APPENDIX B: PORTFOLIO OPTIMIZATION MODEL

Transcription

1 APPENDIX B: PORTFOLIO OPTIMIZATION MODEL PUBLIC UTILITY DISTRICT #1 OF SNOHOMISH COUNTY Prepared by Generation, Power, Rates, and Transmission Management Division Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B1

2 PORTFOLIO OPTIMIZATION MODEL Appendix B provides further information on the Portfolio Optimization model used in the IRP, as well as the substantive assumptions used to inform inputs that the model uses. The content is organized as follows: 1. Methodology 2. Model Inputs: a. SupplySide Assumptions b. Demand Response Resource Assumptions 1. Portfolio Optimization Model Methodology The 2017 IRP uses a Portfolio Optimizer tool developed inhouse to solve for the least cost portfolio that satisfies PUDdefined Planning Standards in a given scenario. The Optimizer requires as inputs, the outputs of the Probabilistic Load Resource Balance Model 1. Specifically, the Probabilistic Load Resource Balance Model outputs that describe the performance of existing resources within the existing portfolio, customer load, and the resulting possible Load Resource Balance positions at different likelihoods and points in time across the study period. The Portfolio Optimizer model uses those inputs and combines it with information about potential demandside and supplyside resources that could be added to a portfolio and then calculates the following attributes of possible new portfolio combinations: The ability to meet PUD Planning standards for resource adequacy on an energy and capacity basis; The ability to meet Renewable Portfolio Standard (RPS) compliance obligations under the target compliance methodology; The incremental carbon dioxide air emissions associated with generation from new supplyside portfolio additions; and 1 Described in detail in Appendix A Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B2

3 The cost of adding new resources, the value of any resulting surplus energy, and the Net Present Value of the resulting portfolio. The purpose of this section is to outline the Portfolio Optimizer tool functions, the assumptions it incorporates, and its role in the analytical framework that resulted in the PUD s Long Term Resource Strategy. As described in other portions of this IRP document, the general analytical process for determining the PUD s Long Term Resource Strategy, is described by Figure 1. Figure B1 Application of Planning Standards in Portfolio Optimization. The Portfolio Optimizer meets planning standards in two ways: it imports data from the Probabilistic Load Resource Balance specific to the Planning Standards, and it creates rules that each portfolio must satisfy in order to be considered a valid portfolio that ultimately gets ranked by Net Present Value to determine the lowest Net Present Value (or least cost) Portfolio. In some scenarios, the existing portfolio of resources are sufficient to meet some Planning Standards in some time periods without any additional resources for that time period. An example of this is Renewable Portfolio Standard compliance, which is satisfied in the first Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B3

4 several years of most scenarios by the renewable generation attributes of resources in the existing portfolio. In other time periods, additional resources are needed to augment the existing resources in the portfolio in order to meet Planning Standards. For example, in all scenarios, there is a need for some form of winter peaking capacity resource beyond what the portfolio of existing resources contains. The Optimizer model requires a potential portfolio to add enough resources to satisfy all Planning Standards in all time periods, or be deemed an invalid portfolio. Figure B2 shows how each of the planning standards are incorporated into the Portfolio Optimizer. Planning Standard Annual Resource Adequacy Annual 0aMW after any needed portfolio additions. No forecast market purchases allowed to meet standard Monthly HLH Resource Adequacy Monthly HLH LRB@P5 0aMW after any needed portfolio additions, Forecast Market Purchases to Address Monthly HLH LRB Deficit 100aMW Monthly Peak Week HLH Resource Adequacy Monthly Peak Week LRB@P5 0aMW after any needed portfolio additions, Forecast Market Purchases to Address Monthly Peak Week HLH LRB Deficit 200aMW RPS Compliant Portfolio Valid portfolio s will meet RPS compliance obligations under the Target methodology by some combination of procuring renewable resources with REC generating attributes or purchasing unbundled RECs, to augment RECs produced by the existing portfolio. Figure B2 Probabilistic Model Data Inputs Annual LRB@P50 of the existing portfolio before new additions for all years Monthly HLH LRB@P5 of the existing portfolio before new additions for 4 indicator months per year (Dec, March, April, August) for all years ( ) Monthly Peak Week LRB@P5 of the existing portfolio before new additions for the peak week period of 4 indicator months per year (Dec, March, April, August) for all years ( ) Existing annual portfolio is imported from the Probabilistic LRB model for eligible resources. Annual is also imported from the LRB model to set the annual RPS target in MWh for each year Optimization Constraints (Model Rules) Rule will mark as invalid any portfolios with Annual LRB s less than 0 amw in any year, after portfolio additions. Rule will mark as invalid any portfolios with Monthly HLH LRB s less than 100 amw in any month and any year, after portfolio additions (forecast market purchases allowed to cover LRB deficits between 0aMW to 100aMW). Rule will mark as invalid any portfolios with Monthly Peak Week HLH LRB s less than 200 amw in any month and any year, after portfolio additions, (forecast market purchases allowed to cover LRB deficits between 0aMW to 200aMW). Rule will not mark as invalid any portfolios with insufficient RECs (either through portfolio resources or unbundled REC purchases) to meet the RPS target in any given year Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B4

