How Fast Will Electricity Demand Grow?

Source: Dominion Energy Virginia 2018 Integrated Resource Plan

Dominion projects that demand for electricity will increase 1.4% a year over the next 15 years. How accurate is that forecast? Billions of dollars ride on the answer.

The single-most important forecast Dominion Energy Virginia had to make when compiling its 2018 Integrated Resource Plan (IRP), released earlier this week, was to project electricity consumption in its service territory over the next 15 years. All of the company’s calculations regarding the need for new generating capacity spring from that forecast.

Dominion projected that overall electricity demand will increase 0.8% annually on average and that peak demand will increase 1.4%. Based on that forecast, the authors of the plan had to figure out how to increase the peak supply by more than 20%. However, if Dominion is over-estimating load growth and builds a power portfolio to match, it could spend billions of rate payers’ dollars on unneeded capacity.

The company explains its forecast methodology in considerable detail in the IRP. It uses two econometric models, one being a “customer class-level sales model,” and the other a “system-level hourly load model.” These models have been “developed, enhanced, and re-estimated annually for over 20 years,” the IRP says.

The models used this year have been tweaked to represent the findings of the most recent customer appliance survey, which observed a significant increase in the penetration of light-emitting diode (LED) lighting among residential customers. To reflect this trend, Dominion modified its models to reduce forecasts of residential lighting load. Accordingly, residential lighting usage is expected to fall by roughly half between 2018 and 2033.

Another key assumption is the rate of economic growth. Dominion observed that the Virginia economy continue to grow despite federal budget sequestration between 2013 and 2017 by about 1% per year on average (adjusted for inflation).  But Virginia has a favorable business climate, in Dominion’s estimate, and growth should pick up. Looking ahead, Moody’s Analytics projects that Virginia’s Gross State Product will expand at a compounded annual growth rate of 1.99%.

(Moody’s growth forecast is extremely close to that of the Congressional Budget Office, which projects a 1.9% compounded annual growth rate nationally over the next 10 years.)

Also, notes the IRP, Virginia should continue attracting more data centers, drawn by access to fiber optic networks as well as “low-cost, reliable power sources.” Data centers are far more energy-intensive than typical commercial or industrial operations.

The Dominion-PJM gap

Dominion acknowledges that its DOM zone forecast is higher than that of PJM Interconnection, the organization that operates wholesale electricity markets in a 14-state region. PJM has no axe to grind either ideologically or in terms of self-interest when it prepares its forecasts, so its estimates are widely regarded as credible. Dominion critics make frequent note of the discrepancy.

The first thing to note, says the IRP, is that the gap difference the two forecasts has shrunk in the past year. Here I’ve laid the forecast for “energy” (total annual electricity consumption) and “peak” (peak one-day consumption) side by side:

Energy forecast
Dominion — 0.8% annual increase
PJM — 0.9% annual increase

Peak forecast
Dominion — 1.4%
PJM — 0.8%

Thus, if I’m reading the IRP correctly, Dominion’s energy forecast is slightly more conservative than PJM’s — a fact that I’m surprised Dominion didn’t highlight in its IRP.

A big gap still remains between the Dominion and PJM peak forecast. PJM’s forecast, issued Dec. 28, 2017, was rendered out of date almost immediately by 12 consecutive days of bitter cold weather across the northeastern quadrant of the United States in December and January. PJM’s 90/10 projection of peak demand for the DOM zone was 19,512 megawatts. Actual demand exceeded  PJM’s forecast by 1,400 megawatts on five different occasions, says the IRP. (To be fair, PJM’s 90/10 projection did allow for a one in ten chance that the actual peak might exceed the forecast.)

States the IRP: “The Company understands that PJM has recognized this issue in its load forecasting and is in the process of revising their load forecasting methods.”

The critics respond

Dominion critics don’t find the utility’s analysis persuasive. “For at least a decade Dominion has predicted steady robust load growth, and has been wrong every year,” said Southern Environmental Law Center attorney Will Cleveland in a press release yesterday. “It’s time for Dominion to change its model and assumptions to reflect reality—there is less load growth than predicted, and what load is coming to the Commonwealth comes from companies demanding renewable energy options.”

