 250 and 1000 MW), the (scarcity) multiplier will be 6. This value is intended to allow for scarcity to be reflected in the energy price as a result of energy offers. In the AESO s view, this ensures that in tight market conditions, suppliers have strong incentives to produce power, importers have strong incentives to schedule imports, and loads have strong incentives to avoid consumption. The AESO considers these incentives to be essential for the market to adequately support the reliable operation of Alberta s power system. The lower bound on the supply cushion range was selected on the basis of 250 MW being an approximation of the historical amount of priceresponsive load in Alberta (see subsection (b) below). The AESO intends to conduct further analysis and consultation on the level of the scarcity multiplier. The elements of this proposal are discussed further below. Finally, the AESO notes that its consideration of these elements reflects its prior analysis of the effect of increased net demand variability. A summary of the AESO s proposal is set out in Table 2. Table 2: Supply cushion and price multiplier range Supply Cushion ASRP Price Multiplier >1000 3x x <249 No-look / screen lifted Market-wide marginal cost multiplier of 3 when supply cushion is above 1000 MW (a) The rationale for the determination that the market-wide marginal cost multiplier of 3 when supply cushion is above 1000 MW is based on the following evidence and arguments, each of which is discussed in greater detail below: a) Mitigation of offer prices to three times marginal cost allows operating costs to be recovered from the energy market; and b) Mitigation of offer prices to three times marginal cost stabilizes net energy revenue across a variety of market conditions. a) Mitigation of offer prices to three times marginal cost allows operating costs to be recovered from the energy market. The AESO is satisfied that mitigation to three times short-run marginal cost under these circumstances ensures that mitigated firms are able to recover their operating costs through the energy market while still mitigating market power by placing an upper bound on the offer prices of firms with market power In particular, the AESO s analysis indicates that a lower multiple provides insufficient revenue to recover operating costs some circumstances. The AESO analyzed historical generator run-time data for the period of January 1, 2013 to December 31, Figure 7 shows the distribution of these run-times all for simple cycle units in Alberta. The mean and median run-times were found to be 271 and 130 minutes, respectively. The figure shows that a generator s run-time was less than 30 minutes for only 16% of starts. Page 22 of 35

121 Based on this analysis and other analysis conducted by Brattle 7, the AESO is of the view that using a run-time of 30 minutes for the selection of the multiplier is appropriate. Figure 7 - Run-times for simple cycle units in 2013 and 2014 The rightmost column of Table 3 below provides the ratio of average-to-marginal cost for each combination of generator and assumptions. This ratio is highly sensitive to the assumed run-time, where a longer assumed run-time lowers the average-to-marginal cost ratio. The reason for this is that the start-up and shut-down costs are averaged over a larger amount of production. 7 Brattle (2018), Market power screens and mitigation options for AESO energy and ancillary services markets. Page 23 of 35

122 Table 3 - Sensitivity of ratio of average-to-marginal costs to run-times *2013/14 historical run time data was used as natural gas levels during this period were moderate. Run times shown are the base 30 minute data from Brattle plus Alberta s historical median and mean run time values. Simple cycle generators typically have the highest average-to-marginal cost ratio. The AESO expects investment in this type of generator will occur in the future and that they will be an important source of production in the future. As such, their ability to recover their operating costs from the energy market is important. Based on an assumed run-time of 30 minutes, the ratio is 2.73 and then a reasonable value for the multiplier would be three. All other generation types would also able to recover their operating costs under this approach. The AESO considers this to be important to ensuring that the market supports the reliable operation of Alberta s power system. b) Mitigation of offer prices to three times marginal cost stabilizes net energy revenue across a variety of market conditions Table 4 reports Brattle s generator revenue impact analysis associated with various market power mitigation rules. The assessment considers the net energy revenue and net-cone using historical data for the period 2013 to 2016 for two types of generation technologies. While this analysis shows that each type of generation technology is able to recover its operating costs from the energy market under a variety of different mitigation approaches (i.e., net energy revenue is positive), mitigation reduces the variability of net revenue across a range of market conditions that are exemplified by use of a set of past years in the Alberta market in which market conditions varied significantly. This result is driven by the fact that one effect of mitigation reduces market prices. Note that the AESO considers that, in the context of a market design with singlepart offers and no uplift mechanism to recover start-up or other similar costs, mitigation of offers to levels close to marginal cost does not provide sufficient net energy revenue to fully cover cycling costs (see the previous section). One implication of stabilizing net energy revenues is that net-cone values are also stabilized across a wide range of potential energy market outcomes. This supports stability in the location of the capacity market demand curve across a range of potential energy market outcomes (that are Page 24 of 35

