Subscribe now

There are four main components of power plant cost.

Fuel is fuel. Some amount must be used just to keep the plant alive even if it’s not exporting power to the grid, but mostly this follows plant usage.

Non-fuel production expenses are all the operating and maintenance related expenses. Some of these change based on how much the plant is run, while others don’t.

Depreciation cost represents the cost of having built the plant, and ratepayers pay a chunk each year that’s designed to reach zero when the plant is retired (that way everyone who gets that plant’s power pays a piece of the capital cost). That’s a sunk cost, regardless of what you do next, but it’s significant, and every time we build a power plant, we lock that cost in for future ratepayers.

When a utility buys power, they pay a purchase price that lumps all these costs together -- when a utility isn’t the owner of a power plant, they pay either a set rate per kilowatt hour (to most independent power producers) or a percentage of annual costs (Bradley Lake hydro), or a rate set based on the cost of all their other plants (very small independent power producers and consumers with excess home solar production).

Data from 2024 and 2025. Excludes a few very small plants and plants that are net energy consumers.

A handful of heavily-used fossil fuel plants make up a large majority of the roughly $500 million cost, and produce most of the electricity. Around $27.5 million is for backup fossil fuel plants used at less than 10% capacity, and the remaining $60 million is a combination of small renewable plants and purchased power, primarily from Bradley Lake hydro, but also from small hydro, wind, solar, and coal.

Per kilowatt hour, you get a different picture. The least-used backup plants are off the chart because they produce so little power that the per unit price is pretty meaningless. But even the heavily-used plants vary widely. GVEA’s coal and oil plants are very expensive. The gas plants run by CEA are relatively cheap. The weighted average cost is around 11 cents per kilowatt hour. Most of the purchased power is cheaper than the weighted average, from various renewable projects and a coal-fired combined heat and power plant in Fairbanks.

The patterned blue bars on the right side of the sheet include overproduction by net metering customers (excess within a month), and very small renewable projects. The purchase price for these isn’t set through a contract, but changes quarter to quarter and tracks the power costs of the utility it’s sold to through “avoided cost” or the “small facility purchase power rate.”

GVEA

Most of GVEA’s power comes from oil and naphtha-burning plants with huge fuel costs, or from coal power with high operating costs. Since there is no longer excess gas in Southcentral, they can’t dilute these costs by buying power from other utilities. This also means GVEA has to run its backup plants more often to meet peak demand. With a higher overall cost, the price it pays to the smallest producers also tracks higher than other utilities.

MEA

MEA only owns one power plant. That plant’s cost per kilowatt hour has increased in recent years, partly due to gas costs, but largely due to lower use, because MEA operates in a power pool with CEA, and most of its power comes from CEA’s more efficient gas plants. It also buys power from a number of small renewable facilities, some under contracts, and some under the automatically calculated rates.

CEA

CEA has access to most of the cheaper power on the Railbelt. Their two main gas plants are more efficient than the other gas plants (MEA and HEA plants use around 17% more gas per kilowatt hour). The gas itself is also cheaper, since CEA gets most of its gas from the gas field it owns. However, it also spends a significant amount on the backup Beluga gas plant, which is little used. CEA also hasn’t filed a 2025 annual report yet, so gas costs are likely a bit higher than shown here.

HEA

HEA has the most expensive gas contract on the Railbelt. Its primary gas plant has a similar efficiency to MEA’s plant, but is used more heavily. Both its backup gas plants are very expensive. Bernice Lake is a very old plant, while Soldotna is one of the newer plants on the Railbelt (2014), but is used at only 5-10% capacity. Combined, these backup plants make up around 10.5% of HEA’s generation costs, and provide about 4% of its power.

Without depreciation

Depreciation is a useful way to think about the economic consequences of what we’ve already done, but if we want to compare new options to existing ones, it makes more sense to look at the fuel and operating costs, since those costs can still change. And those costs are all over the place. Even quite expensive new generation could pencil out immediately if it replaces very expensive GVEA generation, or the lesser used gas plants. But it would be significantly harder to match the current costs of the cheapest gas plants.

But the cost of those cheaper plants is also likely to go up.

