The Benefits and Costs of Solar Distributed Generation for Arizona Public Service
R. Thomas Beach and Patrick G. McGuire, May 8, 2013 (Crossborder Energy)
This report provides a new cost-benefit analysis of the impacts of solar distributed
generation (DG) on ratepayers in the service territory of Arizona Public Service (APS). On January 23, 2013, the Arizona Corporation Commission ordered APS to conduct a multi-session technical conference to evaluate the costs and benefits of renewable DG and net energy metering (NEM), as part of the ACC’s consideration of the APS Renewable Energy Standard (RES) 2013 Implementation Plan. This report is intended to contribute to the technical conferences and the ACC’s future deliberations on the APS 2013 RES Plan, and to provide a different perspective than the studies on the value of solar DG that APS commissioned in 2009 from R.W. Beck (the “Beck Study”) and in 2013 from SAIC (the “SAIC Study”), which recently acquired R.W. Beck.
The scope of this report is limited to assessing how demand-side solar will impact APS’s
ratepayers. In the context of the cost / benefit evaluations of demand-side programs, this analysis is a ratepayer impact measure (RIM) test. It is not a total resource cost (TRC) test that would look more broadly at whether distributed solar resources provide net benefits to Arizona. Generally, policymakers should look at a variety of cost-benefit tests, including the broad TRC test, in evaluating whether to initiate, continue, or expand a demand-side program.
In assessing the benefits and costs of solar DG from a ratepayer perspective, it is important to use a time frame that corresponds to the useful life of a solar DG system, which is 20 to 30 years. This treats solar DG on the same basis as other utility resources, both demand- and supply-side.
When a utility assesses the merits of adding a new power plant, or a new energy efficiency (EE) program, the company will look at the costs to build and operate the plant or the program over their useful lives, compared to the costs avoided by not operating or building other resource options. A central problem with the Beck and SAIC Studies is that they assess the benefits of solar DG only in a single-year “snapshot,” without considering the long-term benefits of the solar resource over its full expected life.
In addition, solar DG provides significant benefits as a resource that can be scaled easily,
from a system serving a single home to utility-scale plants, and that can be installed with shorter lead times and on a wider variety of sites compared to large-scale fossil generation resources. As APS itself recognizes in its 2012 IRP, DG combines with other small-scale, short-lead-time, demand-side resources such as EE and demand response (DR) programs to reduce APS’s need for supply-side generation, both in the near- and long-terms. The Beck and SAIC Studies do not recognize these benefits of solar DG resources; instead, they first construct "blocks" of solar DG of different sizes, corresponding to different scenarios for solar DG penetration, and then analyze each block as though it were a conventional large-scale power plant. As a result, these studies
calculate few capacity-related benefits from solar DG except in the higher penetration scenarios that are years in the future. In reality, solar DG and APS’s other demand-side programs combine to continuously avoid the need for supply-side resources, and all of these resources should be assigned capacity value commensurate with this role and on a comparable basis.
This report relies on data from APS’s 2012 Integrated Resource Plan (2012 IRP),
supplemented with data from the Beck Study and with data presented in the series of technical workshops that APS held in March and April 2013. Our intent in using this data is to minimize debates over the input assumptions. We also have used a limited amount of current data from the regional gas and electric markets in which APS operates. Our approach to valuing solar DG makes two key changes to the Beck and SAIC studies: first, our analysis is performed over 20 years, instead of just for single years; and, second, we evaluate the benefits of solar DG based on the change in APS’s costs per unit of solar DG installed, without requiring solar DG to be installed in the same “lumpy” increments as large-scale conventional generation. We also draw upon relevant analyses that are standard practice in other states, including the avoided cost “calculator” for demand-side programs adopted by the California Public Utilities Commission (CPUC), as well
as new studies such as the value-of-solar analysis that Clean Power Research (CPR) used in developing the solar tariff for Austin Energy.
The costs of solar DG for APS ratepayers are principally the lost revenues from solar DG
customers who use their on-site solar generation to serve their own loads and who export excess output back into the grid, thus running the meter backward using net energy metering (NEM). For the costs of solar DG, we rely on data that APS reports on the 20-year levelized rate credits that both residential and business customers who install solar DG will realize from the output of their net-metered systems. Finally, on the cost side we also include APS’s remaining DG incentives and the utility’s calculated costs to integrate intermittent solar generation into the grid.