5 How the Portfolio Optimizer Works. Broadly speaking, the portfolio optimizer works from a starting point of the Load Resource Balance of the existing portfolio at a given performance likelihood (ex: Monthly HLH P5), that then adds and evaluates different combinations of possible resources to satisfy Planning Standards. The Optimizer calculates the incremental cost of each resulting portfolio, the incremental carbon emissions associated with each portfolio, and the optimal combination of renewable resources and unbundled RECs to meet RPS requirements. For example, a given scenario may have an annual energy need of 15aMW in the years that needs be fulfilled in order to satisfy the planning standard that the Annual Load Resource balance be equal to or greater than 0aMW. When the optimization runs, the Optimizer will cycle through all of the possible resources from the resource menu, including both supply and demand side resources, and evaluate them at different possible delivery dates (a 15aMW biomass delivered in the year 2027 for example). As the optimizer evaluates different resources at different delivery dates, it determines whether the planning standard has been met, and what the NPV of the resulting portfolio will be. In the given example, it would continue to optimize by trying different combinations of resources until it has found the most costeffective way to meet the annual energy need of 15aMW. In the actual model, all planning standards and portfolio needs are simultaneously considered across the study period. Figure 2 shows an example of how the NPV of the portfolio is improved over the course of optimization by finding incrementally better combinations to address portfolio needs. In the case of the pictured optimization, the original NPV of the portfolio was $3.15 billion, but was reduced to $2.37 billion after the optimizer had cycled through almost 50,000 iterations of possible portfolio combinations. Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B5

6 Figure B3 An internal algorithm helps the Optimizer efficiently select portfolio components based on previous iterations and their success in helping a portfolio either lower portfolio costs or satisfy a component of planning standard requirements. Each Optimization typically requires tens of thousands of iterations to solve, and in most cases more than 100,000 iterations are required. The optimizer algorithm helps this process to be completed in a timely fashion most scenarios can be solved with an optimal portfolio in around 2 hours. Staff considers an optimization complete when after an optimization run has found a solution, a subsequent run of the same optimization cannot find a solution with a lower NPV. Model assumptions that affect calculated Portfolio Resource Adequacy As the Portfolio Optimizer runs, one of its primary functions is to determine whether a potential portfolio satisfies resource adequacy planning standards. These planning standards for annual energy, Monthly HLH capacity, and Monthly Peak Week HLH capacity represent the PUD s maximum willingness to be short the resources needed to serve demand, and be exposed to Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B6

