360 likes | 621 Views
Net CONE for the ISO-NE Demand Curve. 3 rd Response to Stakeholder Comments and Draft Proposal. NEPOOL Markets Committee. Samuel Newell, Brattle Chris Ungate , Sargent & Lundy. February 27, 2014. Agenda. Responses to Stakeholder Questions and Comments
E N D
Net CONE for the ISO-NE Demand Curve 3rd Response to Stakeholder Comments and Draft Proposal NEPOOL Markets Committee Samuel Newell, Brattle Chris Ungate, Sargent & Lundy February 27, 2014
Agenda • Responses to Stakeholder Questions and Comments • Principles for Selecting the Reference Technology • Capital and FOM Cost Estimates • CONE Calculation • E&AS Revenue Offset • PER/PFP • Review of Reference Technologies • Draft Recommendation • Next Steps
Principles for Selecting a Reference Technology • Objective • Estimate Net CONE that supports prices just high enough to attract sufficient new investment to meet resource adequacy objectives • Criteria for selecting the Reference Technology to meet the objective • Reliably able to help meet load when installed capacity is scarce • Complies with all environmental regulations • Dispatchable technology that is available to generate during system peaks • Likely to be economic • Available as a utility-scale commercial plant • Lowest or near-lowest estimated Net CONE • Demonstrated commercial interest by developers, as evidenced by projects recently completed, under construction, or in the queue in New England or the rest of U.S. • Can estimate Net CONE with low uncertainty • Cost estimates based on established technologies • Less E&AS uncertainty relative to other technologies
Permitting of F-Class Frame CT • Based on stakeholder input, we investigated further whether a frame CT with a lower efficiency (and higher CO2 emissions) relative to aeroderivatives would be able to receive an air permit • As no F-class frame-type CT has been proposed recently in New England, there is no record of a Frame CT being approved or refused an air permit • We discussed permitting a Frame CT with MA and Connecticut environmental officials, who said that the permit would not be denied solely based on technology • Recent Footprint/CLF settlement may raise perceived risks of getting a Frame CT permit in MA, although capacity factor is much lower than CC • Although permitting risks appear to exist (especially in MA), we do not believe that these risks should disqualify the Frame CT as the Reference Technology • There has not yet been a permit refused • The plant could be located in other states
Electrical Interconnection Assumptions • We completed our estimate of electrical interconnection costs • Assume no network upgrades needed, consistent with recent projects • Conducted a bottom-up analysis of direct interconnection costs • Determined that direct interconnection needs would be the same for all reference technologies • Assumptions on direct interconnection needs for all technologies: • 345 kV interconnection sized for 200 – 700 MW generating plant • Point of interconnection is existing 345 kV open air substation • 0.5 mile transmission line between plant and expanded substation • A position is available in the existing switchyard for connection of the line from the plant • Adequate space is available in the existing control building for the additional panels • Battery and charger have sufficient capacity for the additional load • Modifications at two remote switchyards for a new relay panel for line protection • Based on input from ISO-NE, S&L established a list of necessary equipment • Items include disconnect switches, circuit breakers, capacitance voltage transformers, relay and metering panels, steel and supporting structure, foundations, aluminum bus and supports, insulators, power and control cables for connecting the equipment, panel wiring, conduit, high voltage jumpers, hardware, and connections to the ground grid
Electrical Interconnection Costs • S&L estimated the direct interconnection costs • Switchyard modifications at $6.0 million • 0.5 mile transmission line at $1.1 million • Total is $7.1 million for all reference technologies • We added these costs into our analysis, replacing the placeholder values previously included, which had the following impact on CONE
