Skip to main content
power line towers

E3’s Finding that Gas Peakers are Least Cost and Needed for PNW Reliability is Outdated and Flawed

By: Cynthia Mitchell & Sashwat Roy**

This article is responsive to the E3 report entitled, “Resource Adequacy in the Pacific Northwest” dated March 2019, which was most recently referenced in Matthew Bandyk’s December 17 article[i]. We find the report is in large part, the underlying basis for the uncertainty utilities and policymakers are facing in moving away from fossil fuels for electricity generation. E3 claims that gas generation is the most economic source of firm capacity, and that absent of technological breakthroughs (E3’s emphasis), it would be extremely costly and impractical to replace all carbon-emitting firm generation capacity with solar, wind, and storage, due to the very large quantities of these resources that would be required. While E3 inflates the capital costs for emerging technologies such as battery storage, hybrid resources and demand response to support its arguments, additionally, it bypasses the significant economic risks assumed in constructing a new gas power plant in a future where clean energy will dominate energy policy pathways due to lower economic and environmental costs and targeted policy measures.

As part of my utility advocacy work in Washington State,[ii] I reviewed the E3 report when it came out in 2019, and revisited E3’s initial analysis[iii] and subsequent January 2020 update[iv]. I contrasted E3’s assumptions and findings for the Pacific Northwest (PNW) to E3’s work in other states, the analyses of other leaders in the field, and the cost and performance of various renewable and storage projects. What I found is that E3 analyses for the PNW are fundamentally outdated and flawed. Technological breakthroughs in the cost and capabilities of utility scale wind and solar and battery storage have already occurred at such a rapid pace as to increasingly displace gas generation in utility resource planning and operations. In addition, numerous industry and research reports conclusively prove that as the level of solar penetration increases, the ability of standalone battery storage to provide peaking capacity benefits increases and does so at similar or lower lifecycle costs[v].  

Also, E3 turns a blind eye to the ability for utilities to modify energy use, particularly peak demand, through the vast array of distributed energy resources, starting with demand response. E3’s consideration of demand response is convoluted, coming at the tail end of all existing and new generation, and is highly constrained to the bare bones of its potential.

E3’s framing of PNW resource adequacy is essentially a throwback to the 1980’s of uncontrolled electric load, without consideration of the plethora of opportunities to optimize the grid through DERs for critically needed resiliency and reliability. Claiming to have been at the forefront of DER assessment since 1989,[vi] including California whereby law all cost-effective energy efficiency and demand response are to be procured before other resources, including renewable generation, E3 knows better. From my near two decades as The Utility Reform Network’s (TURN) utility energy efficiency consultant, I rubbed shoulders with E3 over many California projects and analyses. I know they are capable of much better work.

Going back to Amory Lovin’s 1976 Soft Energy Paths: Toward a Durable Peace, we know that how we get clean energy is through a combination of renewable energy and energy efficiency and demand response. Nearly fifty years ago, Lovin viewed the energy problem as one of inefficient energy use, coupled with a lack of development of renewable energy sources. One of his trademark Amory-isms was “It’s not the kilowatt hours per se, but the end use services – cold beer and warm showers – that consumers want.”

Costs of Wind, Solar, and Battery Storage

Unpacking the shortcomings in E3’s work starts with comparing their assumptions on the cost of wind, solar, and battery storage with that of other credible sources and best practices. In their 2020 PNW update, E3 forecasts the 2045 capital cost (discounted to 2018$) of solar PV at $980/kW and 4-hour Li-ion Battery at $590/kW.[vii]

Compare E3 to Lazard’s latest Levelized Cost of Storage Analysis (LCOS 6.0)[viii] which shows that PV+Storage projects are becoming increasingly competitive as utilities look for ways to supplement retiring conventional generation resources while avoiding investments in new peaking power plants. Lazard’s LCOS analysis for a 50 MW Storage/200 MWh Capacity utility scale PV + Storage system comes in at $188 - $320/kW.[ix] Not only is Lazard’s combined PV + storage cost significantly less than E3’s 2020 separate cost assumptions for solar and storage shown above, Lazard’s PV + Storage cuts the knees out from under E3’s 2020 gas CT cost of $850/kW.[x]