7 wholesale market prices to serve demand over the study period. These planning standards are given in Figure B2. The determination of portfolio adequacy is based in large part on assumptions of existing resource generation, and assumptions of generation of potential resource additions. Existing resource generation is based upon simulations done in the Probabilistic Load Resource Balance Model for each scenario. This model produces estimates for load, resource generation, and the resulting portfolio position as a variety of likelihood s and time periods. Embedded in that model are the following assumptions: 1. All contracted resources with contracts set to expire over the study period, will expire as specified in the contract without extension. The associated resource generation does not contribute to the PUD s portfolio after that time in the portfolio model. Over the study period, three wind contracts and two smaller biofueled projects are set to expire. 2. The PUD s BPA contract represents a significant proportion of its resource portfolio, and the current terms of the PUD s agreement are set to expire in While it is possible that terms of that agreement could change post2028, the IRP models that in the period, the BPA contract functions as if under the existing terms of the agreement. There was not sufficient information available to assume any specific deviations from the terms of the current agreement. 3. With the exception of the Climate Change scenario, which includes anticipated changes in local and regional hydrology due to climate change, existing resource generation is based on probabilistic simulations of their historic generation under current operating parameters. As a result, an underlying assumption of the existing portfolio load resource balance used by the Portfolio optimizer is that resources perform within the bounds of their historical production and operating constraints of those resources don t change dramatically. Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B7

8 2. SupplySide Resource Assumptions The Portfolio Optimization Model considers supplyside resources and demandside resources in an integrated portfolio approach to arrive at candidate portfolios for each scenario. The purpose of this section is to outline how supplyside resources were modelled for consideration in the Portfolio Optimization Model. The Portfolio Optimization Model s purpose is to find the most cost effective resource combination to meet portfolio needs in a scenario, such that all Planning Standards are satisfied. Planning standards measure energy, capacity and Renewable Portfolio Standard compliance and as a result these attributes need to be assigned to potential resources, along with their costs of acquisition. Annual Energy Attributes Annual Energy attributes for supplyside resources were given by the equation: Annual Energy (MW)= Resource Nameplate (MW)*Capacity Factor@P50 Generation(%)*Annual Dispatch(%) Figure B4 lists supply side resources options, their nameplate, capacity factor, annual dispatch and annual energy attributes. Note that some storage supplyside resources that transfer energy from one time period to another have no net annual energy attributes (marked with an * below). Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B8

9 Figure B4 Annual Energy Attributes of Supply Side Resources Resource Nameplate (MW) Effective Capacity Factor Assumed Dispatch Rate or Equivalent Avg Annual Energy Simple Cycle Combustion Turbine (SCCT) % 10% MW Short Term Capacity Contract (5 year) % 10% 2.5 Dual Fuel Reciprocating Engine % 10% 5.0 Pumped Storage Hydro Low 100 N/A 18% Pumped Storage Hydro High 100 N/A 18% Landfill Gas 10 85% 100% 8.5 Biomass 15 85% 100% 12.8 Geothermal (traditional) 25 90% 100% 22.5 Energy Storage Battery 25 N/A 100% Long Distance Wind (Montana) 50 44% 100% 21.8 Run of River Hydro (small hydro) 30 45% 100% 13.6 WA/OR Wind* 50 35% 100% 17.5 Customer Owned DG 15 11% 100% 1.7 Utility Scale Solar (U/S Solar E Wash) 25 27% 100% 6.8 Utility Scale Solar (U/S Solar W Wash) 5 13% 100% 0.6 Monthly Capacity Attributes Monthly capacity attributes during Monthly OnPeak hours for supplyside resources were given by the equation: HLH Capacity (MW)= Resource Nameplate (MW)*Capacity Factor@P5 Generation(%)*Annual Dispatch(%) The Monthly Capacity equation differs from the Annual Energy equation only by the assumed generation during the Monthly HLH hours. In order to match the Planning Standard conditions of a PUD Load Resource Balance of P5 (the Load Resource Balance that would be exceeded 19 out of 20 years), Monthly HLH generation is measured as would be expected in at P5 generation conditions for the resource. Figure B5 lists supply side resources options and their nameplate, capacity factor, annual dispatch and December HLH capacity attributes. Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B9