“These misleading projections used as the justification for spending customer money on the Atlantic Coast Pipeline will come back to haunt us,” Greg Buppert, also an SELC attorney. “If Virginia fails to force Dominion to plan around true projections we can expect this modeling to lead to more unneeded projects that will cost Virginians money, their water, land, and air quality.”

Two or three days before Dominion published the IRP, I communicated with William Shobe, director of the Center for Economic Policy studies at the University of Virginia for a related article I was working on. He was in India at the time, and we exchanged emails.

“Actual electricity demand growth over the next several years will not come close to Dominion’s inflated 1.3% growth” as forecast in its 2017 IRP, wrote Shobe. “Something like 0.5% to 0.7% is much more likely. And this is with data center demand.”

Residential demand has been declining the past few years due to improvements in the building stock, he says. Industrial demand is declining as companies pursue energy-efficiency measures. Commercial demand is rising, but only because of the data centers. “Without a growth in demand,” he says, “Dominion’s fossil capacity is over-built.”

Share this article


(comments below)


(comments below)


15 responses to “How Fast Will Electricity Demand Grow?”

  1. LarrytheG Avatar

    re: ” The company explains its forecast methodology in considerable detail in the IRP. It uses two econometric models, one being a “customer class-level sales model,” and the other a “system-level hourly load model.” These models have been “developed, enhanced, and re-estimated annually for over 20 years,” the IRP says.

    two questions:

    why use a model that is different from other existing models or at least compare and contrast Doms model with the other industry models – ?

    If their model is 20 years old – how did it’s forecasts do for the past years?

    More than this:

    if Dominion is buying power from PJM… in past years.. has the amount they are buying – gone up?

    How about the other Co-ops in Virginia? Are they forecasting similar to Dom? Some of the “rural” Co-ops are now serving urbanized areas not rural areas so their growth demands should confirm Doms …

    Finally – if you had a map of Virginia – where are they forecasting growth ?

    As far as I can tell – none of the other states are seeing anything like the increases Dominion is forecasting… and Virginia is not exactly the top of the heap in economic growth – anyhow.

  2. djrippert Avatar

    For once I can see Dominion’s point. Population growth and ever more electric cars seem like drivers of increased energy consumption. Each electric car (driven 15,000 per year) adds $45 to your monthly electricity bill.

    1. TBill Avatar

      Electric cars will not automatically “take off” because there are inconveniences related to them. But we can probably expect national mandates to force EV adoption, and of course the utilities will be forefront of lobbying for the mandates. Basically a new ethanol-type lobby emerging,

      1. djrippert Avatar

        I thought this was a reasonable discussion of the near future of electric cars. The consensus is 15% market share for electric vehicles by the 2030s. Electric cars constitute 1% market share in the US today.

        Dominion is providing a 15 year forecast, so 2019 – 2034 (or so).

        If electric cars grow from 1% market share to 15% market share over that period I guess it will affect electricity use.

        I thought the discussion of the costs of operating electric cars and the interplay between autonomous and electric cars was particularly interesting. Electric cars are designed to last 500,000 miles with 1m miles on the drawing board? Wow.

  3. Steve Haner Avatar
    Steve Haner

    I know the thinking is that if Dominion’s growth forecast is off by a couple of hairs, compounded over time that means there is no need to build one or more of the planned natural gas generation plants, or extend the nukes. I get less worked up about this because the IRP is not actually a chiseled-in-stone capital plan. As the application for each facility comes up in future years, the utility still has to make a case that it is really needed — with then-current data — and the SCC still will have full authority to say nope on that basis (I think.) It is fair to read it as “if this happens, this is what we’ll do” rather than “we know this is going to happen and we are gonna do this no matter what.”

    So it is also very fair to ask, what will they do if the growth does not occur? Which facilities are the most likely to fall off the plan? And it is fair to debate native load vs. buying more through PJM. But again, even if the IRP is adjusted – it’s still just a planning document.

  4. LarrytheG Avatar

    re: planning document… I agree…

    re: electric cars – is Dom forecasting that demand?

    but I’ll go along with the ” we might have to do this and if that time comes – the we’ll have to make the case to the SCC”…. tentatively , given the GA’s latest penchant for “undoing” regulation….

    But also.. essentially – Dom seems to be saying that they’d rather make their own power than buy it from PJM and/or have other independent generators make it and sell it to Dominion via PJM… … as opposed to forecasting a demand and laying out some options for meeting that demand – which might
    well include buying it from PJM..