123 characterized here by different years when there were very different conditions in the Alberta electricity market). The AESO is of the view that to the extent that the market power mitigation framework can support predictability (stability) in energy market outcomes, expectations about energy market outcomes will be formed with greater confidence. This approach hopefully results in better informed offer prices in the capacity market while reducing the likelihood of the capacity market clearing on the basis of expectations of minimal exercise of market power in the energy market, only to have substantial market power be exercised in the energy market, with consumers having to pay twice for capacity (once in the capacity market and then again in the energy market). As discussed at the beginning of this section, this would not be consistent with the evolving purpose of the energy and ancillary services markets. Taken together with the other evidence, the AESO considers that mitigation based on restricting offer prices of firms with market power to be no greater than three times the short-run marginal cost, is appropriate. Table 4 Analysis of various market-wide cost multipliers 8 Scarcity multiplier of six when supply cushion is between 250 and 1,000 MW (b) Scarcity pricing is a critical element for investment, retirement, and decisions related to consumption and production in the energy market. The Final CMD proposal adjusts the marketwide marginal cost multiplier for scarcity market conditions using a scarcity multiplier. This approach improves the outcome that any resulting high prices were a result of competitive forces and not as a result of market power. Additionally, the scarcity multiplier approach allows for competitive scarcity offers as the supply cushion starts to tighten. Rather than using a 500 MW supply cushion to lift the market power screen, the supply cushion threshold is used to adjust the asset-specific reference price by using a scarcity multiplier. CMD Final proposes further graduation to the cost multiplier at less than 1000 MW supply cushion and a no-look scarcity threshold of 250 MW to ensure that the market has room for price formation as the market tightens. The AESO recognizes the importance of real-time prices signaling the economic scarcity of energy and related services such as ramping capability. It is 8 Source: Brattle, Assessment of bid mitigation options. November 21, Available online at: Page 25 of 35

124 the AESO s view that economic scarcity does not require physical scarcity, i.e., energy supply need not be physically exhausted for energy to be considered to be in short supply. Allowing competitive outcomes and price formation expectations with market fundamentals is critical so that the value for flexibility and ramp is reflected in the energy market price signal. This is especially important given future increases in net demand variability. Exception request for an asset-specific reference price Asset-specific reference prices will be calculated in a manner to ensure that an asset s operating costs are recoverable from the energy market under reasonable circumstances. As discussed above, these prices will be calculated formulaically by the AESO. Firms will be able to make an asset-specific request for an exemption from the asset-specific reference price should they consider that they would not be able to recover their asset s operating costs from the energy market under reasonable circumstances. The AESO will approve or reject the request, based on the evidence provided. No resource s reference price will ever be less than $25/MWh A minimum asset-specific reference price of $25/MWh will apply to all generation offers, irrespective of their type or other characteristics. There are two reasons to set a minimum asset-specific reference price at this level: (a) to avoid rare circumstances where the formula-derived reference price could be set extremely low, perhaps even negative; e.g., natural gas prices could be negative on a given day; and (b) based on the cost characteristics of assets in that exist in the Alberta market, offers at this price are extremely unlikely to reflect the exercise of market power. No mitigation of offers that are below the asset-specific reference price As described in subsection above, an offer fails the market power mitigation test if the following three conditions are satisfied: (a) a specific firm has market power (net of obligations); (b) the energy market is not sufficiently tight (i.e., the no-look level does not occur); and (c) an offer price is above the relevant asset-specific reference price. If a firm has market power, but submits an offer for an asset that is below the relevant assetspecific reference price, the offer will not be changed. Market power screen and mitigation applies irrespective of whether an asset has a capacity commitment Market power mitigation in the energy market will occur based on the conditions in that market and is not dependent upon the outcomes of the capacity market, specifically which assets (and to what extent) have a capacity commitment. Market participants without capacity obligations are equally able to possess market power and impact price levels. Accordingly, there is no reason why an asset without a capacity commitment should be excluded from market power mitigation. Ex ante monitoring and mitigation is expected to continue The existence of mitigation, whether it affects offer prices in a given delivery hour or not, does not remove the role of ex post monitoring. Further, since the mitigation scheme is intended to be part Page 26 of 35