Our work concludes that the benefits of DG on the APS system exceed the cost, such that
new DG resources will not impose a burden on APS’s ratepayers. The following table
summarizes our results. The benefits exceed the costs by more than 50%, with a benefit / cost ratio of 1.54. The benefits also exceed the costs in both the residential and commercial markets considered individually. Based on SAIC’s projection of 431,000 MWh of incremental solar DG in 2015, these benefits amount to $34 million per year for APS’s ratepayers…
Benefits of Solar DG
APS’s 2012 resource plan makes very clear that the utility’s marginal sources of generation are principally natural gas-fired resources. In addition, APS expects renewable generation to compete with, and potentially to displace, a portion of these future gas-fired resources:
APS foresees the ability to treat natural gas and renewable energy resources as
competing levers during this time period, and resource decisions can be modified
from the current plan based on the relative tradeoffs between those fuel sources
throughout the intermediate-term stage. For example, APS plans to add over 3,700
MW of natural gas generation capacity and 749 MW of renewable coincident-peak
capacity during this stage. In the event that solar, wind, geothermal, or other
renewable resources change in value and become a more viable and cost-effective
option than natural gas, future resource plans may reflect a balance more
commensurate to the Enhanced Renewable Portfolio.5
In the future, to the extent that APS’s customers invest in demand-side resources, including on-site solar DG, the resources displaced will be new gas-fired generation.
Accordingly, APS’s future avoided energy costs are the energy costs of APS’s long-term
gas-fired generation resources. To estimate these avoided costs, we first develop a long-term forecast of APS’s burnertip cost of gas at its power plants. This forecast uses current (April 1, 2013) forward gas price data from the NYMEX Henry Hub market, the basis differential from the Henry Hub to the Permian basin, plus variable delivery costs over the El Paso Natural Gas (EPNG) system to APS’s plants in Arizona. Figure 1 compares this projection to APS’s 2012 IRP cost of gas forecast6 and to the APS gas cost forecast for 2015, 2020, and 2025 (based on the December 31, 2012 forward market) which SAIC has used. Our gas cost forecast is very similar to the SAIC forecast.
Because our forecast is based on forward market natural gas prices, it represents a cost of
gas that APS could fix for the next 20 years. This captures the fuel price hedging benefit of renewable DG, which has no fuel costs and thus avoids the volatility associated with generation sources whose cost depends principally on fossil fuel prices.
Figures 5-3 and 5-5 of the Beck Study show that solar DG systems on the APS system
typically displace combustion turbine (CT) generation during the four peak summer months (June-September) and combined-cycle (CCGT) generation in other months. We assume that solar DG avoids generation from new, efficient, state-of-the-art gas plans, with heat rates of 9,400 Btu/kWh for CTs and 7,300 Btu/kWh for CCGTs, plus the corresponding variable O&M costs for such generation.8 We use our gas price forecast as the fuel costs for these avoided resources. We note that the resulting avoided energy costs in the near term (2014-2015) are close to current forward market prices for the Palo Verde trading hub, as shown in Figures 2 and 3. We also include APS’s 2012 IRP forecast of greenhouse gas (GHG) allowance costs ($15 per metric ton, starting in 2019) as an adder to the gas price forecast,9 using the standard natural gas CO2 emission
rate (117 lbs/MMBtu). Finally, we assume that APS will avoid marginal line losses of 12.1%, based on the detailed analysis of the loss impacts of solar DG that is in the Beck Study.10 W ith these inputs, our Base Case forecast of APS’s avoided energy costs for solar DG is a 20-year levelized value of 7.1 cents per kWh, in 2014 dollars.
In addition, we have modeled two sensitivity scenarios for APS’s avoided energy costs for 2019 and subsequent years. The first is a High Case which assumes APS’s High projection of GHG costs from the 2012 IRP. The second sensitivity is a Low Case with zero GHG costs for the next twenty years, which is the Low GHG scenario from the 2012 IRP.
Figure 2 shows our Low, Base, and High avoided energy cost forecasts for the peak
months of June – September; Figure 3 presents the results for the off-peak months of October through March. Table 2 summarizes the resulting 20-year levelized avoided energy costs for solar DG in APS’s service territory, including avoided line losses.