Effects of Greenfield Assumptions • Stakeholders asked how costs would differ for a brownfield unit • We estimate capital costs would be 3-6% lower for a brownfield location by assuming the following infrastructure is in place, requiring no or limited modifications: • Water well infrastructure • Fire pumps • Fire/makeup water storage tanks • CO2 bulk gas storage • Raw water and fire protection piping and piping components • Buildings: Control, Warehouse, Admin • Sanitary sewer system and waste treatment system • Various site works (fencing, roads, etc.) • Actual savings is very site specific—could be lower or higher • However, we continue to assume a greenfield site due to the generic nature of the reference resource and because brownfield sites could be limited
CC Dual-Fuel Specification • Stakeholders asked whether it was necessary for dual-fuel capability to be included on a CC • Ensuring enough guaranteed fuel supply to meet reliability objectives during cold snaps is a major concern for ISO-NE • ISO-NE sponsored a separate analysis which indicated that the proposed PI incentives would justify the costs of dual-fuel capability • We include dual-fuel capability on all plants • The incremental cost is about $17.5 million (2013$) for the CC • Includes equipment, labor, and materials, indirect costs, and fuel inventory • Dual-fuel costs contribute $0.5/kW-mo to the Net CONE
LM6000 Cost Estimate • In response to stakeholder requests, we completed a cost estimate for an LM6000 assuming a 4x0 configuration (174 MW) to take advantage of economies of scale • Cost of many common facilities does not increase or increase significantly with extra units, such as buildings, water treatment, site works, interconnections, etc. • There could also be discounts on higher volume orders of turbines, but our estimate does not include this • The LM6000 has higher Gross CONE than the LMS100 and will be expected to have E&AS margins less than the LMS100 due to its higher heat rate
Comparison to Actual Plant Costs • Based on stakeholder requests, we compared our aeroderivativeCT cost estimates to the actual costs of the turbines in Connecticut’s peaker solicitation • One single unit LMS100 bid @ $1,449/kW • Seven 2-10 unit LM6000 bids ranging from $1,046/kW to $1,292/kW • An apples-apples comparison is not straight-forward, due to: • Escalation to 2013 dollars • Brownfield sites and Number of units • Chillers and Gas/electrical interconnection costs • Lack of detail on equipment pricing, owners cost, fuel inventory, spare parts, working capital, financing fees, etc. • Outside of items for which no detail was available, the biggest reasons for the lower costs of the CT peakers are economies of scale, escalation, and brownfield • After accounting for all of those differences, the adjusted costs are comparable, but we are slightly higher • Hence we reduced some of our soft costs that are based on judgment and calibration
Capital Cost Adjustments • Based on those comparisons, we made the following changes to our capital cost assumptions • Reduced EPC Contingency from 12% to 10% • Reduced Owner’s Costs (Services) from 7% of EPC Costs to 6% • Reduced Owner’s Contingency from 10% to 8% • Adjusted method of escalating costs from 2013$ to 2018$ to better account for when costs are incurred in the drawdown schedule • These changes have the following impacts on Overnight Costs (2018$/kW) and Net CONE (2018$/kW-mo)
Cost of Capital Adjustments • Based on conversations with stakeholders, we adjusted the assumed capital structure to better reflect typical projects and their associated COE and COD • Yet we are maintaining the 8.0% ATWACC we already established using multiple reference points, hence CONE will not change • = DF×COD×(1-T) + (1-DF)×COE • We were also asked to clarify the treatment of taxes corresponding to our cost of capital. We use a very standard approach • Apply the ATWACC to cash flows after deducting corporate income taxes (after accounting for depreciation deductions) • Treat these cash flows as “all-equity” cash flows, such that interest on debt and the debt tax shield are accounted for through the ATWACC, not the cash flows • Our cost of capital will appear lower than equivalent costs expressed in pre-tax terms
Economic Life • Stakeholder asked that we consider a longer economic life for the reference technologies • After reviewing the economic life for calculating Net CONE, we will maintain the assumption at 20 years for all technologies • Reasons for longer economic life • Stakeholders view power generation plants as 30+ year assets • Longer economic life is consistent with our assumptions for O&M costs • No major equipment replacements are required until rotor replacement at 25 years or later (depending on hours of operation) • Reasons for shorter economic life in financial modeling • Market risks, including lower cost capacity resources entering market • Risk of market interventions that depress prices