Lazard’s latest annual Levelized Cost of Energy Analysis (LCOE 14.0) showing that as the cost of renewable energy continues to decline, certain technologies (e.g., onshore wind and utility-scale solar), which became cost-competitive with conventional generation several years ago on a new-build basis, (emphasis added) continue to maintain competitiveness with the marginal cost of selected existing conventional generation technologies.[xi] Lazard’s data pegs solar at $29 – $107/MWh relative to a gas combustion turbine at $124 – $162/MWh, and wind at $23 – $46/MWh relative to a gas combined cycle unit at $41 – $61/MWh.[xii]

Western region cost data for wind, solar, and battery storage projects comes in with the lower range of Lazard’s. In the figure below, on the left are the summary results from Colorado Excel’s competitive solicitation December 2018. Coal and gas variable costs (fuel and O&M) are at $37 and $30 per MWh respectively, with solar and wind total costs (all up-front capital costs and O&M costs) at $30 and $18 per MWh. Adding in gas generation capital costs can easily double the total cost of gas generation relative to solar and wind. This means that per the Xcel bid results two years ago, solar and wind not only cost less than the total cost of gas generation but were also less costly than the operating cost of gas generation. And, with no fuel costs and minimal operational costs throughout the lifecycle, solar and wind are a consumer hedge against future utility rate increases.

The same figure (Figure 1) shows on the right 2020 costs for solar generation with battery storage. At around $30 MWh, renewable generation with storage is increasingly reliable relative to gas generation, very cost competitive, and solving the renewable intermittent matter.[xiii]


Figure 1. Cost of Utility-scale renewables from actual bids.

Cost of Utility-scale renewables from actual bids
Source: Regulatory Assistance Project


Capabilities of Wind, Solar, and Battery Storage

E3’s modeling of wind and solar generation fails to reflect how battery storage is being used to address renewable intermittent issues, thereby creating clean energy that “walks and talks” like gas combustion turbines and combined cycle units. For instance, in E3’s 2020 report, the total installed capacity for the PNW in 2045 doubles when going from the reference case with gas to the 100% renewable case – from about 75 GW to about 160 GW. The combined capacity of wind and solar increases 5-fold (18 GW to 99 GW), with only 10 GW of battery storage, (or 10% storage to solar), and with storage duration to 4 hours in 2045.[xiv] With the required installed capacity of renewables doubled, and battery storage woefully underutilized and constrained, it’s no surprise that E3’s 2045 PNW 100% renewables scenario fails on cost and performance relative to E3’s 2045 PNW gas generation reference case. In fact, a recent joint report filed by three major IOUs before the CPUC shows that ELCC values for solar paired with storage resources remain well above 95% for study years 2023 and 2026[xv]. This can be attributed to the ability of a solar resource to consistently charge a 4-hour battery storage device prior to the hours when energy is most critical to grid reliability or what is now the “net peak load.”      

Apart from technical capabilities of renewables and storage resources, recent effort to develop a resource adequacy program by the Northwest Power Pool (NWPP) and a set of 18 utility members has been underway since October 2019. This multi-state program, which is currently in its ‘detailed design phase’ spearheaded by the Southwest Power Pool (SPP), can unlock the geographical and resource diversity in the region and enable a uniform set of reliability standards paired with an operational resource sharing program to provide substantive economic and reliability benefits in the region. In fact, E3 was commissioned by the NWPP to provide a study of a potential RA program, resulting in a report[xvi] published in October 2019. In the report, they conclude by stating that a regional RA program would ensure that sufficient generation is available to reliably serve demand during periods of grid stress and could also produce cost savings by allowing utilities to rely on other entities’ resources rather than building their own at higher cost.

As a respected firm with a national presence, in other jurisdictions E3 has come to a different set of findings regarding the capabilities of solar and storage. For instance, in work for First Solar on the economic benefits that might result if solar plants operated flexibly, E3 argues that What’s needed most is a shift in thinking – not only about how resources can and should be used, but also in underlying business practices. Current contracts typically exclude solar from providing reliability services, for example, and deny revenues during curtailment… Utilities and grid operators should stop thinking of solar as a problem to be managed and start thinking of it as an asset to be maximized.” E3’s 2018 First Solar work garnered several awards and podcasts.[xvii] A similar study commissioned by the California ISO in partnership with Avangrid, NREL and GE, conducted subjective tests on Tule Wind Farm, located in eastern San Diego County, to demonstrate that large, utility-scale wind plants can provide essential grid services (reg-up, reg-down, voltage control, frequency control) to the grid that currently are being supplied by conventional thermal resources[xviii].       

