Memorandum on Developing Cost of New Entry (CONE) Study

Transcription

1 Memorandum on Developing Cost of New Entry (CONE) Study Choice of Technology There is no theory on the choice of technology for the CONE. But PJM, NYISO, ISO-NE adopted the natural gas combustion turbine (CT) as a matter of historical practice. The historical practice evolved out the idea that a load serving utility, if it was short of meeting its obligation to hold sufficient resources to meet the installed reserve margin requirement (IRM), would presumably have no other options of supply available but to build something: no other generation resources available, no demand-side management (DSM). The idea that the quickest and easiest ad lowest capital cost option would be to build a CT. Moreover, the levelized cost of a CT would be the penalty for failing to meet the IRM target and would be equivalent to a self-build option. Another way of viewing the choice of technology is it is a benchmark against which new entry can take place. Historically and incorrectly, many generation owners have viewed the CONE, and Net CONE (CONE minus energy and ancillary service revenues) as benchmark by which new entry should or should not take place. The better way of looking at CONE and Net CONE is a benchmark to beat for new entry. If new entrants have a lower cost, more efficient technology, they will enter the market at prices below Net CONE. What CT would be appropriate for Alberta? Since Alberta is a relatively small market with a nearly 12,000 MW peak and high load factor, taking the CT choices from NE US RTO markets would not be appropriate as such a choice would constitute such a large addition to the AESO market that it would create a potential huge over-supply. For reference, PJM and NYISO have recently used a GE 7FA.05 vintage CT with 2 CTs being built as part of the facility to take advantage of economies of scale in building the CT facility which implies a lower cost in $/kw installed. So, smaller CTs such as aero-derivative CTs like the GE LM6000 or the GE LMS 100 or smaller turbine sets offered by Siemens or Mitsubishi (mention of these vendors in no way constitutes endorsement) may be the better way to go and would not overwhelm the supply of the AESO market. Is a combined cycle configuration appropriate for CONE? The answer to this question depends on the philosophy AESO wishes to take. As a last resort technology, it may not make sense as discussed above. But if one wants to examine the lowest cost generation technology, or go with what is actually being installed on a general basis, this could make sense. ISO-NE has opted for combined cycle based upon what has been installed, and as it turns out, it is a more expensive Net CONE than the CT in their market context. In either case, it may be worthwhile as part of the CONE study to examine the cost of a combined cycle configuration that goes with the choice of CT to get an estimate of the cost and Net CONE impact as PJM has done previously. In PJM s case, the combined cycle configuration was used as part of implementing the Minimum Offer Pricing Rule (MOPR), but this is likely not relevant for the AESO market.

2 Other generator Technology Considerations There are a variety of other considerations for the CONE technology that should be examined possibly included. These are the main issues faced by PJM going through multiple CONE studies over the past decade, though they may not be exhaustive. 1. Power Augmentation: This is a way of increasing power output through water/steam injection to keep the turbines cooler to increase output. In the context of Alberta, this is not a problem in the winter, but could add needed capacity in the summer. 2. Environmental controls: Will the new facility need an SCR for NOx controls, or water injection systems if it is a CT running on diesel (#2 fuel oil)? This can be answered through knowledge of whether the air permit would require such controls and the choice of technology and the NOx emissions with only low NOx burners. 3. Cooling system: Most generators use water to cool their turbines, but in some areas, water may be difficult to access or there are environmental restrictions. If water cooling is to be used, will it be once through cooling (open loop), or closed loop cooling (use of towers to cool water and reuse it again). In some areas, air cooling is now being used due to water restrictions. 4. Back-up Fuel: In the PJM, NYISO, and ISO-NE contexts, natural gas is often much more expensive or hard to get during the winter heating season. In this case, many units have backup fuel (#2 fuel oil) stored on-site to run in case they cannot get gas. It is not clear this is an issue in Alberta, or could become one, but it is yet another consideration for thinking about the characteristics of the CONE resource. Generator Siting This can become an issue as not all potential sites are created equally. Ideally, a good site for a new resource would be close to a gas pipeline, robust electric interconnection, and have water for cooling. It is not clear whether all such sites have already been taken by incumbents or other possible ideal sites remain. If there are a variety of different sites, they all are likely to have differences in property taxes, water rates, and the price of land could also be quite different. Moreover, the cost of generation interconnection and building a gas pipeline lateral may also differ significantly by site. The CONE study should examine a variety of sites and examine all these features, and rather than choosing a single site, maybe blending the characteristics of multiple sites to come up with an average or representative site. Cost of Labor and Materials This is an issue that has come up in CONE studies historically. And generally, with arguments over special local needs and considerations. It is essential the CONE study account for such regional differences that would apply to Alberta, that may not apply generally across Canada. For example, many trades that would be involved in building new power plants, are also involved in the oil and gas supply chain in some manner, thus leading to potentially higher labor costs. The same may also be true of some material costs like concrete and steel. But while accounting for local factors is critical, the numbers cannot just be made up with some sort of documentation that cannot be verified by the AESO staff, AUC, or stakeholders. A good check in the US has been the government statistical databases that collect very specific information on labor and

