Every two years, Duke Energy is required to file a plan with utility regulators that outlines different portfolios of new and existing resources that will be available to meet anticipated future energy demand while also attempting to meet carbon reduction targets. This Carbon Plan is developed with computer modeling software (called EnCompass) that is highly sensitive to input assumptions.
After Duke’s proposed Carbon Plan is filed, advocates and interested parties can examine and challenge Duke’s modeling and assumptions. This post gives a detailed look at testimony that identifies points of bias in Duke Energy’s North Carolina Carbon Plan Integrated Resource Plan (CPIRP).
SACE and our allies (Sierra Club and NRDC, represented by SELC, and in partnership with NCSEA) hired Dr. Maria Roumpani, an independent consultant, to examine Duke’s plan and the modeling assumptions. Dr. Roumpani’s extensive analysis identified numerous issues that bias Duke’s plan against the swift replacement of aging, dirty coal plants with renewable energy, and instead cause the plan to favor a major new fleet of fossil gas plants.
Duke presented three “Pathways” that attempt to meet its increasing load forecast, with Pathway 1 retiring coal the earliest and overall being the cleanest of the three, and Pathway 2 being an intermediate option. Pathway 3, Duke’s preferred portfolio, includes 6,800 MW of new combined cycle gas plants, 2,100 MW of new combustion turbines (sometimes called “peakers”), the delayed retirement of parts of its coal fleet, and a five-year delay in complying with the 2030 North Carolina carbon reduction requirements.
Dr. Roumpani’s findings show that Duke overestimates the reliability of fossil resources, underestimates reliability risks and regulatory costs of fossil resources, overestimates the costs of clean energy resources, artificially limits the performance potential of clean energy resources, and completely ignores additional resources that can be utilized to decarbonize the system while reliably meeting the forecasted demand. The result is an artificial cost advantage for Pathway 3 (which proposes delayed climate action) over Pathway 1 (which would include swift coal plant retirements). Dr. Roumpani found that Duke’s artificial modeling limitations make this result “almost pre-determined.”
Solar Build Limits: Within its computer model, Duke set annual build limits on how much solar, wind, and batteries can be added to the grid each year, with the most restrictive limits in the near term. Duke cites interconnection limitations as a reason to limit solar, but Dr. Roumpani notes that they do not include strategies to eliminate these limitations, such as demand side resources, load management options, transmission enhancements, and consideration of alternative load forecasts. (pp. 12-13)
Clean Portfolio Premiums: Duke placed a 20 percent “cost risk premium” on all capital costs in Pathway 1 – the cleanest of the three portfolios. As Dr. Roupmani states, “(T)he Companies take an extra step to undermine the one portfolio that includes higher levels of renewable resources…. This approach is not reasonable, especially because the Company has chosen not to quantify other risks…. The sole purpose of this adder seems to be to undermine P1 when comparing the costs with P2 and P3.” (pp. 78-79) Duke also includes an 8 percent cost adder, declining until 2030, on all supply side resources in all portfolios to reflect cost uncertainties. This adder disappears in 2030, so it only minimally impacts new gas units, but it penalizes faster deployment of clean resources like solar and battery storage.
Reliability Penalty on Renewables: Duke uses a reliability metric called Effective Load Carrying Capability (ELCC) that sharply discounts the value of solar, wind, and batteries. ELCC is a measure of a resource’s ability to send energy to the grid when there may be energy supply shortfalls. Duke does not apply this same measure to coal and gas plants in its EnCompass modeling. Dr. Roumpani notes this results in an uneven playing field. (P. 67) Instead, Duke models coal and gas as if they are almost completely reliable, when in fact they experience outages and are particularly prone to failure during extreme weather. Because Duke’s model assumes that the coal fleet is reliable, when coal retires it overestimates the amount of solar, wind, and batteries that would be needed to take the place of coal.