10 Figure B5 Monthly Capacity Attributes of Supply Side Resources for December On Peak (HLH) Assumed Effective Dec Nameplate Dispatch Resource Capacity HLH (MW) Rate or Factor Capacity Equivalent Simple Cycle Combustion Turbine (SCCT) % 97% MW Short Term Capacity Contract (5 year) % 10% 2.5 Dual Fuel Reciprocating Engine % 97% 48.5 Pumped Storage Hydro Low % 100% 95.0 Pumped Storage Hydro High % 100% 95.0 Landfill Gas 10 85% 100% 8.5 Biomass 15 85% 100% 12.8 Geothermal (traditional) 25 90% 100% 22.5 Energy Storage Battery 25 50% 100% 12.5 Long Distance Wind (Montana) 50 23% 100% 11.6 Run of River Hydro (small hydro) 30 14% 100% 4.3 WA/OR Wind* 50 12% 100% 5.9 Customer Owned DG 15 3% 100% 0.4 Utility Scale Solar (U/S Solar E Wash) 25 3% 100% 0.8 Utility Scale Solar (U/S Solar W Wash) 5 1% 100% 0.1 Peak Week Capacity Attributes Monthly Peak Week capacity attributes during Monthly Peak Week hours for supplyside resources were given by the equation: Peak Week HLH Capacity (MW)= Resource Nameplate (MW)*Capacity Factor@P5 Generation(%)*Annual Dispatch(%) The Monthly Peak Week Capacity equation differs from the Monthly Capacity equation only by the assumed generation during Peak Week hours. In order to match the Planning Standard conditions of a PUD Load Resource Balance of P5 (the Load Resource Balance that would be exceeded 19 out of 20 years), Peak Week HLH generation is measured as would be expected in Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B10

11 at P5 generation conditions for the resource during the Peak Week time period. Figure B6 lists Supply Side Resources evaluated, their nameplate, capacity factor, annual dispatch and December Peak Week (on peak) capacity attributes. Figure B6 Monthly Peak Week Capacity Attributes of Supply Side Resources December Peak Week (MonFri, 6x16) Resource Nameplate (MW) Effective Capacity Factor Assumed Dispatch Rate or Equivalent Dec Peak Week Capacity Simple Cycle Combustion Turbine (SCCT) % 97% MW Short Term Capacity Contract (5 year) % 10% 2.5 Dual Fuel Reciprocating Engine % 97% 48.5 Pumped Storage Hydro Low % 100% 97.0 Pumped Storage Hydro High % 100% 97.0 Landfill Gas 10 97% 100% 9.7 Biomass 15 97% 100% 14.6 Geothermal (traditional) 25 97% 100% 24.3 Energy Storage Battery 25 50% 100% 12.5 Long Distance Wind (Montana) 50 23% 100% 11.6 Run of River Hydro (small hydro) 30 14% 100% 4.3 WA/OR Wind* 50 3% 100% 1.7 Customer Owned DG 15 1% 100% 0.2 Utility Scale Solar (U/S Solar E Wash) 25 2% 100% 0.4 Utility Scale Solar (U/S Solar W Wash) 5 1% 100% 0.0 Renewable Portfolio Standard Compliance Attributes Renewable Portfolio attributes for supplyside resources were measured in Renewable Energy Credit MW and given by the equation: REC (MW)= Resource Nameplate (MW)*Annual Capacity Factor@P50 Generation(%)*Annual Dispatch(%)*REC Multiplier(0 if Ineligible, 1 if Eligible for 100% of MWhs, 2 if Eligible for MWh*2) Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B11

12 Figure B7 lists supply side resource options, their Annual Energy, REC multiplier, and REC MW attributes: Figure B7 Annual Renewable Energy Credit Attributes of Supply Side Resources Avg REC Annual Resource Annual Multiplier REC MW Energy Simple Cycle Combustion Turbine (SCCT) 25 MW Short Term Capacity Contract (5 year) Dual Fuel Reciprocating Engine 23.9 Pumped Storage Hydro Low Pumped Storage Hydro High Landfill Gas Biomass Geothermal (traditional) Energy Storage Battery Long Distance Wind (Montana) Run of River Hydro (small hydro) WA/OR Wind* Customer Owned DG Utility Scale Solar (U/S Solar E Wash) Utility Scale Solar (U/S Solar W Wash) The following describes the way environmental attributes or renewable energy credits (RECs), associated with a renewable resource, were modeled and made available in the 2017 IRP analysis and satisfy the PUD s annual EIA renewables compliance requirement. Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B12