    When Dominion calculates it’s carbon footprint – does it include the carbon emitted by generators of purchased power? How about the rest of Virginia and the RECs – do they also forecast their demand, calculate their “footprint”, yadda yadda?

    I think Dominion serves about a million homes… and there are maybe 3 million in Virginia. no?

  5. I have not had a chance to review in detail the load growth explanation in this year’s IRP. However, I wanted to make some general comments before I get too far behind with the other posts.

    The data centers are the primary driver of growth in demand for Dominion. The Quanta report commissioned by Dominion concludes that the growth in smaller Category I data centers “will mature and growth . . . will slow”. They go on to say that “it is likely that there will be little warning if and when a slow-down occurs”. “After 2023, no more Category II data centers are forecast to develop, and the load for this group is projected to grow at .7% annually”, according to the Quanta report. After 2023, even the Quanta report projects a tapering off in growth. The study warns that even a significant decline in electricity usage from data centers is possible if the ongoing reductions in power usage by computer chips occur, or new computing technologies are introduced. Dominion takes what it wants from the report, then ignores the rest and projects an unsubstantiated growth in data center demand after 2023.

    Dominion does not include the contribution from customer-sited solar units. Last year this was estimated to be 2,000 MW by 2020. Dominion’s forecasts assume that it will not exist.

    The Dominion forecast does not appropriately consider the contribution of energy efficiency. Dominion claims that its actual data for appliance saturation and efficiencies is lower than the EIA data for the South Atlantic Census Region used in the PJM forecast. Dominion believes that it is more appropriate to adjust its load growth upward to reflect the lagging performance in Virginia rather than adopt the modern practices used in the eight-state region (plus D.C.) of which Virginia is a part. It is understandable that Dominion has not factored in the full impact of the $870 million investment in energy efficiency authorized by the recent energy bill. But Dominion has consistently underestimated the effect of energy efficiency. That is one of the reasons it consistently over-forecasts growth in electricity demand.

    Dominion’s forecasting model equally weights every year’s contribution over the past 30-years for a variety of factors affecting electricity use. The recent dramatic changes in usage patterns get lost in the 30-year average. Many other forecasters believe that recent developments should be given greater weight so that future forecasts reflect recent advances instead of old outmoded habits and technologies.

    Dominion still assumes that a Y% increase in population or GDP will result in an X% increase in electricity usage, even though the past ten years have shown that growth in electricity use is now decoupled from growth in population and economic activity.

    The IRP accurately reflects Dominion’s modeling results, but those results reflect the utility habit of thinking the future will be mostly like the past. Dominion has not significantly revised its modeling techniques from what served them well in the 20th century. But we should not make 21st century plans based on outmoded techniques and assumptions.

    It serves Dominion’s interests to show that more new generation is needed. We should not expect to see different forecasts until we remove the incentive for them to build more.

    The year after year overstatement of growth in electricity usage sounds like a source of pride – look Virginia is growing! This is an old-fashioned way of thinking. If the U.S. can produce more goods and services using less electricity, then Virginia is lagging behind not leading with projected increases in electricity use.

    Electric vehicles can be a great complement to expanded use of solar. Batteries already paid for in the price of the car can be used to soak up excess solar generation during daylight hours. This helps solar producers and EV owners. The energy can be used later for transportation or a portion can be returned to the grid when it is most needed.

    Instead of preparing for a modern future, Dominion wants ratepayers to repay $12-$16 billion to extend our existing nuclear facilities for another 20 years of operation. This would incentivize electric vehicles to recharge at night, completely missing out on the cleaner, lower cost use of solar for EV charging that will be occurring in other states.

    We need a comprehensive way of planning our energy future in a way that serves all Virginians and our utilities. The GA has tied the hands of the SCC. The regulator can only deal with what is in front of them, and the IRP is only a planning document, not a specific request to build something. But it does indicate that Virginia is far out of touch with the national trend in energy usage and how other states are dealing with it.

    Our utilities can be an important part of a modern energy system given the right guidance and incentives. Until that occurs we will continue to review these “stories” that have little value in making good long-term energy decisions for Virginia.