3 material costs. The same is true in Canada through the use of Statistics Canada information to corroborate the labor and material costs or to find ways to adjust these costs. Financial Assumptions The cost of capital and terms of financing are large factors in coming up with the CONE. The higher the cost of capital, all else equal, the higher is the CONE value. The longer is the term of financing, the lower is the levelized CONE value. Also, the longer is the assumed economic life of the project, or the number of years over which costs can be recovered, the lower is the levelized CONE value. Cost of Capital The cost of capital will depend upon the cost of equity and the cost of debt, and is weighted by the relative shares of equity and debt. This is the weighted average cost of capital (WACC). The relative cost of equity and debt will depend upon whether the CONE technology is being balance sheet financed or non-recourse financed. Generally speaking, balance sheet financing relies upon the capital and the entire balance sheet of the parent company to back up the project, and may be seen as lower risk and may have a lower cost of capital. Conversely, non-resource, merchant financing is seen as riskier and is seen to have a higher cost of capital to make a project happen. Often merchant financing usually uses a debt/equity ratio of anywhere from 40/60 to 60/40 Given the lack of merchant financing in Alberta to date, getting a good handle on what this cost of capital will be is not an easy task. Cost of Equity To date, CONE studies in PJM have used the Capital Asset Pricing Model (CAPM) to determine the cost of equity. This model requires information on the risk-free rate of interest (usually a government bond), the risk premium, beta, and the historic equity premium over debt. Fortunately, much of this information is available. The cost of Canadian government debt is well know, the historic equity risk premium is known for many advanced economies (4-6%), and the benchmark beta is equal to 1 based on a portfolio of stocks. The hard part will be determining beta for balance sheet financed projects (likely less than 1) versus merchant developer project (likely greater than 1). But in thinking about the cost of equity, it is essential to consider that capital will flow to where it believes it can get the best risk-adjusted returns. Capital does not abide by sectors or international boundaries. In the US markets, there has been a large flow of capital from other parts of the world (Japan, Korea, Australia, New Zealand to name some of the largest sources) looking for returns higher they can get at home. From a new entry perspective, this has reduced the cost of equity capital, and as such should be considered as possible in the Alberta context. Empirically, the cost of equity has been taken from publicly traded independent power producers (of which there are very few today in the US, and no true IPPs in Alberta). This will be an issue as the only data point for the Alberta market are integrated companies that finance on balance sheet. So there may need to be some creative ways to find empirical evidence to support the number chosen for merchant, non-recourse financing project.

4 Cost of Debt The risk-free rate of debt is well known, but the cost of debt for a new balance sheet financed project versus a merchant project can differ substantially. The cost of debt on balance sheet financing will depend in the credit rating of the parent company, but is likely relatively low compared to merchant financing. Merchant debt financing in the US context has come through the Term Loan B markets, with consortia of banks competing for providing debt capital to various projects. This has resulted in a reduction in the cost of debt overall from a new entry perspective. Additionally, merchant developers are utilizing financial hedging strategies such as revenue put options that look like mortgage insurance home buyers often pay. The revenue put is purchased by the developer up front, adding to the initial cost of the project, but in return get lower debt rates while still being able to receive the market upside of the project. Such instruments have not been considered in CONE studies previously, but should be considered as a financing option available for the CONE technology. Empirically, the cost of debt has been estimated by examining publicly traded independent power producers (of which there are very few today in the US, and no true IPPs in Alberta). This will be an issue as the only data point for the Alberta market are integrated companies that finance on balance sheet. So there may need to be some creative ways to find empirical evidence to support the number chosen for merchant, non-recourse financing project. Economic Life of the Project In the US market context, the economic life of the project, or the number of years over which the project is allowed to recover its costs has been 20 years. This is certainly at the lower end of the actual operating life of CTs and combined cycle facilities. But in the US context, it also matches the allowed depreciation of this class of asset. If Canadian tax law allows for depreciation of a longer or shorter period for generation projects of this kind, there should be consideration of using the Canadian tax treatment. Beyond tax treatment, there is also the issue of how long the resource is likely to remain in service. In the PJM discussions with stakeholders, some load groups want a 30-year life which would reduce the levelized cost of the CONE resource, while merchant developers would prefer a shorter recovery period that would raise the CONE value. Levelization Method

5 Translating project investment and fixed operations and maintenance costs for new generation resources over the expected economic life of the resource into a levelized annual cost is standard practice in the utility industry. The levelized annual cost provides information to the project developer, regulators, and counterparties concerning the constant stream of revenues needed each year to cover the cost of the project including returns on capital. That constant stream of payments can be expressed in either real or nominal terms. As shown in the figure above, expressing the constant stream of payments in nominal terms, ( nominal levelized ), means that the payment in each year is the same regardless of inflation. Under nominal levelized, the project developer would receive the same dollar amount (e.g.$120,000/mw-year) in each year over the life of the project regardless of the assumed rate of inflation over the life of the project. Expressing the constant stream of payments in real terms ( real levelized ), means that the payment each year, while the same on an inflation-adjusted basis, increases each year over the life of the project by the rate of inflation. The contrast between using nominal or real levelized values is that initially, the CONE value will be different in the early years. And if the CONE values are reset periodically, the real value will likely remain below the nominal value. Clearly, capacity buyers like to see the real levelized value because it is lower cost while capacity sellers like to see the nominal levelized value. For any given assumed rate of inflation, the present value of the stream of payments under either nominal levelized or real levelized is exactly the same. What differs is the trajectory of the payments in nominal terms. Under nominal levelized cost recovery, the payments made in the early years are greater than the payments in the early years under real levelized cost recovery. However, in the later years the nominal levelized payments are less than the real levelized payments.

6 In the end, the choice of using nominal or real levelized values will depend upon the way risk is viewed in developing the capacity market, and whether supply adequacy would be best served by nominal or real levelized CONE values. Arguments have been made in PJM that supply adequacy would be best assured with nominal levelized CONE since the higher nominal value provides a greater incentive to invest upfront by reducing the value of the real option to wait for more complete information before making the investment decision.