13 The environmental attributes or RECs associated with energy produced by a Washington state eligible renewable resource can be purchased separately from the energy itself. The assumption for unbundled RECs is that the seller of the REC owns or contracts for the renewable resource and may have RECs surplus to their own compliance need or are trying to maximize the revenue from the energy and REC streams for their project portfolio. Today, the Northwest has a reasonably liquid bilateral market for unbundled RECs, with REC prices forecast for the 2018 through 2022 period near $5 per REC as shown in the table below: Anticipated Average REC Cost $2.00 $ 3.00 $ 5.00 $ 5.00 $ 5.00 Staff anticipate that as renewables compliance requirements in Washington state increase from 9% to 15% of load in 2020, and both California and Oregon experience increases in their annual renewables requirements, the availability of surplus RECs in the region will diminish, increasing the cost of unbundled RECs beyond the shortterm REC market s price range today. Because the long term market for RECs in the 2020 s through 2037 is less liquid and price discovery is limited, the PUD chose to be more conservative and not forecast future compliance RECs prices that reflect near term prices across the 20 year IRP study period. REC prices were instead modeled to address the question if REC prices over the 20 year IRP study period were modeled at their theoretical maximum, would the PUD have a preference to purchase unbundled RECs or to invest in a renewable resource (which includes energy, capacity and REC characteristics) as a future generating asset? The addition of unbundled RECs at the maximum REC price was found to be a more optimal solution in the scenarios and sensitivities examined, except in the High Growth scenario; the High Growth scenario had annual energy portfolio needs. The High Growth scenario met the renewables requirement with a combination of renewable generating assets and unbundled REC contracts. Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B13

14 Maximum unbundled REC prices were modeled at the cost to develop a new Northwest wind project, where project development costs were reduced by the value of the project s energy sold at forecast market prices over a 25 year asset life. 2 The remaining value was allocated to the environmental attributes or RECs associated with the project s annual production. The Business As Usual (BAU) case was found to produce the highest unbundled REC prices, and as a result was used as the REC price for all scenarios. A sample set of years showing the REC price calculation components and REC prices used are given in the table below Nominal Wind Cost per MWh for a New WA Wind Project Delivered in the Calendar Year listed $70.75 $ $ $ $ $ Nominal Energy Value $ $ $ $ $ $ Nominal Fundamental REC Price $ $ $ $ $ $ The volume of unbundled RECs assumed to be available from the market were modelled as 100,000 MWh increments available under a 25 year contract term, with a modeling limitation that a maximum of 1,000,000 MWh of unbundled RECs could be accumulated during any calendar year. In every scenario but the High Case, it was more cost effective to meet annual renewables requirement using the target methodology with a strategy of at least 95% procured unbundled RECs. Figure B8 shows a comparison of unbundled REC purchases and new renewable resource generation, expressed in the 20year average annual amw of REC generation. The smaller renewable resources contributing to the Business As Usual w/ CA Carbon and Climate Change w/ Low Societal Cost of Carbon Cases were small, local solar installations, and customer owned distributed generation resources respectively 3. The Long Term Resource Strategy 2 New wind development costs did not assume extension of the Federal production tax credit beyond These resources were likely selected as a result of the modeling construct that unbundled REC contracts could only be purchased in 100,000 MWh contract increments, as each resource contributes to addressing a remaining need after a REC contract add, and the PUD has no annual energy need for the energy they provide in these scenarios. Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B14

15 20Year Average amw purchased ( Climate Change with Low Societal Cost of Carbon ) anticipates meeting about 95% of its RPS compliance needs with unbundled RECs over the study period. Figure B8 Comparison of Renewables Requirement Portfolio Additions by Scenario Year New Renewable Resource Generation RECs (in average amw) 20Year Unbundled REC Purchases (in average amw) Low Load Growth Case Business As Usual w/ca Carbon Business As Usual w/no Carbon Climate Change w/ Low Societal Cost of Carbon High Load Growth Case As a result of the more conservative approach to the fundamental price of RECs in the IRP analysis, the cost of procuring RECs is a large proportion of the overall cost of meeting future needs. The cost of procuring RECs at the fundamental REC price in the Long Term Resource Strategy represented 50% of the total portfolio costs, but was still more cost effective than procuring an equivalent amount of bundled energy and RECs through new renewable resource additions, since the majority of the resource s output would have been surplus to the PUD s annual energy needs, and not sufficient to address the seasonal capacity needs. Figure B9 shows optimized portfolio cost and revenue components at three different REC price levels. Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B15