The unreliability of the coal and fossil gas fleet was included one particular calculation called the reserve margin, but it was not reflected in the remainder of its modeling. The reserve margin is a percentage of extra generation above peak forecasted demand that can be available if power plants or transmission lines are down. If a utility has an efficient and well-maintained fleet, it should have a lower reserve margin, which then lowers the cost to ratepayers. In this instance, however, Duke has incorporated the fleet failures from Winter Storm Elliott into its reserve margin calculation, and Dr. Roumpani noted that this element alone inflated the reserve margin by 2.5 percent (p. 37). So the reliability risk was incorporated where it supported a higher reserve margin, but it was not incorporated in the modeling where it would lower the amount of fossil fuels in the plan. To put some numbers on the impact: Duke has projected a combined revised peak load of over 3,700 MW, so a reserve margin that is 2.5 percent higher would lead to one additional 900 MW gas plant in the plan.
Battery Storage: Duke limits the role of battery energy storage by imposing annual build limits in its modeling, overstating costs, ignoring the grid benefits provided, assuming a 20 percent cost risk premium (mentioned above) to capital costs in the cleaner Pathway 1, and completely omitting long-duration energy storage.
Duke also added “integration costs” for solar and solar plus storage but did not include the flexibility savings that pairing solar with storage provides, thus overstating the cost of these resources. (p. 82) Energy storage that is integrated with solar saves the gas or oil fuel costs that would be incurred by ramping a peaker up and down to manage the variability of the solar.
In addition, Duke has chosen to rely on capital-intensive emerging technologies, such as Small Modular Reactors (SMRs) and hydrogen, while ignoring the rapid development and adoption of more nimble resources such as long-duration energy storage (LDES) technologies. SMRs and gas/hydrogen turbines perpetuate a rigid supply system that cannot adapt to a rapidly changing technology and policy landscape. (Read more about the problems with this rigid plan here.) This locks ratepayers in to both infrastructure costs and fuel supply risks. Duke included hydrogen in its model, but not LDES.
And when Duke vetted the modeling outcomes for reliability, only gas resources were allowed to fill any gaps. Battery storage was not considered, nor were the additional grid benefits that storage provides. (P. 70)
Coal: In Pathway 1, coal retirements are condensed to earlier years where they coincide with strict clean resource build limits, forcing the model to select new gas units because 1) the capacity of retiring coal exceeds Duke’s annual build limit for clean resources and 2) additional options such as long-duration energy storage and demand-side resources are not a selectable option in the model. In modeling of all Pathways, Duke did not allow any coal retirements before 2029, the period with the strictest limits on clean resources. Roumpani noted “Even if one coal unit could economically retire in 2028 and be replaced by solar plus storage, this retirement would not be reflected in the results given the Companies’ modeling constraints.” (p. 21)
Certain coal retirements were artificially delayed in the model in order to wait specifically for new proposed gas capacity to come online rather than opening that replacement capacity up to all resources. In addition, Duke artificially delayed the retirement of the Belews Creek coal plant until 2036 because the site is “well suited” for Advanced Nuclear, an unproven, risky, and likely expensive option. Ratepayers could pay for the most polluting, least reliable resource (coal) while waiting indefinitely for an expensive, never-proven replacement (Advanced Nuclear) instead of converting quickly to well-known solar, wind, storage, and demand-side resources.
Duke’s coal fleet has grown increasingly unreliable as it ages, but this is not captured fully in the modeling. In addition to increasing maintenance issues, the coal fleet has weather-related reliability issues. Coal piles and mechanical parts freeze during extreme low temperatures. As this analysis of Winter Storm Elliott shows, the majority of the power plant failures on the Duke system during that major reliability event occurred within its aging coal fleet:
Source: Roumpani Testimony p. 35, created by South Carolina Office of Regulatory Staff
In addition to these technical biases, Roumpani identifies risks related to coal that are inherently not captured in the modeling, including risks caused by a declining workforce, a supply chain that does not respond quickly to demand volatility, an increased need to rely on higher sulfur coal with related higher environmental compliance costs, reduced economies of scale, and increasing mining costs and rail transportation disruptions. (pp. 28-29)
Finally, Dr. Roumpani points out that the new EPA carbon pollution standards were not incorporated into the modeling, rendering its coal retirement schedule noncompliant. For instance, Duke’s plan would retain two coal-fired units at Roxboro past the 2032 deadline that would require a huge and unaccounted-for financial investment in carbon capture and storage in order to continue operating. (pp 26-27)
Gas: Dr. Roumpani notes that the selection of new gas capacity in the model “stems from an artificial lack of alternatives at a time of high load growth” (emphasis added, p. 47). The annual build limits for solar and battery storage, mentioned above, handicap clean resources in the modeling and result in an overbuild of fossil resources. Dr. Roumpani notes that Duke’s modeling consistently hit predetermined build limits set by Duke for clean resources, suggesting that removing or easing those limits would lead to the selection of additional clean resources instead of gas.