Consider also E3’s 2019 work in Minnesota (the same latitude as Washington state, but very little hydro) on the ability for storage to replace new peaking capacity. E3 found that “…solar plus storage is cost-effective today and stand-alone storage is cost-effectives in 2025…for new peaking capacity.”[xix] E3 also found that 324 MW of Minnesota’s peaking capacity has an operational profile that would allow it to be “mimicked” by 4-hour energy storage. And E3 cites the long-term potential for 4-to-6-hour storage will likely grow substantially in the state, particularly under higher levels of renewables: in 10 years from the current queue of about 160 MW to up over 1,000 MW. I particularly like the visual E3 creates in referring to energy storage as the “Swiss Army knife” of the electricity system in recognition of the many services it can perform.[xx] E3’s Minnesota work is more in keeping with Lazard’s 2020 view of long duration battery storage (>6 hour) is gaining traction as a commercially viable solution to challenges created by intermittent energy resources such as solar or wind “As regional grids achieve higher penetration of renewable energy generation, long-duration storage is well-positioned to take advantage of the corresponding increase in the potential for curtailed and low price generation.”[xxi]

National and International PV and Storage Projects

There are any number of projects where storage is more appropriately sized to the address the intermittent nature of renewables, and where storage is used on a standalone basis for peaking needs and grid reliability.[xxii]

  •  PG&E, San Francisco, CA 1 MW/2 MWh battery standalone, 2-hour storage duration; and 0.5 MW/2 MWh battery with 1 MW PV, or 50% storage to solar and 4-hour storage duration
  • California Independent System Operator (CAISO) Los Angeles, CA, 50 MW/200 MWh battery standalone and 4-hour storage duration
  • Independent Electricity System Operator (IESO) Ontario, Canada, 1 MW/2MWh battery standalone and 2-hour storage duration
  • ISO New England 10 MW/60MWh battery standalone and 6-hour storage duration
  • Electric Reliability Council of Texas (ERCOT) Corpus Christi, TX, 100MW/200MWh battery and 100 MW PV, or 100% storage to solar and 2-hour storage duration
  • National Electricity Market (NEM) Queensland, Australia, 100 MW/200 MWh battery and 100 MW PV, or 100% storage to solar and 2-hour storage duration
  • Victoria, Australia 0.5 MW/2MWh battery and 1 MW PV, or 50% storage to solar and 4-hour storage duration

PacifiCorp’s recently issued RFP stemming from their 2019 IRP preferred portfolio show significant amount of renewable energy and battery storage resources to meet energy and capacity needs amounting to almost 7 GW by 2025[xxiii] in light of transition to a coal-free future. Other PNW utilities like Portland General Electric (PGE) and Idaho Power have also taken major steps to invest in non-emitting generation paired with storage to meet capacity needs with PGE’s  380 MW solar + wind + battery project (Wheatridge Energy Facility) came online to deliver energy in December 2020.  

What’s interesting with these examples is not just the projects’ competitiveness as measured levelized costs, but also the high internal rates of return (IRR), ranging from 8.1% for New England ISO T&D deferral, to 33.7% for CASIO peak capacity.[xxiv] Washington state is joining the ranks with the state’s first utility-scale solar and battery storage site was just com­pleted in City of Richland. In addition to generating electric­ity, this facility will offer a train­ing program for solar and battery storage technicians.[xxv] Also, a Colorado energy company is in­vesting in south­east Washington, announcing plans for the 850 MW Horse Heaven Wind, Solar and Battery Farm near the Tri-Cities.[xxvi]

This is just a very small sample of how the rapid development of renewable generation coupled with battery storage and standalone energy storage, is competing with existing and new gas generation on cost and reliability. The US Department of Energy’s “Energy Storage Grand Challenge: Energy Storage Market Report”, December 2020, finds the outlook for energy storage beyond robust. “By 2030, stationary and transportation energy storage combined markets are estimated to grow 2.5 – 5 terawatt-hours (TWh) annually, approximately three to five times the current 800-gigawatthour (GWh) market.”[xxvii]

Distributed Energy Resources: Energy Efficiency, Demand Response, Energy Storage

Demand for electricity varies by season, day, and time; and resources to meet needs are not a one size fits all. In much of the PNW, the electric utilities experience peak electricity demand during winter cold spells. This winter peak is of limited dura­tion, driven by predictable weather events and has been man­ageable by using natural gas peaking plants.