16 Conservation and capacity resource choices did not change in these optimizations, but the cost of Procuring RECs and the overall Portfolio NPV were significantly reduced. Figure B9 Comparison of Climate Change with Low Societal Cost of Carbon at Various REC Price Levels Fundamental Price per REC Average Price of $15/REC Average Price of $10/REC Cost for New Conservation $ 147,987,664 $ 147,987,664 $ 147,987,664 Cost of New Capacity Additions $ 259,431,989 $ 259,431,989 $ 259,431,989 Cost of New DG Additions $ 34,781,279 $ 34,781,279 $ 34,781,279 Cost of Procured RECs $ 453,201,140 $ 139,842,011 $ 93,228,007 Total Portfolio Costs $ 895,402,072 $ 582,042,943 $ 535,428,939 Less Portfolio Revenues $ 473,243,503 $ 473,243,503 $ 473,243,503 Portfolio NPV $ 422,158,569 $ 108,799,440 $ 62,185,437 Supply Side Resource Costs Supply side resources were priced based upon peer review and inhouse modeling of development and operations costs for different resource types. Annual supply side resource cost components include: Debt service costs associated with developing the project Any associated fuel costs needed to generate electricity (this includes consideration of carbon pricing in fuel costs in scenarios with carbon pricing for fossil fueled resources) Any transmission costs needed to bring the resource s electricity to the PUD busbar Any anticipated variable integration charges associated with adding the resource Any Operation and Maintenance (O&M) costs associated with operation of the resource Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B16

17 Different resources have different lifespans and are available at different times over the course of the IRP study period of For instance, a new Wind project is assumed to have a useful life of 25 years, while a new Pumped Storage Hydro project is assumed to have a useful life of 50 years. Both hypothetical projects could also be delivered in many possible years over the course of the study period. In order for resources to be considered on their overall cost effectiveness within the portfolio, projects were adjusted by delivery date, and end effects were added to compare resource costs over the same period of time. To value resources delivered in different time periods, individual resource cost estimates were developed for each year the resource was available. Fuel forecasts (including carbon pricing forecasts), an assumed cost escalation rate of 2.5% and a discount rate of 5% were used to inform all resource cost estimates. To compare resources with different lifespans, end effects were added to resources with lifespans shorter than the longest possible resource life in the supply side resources studied. The potential resource with the longest useful life in the Supply Side Resource Option list was a Pumped Storage Hydro project delivered in the year With a useful life of 50 years estimated, a Pumped Storage Hydro project delivered in 2037 would last until the year For resources with useful lives shorter than this (including potential Pumped Storage Hydro Projects delivered before the year 2037), an equivalent amount of annual energy to the resource being considered was replaced by market purchases for each year after the resources useful life expired, until For example, a project delivered in 2027 with a useful life of 20 years, and an annual energy output of 10aMW, would have end effects equal to the sum of 10aMW*Market Price*8760 hours of every year from 2047 to In this way, projects of differing lives can be considered on the same basis, and the PUD is less at risk of selecting a supply side resource using only information about its potential shorterterm contributions to the utility. Figure B10 shows the resource cost components, and end effect cost components of a 5MW local solar project delivered at different times across the study period. Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B17

18 Figure B Resource Costs $15,630,864 $12,661,687 $9,950,324 End Effects $2,953,268 $1,991,844 $1,183,006 Total Costs $18,584,132 $14,653,530 $11,133, Demand Response Resource Assumptions The 2017 IRP included consideration of five demand response programs that could be included in candidate portfolios using the integrated portfolio approach. These programs are described in Figure B11 below. This section describes how the Demand Response programs were treated in the Portfolio Optimization Model for inclusion in candidate portfolios. Program Residential Direct Load Control Space and Water Heat Residential Direct Load Control Water Heat Residential Direct Load Control WiFi Thermostat Residential Smart Water Heaters Commercial & Industrial Curtailment Figure B11 Summary of Demand Response Program Options Description Curtails enrolled resident s space and water heaters for timelimited event calls to reduce utility s Peak Loads Curtails enrolled resident s water heaters for timelimited event calls to reduce utility s Peak Loads Curtails enrolled resident s space heaters for timelimited event calls to reduce utility s Peak Loads Allows the utility to program techenabled Water Heaters to primarily heat water during offpeak hours for use in onpeak hours. Not event limited on a monthly or annual basis. Curtails enrolled all or a portion of Commercial and Industrial customer loads during time limited event calls to reduce utility s Peak Loads Demand Response programs are different than both supplyside and demand side resources in a number of important ways. From the PUD s perspective, currently available Demand Response programs provide benefit by temporarily shifting load from one time period to another time period. This is useful because the difference between onpeak hour load and offpeak hour load is substantial, and shifting onpeak load to offpeak hours is one approach that could contribute to meeting onpeak needs during capacity constrained periods. However, unlike conservation and supplyside resources, demand response programs are often limited in the amount of total times Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B18