  6. CleanAir&Water Avatar

    Columnist Michael Skapinker, a voice of the global business establishment, recently wrote in the Financial Times, “Taking the Long View is Crucial. The shareholder-first model has damaged public support for business and capitalism.”

    I would love to see our primary utility “take the long view” in their annual IRP. One way to do that would be to begin preparing for EVs. Here’s why … “Recent forecasts have suggested that at a 30% annual growth rate in EV sales would result in as much as 2 million b/d of oil demand being displaced by 2028, while at the same time adding 2,700 TWh to electricity demand globally by 2040.”

    Oil major ExxonMobil sees 50 million EVs on the road by 2040, or an annual 14% increase, compared with BNEF’s 400 million EV’s on the road. Evidently Exxon is counting on hybrids as the next step even though Ford recently announced it will “more than double its investment in EVs, pledging $11 billion through 2022 for the effort. Within the next five years, Ford says it will unveil 16 fully electric vehicles, with the first ones scheduled for 2020.” Probably spurred by their global insight, it seems Ford is “Taking the Long View”.

    Dominion could start preparing for EV adoption, because EVs are the only major change that will actually increase electricity use. “Utilities can install this infrastructure without waiting for connection requests from new charging stations. Equipment that serves the specific operational needs of charging should still be considered part of the grid and rate based.” That would be forward looking and good for Dominion.

    Demand growth numbers from the IRP do not hold up. Moody’s Analytics projects that Virginia’s Gross State Product will expand at a compounded annual growth rate of 1.99%. However, increased energy use and growth in the economy is now acknowledged to be uncoupling. Since 2008, primary energy usage has shrunk 1.7% even as GDP has accelerated by 15.3%”

    Here is the problem as stated by Brent Nelson , an Associate Professor of Mechanical Engineering, Northern Arizona University and fellow blogger on Our Energy Policy … while there is still a positive correlation … “What you do see is that since about 1980, GDP has grown faster that energy consumption, and over the last decade plus, the divergence has accelerated. This divergence is driven by both energy efficiency improvements and sectoral shifts in the economy (becoming more services-intensive and outsourcing manufacturing to other countries).

    Relying on data center growth does not grow demand as Dominion wants either. Many IT companies tha have accepted the pledge at RE100 including Hewlett Packard, Google, Facebook, Equinix, eBay, Autodeck, Apple, Adobe and Microsoft. “RE100 is a collaborative, global initiative uniting more than 100 influential businesses committed to 100% renewable electricity, working to massively increase demand for – and delivery of – renewable energy.”

    One example … “Google is committed to sourcing 100% renewable electricity and is on course to reach this goal in 2017.”

    Finally, from the blog article …
    “Based on that forecast, the authors of the plan had to figure out how to increase the peak supply by more than 20%.”
    On-site solar could cut a large part of peak demand but coupled with offshore wind the synergy between the different peaks of generation fills the utility peak demand. If Dominion could take the LONG VIEW, a Brattle Group report showed electricity demand from a fully electrified transportation fleet in 2050 dwarfs potential lost sales from distributed solar generation by a factor of five.

  7. Let’s set aside differences of opinion about the forecast for a moment.

    When a company in any industry is faced with a gap in an important resource, they have a “make or buy” decision. Dominion always selects the “we will build it” choice, because it highly favors the shareholders.

    The SCC and the AG’s office need to get back in the game. Someone has to speak on behalf of the customers. The state legislators have defaulted on their responsibilities to represent the interests of families and businesses in their districts regarding energy issues.

    We need to return to balance. Other states have realized that the lowest cost, most economically stimulating way to develop a modern energy system is to open it up to participation by other organizations in addition to utilities.

    Utilities have an important role to fill, and they should be fairly paid to do it. They should also be incentivized to do it in an innovative and cost-effective way. But it creates a poorer outcome when the utilities are allowed to own and control it all. Market forces and customer choice have a place in the modern energy system.

    Overbuilding new gas-fired plants has created a surplus of generation in PJM and a low wholesale price. Virginia customers could obtain whatever extra power might be needed much less expensively by buying some of the PJM surplus, rather than allowing Dominion to invest in all of the new projects that ratepayers will end up paying 3-4 times the investment price for.

    Choosing this method gives us more time to see how the trends develop. Solar can be deployed quickly and customer-sited distributed solar costs the ratepayers nothing and benefits the grid more than utility-owned solar added at the transmission level. We should have some of each – balance.