She also reveals that the net cost to upgrade new and existing fossil plants to meet the requirements of the new EPA carbon pollution standards is not reflected in the three portfolios. In an earlier filing, Duke did develop two supplement portfolios that modeled 1) running fossil gas units below the level that would invoke EPA compliance costs and 2) running fossil gas units on hydrogen. The cost of those portfolios increased Duke’s present value revenue requirement by $3.6 billion and $10.5 billion, respectively. These cost impacts were not included, however, in Duke’s most recent filing. (p. 52)
The fuel supply risk of gas is also overlooked. An electricity system fueled by fossil gas is dependent upon the gas supply system. But while reliability of the electricity supply system is overseen by the Federal Energy Regulatory Commission (FERC) and North American Electric Reliability Corporation (NERC), there is no such equivalent agency overseeing the reliability of the fossil gas supply system. In addition to issues at the plant itself, gas power plants can prove unreliable if they do not have fuel because supply or pipeline systems are impacted by extreme weather.
Dr. Roumpani’s recommendation to Duke and to the NCUC is clear: “(T)he Companies should invest in a no-regrets, flexible portfolio, including demand side resources and transmission enhancements, while primarily consisting of modular, scalable, and quickly deployable clean energy resources that mitigate ratepayers’ exposure to fuel price volatility, and the quickly changing market and policy environment.” (p. 16)
The post Duke’s Carbon Plan: Part 2: Flawed Modeling Assumptions Produce Fossil Fuel Bias appeared first on SACE | Southern Alliance for Clean Energy.
After Duke’s proposed Carbon Plan is filed, advocates and interested parties can examine and challenge Duke’s modeling and assumptions. This post gives a detailed look at testimony that identifies points of bias in Duke Energy’s North Carolina Carbon Plan Integrated Resource Plan (CPIRP).
Computer Models are Only as Good – or as Bad – as the Information They’re Fed
SACE and our allies (Sierra Club and NRDC, represented by SELC, and in partnership with NCSEA) hired Dr. Maria Roumpani, an independent consultant, to examine Duke’s plan and the modeling assumptions. Dr. Roumpani’s extensive analysis identified numerous issues that bias Duke’s plan against the swift replacement of aging, dirty coal plants with renewable energy, and instead cause the plan to favor a major new fleet of fossil gas plants.
Duke presented three “Pathways” that attempt to meet its increasing load forecast, with Pathway 1 retiring coal the earliest and overall being the cleanest of the three, and Pathway 2 being an intermediate option. Pathway 3, Duke’s preferred portfolio, includes 6,800 MW of new combined cycle gas plants, 2,100 MW of new combustion turbines (sometimes called “peakers”), the delayed retirement of parts of its coal fleet, and a five-year delay in complying with the 2030 North Carolina carbon reduction requirements.
Duke’s Biases Lead to Skewed Results:
Dr. Roumpani’s findings show that Duke overestimates the reliability of fossil resources, underestimates reliability risks and regulatory costs of fossil resources, overestimates the costs of clean energy resources, artificially limits the performance potential of clean energy resources, and completely ignores additional resources that can be utilized to decarbonize the system while reliably meeting the forecasted demand. The result is an artificial cost advantage for Pathway 3 (which proposes delayed climate action) over Pathway 1 (which would include swift coal plant retirements). Dr. Roumpani found that Duke’s artificial modeling limitations make this result “almost pre-determined.”