Impact of “Lumpy” New Entrants • We initially assumed CONE based on a “level-real” assumption (total revenues increase at inflation) • Some stakeholders said CONE should recognize that new entrants will opt for the 5-year lock-in and depress the price other resources receive • To account for the “lumpiness” effect, we estimated future capacity prices as % of Net CONE, assuming: • Lock-in keeps prices constant nominally for 5 years • NICR increases 381 MW/yr due to load growth • New entry from 715 MW CCs when capacity prices rise above CC Net CONE • For an entrant to earn its required return, it would have to offer $0.64/kW-mo above level-real Net CONE • The impact could be smaller if retirements absorb the overhang, if entrants are smaller, or if E&AS revenues increase; but other factors such as energy efficiency or new renewables or transmission could go the other way. Overall, we believe the adder is justified. Cash Flows with Lumpy Investments and 5-yr Lock-In
CC Historical E&AS with Daily Fuel Costs CC Historical E&AS Margins • Stakeholders expressed concerned that our analysis of historical CC E&AS margins included very low or zero margins in Winter 2012/2013 • We revised fuel costs based on daily historical fuel burn and daily Algonquin Citygate (ACG) gas prices instead of monthly averages • We eliminated plants with firm gas arrangements,as their fuel costs likely differ from the daily ACG prices • This decreases projected 2018/2019 E&AS margins by $0.12/kW-mo
CC E&AS Projections with Improved Performance • Based on stakeholder concerns that CC E&AS are too low, we assessed whether we should: • Adjust for heat rate improvement in the technology • Remove any poor performing plants from our sample of representative units • CC heat rates have improved by ~200 Btu/kWh over the past 10 years; including improvement increases CC E&AS margins by $0.43/kW-mo • We removed plants with very high heat rate or low capacity factor, which increases CC E&AS margins by an additional $0.94/kW-mo
Comparison to PJM E&AS Margins Monthly CC E&AS Margins • Stakeholders asked us to compare CC E&AS margins to those in PJM • To do an apples-apples comparison, we used a virtual dispatch model with: • Heat rate = 7,350 Btu/kWh • Min Up and Down Time = 4 hrs • VOM = $2.35/kWh • Forced Outage = 2.5% • The results show that PJM margins are in fact significantly higher than ISO-NE Note: Virtual dispatch results for ISO-NE appear higher than historical actuals we are using for E&AS; this appears to be because of idealizations in the virtual dispatch, as indicated by the 73% modeled capacity factors vs. 62% actual in our sample. Note: PJM gas correspond to Transco Zone-6 Non-NY (Ventyx), and PJM electricity prices correspond to the PSEG zone (Ventyx) Annual Average CC E&AS Margins
Representative PeakerOperations • Stakeholders noted that units providing FRM would generate rarely, only in real-time, and then primarily with oil due to difficulties getting gas with no notice • We agree and changed our sample of representative units to the new peakersproviding FRM in Connecticut E&AS Net Revenues for Representative CTs
Peaker E&AS Margins We updated CT E&AS analysis accordingly • Similar to the CC, we averaged their historical revenues over a three year period (Oct 2010 – Sept 2013) to capture all recent revenues • This includes a few months of the more recent higher FRM prices (but not the earlier Connecticut-specific LFRM) The updated CT E&AS margin is $2.72/kW-mo, incl. $1.22/kW-mo FRM We applied this margin to both the LMS100 and Frame CT due to their ability to provide fast-start
Electricity Futures • At stakeholder request, we reviewed longer-term electricity and gas futures • Using NYMEX futures and OTC trades (Ventyx) through 2019, we found significantly lower gas & electric prices in winter than the “extended near-term futures” projection we had been using • We updated our analysis using the average electricity prices from the other sources as the basis of our E&AS projections, resulting in a reduction of average 2018/2019 prices from $77.5/MWh to $52.5/MWh • The change in electricity futures reduces E&AS margins by 0.77 to 1.28 $/kW-mo depending on the technology
Peak Energy Rent (PER) Deduction • Stakeholders asked us to quantify the impact of PER on Net CONE • PER depends on number of scarcity hours • We made use of ISO-NE’s analysis conducted for PI; scarcity hours depend on assumptions about ties and DR dispatch • We chose an intermediate value of 10.9 because forward prices used to calculate the E&AS offset do not seem to be pricing in 21 hours of shortages per year • At H = 10.9, the PER deduction will be expected to be $0.87/kW-mo for all technologies, which will result in a net increase of Net CONE (see the appendix for a detailed calculation of PER)