With high/imported gas costs

Operating and fuel costs are a moving target. All costs will increase over time, but a major current driver is the Cook Inlet gas crisis. This is what the same data would look like if all natural gas cost $15/Mcf (current costs are around $8-10/Mcf, depending on utility). The weighted average would be around 14 cents per kilowatt hour, and likely higher when shifts in operating costs and other fuel costs are included.

Economic Dispatch

There’s been a lot of talk in recent years about “economic dispatch” on the Railbelt -- where plants would be turned on and used in the order of which is cheapest, rather than which utility owns them. Clearly, the plant costs vary wildly. And one study has shown around $42 million/year in potential benefits -- which is small compared to the overall cost of running the Railbelt (around $975 million), but still meaningful.

However most of these savings come from shifting some power production from expensive GVEA oil and coal plants to cheaper gas plants. That requires not just an organizational solution, but also a solution to the Cook Inlet gas crisis. If there’s no gas to spare (like today), none of that shifting can happen. If the gas is expensive (like in my final chart), the shifting will do much less good. Economic dispatch could set up a useful framework, but likely also requires different generation sources to have a major impact.

Subscribe now

Data notes

Operating costs for each power plant are from RCA annual reports (FERC form 1). Purchased power and fuel costs for each plant are from the same annual reports or from quarterly COPA filings. Depreciation costs are from each utility’s most recent depreciation or cost of service study. I only included costs directly attributed to a specific power plant, which excludes interest payments for debt due to power plants, transmission, and any share of a utility’s more general costs. Most plants have been updated to 2025 values, but some are 2024 based on lagging annual reports from CEA and GVEA. Eva Creek hasn’t reported operating costs since 2022, so I used 2022 operating costs combined with 2024 depreciation and generation numbers.

Subscribe now

This is the third time I’ve written about net metering (read 1 and 2 here), and as a numbers person, I find it frustrating because the debate seems to endlessly circle the same place. In short: “We can’t allow people to get full credit for their solar energy! Look at how they decided to stop that in Hawaii!” vs. “We are not Hawaii!” Read on for the weeds and the math, but yeah, we really aren’t Hawaii, and no one in Alaska needs to worry about their neighbor installing solar panels.

A quick primer: Net metering is the system where a home or business gets charged for the “net” amount of electricity they consume -- the amount they take in from the grid minus the amount they export to the grid. On the Railbelt, this true-up happens every month. In the summer months, this means some people produce more than they use. This extra production gets reimbursed at a lower wholesale rate. Annual net metering basically makes the true-up happen annually, so extra production in the summer can offset use in the winter.

With the loss of federal tax credits, solar panels don’t pay off economically as well as they did before -- annual net metering would help keep them more attractive. On average, annual net metering would put the payoff for folks installing solar panels roughly back to where it was with tax credits and monthly net metering.

In addition to being a big deal for the folks that have them, these solar panels provide a small but real benefit to the system as a whole, reducing the amount of oil and gas burned for power. While utility-scale systems could theoretically have a much bigger benefit, the reality is these home solar systems are one of the few bright spots on a grid that otherwise isn’t doing much of anything at all. Most of the renewable energy added to the Railbelt in the last 5 years is from these distributed solar systems, and they produce around twice as much energy as our utility-scale solar projects (utility-scale solar produced around 8 million kilowatt hours in 2025, while small scale solar produced over 17 million kilowatt hours).

The debates about net metering generally aren’t talking about that. Instead, they’re talking about how to stop net metering from raising other people’s rates.

Why? The basic structure of electric rates means that all else being equal, people using less electricity will make rates go up. I’ve written extensively about this before -- it doesn’t mean that people with solar panels are always paying less than their fair share, that conservation is a subsidy, or that a state should start paying utilities for lost revenue any time it funds a weatherization program or insulates some school buildings.

But those are philosophical arguments that do not matter in Alaska right now, and possibly not for the next hundred years.

We are not Hawaii

You can tell a story about how other states had net metering, which reduced electricity sales so much that they passed new laws to change the rules. It’s a true story -- both California and Hawaii did this. But the key plot point is when they changed those rules -- when small-scale solar was generating around 5 or 6 percent of total energy on the system. Whatever you believe about fairness, that’s enough energy to change the math.

The lesson here isn’t “net metering is a bad idea” -- it’s “net metering is a good incentive system for low levels of solar penetration.” Like Alaska levels.