Solar Build Limits: Within its computer model, Duke set annual build limits on how much solar, wind, and batteries can be added to the grid each year, with the most restrictive limits in the near term. Duke cites interconnection limitations as a reason to limit solar, but Dr. Roumpani notes that they do not include strategies to eliminate these limitations, such as demand side resources, load management options, transmission enhancements, and consideration of alternative load forecasts. (pp. 12-13)
Clean Portfolio Premiums: Duke placed a 20 percent “cost risk premium” on all capital costs in Pathway 1 – the cleanest of the three portfolios. As Dr. Roupmani states, “(T)he Companies take an extra step to undermine the one portfolio that includes higher levels of renewable resources…. This approach is not reasonable, especially because the Company has chosen not to quantify other risks…. The sole purpose of this adder seems to be to undermine P1 when comparing the costs with P2 and P3.” (pp. 78-79) Duke also includes an 8 percent cost adder, declining until 2030, on all supply side resources in all portfolios to reflect cost uncertainties. This adder disappears in 2030, so it only minimally impacts new gas units, but it penalizes faster deployment of clean resources like solar and battery storage.
Reliability Penalty on Renewables: Duke uses a reliability metric called Effective Load Carrying Capability (ELCC) that sharply discounts the value of solar, wind, and batteries. ELCC is a measure of a resource’s ability to send energy to the grid when there may be energy supply shortfalls. Duke does not apply this same measure to coal and gas plants in its EnCompass modeling. Dr. Roumpani notes this results in an uneven playing field. (P. 67) Instead, Duke models coal and gas as if they are almost completely reliable, when in fact they experience outages and are particularly prone to failure during extreme weather. Because Duke’s model assumes that the coal fleet is reliable, when coal retires it overestimates the amount of solar, wind, and batteries that would be needed to take the place of coal.
The unreliability of the coal and fossil gas fleet was included one particular calculation called the reserve margin, but it was not reflected in the remainder of its modeling. The reserve margin is a percentage of extra generation above peak forecasted demand that can be available if power plants or transmission lines are down. If a utility has an efficient and well-maintained fleet, it should have a lower reserve margin, which then lowers the cost to ratepayers. In this instance, however, Duke has incorporated the fleet failures from Winter Storm Elliott into its reserve margin calculation, and Dr. Roumpani noted that this element alone inflated the reserve margin by 2.5 percent (p. 37). So the reliability risk was incorporated where it supported a higher reserve margin, but it was not incorporated in the modeling where it would lower the amount of fossil fuels in the plan. To put some numbers on the impact: Duke has projected a combined revised peak load of over 3,700 MW, so a reserve margin that is 2.5 percent higher would lead to one additional 900 MW gas plant in the plan.
Battery Storage: Duke limits the role of battery energy storage by imposing annual build limits in its modeling, overstating costs, ignoring the grid benefits provided, assuming a 20 percent cost risk premium (mentioned above) to capital costs in the cleaner Pathway 1, and completely omitting long-duration energy storage.
Duke also added “integration costs” for solar and solar plus storage but did not include the flexibility savings that pairing solar with storage provides, thus overstating the cost of these resources. (p. 82) Energy storage that is integrated with solar saves the gas or oil fuel costs that would be incurred by ramping a peaker up and down to manage the variability of the solar.
In addition, Duke has chosen to rely on capital-intensive emerging technologies, such as Small Modular Reactors (SMRs) and hydrogen, while ignoring the rapid development and adoption of more nimble resources such as long-duration energy storage (LDES) technologies. SMRs and gas/hydrogen turbines perpetuate a rigid supply system that cannot adapt to a rapidly changing technology and policy landscape. (Read more about the problems with this rigid plan here.) This locks ratepayers in to both infrastructure costs and fuel supply risks. Duke included hydrogen in its model, but not LDES.