    This also gives us time to develop a genuine energy plan for Virginia. A policy written by the utility and voted on within a few days with little evaluation is not the way to do it.

  8. LarrytheG Avatar

    Some of it comes down to this. Did Kodak see the future but refused to change… trying to hold on to their decades-long business model?

    You can say the same thing about other companies – and industries.

    Some companies get on the bubble and evolve (IBM) – and others don’t have the leaders or the vision to adapt… to see opportunities rather than defend their current model and they “Kodak”.

    so trying to understand DVP’s behavior. Can they be the same company they are now if they become a company that sells less electricity?

  9. Acbar Avatar

    Steve, it is “just a planning document,” but such a critical one that there’s a an entire chapter in the Virginia Code devoted to it. Code sec. 56-599.C reads: “The Commission shall analyze and review [the utility’s annual] integrated resource plan and, after giving notice and opportunity to be heard, the Commission shall make a determination as to whether an IRP is reasonable and is in the public interest.”

    I’m with Tom here on one key aspect of this. He says, “It serves Dominion’s interests to show that more new generation is needed. We should not expect to see different forecasts until we remove the incentive for them to build more.” Yes, and that’s the root of the problem: the continued ratebasing of generation investment, yielding a risk free regulator-fixed rate of return, guaranteed against obsolescence regardless of whether it runs or not due to uncertain future grid conditions. All this when Dominion sits within the most competitive wholesale electricity markets in the U.S., run by PJM, where Dominion could easily hedge its own forecasts and resources by buying from the markets until there is greater clarity about the future grid.

    Who bears the risk if the IRP forecast is too high and, based on it, unneeded generation is built? Not Dominion Energy. That risk falls on ratepayers when Dominion’s generating units are built with ratepayer funds and put in the regulated ratebase. In contrast, who bears the risk if the forecast is too low? That risk falls on PJM, which is charged with keeping the lights on in the PJM region (and with blowing the whistle to the FERC if PJM’s own forecast shows an impending shortfall of total resources to meet total customer demand). So does Dominion have a reason to be conservative, here? Other than passing the laugh test, I can’t find one. If any players have reason to have the overall public interest in mind (as opposed to financial incentives), it’s PJM and its regulators (the PJM states each have a state utility regulatory commission and they also meet together as “OPSI”).

    Now, as for those forecasts: PJM is divided into “zones” in which a given utility traditionally owns nearly all of the transmission wires and provides nearly all of the retail electric service either directly or indirectly (through middlemen like coops). Dominion’s “zone” covers eastern and central and northern Virginia, but it also includes a large chunk of eastern NC. PJM prepares a forecast for each of its zones, using inputs from the companies involved and also from State Commission and federal sources; and Dominion also prepares one for its zone. The PJM and DVP forecasts are intended to be directly comparable. If they differ, there should be alarm bells at the SCC.

    Above, Jim says, “A big gap still remains . . . . PJM’s forecast, issued Dec. 28, 2017, was rendered out of date almost immediately by 12 consecutive days of bitter cold weather across the northeastern quadrant of the United States in December and January.” I disagree! We simply had an extreme event near the front end of the forecast cycle rather than later. Jim should have noted that even Dominion concedes this:  “”To be clear, the 90/10 forecast represents the 90th percentile forecasted peak demand level that, in theory, should only be realized or exceeded once every nine years on average.” OK, this is once. Then Dominion says something that should be controversial: “During the first week of January 2018, the actual peak demand for the DOM Zone exceeded PJM’s 90/10 winter forecast peak demand level for the DOM Zone on five different occasions . . ..” Really? Did one weather event lasting several days cause a single load forecast excursion or five separate excursions? Putting aside the semantics of load forecasting, it should be clear to most consumers that this was one winter event.

    How did PJM react to that event? On February 26, PJM issued a detailed report on it as well as this statement in a press release: “According to the report, overall, there was a significant reduction of forced/unplanned generator outages compared to the winter of 2014 to 2015. The reduction in forced outages is partially attributed to the wind chill impact being lower during the cold snap than it was in 2014 and 2015. Despite stressed conditions and the situation not being 100 percent comparable to the Polar Vortex, forced outages, both at the plant level and in the area of gas supply, were all significantly reduced.”