On the Railbelt, we get around a third of one percent of our energy from small-scale solar, far less than the 2% national average, and not even in the same universe as California and Hawaii. At the current rate of installation (around 2.5 megawatts of new solar each year), we’ll reach the 5 or 6% level where those other states adjusted their laws in a little over 100 years.

Maybe installation rates will increase enough that that century turns into decades. I hope they do. But that isn’t a problem for the current legislature.

SB 150 holds distributed solar hostage to legislative funding, while HB 164 ties guardrails to actual impacts.

In my original post on annual net metering, I discussed the “I’m scared of solar panels” fund, where the legislature can appropriate money to pay utilities for revenue losses due to net metering (though it’s still unclear whether that applies only to power fed into the grid by those panels, or to an estimate of how much power those people would have bought if they didn’t have panels).

Both bills keep the fund, but take it in very different directions. Every Alaskan knows that the legislature is constantly struggling to fund the most basic functions of government. Which means that despite Alaska Energy Authority having some seed money for it, the fund is probably going to be pretty low on the legislative priority list when that’s gone.

SB 150 uses an empty fund as an excuse to cut off net metering entirely for anyone who doesn’t already have solar panels -- not just annual net metering, but also the current monthly version.

(f) A load-entity is not required to provide a net metering service under this section unless the commission determines that the net metering reimbursement fund established under AS 42.45.015 is sufficiently capitalized to adequately offset revenue lost in excess of avoided costs because of the net metering service.

HB 164 allows annual metering to keep going without the fund, instead adding a guardrail that it can’t raise rates for nonparticipants by more than 2%

(e) If funds are not available in the net metering reimbursement fund established under AS 42.45.015 and revenue losses attributable to a load-serving entity’s net metering service would require the entity to request approval from the commission to increase rates for customers who are not consumer-generators by more than two percent annually, the entity may request approval of the commission to modify rates for consumer-generators. A rate modification under this subsection may not increase rates for consumer-generators more than is necessary to avoid increasing rates by more than two percent annually for customers who are not consumer-generators.

This is a reasonable guardrail. A 2% rate increase would be around half a cent per kilowatt hour on the Railbelt today -- not huge, but not nothing either. Right now, a switch to annual net metering would leave rate impacts at or below 0.1% -- completely irrelevant to other consumers, and more than 20 times below the level set by that clause.

Even worse, SB 150 allows utilities to dismantle the current monthly net metering system

(e) A load-serving entity may establish an innovative residential rate structure, under regulations adopted by the commission, that accurately reflects the entity’s costs to provide electrical service to consumer-generators that receive net metering service.

There’s a lot hiding here in the word “innovative” -- under this bill, utilities could charge high fees to solar panel owners, change the reimbursement rates to drastically drop the credit rates for solar power, set high minimum monthly purchase levels, or more. And the “accurate reflection” of costs isn’t any protection. There is no way to accurately determine the costs to provide service to any customer or group of customers. Instead, the “cost-causer, cost-payer” principle is shorthand for a particular crude way of lumping costs into buckets that’s been used for decades to set electric rates.

This sort of change is how Hawaii and California dismantled their original net metering programs -- once they’d reached levels vastly higher than Alaska has today.

In general, the costs of providing electrical service to someone with solar panels are the same as providing it to any person who bought the same amount of energy. A big house with solar panels and a heat pump pays more in the ‘fixed costs’ utilities like to talk about than a small cabin with no solar panels (my old post has actual examples of this)

HB 164 cuts off new installations in 2037

In addition to the 2% rate increase protection, HB 164 sets a sunset date of 2037 for new installations. I don’t think this is necessary, and the rate protection is a much more elegant way of addressing the hypothetical problem -- who knows how much solar will really get installed in the next decade?

Even if we dramatically increase the installation rate, we’re starting at such a low level that ten years won’t get us anywhere close to 2015 Hawaii. Even installing five times as much solar every year for a decade will leave us below the 2% rate increase threshold. Almost certainly the sunset will kick in first.

In the ‘current installation rate’ scenario I assumed that the percentage of net metering and overproduction equals that of today. In the increased installation scenarios, I assumed only 1/3 of power was self-consumed, with 1/3 each over produced or consumed within the month, to account for larger installations being incentivized.

That said, a decade of annual net metering with a guardrail to keep people comfortable with the risks is a pretty good outcome.