125 of a competitive energy market that allocates resources efficiently, conduct whose purpose is to evade the mitigation scheme would not be consistent with supporting a fair, efficient and openly competitive electricity market Roadmap Reforms in the EAS Markets Pricing This subsection provides an overview of items in the EAS roadmap that may be triggered by events or changes in the market. This roadmap has been developed to provide greater certainty to investors (both incumbents and new) of pending market changes that are being considered so that assessments can be made prior to the 2019 auction. The following pricing designs have been assessed to have limited value or need at this time; therefore, the current pricing methodology will continue in the energy market. Offer cap Above $ (a) The current offer cap is effectively non-binding and will increasingly not be an issue with the introduction of the capacity market. While the majority of revenues will be expected to remain in the EAS markets, the offer cap consultation is not a priority at this time. The price cap does not appear to be limiting especially given pending changes to market power mitigation; however, a pricing signal may be of value to clear surpluses in the energy market. The cap and floor need to be wide enough to allow scarcity pricing to occur and will be further examined if the cap or floor is limiting. Negative pricing (b) The AESO currently employs an administrative mechanism to address supply surplus. Upon reaching $0/MWh in the energy market merit order, the AESO first curtails import assets, then $0 flexible blocks, including renewables, then $0 non-flexible blocks, and finally curtailing generation offline. Negative pricing is widely considered an improved alternative to manage congestion and over-generation that improves market efficiency and liquidity, particularly with increasing variable energy production. However, other jurisdictions may be encountering issues with negative pricing, where subsidized resource offers may be capable of offering even below inflexible generation. As noted in various forms by PJM and the US Department of Energy in 2017, as well as the creation of asset-specific offer floors by the IESO, there may be potential concerns on the impact of subsidies distorting market outcomes and eroding revenue streams. Further considerations and risks for implementing negative pricing: Setting the price floor: A negative price floor may simply move the high level of equal price offers to a new floor thus moving the supply surplus issue to a new price. The price floor must be set low enough to promote additional depth in the merit order. Products indexed to pool price: Active operating reserves are indexed to pool price and currently cannot go below $0/MWh. However, the real power provided for a product (e.g. regulating reserve) during negative pricing would incur a cost to the provider; requires consideration on whether they are isolated from the effect. Importers: If importers continue to submit $0 offers and are not eligible to set pool price, there may be an issue with negative pricing. System changes: Scope of changes to the energy trading system, dispatch tool and settlement processes to be assessed. Page 27 of 35