SAIC used the results of APS’s confidential production cost modeling to estimate avoided energy costs; the SAIC results are shown in the second column of Table 3, below. These modeling results are too low to be credible as long-run avoided energy costs for the resources displaced by solar DG. The final column of Table 3 shows the marginal heat rates that are implicit in these results, based on the SAIC/APS natural gas and GHG cost forecasts. These heat rates are far lower than the heat rates of even the most efficient new gas-fired resources, indicating that APS’s modeling either (1) assumes that solar DG often displaces APS’s existing coal-fired generation or (2) reflects only the low, short-run incremental costs of moving already-operating gas plants in the western U.S. from one loading point to another. Moreover, even if this modeling is realistic, it understates APS’s avoided opportunity costs of selling its excess generation into the
regional energy market at Palo Verde and other trading hubs, as shown in Figures 2 and 3. In sum, these results significantly understate the long-run energy costs avoided by solar DG resources which will completely displace the need for and the full costs of future gas-fired units.
b. Generation Capacity
The 2012 IRP finds that APS does not need new large-scale, fossil resources until 2017.11 However, the 2012 IRP shows continued growth in energy efficiency and demand response programs and in distributed solar resources between 2012 and 2017 (see Table 2), such that the new demand-side resources will contribute 1,150 MW to meeting APS’s peak demands in 2017. Solar DG, along with energy efficiency and demand response, thus contributes to deferring any resource need until 2017. As a result, solar DG installed before 2017 has greater value than just avoiding short-term energy costs. DG also hedges against events that could accelerate the 2017 need, such as unexpected increases in demand (from an accelerating economic recovery) or the loss of existing resources (for example, nuclear plant shutdowns such as the recent problems at the San Onofre plant in southern California).
Combustion turbines are the least-cost source of new utility-scale capacity. CTs are the
long-term peaking resource typically displaced by solar DG, and are the resource that APS expects to add in 2017. The Beck and SAIC Studies use the fixed costs of a new CT to calculate solar DG’s generation capacity value. The CT fixed costs in the Beck Study were based on a CT capital cost of $1,088 per kW in 2008, times a fixed charge rate of 11.79% to convert to an annual levelized value.12 The 2012 IRP cites CT capital costs in a range of $600 to $1,400 per kW, with heat rates from 8,900 to 11,900 Btu/kWh for a variety of brownfield and greenfield projects.13 SAIC is using a CT capital cost of $1,136 per kW, plus $206 per kW in gen-tie transmission.14 Following the Beck and SAIC Studies, we also have included (and updated) the fixed O&M costs and the El Paso Natural Gas pipeline reservation costs for a new CT built in APS’s service territory. As shown in Table 4, we calculate that APS’s levelized avoided capacity costs are $190.10 per kW-year in 2014 dollars.
The CT fixed costs are multiplied by the effective load-carrying capacity (ELCC) of PV
generation. At the present level of solar PV penetration, this adjustment is 50% for a fixed array and 70% for an array with single-axis tracking. APS used these adjustments in the 2012 IRP to determine the firm capacity of solar resources, including resources that will be developed in the 2013-2015 time frame.15 The resulting avoided generation capacity costs are shown in Table 4.
This analysis focuses on the value of solar to be developed in the next several years
(2013-2015). The Beck and SAIC Studies indicate that, if solar penetration increases
significantly, the capacity value of solar that is installed in 2020 and 2025 may be lower than today, as the increased amounts of installed solar resources shift APS’s afternoon peak to later in the day. This possibility does not diminish the capacity value of solar installed today; indeed, the decline in capacity value in 2020 and 2025 will not occur unless substantial amounts of solar are installed over the next twelve years. Finally, the Beck / SAIC result that the capacity value of solar will decline over time assumes that the future will look like today, only with more solar. This is unlikely to be true. For example, other trends, such as hotter summers resulting from climate change, could increase future peak demands by more than expected, and offset the impact of solar additions. Customers also can respond to the changing mix of resources. If additional solar reduces the price for grid power in the afternoon, if those prices are conveyed in accurate price signals, and if customers have greater choice and control over when and from where they consume electricity, consumers will respond by shifting consumption from the evening to the afternoon – i.e. the opposite of what DR tries to achieve today – pre-cooling homes, running appliances remotely, and filling batteries in the afternoon instead of the evening.
c. Ancillary Services and Capacity Reserves
The Beck Study found that the intermittency of solar DG is unlikely to increase the
ancillary services or operating reserves that APS must supply to ensure reliable service, given the geographically dispersed nature of DG systems.16 The study did not consider, however, the fact that DG will result in a reduction in the loads that APS will serve, because the majority of DG output will serve the on-site load of the DG host customer or will run the customer’s meter backward if power is exported. WECC reliability standards require control area operators to maintain operating reserves (spinning and non‐spinning) equal to 7% of the load served by thermal generation. As a result, load reductions from DG will reduce APS’s requirements to procure operating reserves. In addition, APS must maintain a capacity reserve margin of 15%. Thus, each kW reduction in APS’s peak demand from DG will reduce the utility’s capacity requirements by 1.15 kW. We model these avoided ancillary service and capacity reserve requirements as 7% of Base Case avoided energy costs from Table 217 and 15% of the south-facing avoided generation capacity costs from Table 4. These avoided ancillary service and capacity reserve
costs are summarized in Table 5.