19 they can be used, and how clustered those uses can be. For example, while an Industrial customer may enroll in a voluntary curtailment program to curtail its load for a total of 120 hours in a year, it is less likely that customer would be willing to curtail all 120 hours from Monday Friday in one specific week, or 500 hours over the course of a year. For this reason, Demand Response programs required a tailored modelling methodology to be considered in the Portfolio Optimization Model. Another characteristic of Demand Response programs is that they don t reduce overall demand like conservation instead they primarily shift demand from one time period to another. The Demand Response programs in the Demand Response Potential Assessment included information on the assumed MW capacity of each program for a peak hour in every year of the study period, and the duration of the call for that particular Demand Response program type. Those details for the year 2037 are provided in Figure B12 below. Note that the Smart Water Heater Program is not call limited and can provide load shifting from onpeak periods to offpeak periods at all times due to its program design. Figure B12 Comparison of Demand Response Program Event limitations Program Peak Hours/Year Avg Event Length 2037 MW Limit (Hours) Achievable DLC WiFi TStat DLC Air & H20 Heat DLC H20 Heat C&I Curtail Smart H20 Heat N/A N/A 34 The 2017 IRP considered the Planning Standards applicable to the Peak Week onpeak hours during a Month, and the OnPeak Hours during the Month. Both of these time horizons (80 hours in a peak week, and at least 372 onpeak hours in a month) are broader than the duration of an average event length for any of the demand response programs considered (typically four). As a Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B19

20 result, some modeling assumptions were needed to measure the potential impact of an available demand response program across the planning standard time periods. The most significant modeling assumption made was that all Demand Response program call events could occur within each annual December Planning Standard period. This assumption supposes that all 50 annual program hours could occur during a December Peak Week, and would therefore also occur during the December OnPeak hour period. While this assumption may not be realistic, it allowed the PUD to see if generic demand response programs could be costeffective under favorable conditions, in order to provide the PUD enough information to determine if it should explore specific deliveries of potential demand response programs. Demand Response Potential Calculation for December Peak Week The formula that provided the potential achievable amw for the December Peak Week period for each demand response program was given by the following equation: Annual Achievable MW* (Peak Hours per Year Limit/Peak Hours in Week[80]) The results of the calculation for each program is provided in Figure 3 below. The values shown represent the contributions to December Peak Week portfolio needs provided by each demand response program in the Portfolio Optimization Model. Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B20

21 amw 250 Figure B13 Achievable Potential of Demand Response Programs for December Peak Week by Year DLC Wifi TStat DLC Air & H20 Heat DLC H20 Heat C&I Curtail Smart H20 Heat Demand Response Potential Calculation for December OnPeak Hours The formula that provided the potential achievable amw for the December OnPeak period for each demand response program was given by the following equation. Available MW*( Available Hours/Peak Hours in Dec HLH[~400]) The results of the calculation for each program is provided in Figure B14 below. The values shown represent the contributions to December OnPeak portfolio needs provided by each demand response program in the Portfolio Optimization Model. Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B21

22 amw Figure B14 Achievable Potential of Demand Response Programs during December OnPeak Hours (HLH) by Year DLC Wifi TStat DLC Air & H20 Heat DLC H20 Heat C&I Curtail Smart H20 Heat Snohomish County PUD DRAFT 2017 Integrated Resource Plan Appendix B B22