    Now let’s also put that in proper context: PJM overall is a summer peaking region, meaning, the peak load that generation and other resources must be planned for will occur on the hottest summer day. There should always be enough generation in PJM for a winter event so long as the summer peak is substantially higher. If a generation owner wants to take a unit out of service for maintenance, it must ask PJM to coordinate with all the other owners a few months in advance; accordingly there is a lot of planned maintenance done in the winter, when demands are lower than summer. Importantly, those resources are still there but simply scheduled to be unavailable, and that’s a short term decision that PJM and the owner often can change if the forecast changes. As the press release indicates, PJM had a relatively good experience with generator availability and scheduling during the January event.

    Another important difference is the future of customer-owned solar generation in Virginia. The IRP says in its load forecasting overview that Dominion makes “an adjustment to PJM’s behind-the-meter generation (“BTMG”) solar forecast” for the Dominion zone. You bet it’s an adjustment: from 2000 MW down to zero! See note 6 in Appendix 2I: “Future BTMG . . . is not included in the Base forecast.” It’s not included as an adjustment, either. DVP enters a placeholder of 416 MW for years 2018 through 2033 — is that the amount of rooftop solar that already exists on the DVP system? — with no growth thereafter, and does NOT reduce its forecast loads by this amount; but PJM forecasts DOM Zone growth offsets (and DOES reduce its forecast loads by these offsets) up to 547 MW in the “distributed solar” category.

    There are other differences, including differing forecasts of data center load growth, but the bottom line is, the load forecasts reflected in this IRP need serious scrutiny and inconsistencies resolved by the State Corporation Commission. And we haven’t even discussed Dominion’s proposed generation construction program, yet.

  10. LarrytheG Avatar

    Acbar’s crystal clear commentary makes me wonder about PJM and DOminion in terms of who they are and what they do , their incentives and competencies.

    Dominion is clearly a for-profit trying to maximize it’s profitability.

    PJM is basically a quasi-govt organization with no profit motive.

    Conventional wisdom from “pro-business” folks will assert that the private-sector is efficient and by it’s nature makes cost-effective decisions in the best interest of the company -and it’s customers.

    Conventional wisdom from “pro-business” is that govt-organized institutions are bureaucratic, inefficient and not tuned in to economic realities…

    the second argument is the one the GA has used to try to neuter the SCC.

    So here’s the question.

    Why do we trust and rely on PJM when clearly it’s one of those awful Govt entities that just doesn’t have the incentives that a for-profit private sector entity – like Dominion has?

    Bonus Question – If PJM wrote the IRP for Virginia would we trust it and rely on it for planning?

  11. LarrytheG Avatar

    In one way , I might agree with Dominion in the area of what KIND of gas plants we will need for incorporating more solar and that is the smaller CIT “peakers” type plants rather than base-load behemoths.

    It’s always been a curiosity for me in understand how much 24/7 baseload Virginia requires verses how much total we need under peak load conditions.

    Base load runs all the time 24/7 – and needs to – especially at night and on mild weather days where there are not excess demands due to high or low temps.

    so what is the “split” for Virginia?

    It would seem that we MUST have some minimal amount of baseload to power the state 24/7 and we need gas for the times when we’d expect to be able to harvest solar but it’s not available.

    I think even the enviros call for smaller, regional gas plants…. and more distributed grid and less central utility grid.

    If, in the end – Dominion builds and uses smaller, more distributed gas plants AND more solar – and operates the big baseload plants less – then isn’t that a better way?

    1. Acbar Avatar

      Read this:
      And this:

      These are generic articles. The daily load shapes for PJM are available and used by generation developers to decide whether to build new generation, and what kind. Dominion’s load shape is close to the PJM average. As long as a new generator is built within PJM it really doesn’t matter where, as the generator sells into the same wholesale market and faces the same competition; therefore the cost of land and cheap transmission access and fuel transportation are the most important locational factors for new generation.

      Combustion turbines are cheap to build and very expensive to run. They are a poor choice if you are going to run them more than a short time and more than infrequently. They will get dispatched by PJM very little. But, they satisfy the requirement for capacity that’s available to run, and they look pretty.

  12. […] point of unanimity: Everybody ganged up on Dominion over its forecasted healthy growth in demand for the next 15 […]

Leave a Reply