126 Impact on transmission constraint management (TCM): TCM may require adjustments to accommodate negative pricing; further assessment required. Dispatch: Any issues related to administratively dispatching efficiently or clearing clearly the energy at the price floor. The supply surplus events are currently cleared administratively and few issues have resulted. The introduction of negative pricing may introduce challenges that will need to be reviewed. However, the AESO does not consider negative pricing to be a priority at this time. As the frequency and impact of supply surplus increases, negative pricing may assist in addressing supply surplus. This enables a market-based approach to address surplus rather than the current administrative curtailment mechanism. However, given the additional risks identified in a negative pricing model and taking into consideration the increase of renewables on the system with competing pricing incentives, the negative pricing model will be evaluated when dispatch related to clearing MWs during supply surplus events at $0 become an issue or inefficient. Administrative shortage pricing (c) Administrative shortage pricing provides a mechanism for increasing EAS market prices above offered prices during times of supply shortage typically measured by the release or depletion of operating reserves. The purpose of administrative shortage pricing is to enhance market price signals for response to these events, and to provide an enhanced investment signal for quickstart and fast-ramping assets (such as peakers and demand response), which are designed to avoid loss of load while capacity is tight in the energy market. Administrative shortage pricing has been adopted by many independent system operators with centralized wholesale markets. For clarity, administrative shortage pricing is separate from scarcity pricing which can occur within the market when offer prices are higher than resources actual short-term marginal costs. Considerations in the design of administrative shortage pricing include: There may be little or no regulatory tolerance for energy and ancillary services prices above $1,000/MWh. Price tolerances like this are not uncommon in other jurisdictions. Shortage pricing will be activated infrequently (0 50 hours per year, depending on planning and operating reserve levels). The magnitude and frequency of shortage pricing (when it is triggered and what price it is triggered to) is dependent on market design objectives. Capturing an effective price signal for flexible investment and operational behavior. Shortage pricing levels and maximum price levels ($/MWh). The entire price signal (from all markets, not just energy and ancillary services) in the worst shortage conditions when involuntary load-shedding occurs should theoretically reflect cost of that load-shedding (i.e., value of lost load). Supply shortfall events are also managed administratively and with few issues arising. Further, it is anticipated that with the introduction of a capacity market, the frequency of shortage events will be lessened. However, this concept may be further reviewed as required to incent price responsive behavior near shortages but is not a priority at this time. Depending on the type, and level of offer mitigation, and the degree of in-market scarcity pricing resultant from mitigation, additional administrative shortage pricing may need to be considered to enable an effective price signal. The AESO understands the value associated with shortage pricing but has determined that it will be examined if the market pricing during scarcity become inefficient. Page 28 of 35

127 Dispatch and flexibility The net demand variability (NDV) is expected to increase materially by 2030 due to the expected increases in variable renewable energy. The NDV analysis indicated that this change may be manageable with current requirements, assuming there is no change to average ramp behaviour. The forecast fleet does have the capability to meet forecasted flexibility needs, but changes to the dispatch tolerance requirements may be required to ensure dispatch certainty. The following Figure 9 shows materially higher variability swings when comparing 2015 to As shown, there are more events of high NDV, with increasingly larger ramps - both time and rates. The occurrence of these increased ramp events are dependent on the timing for increased variable resource additions without the ability of the system to manage these variations through greater dispatch certainty or associated products. Figure 9 Histogram of 60-minute NDV, based on 10-minute data Consultation on dispatch certainty is expected to continue in If dispatch certainty is resolved, the issues related to the ability to tolerate more variability may be delayed. Dispatch certainty dispatch tolerance and ramping (a) The current dispatch tolerance requirements create uncertainty for the AESO to operationally manage the grid. The AESO must rely on historical ramp behaviour and knowledge of the assets when dispatching. Under the current requirements, pool participants have a fairly wide range of dispatch tolerance to use at their discretion when meeting dispatch instructions. The dispatch tolerance requirements provide the following ranges: ramp rate: +/- 40% of submitted ramp rate; time to respond: 0 to 10 minutes; and dispatched target MW: +/- 10 MW for assets with maximum capability (MC) greater than 200 MW, and +/- five MW for assets that have an MC less than or equal to 200 MW. In the following Figure 10, the green area depicts the dispatch tolerance window, and the area of dispatch uncertainty the AESO must manage when dispatching. Page 29 of 35