And when Duke vetted the modeling outcomes for reliability, only gas resources were allowed to fill any gaps. Battery storage was not considered, nor were the additional grid benefits that storage provides. (P. 70)
Coal: In Pathway 1, coal retirements are condensed to earlier years where they coincide with strict clean resource build limits, forcing the model to select new gas units because 1) the capacity of retiring coal exceeds Duke’s annual build limit for clean resources and 2) additional options such as long-duration energy storage and demand-side resources are not a selectable option in the model. In modeling of all Pathways, Duke did not allow any coal retirements before 2029, the period with the strictest limits on clean resources. Roumpani noted “Even if one coal unit could economically retire in 2028 and be replaced by solar plus storage, this retirement would not be reflected in the results given the Companies’ modeling constraints.” (p. 21)
Certain coal retirements were artificially delayed in the model in order to wait specifically for new proposed gas capacity to come online rather than opening that replacement capacity up to all resources. In addition, Duke artificially delayed the retirement of the Belews Creek coal plant until 2036 because the site is “well suited” for Advanced Nuclear, an unproven, risky, and likely expensive option. Ratepayers could pay for the most polluting, least reliable resource (coal) while waiting indefinitely for an expensive, never-proven replacement (Advanced Nuclear) instead of converting quickly to well-known solar, wind, storage, and demand-side resources.
Duke’s coal fleet has grown increasingly unreliable as it ages, but this is not captured fully in the modeling. In addition to increasing maintenance issues, the coal fleet has weather-related reliability issues. Coal piles and mechanical parts freeze during extreme low temperatures. As this analysis of Winter Storm Elliott shows, the majority of the power plant failures on the Duke system during that major reliability event occurred within its aging coal fleet:
Source: Roumpani Testimony p. 35, created by South Carolina Office of Regulatory Staff
In addition to these technical biases, Roumpani identifies risks related to coal that are inherently not captured in the modeling, including risks caused by a declining workforce, a supply chain that does not respond quickly to demand volatility, an increased need to rely on higher sulfur coal with related higher environmental compliance costs, reduced economies of scale, and increasing mining costs and rail transportation disruptions. (pp. 28-29)
Finally, Dr. Roumpani points out that the new EPA carbon pollution standards were not incorporated into the modeling, rendering its coal retirement schedule noncompliant. For instance, Duke’s plan would retain two coal-fired units at Roxboro past the 2032 deadline that would require a huge and unaccounted-for financial investment in carbon capture and storage in order to continue operating. (pp 26-27)
Gas: Dr. Roumpani notes that the selection of new gas capacity in the model “stems from an artificial lack of alternatives at a time of high load growth” (emphasis added, p. 47). The annual build limits for solar and battery storage, mentioned above, handicap clean resources in the modeling and result in an overbuild of fossil resources. Dr. Roumpani notes that Duke’s modeling consistently hit predetermined build limits set by Duke for clean resources, suggesting that removing or easing those limits would lead to the selection of additional clean resources instead of gas.
She also reveals that the net cost to upgrade new and existing fossil plants to meet the requirements of the new EPA carbon pollution standards is not reflected in the three portfolios. In an earlier filing, Duke did develop two supplement portfolios that modeled 1) running fossil gas units below the level that would invoke EPA compliance costs and 2) running fossil gas units on hydrogen. The cost of those portfolios increased Duke’s present value revenue requirement by $3.6 billion and $10.5 billion, respectively. These cost impacts were not included, however, in Duke’s most recent filing. (p. 52)
The fuel supply risk of gas is also overlooked. An electricity system fueled by fossil gas is dependent upon the gas supply system. But while reliability of the electricity supply system is overseen by the Federal Energy Regulatory Commission (FERC) and North American Electric Reliability Corporation (NERC), there is no such equivalent agency overseeing the reliability of the fossil gas supply system. In addition to issues at the plant itself, gas power plants can prove unreliable if they do not have fuel because supply or pipeline systems are impacted by extreme weather.
No Biases, No Regrets
Dr. Roumpani’s recommendation to Duke and to the NCUC is clear: “(T)he Companies should invest in a no-regrets, flexible portfolio, including demand side resources and transmission enhancements, while primarily consisting of modular, scalable, and quickly deployable clean energy resources that mitigate ratepayers’ exposure to fuel price volatility, and the quickly changing market and policy environment.” (p. 16)
The post Duke’s Carbon Plan: Part 2: Flawed Modeling Assumptions Produce Fossil Fuel Bias appeared first on SACE | Southern Alliance for Clean Energy.