DERs can reduce peak de­mand on the electric grid by a process called demand re­sponse, which can eliminate the need for natural gas peaking plants. The term demand response is decades old and was orig­inally used to describe pro­grams where factories or buildings could manually shut down electric loads in a grid emergency. As pro­cesses became more auto­mated and flexible, demand response evolved to control­ling loads from appliances that do not have to be running constantly and activities such as charging a battery, which can be delayed until loads decrease.[xxviii] 

Technologies such as demand response, and in combination with energy storage (e.g., thermal energy in water heat­ers and electrical energy in batteries), can be used to manage peak load more cost effectively, more quickly, and with a much lower carbon impact. Energy storage (like batteries) are valuable dual-purpose devices. Energy storage can either supply energy to the grid or demand energy from the grid. The key is timing in order to maintain an optimal balance of supply and demand.

As mentioned earlier, E3 considers demand response after all renewable and fossil generation, contrary to its longstanding “first loading order” resource status. E3 in its 2019 PNW report discounts demand response for “effectiveness” by 50%, resulting in only a 1.5% contribution to peak load, and constrains demand response to a maximum of 10 calls per year, with each call lasting for a maximum of 4 hours, resulting in essentially no meaningful DR contribution to capacity and reliability requirements. Interestingly, E3’s 2020 PNW report in moving from the 2045 reference case with gas generation to 100% renewables, DR is cut by about one-half.[xxix]

At minimum, E3 should have considered the following approaches to DERs from Jim Lazar’s “Teaching the Duck to Fly.”[xxx]

  • Target Energy Efficiency to the Hours When Load Ramps Up. Acquire energy efficiency measures with a focus on measures that provide savings in key hours of system stress.
  • Implement Aggressive Demand-Response Programs Strategy. Deploy demand-response programs that shave load during critical hours of the year, only on days when system stress is severe.
  • Control Electric Water and Space Heaters to Reduce Peak Demand and Increase Load at Strategic Hours. Control electric water and space heaters to increase electricity usage during night hours and mid-solar-day hours and decrease usage during morning and evening peak demand periods. Electric water space heaters controlled to provide high reliability of hot water and space service, low draw of heaters during key morning and evening hours, and increased draw of heater loads during off-peak and solar-day hours.
  • Deploy Electrical Energy Storage in Targeted Locations. Identify locations where electricity storage, including batteries, can provide more than one function. Deploy storage to simultaneously reduce the investment needed for transmission and distribution, and to provide intra-day storage of energy from intermittent renewable resources.

Given the significance of electric water heating load in the PNW, its worth considering further the benefits of load control. Jim Lazar explains that control of one million electric water heaters means that up to 4,400 MW of load could be “turned on” as needed to absorb wind or solar energy, and that up to 1,000 MW of water heating load that occurs during periods of high demand could be “turned off” as needed to manage peak loads. In addition, up to 10,000 MWh per day of electricity consumption could be shifted between time periods as needed. Water heater control of this sort up and down the Oregon-Washington I-5 corridor would be a great place to start!

By E3’s way of thinking about the PNW, demand response is nothing more than “lipstick on a pig”, and far out of step with regional and national trends. For instance, the Northwest Power and Conservation Council, 7th Plan, 2016, identified more than 4,300 megawatts of regional demand response potential.[xxxi] A significant amount of this potential, nearly 1,500 megawatts, is available at relatively low cost; less than $25 per kilowatt of peak capacity per year. When compared to the alternative of constructing a simple cycle gas-fired turbine, demand response can be deployed sooner, in quantities better matched to the peak capacity need, deferring the need for transmission upgrades or expansions.