The Beck Study reported that APS incurs $125 million in high-voltage transmission costs
for every 400 MW increase in peak demand, and $7 million in lower-voltage subtransmission costs per 30 MW of load growth.18 The SAIC April 11 presentation, at slide 63, shows $29.5 million in deferrable subtransmission costs for a 130 MW decrease in peak demand. In the long-run, solar DG combines with EE and DR resources to defer such costs even if, over a short-term period such as a three-year transmission planning cycle, none of these small-scale resources individually amounts to 400 MW or to the smaller amounts in specific areas that is required to defer subtransmission projects. Given that EE, DR, and DG resources will combine to reduce APS’s peak demands by 1,150 MW in 2017, it seems clear that, in aggregate, these resources will avoid significant transmission costs on the APS system. Escalating these avoided transmission and
sub-transmission costs to 2014 and using the current APS carrying charge of 11.05% for
transmission yields a levelized avoided transmission cost of $65.14 per kW-year, as shown in Table 6. As with avoided generation capacity costs, we apply the solar ELCC values to the avoided transmission costs, in recognition that peak solar output does not necessarily coincide with system peak demands.
The Beck Study examined a range of possible DG impacts on distribution system costs. These impacts are more location-specific than the effects of DG on the generation or transmission systems. The Beck Study concluded that distribution capacity cost savings are possible if demand reductions from DG exceed load growth on distribution feeders or substations, and if solar DG can be targeted to specific locations where circuits would otherwise need an upgrade.19 The study valued these reductions using a distribution avoided cost of $115,000 per MW of DG ($115 per kW).20 SAIC has now backed away from these results, arguing that it could identify only 5-9 circuits on which installed PV capacity reduced the circuit peak to below the 90% of capacity threshold at which the utility begins to plan an upgrade.21 Yet this appears to be an appreciable fraction of the 30-40 circuits that APS upgrades each year.22 Moreover, even on a circuit whose
loading is below the 90% threshold today, PV can reduce the peak loading and defer the future date when that circuit’s loads exceed the 90% threshold, a date that may be beyond the current distribution planning period but well within the lives of the installed PV systems. The Beck Study reported that 50% of the feeders modeled show potential for reducing peak demand and deferring capital improvement projects.23 Avoided distribution capacity costs can be valued using the same approach applied to transmission costs in Table 5, with the additional assumption that PV can avoid distribution costs on 50% of circuits. Table 7 presents these results.
With the exception of greenhouse gas emissions, the Beck and SAIC studies have not
quantified any of the environmental benefits of renewable generation, such as reductions in criteria air pollutants (SO2, NOx, and PM 10) and decreased water use for electric generation. APS did quantify these benefits in the 2012 IRP, however. The utility calculated both the reduced emissions of these pollutants and the lower water use, per MWh of renewable generation,24 and included estimates of the dollar value of such reductions.25 Table 8 summarizes these environmental benefits.
g. Avoided Renewables Costs
Solar DG helps APS to meet Arizona’s Renewable Energy Standard (RES) requirements.
The Arizona RES regulations include a requirement that APS must procure renewable generation equal to a certain percentage of its sales, with the percentage increasing from 4.0% in 2013 to 10% in 2020 and 15% by 2027. The RES requirement also provides that, after 2011, 30% of the new renewable generation meeting the RES standard must be DG resources. Pursuant to Arizona Corporation Commission (ACC) Decision No. 71448. APS also must procure an additional 1,700,000 MWh of incremental renewable generation by December 31, 2015.26 The Beck Study did not attribute any value to DG’s contribution to meeting APS’s RES requirements. However, because it is customers who make investments in DG resources, not APS, such customer-owned resources allow the utility to avoid the higher capacity-related costs of renewable power.
APS has also argued that solar DG does not avoid the costs of other renewable resources
because APS already has procured adequate renewables to meet its RES requirement. However, all of these resources are not yet on-line, so solar DG may hedge against the failure of some of the utility-scale renewables with which APS has contracted. Moreover, APS itself recognizes that, in the long-run, it may have to procure renewables beyond today’s RES requirements. The 2012 IRP includes an Enhanced Renewable Portfolio which assumes that APS increases the contribution of renewable energy to 30% of retail sales by 2025 and meets 90% of load growth with emissions-free resources. In addition to further reductions in emissions of greenhouse gases and criteria air pollutants, there are economic reasons to procure additional renewables. For example, the 2012 IRP notes that, in both the intermediate- and long-terms, “renewable resources have the ability to diversify the overall portfolio of resources and provide mitigation against the inherent price volatility risks associated with a natural gas-dominated energy mix.”27
Renewable generation also results in a number of difficult-to-quantify benefits, including:
• Price mitigation benefits. Lower demand for electricity (and for the gas used to produce the marginal kWh of power) has the broad benefit of lowering prices across the gas and electric markets in which APS operates.28
• Grid security. Renewable DG resources are installed as many small, distributed systems and thus are highly unlikely to fail at the same time. They are also located at the point of end use, and thus reduce the risk of outages due to transmission or distribution system failures. This reduces the economic impacts of power outages.