128 Figure 10 Visual graphic of the current dispatch tolerance requirements The next Figure 11 illustrates the historical ramping-up movements and delay times for a sample asset from July 1, 2015 to June 30, These ramping events only apply to dispatch directives with a dispatch delta close to 50 MW. The gray shaded area shows the dispatch tolerance available to the unit, assuming a 50 MW dispatch. Historical data shows that most of the time, the asset ramped up to the required level in less than 10 minutes. There were some exceptions in which the unit deviated from its normal behaviour. Delay time was five minutes or less during 95 percent of the ramping-up events. Figure 11 Visual graphic of historic dispatch tolerance behavior versus the current dispatch tolerance requirements The above Figure 11 illustrates the large area of uncertainty the current dispatch tolerance requirements create for the AESO to manage. This added operational complexity has the potential to become quite significant considering future forecasted high NDV. The AESO does recognize that all assets have different characteristics in terms of ramp response and operational requirements, and considers that improved certainty of how each asset will respond to dispatches can support efficiency in dispatching for system requirements. Analysis has also shown that, in some cases, there is a difference in submitted versus historical average ramp rates, which highlights the importance of defining requirements that allow the AESO to verify that actual ramp rates remain close to the ramp rates declared by pool participants. Page 30 of 35

129 Options that may be considered to improve dispatch certainty and subject to further consultation include: Ramp by block: this would allow participants to submit a different ramp rate for each block in their energy market offer to more accurately reflect the assets capability at each MW level; and Distribution based tolerance rule: ramp rates would be determined based on the relevant operating state of the asset (i.e., from cold, from MSG, from hot). Tolerances could be established at each state, providing a more clear association between operating state and the associated ramp characteristics of that state. Improving dispatch certainty is both a current proposal and a roadmap item that may be implemented earlier than the other items on the roadmap. Dispatch tools to better analyze ramp expectations are being explored further as roadmap items; these tools would provide the AESO with enhanced ability to assess and anticipate ramp expectations. Ancillary services products are being explored as an option, but from the AESO s preliminary assessment, may not be required at this time. One ancillary services option being explored is a ramp product, as described in next subsection. Ramping product (b) The AESO currently uses regulating reserve (RR) to manage the moment to moment imbalances in supply and demand, and additional volume of regulating reserve is procured during superpeak hours to manage ramp requirements during predetermined superpeak periods. A ramp product is under consideration; however the need for creating a separate ramp product to manage increased flexibility requires further exploration. The design of the ramp product, procurement options, and use of a ramp product, in comparison to other mechanism to manage flexibility, requires further analysis that may be considered upon the occurrence of a set of predetermined conditions, or trigger event such as reliability or operational issues created by the lack of flexibility. The AESO considers it a priority to ensure that the overall system is flexible enough to manage variability instead of trying to carve off a ramp product that exclusively manages the magnitude of the ramp. The ramp product will be considered as part of upcoming consultations. Shorter settlement (c) The AESO has explored the concept of shortening the settlement interval from a pool price calculated as an hourly average of 60 system marginal prices to a 15 minute calculation. The AESO has performed preliminary analysis using historical data that compared unit revenue from a 60 minute settlement model to a 15 minute settlement model. 9 This analysis found that shortening the settlement interval to 15 minutes would provide a financial incentive to: respond faster to dispatch instructions; reduce load in response to high pool prices; and change overall revenues for different asset types (i.e., change the investment price signal). In general, shortening the settlement interval to 15 minutes will improve price fidelity as the settlement price will be closer to the value of the energy at the time when it is needed and may provide financial incentives for market participants to respond more quickly to dispatches. 9 Page 31 of 35