In Conclusion -- the legislature should pass HB 164

This is a way easier conclusion than most. There are two bills, one is helpful, and one is worse than status quo. We should pass the better one!

I really hope that we’ll start seeing some meaningful increases in utility-scale renewable generation on the Railbelt. In the meantime, we should help homeowners and businesses keep chipping away at the problem, a few kilowatts at a time.

Subscribe now

An IRP is a roadmap for the electric grid. It’s supposed to model various scenarios and then choose a set of stuff -- power plants, power lines, batteries, efficiency programs, load management programs -- which will be the option with the ‘greatest value’ for the next 20 years, updating every two years along the way.

There have been various Railbelt-wide planning processes over the years, none of which actually got implemented. The post-mortem on these has largely been about the framework -- whether and how the different utilities worked together. For more on that, Phil Wight has an excellent article about the entirety of Railbelt grid history.

Leaving that aside for the moment, what about the plans themselves? What did those 2008-2015 planning documents say about the future we’re living in now in 2026? What did they get right or wrong in their assumptions? Would we be better off if we had followed them?

I’ve been involved in one of the working groups for this new version of the IRP (filling the “small consumer” stakeholder role for AKPIRG in the objectives working group), and I like homework, so I tried to figure it out.

The plans built more stuff than we actually did, for a load bigger than we actually have - the result in 2026 would have been a power cost higher than we pay today. Those costs were front loaded and expected to decrease into the future, but to remain above 2010 costs. In contrast, inflation-adjusted bills have risen around 8% for the average household. The plan depended heavily on large hydro and geothermal projects later determined to be infeasible. It assumed we’d have a gas line already, and that gas would be both more expensive and more available.

Subscribe now

In 2010, we proposed big, expensive, infeasible projects.

I dug up the tables and graphs from the 2010 Railbelt Integrated Resource Plan and did my best to compare the first chunk of that plan (which went to 2060) to what actually happened from 2011-2024.

2010 IRP capital costs from Appendix E. Actual capital costs approximated based on public announcements for each project.

The 2010 plan had us building nearly $5 billion dollars of power plants in the last 15 years. If that sounds like a lot to you, it sounded like a lot to the modelers as well. The financial analysis section (appendix B), admits there is no possible way for the utilities to fund the plan, and suggests that the proposed all-Railbelt organization that was under consideration at the time (GRETC) would sell $5.9 billion in bonds, which could be reduced somewhat if the state chipped in a $2.4 billion loan. This would have added around 7 cents per kilowatt hour in capital costs to rates.

That plan was anchored by several large projects that were soon declared infeasible. Nine months after it came out, the state abandoned work on the Chackachamna hydro project -- the largest project in the plan -- in favor of the Susitna hydro project, citing worse environmental impacts for Chackachamna, more complex construction, and the need to save more water for fish, which would reduce energy production. Susitna never happened either. In 2014, the geothermal company Ormat abandoned its exploration on Mt. Spurr, determining that it wasn’t commercially feasible. The third largest project; Glacier Fork hydro, was never more than a preliminary permit application. It would have been nearly the size of Bradley Lake hydro.

We knew, at the time, that these were all very uncertain: “Mt. Spurr, Glacier Fork, Chakachamna and Susitna should be pursued further to the point that the uncertainties regarding the environmental, geotechnical and capital cost issues become adequately resolved to determine if any of these projects could actually be built.”

Even if the plan had happened, the largest hydro wasn’t scheduled to come online until 2025. Instead, we’d have spent the last 15 years shutting down oil power in the north (taking advantage of that 2018 in-state gas line!), and replacing some of the gas and coal with hydro, wind, and municipal solid waste, as well as “purchases” which I suspect is intended to include the Fire Island wind project.

Instead we doubled down on Cook Inlet gas

Gas plants were the one thing we spent more on than the 2010 plan suggested. They cost over a billion in total, but were small enough to be financed by individual utilities. Each gas-dependent utility bought or upgraded its own.

Did we build too much gas infrastructure?

The 2010 IRP worried we would. They thought the utilities’ go-it-alone approach would cost 5.6% more than the IRP plan. Afterwards Chugach’s 2015 analysis argued that this lack of coordination led to the Railbelt building $200-$500 million more in gas plants than was needed.