• Economic development. Renewable DG produces more local job creation than fossil
generation, enhancing tax revenues.
We assume that the additional cost of renewable generation provides a proxy for these benefits. These benefits have been calculated separately in at least one study, which estimated these benefits collectively to be from $100 to $140 per MWh in several eastern U.S. markets.29
For the APS system, the 2012 IRP includes APS’s estimates of the incremental cost of
renewables. The Enhanced Renewable scenario in the 2012 IRP features additional purchases of renewables in the 2017-2026 time frame, totalling 4,532 GWh of additional renewable generation by 2026 compared to the Base case (about 500 GWh per year in additional renewable generation).30 The 2012 IRP includes annual revenue requirements for both the Base and Enhanced Renewable scenarios; the difference between these revenue requirements allows one to calculate the annual cost premium for the incremental renewables in the latter scenario.31 The cost premium for these purchases averages $46.55 per MWh from 2017-2026 ($45.27 per MWh on a 10-year levelized basis).32 We use this premium as the measure of the costs which APS will avoid if APS’s customers invest in solar DG, reduce the future need for APS to purchase additional wholesale renewable generation, and provide the benefits listed above. This appears to us to be a
conservative estimate of the value of additional customer-driven renewable generation on the APS system over the next 20 years.
3. Costs of Solar DG
The primary costs of solar DG are the retail rate credits provided to solar customers
through net metering, i.e. the revenues that the utility loses as a result of DG customers serving their own load. Data responses from APS to the ACC staff in the 2013 RES case33 include calculations of the 20-year levelized retail rate credits (i.e. the lost revenues for APS) resulting from DG, as well as the costs of the current incentives paid to customers who install DG. In the technical workshops, APS also has provided Vote Solar with its estimates of residential and commercial lost revenues. For residential customers, the retail rate credits amount to 15.5 cents per kWh; for business customers, the credits are 7.1 cents per kWh.34 APS has assumed a retail rate escalation of 2.5% per year and an 8% discount rate.35 These assumptions produce 20-year levelized retail rate credits of 19.7 cents per kWh for residential and 9.0 cents per kWh for commercial (2014 $). Assuming the current mix of residential and commercial systems, the average rate credit is 13.7 cents per kWh.
With respect to incentive costs, the 20-year levelized cost of the current 10 cents per watt
residential upfront incentive is 0.6 cents per kWh. We understand that APS has proposed to eliminate these residential incentives, so they may be zero in the future. APS also has eliminated business incentives, except for school and government projects.
Finally, we add an estimate of solar integration costs using a recent study which APS
commissioned which estimated integration costs of $2 per MWh in 2020 and $3 per MWh in 2030.36 We assume that these costs scale to other years as a function of gas costs. Table 1 and Table 9 summarize all of these costs of DG for APS’s ratepayers.
4. The Context for this Cost / Benefit Analysis
The Beck and SAIC Studies calculate the benefits of DG – i.e. the “value of solar.” These
benefits could be used in a cost-benefit evaluation of solar DG, such as is presented in the report. The Beck and SAIC Studies do not discuss the cost side of the equation, or attempt to apply any of the standard cost-effectiveness tests to DG. We assume that APS will use a new calculation of the benefits of DG in a ratepayer impact test, such as the one presented in this report.37 The conclusion of this report is that solar DG with net metering is cost-effective for non-participating ratepayers in APS’s service territory.
We emphasize that the ratepayer impact perspective should not be the only one which
policymakers examine in deciding on future policies affecting solar DG in Arizona. The RIM test often is considered the most rigorous of the cost-effectiveness tests for demand-side resources; passing the RIM test with a benefit / cost ratio greater than 1.0 means that there are “no losers” from a demand-side resource. Nonetheless, a full analysis of solar DG as a resource also should consider additional cost-effectiveness perspectives, such as societal, total resource, participant, and program administrator tests.38 Other demand-side programs typically are evaluated from these multiple perspectives, and policymakers should take a similarly broad view in assessing distributed generation programs.