130 Any continued analysis related to shorter settlement would include an examination of the forward looking impact of shorter settlement with increasing variability to assess value. Examination of the shorter settlement will continue and be assessed as part of the roadmap Out of scope reforms in the EAS markets The AESO has determined that the following design changes will not be included as part of the capacity market design or market roadmap, although they may be considered as part of a separate evaluation at another time as the need arises. Locational marginal pricing (a) With the current policy related to unconstrained and recent system build out, pricing on the transmission grid is not required at this time. Security constrained unit commitment (b) The current self-commitment model continues to work in Alberta, given that directives issued for reliability reasons are rare. This means that assets have been able to respond to market signals to self-position their assets in the self-commitment model. Additionally, as shown below, the centralized commitment model yields similar results as the self-commitment model in terms of efficiency. Given that the centralized commitment model, in comparison to the self-commitment model, costs more to consumers for the same energy delivery by shifting the risk from generators to loads (through uplift payments), it was determined that the self-commitment model should continue in Alberta. An asset owner is in the best position to manage its commitment decisions. Accordingly, there is no need or value to shift away from the self-commitment model. Self-commitment is supported by the following incentives and requirements: All resources have the proper market incentives to position their assets to deliver energy and manage their own operational cost. Capacity resources have additional financial incentive to be available during system stressed periods in order to avoid performance payment adjustments and earn marketbased revenues. The AESO publishes the short-term adequacy assessment report to provide a signal to market on supply tightness. Directives have not been required for reliability reasons; however, if required, the AESO may direct long lead time assets to start up to provide energy. If required, the AESO may direct resources to start up, and to provide energy for reliability reasons through reliability unit commitment rules. The metrics for system controller intervention will be reviewed to ensure they do not interfere with the incentives required to self-commit. The AESO net demand variability study indicated increased cycling of large commitment units (300 MW, and larger) from current level of about five on/off starts per unit in 2017 to about 50 on/off starts per unit in 2020/2022, and to about 90 on/off starts per unit in 2030 under the current market model. However, all scenarios tested show sufficient flexibility provided by assets assumed to be part of the overall system fleet in order to manage this flexibility within current market requirements. Additional rule changes are under consideration to provide incentives and value for flexibility which in concert with current requirements will improve the ability of the self-commitment model to continue even with future variability changes on the grid. The AESO commitment modelling study indicated potential but limited efficiency gains from switching to a centralized commitment model. This study showed that assets with a forward view of the market will self-position their assets to respond to expected changes. Page 32 of 35

131 The modelling showed a reduction in unit starts in the centralized commitment model compared to the self-commitment model. Analysis results showed that production cost in the self-commitment model was higher than the centralized commitment model by approximately 6% higher in later years (mid 2020s). See Figure 12 below. Figure 12 Production (input) cost Further analysis of the 2025 period indicated that the estimated increase in production costs in the self-commitment model was related to lower efficient units, such as converted coal to gas units, staying online because of higher start costs and unit operational characteristics. These units displaced more efficient units, resulting in higher fuel, emission, and start-up costs. As such, the self-commitment model has higher costs related to lower efficiency commitment units producing more energy. However, these costs remained with the generator owner based on their decisions and were not assigned to loads through uplift in a centralized model. Figure 10 below shows the input cost per units modelled to be online in 2025 and their contribution to overall production costs. Figure production (input) cost by technology The modelling illustrates that the self-commitment model can continue to support reliability objectives as participants will manage their assets to stay online, even during variability. Additionally, the AESO will continue to have the ability to direct units for reliability unit commitment, and long lead time assets as required. Further analysis may be required to ensure the compensation rules provide the right incentives for accepting dispatches when they occur. The AESO will continue to monitor the impact of NDV on the fleet and continue to assess whether the requirements are sufficient in the longer term. Moving to a centralized commitment model will require a more complicated bid offer structure (i.e., a three part bid model that requires prices for startup, minimum run and no load in addition Page 33 of 35