Pay for Performance (PFP) Net CONE Impact • Stakeholders also requested that we present an analysis of how new entrants would be impacted by PFP • Our analysis of PFP shows: • Each of the new technologies would expect small performance payments on net, reducing Net CONE slightly • At H = 10.9, we calculate PFP will result in net performance payments of $0.09/kW-mo for both CCs and CTs • See appendix for details
Should the Demand Curve be Based on Gross CONE with an E&AS Adjustment Made on an Ex-post Basis? There are several possible ways to make ex-post E&AS adjustments, but all are problematic: • Approach 1: Deduct Asset-Specific E&AS Actually Earned by Each Resource:This is the most problematic approach, since its effect is to pay a different capacity price to each resource (for providing the same product). It also nullifies day-ahead and real-time signals from the energy and AS markets to operate and invest in such a way that provides those products at least cost. For asset classes whose gross CONE exceeds the reference unit’s gross CONE (e.g. baseload units with high CONE and high E&A/S) the unworkable effect could be a negative capacity payment for an entire asset class. • Approach 2: Deduct Generic E&A/S Offset Based on Asset-Class Proxy Units: The second most distortive approach is to deduct a different amount for each type of technology, based on estimated E&AS margins for proxy units with the same fuel type and generic asset characteristics. This has similar problems but distorts only investment decisions, not operational ones. • Approach 3: Uniform E&A/S Deduction for All Resources: Therefore, any ex-poste deduction would have to be uniform, with the E&AS margins for a single reference technology applied to all resources with a CSO. In this case, however, the deduction poses risks for any other technology whose E&AS margins may be lower (e.g., higher in the merit order, such as DR or super-peakers) or weakly correlated (e.g., existing coal plants) to those of the reference unit. All suppliers would have to add their estimates of the reference technology’s E&AS margin to their capacity market offers, with the disproportionate risks faced by units dissimilar to the reference resource needing to inflate their offers even higher. If actual market conditions differed from their expectations, they could earn much less than their reservation price. Finally, even an ex-post administrative calculation of EAS margins may deviate substantially from the margins received by a unit similar to the reference unit due to locational issues, dispatch in DA vs RT markets, location-specific and time-specific fuel costs, and assumed unit parameters. We remain convinced that the best approach is to deduct E&AS margins from the demand curve Net CONE on a forward basis, based on a technology that fits the Principles we’ve outlined
Net CONE Summary Table Net CONE Summary ($2018)
Draft Proposal • Our current thinking is to propose the 2x1 CC as the Reference Technology • Clear signals from developers that CCs are economic and will be a part of the future capacity mix – so how wrong could choosing it be? • Near-lowest Net CONE • Not demonstrably higher E&AS uncertainty than CTs • The concept of averaging more than one reference technology is compelling, but only if both technologies are good reference technologies • The lack of Frame CT projects suggests the possibility of risks or costs that are not captured in our analysis • Averaging in the Frame CT would be betting the market/reliability on a technology with little commercial demonstration; our demand curve analysis showed that the reliability risks of understating True Net CONE are much more serious than over-procurement risks of overstating True Net CONE • The aeroderivativepeakers’ Net CONE is too high be considered economic • Choosing the CC would set the Net CONE at $11.71/kW-mo
Next Steps • Additional feedback must be submitted right away to be considered in our final Net CONE analysis and proposal • March 6: we will post materials for the Mar 12/13 meeting • March 12/13: we will present our proposal for Net CONE • March 21: MC will vote on sloped demand curve and Net CONE
ISO-NE Analysis of Scarcity Hours Source: ISO-NE Memo to Markets Committee, Operating Reserve Deficiency Information – At Criteria And Extended Results, July 5, 2013.