 In particular, demand response is the least expensive means to maintain peak reserves for system adequacy. Its low cost is especially valuable because the need for peaking capacity in the region largely depends on water and weather conditions. As NW runoff increases in winter and spring due to climate change, this makes possible more hydroelectricity for green hydrogen production through electrolysis, or for peaking production in a prolonged cold snap. Instead of wasting excess winter runoff over spillways, produce hydrogen to increase economic competitiveness in the clean energy economy. And if a deep cold snap occurs, hydrogen production is viewed as a "flexible load" (or demand response) that can be shed when needed. Win-win.[xxxii]

Analysis by Brattle[xxxiii] suggests it is far cheaper to control load to follow available generation once the proper infrastructure is in place. See June 27, 2019, for instance, “Brattle Study: Cost-Effective Load Flexibility Can Reduce Costs by More Than $15 Billion Annually.” Brattle economists have released a study that identifies nearly 200 GW of cost-effective load flexibility potential in the U.S. by 2030. This load flexibility potential, which equates to 20% of estimated U.S. peak load in 2030, would more than triple the existing demand response (DR) capability and would be worth more than $15 billion annually in avoided system costs.


There is overwhelming evidence that wind and solar coupled with battery storage and distributed energy resources, are beating the pants off of gas generation in terms of cost and reliability. Added to this, recent market development efforts provide impetus to unlock the regional diversity of resources and enable significant reliability and economic benefits. E3’s PNW work alleging that the new gas generation is cost competitive and needed to keep the lights on, does everyone, including E3, a disservice.


**Sashwat Roy is a contributor to this article and works with Renewable Northwest as a Technology & Policy Analyst. He recently obtained his Ph.D. in Energy and Environmental Policy from the University of Delaware’s Biden School of Public Policy & Administration.


[ii] Cynthia Mitchell is a 40-year veteran in energy policy and utility regulation, an expert on utility integrated resource planning, focused on sustainability through distributed energy resources (energy efficiency, demand response, energy storage), and large-scale central station renewable energy. As an economist, she has worked for Attorney General Consumer Advocates around the country, including about 20 years in California for TURN, The Utility Reform Network. Living in Washington State 2016-2020, her interests and activities included state and local projects. Cynthia recently moved to Santa Fe. New Mexico. [email protected]

[iii] “Resource Adequacy in the Pacific Northwest”, March 2019,


[v] The Potential for Battery Energy Storage to Provide Peaking Capacity in the United States. Denholm et al, 2019.


[vii] Ibid iv, slide 18.

[viii] Lazard

[ix] Ibid vii, slide 5, “Unsubsidized Levelized Cost of Storage Comparison – Capacity ($/kW-year).

[x] Ibid vi.

[xi] Ibid vii, first screen/page.

[xii] Ibid vii, slide 4, “Solar PV versus Gas Peaking and Wind versus CCGT – Global Market”, see US data.

[xiii] Prepared by the Regulatory Assistance Program (RAP),, using information from the Energy Information Administration at the U.S. Department of Energy.

[xiv] Ibid iv, Slide 55.







[xxi], Slide 12.

[xxii] Ibid xvii, Slides 9, 17-24


[xxiv] Ibid xvii, Slides 10, 25-28.

[xxv] Horn Rapids Solar, Storage and Training Project, Energy Northwest, November 2020;

[xxvi] Richland’s new solar and battery project is first of its kind in Washington state;, Nov 11, 2020, by Carrissa Lehmkuhl.;


[xxviii] There are two primary types of DR: 1) “daily peak shaving” that is done to lower the effective overall system peak and 2) “emergency peak shaving” that goes a step farther, but only in extreme circumstances.

[xxix] Ibid iii, Table 14, 2018 Loads and Resources, and p. 59. Ibid iv, Slide 55.


[xxxi] The new 2021 (8th) Power Plan will be out in several months. They had to revamp the main numerical models to adequately model DERS, especially DR. It is likely that DR potential will be even greater. of power generation, for example. Anyway, the momentum is shifting to solving the climate problems.

[xxxii] A real worry with the hydroelectric generation will be in summer and fall of low water years (and the requirements for fish passage and river temperature). But that is when the long summer days of the NW, particularly on the eastern side of the Cascades, could be a benefit for solar (and maybe wind too).


Copyright © 2021 Renewable Northwest