It’s kind of hard to untangle this -- can we simultaneously have spent billions of dollars less than we planned and hundreds of millions more than we should have?

All the new gas infrastructure made gas power more efficient than envisioned in the 2010 plan. If we’d produced the same amount of power with the older plants, we would have spent somewhere around $450 million more on gas (using Cook Inlet average prices rather than evaluating utilities’ actual fuel costs, and excluding post-upgrade savings attributed to utility consolidation and power pooling). But those savings are still only a little over a third of what those plants cost.

Some of those new plants are barely used

Could we have saved as much fuel with less expense? We have five gas plants that were built or upgraded in that time frame. In 2024, they had an overall capacity factor of 58%, which is a measurement of how heavily they were used (a plant producing maximum power every single hour of the year would have a capacity factor of 100%). The most used plant had a capacity factor of 82%, which is about as much as you can get in the real world (due to maintenance, etc…). But the least used plant, Soldotna, had a capacity factor of just 10%. It seems likely that we could have built less.

(for these calculations I’m not including the “duct burner” capacity of those plants that have it, because duct burners are an inefficient way to increase output, and wouldn’t be used unless they were needed)

We built efficiency rather than diversification

Gas use in reality was nearly exactly the same as it would have been under the plan. Where the plan anticipated load growth and new power sources, we got more efficient gas generation and a shrinking load.

The gas crisis ramped up more slowly than we thought. So we put off solving it.

The projected fuel trajectory at the time is surprisingly similar to what is being talked about today, where a few years of Cook Inlet gas were going to give way to imports (in 2013) and then an in-state gas line (in 2018), both of which would be substantially more expensive.

Although we didn’t build the infrastructure to get any outside gas supply, we spent quite a bit of money propping up the Cook Inlet gas system, through the CINGSA storage facility and the Cook Inlet oil and gas tax credits, which were mostly paid out as cash to Cook Inlet oil and gas companies. As a result, overall fuel-related spending was similar to the plan’s prediction.

Gas storage and tax credit amounts are 47% of total spend to adjust for the electric utilities’ proportion of Cook Inlet gas consumption

Costs were slated to rise much more than they did

It’s not possible to back out exactly what “wholesale power costs” are today on the Railbelt in the exact way they were looking at them. But average retail rates have risen only 10%, suggesting that wholesale rates probably haven’t jumped 75%.

We got lucky on efficiency

The 2010 IRP said that between 2011 and 2024, the utilities would spend over $100 million on energy efficiency programs for their members, categorized as “demand side management” programs. In fact, this was the one case where they didn’t choose the lowest cost plan that their model spit out -- they thought it would be even better to spend twice as much.

The state did spend over $600 million on energy efficiency in that time frame, through the Home Energy Rebate Program and the Weatherization Program, but those recommended measures primarily targeted heating fuel reductions. And while electric utilities didn’t spend money on electric efficiency, consumers did, reducing their use dramatically. This is why, although residential rates increased 15% over that time frame (inflation adjusted), residential bills only rose around 8%.

It was a bad plan. And the framework wasn’t set up to help anyone make good choices.

The plan made a lot of wrong assumptions. Some of them were egregiously wrong -- like the idea that every large barely-sketched-out energy project would be feasible, and that we could pay for them all. Some of them were more understandably wrong, like forecasts that predicted load would grow rather than shrink, that gas prices would rise more quickly, and that we’d figure out some other source of gas beyond Cook Inlet. Predicting the future is hard, but ideally, you’d have thought through your assumptions and sensitivities well enough that the edges of your ranges would at least end up overlapping with reality.

But really bad assumptions isn’t what I’m most worried about in the current IRP framework. Another big problem is that the framework wasn’t actually set up to help us choose between plans -- and making more reasonable assumptions won’t fix that on its own.

I’ve been critiquing this old plan based on the “1a/1b” scenario highlighted throughout most of the document. They actually had two other scenarios and 17 different sensitivities they modeled for the first scenario.

However, the two other scenarios are irrelevant, because they are scenarios for a dramatic increase in load that never happened. And while the sensitivities help illustrate some risks, they don’t suggest alternatives.

One set of sensitivities varies the assumptions, but keeps the plan constant. For example, if gas prices and capital costs go up, the suggested plan costs more. If there are tax credits for renewable energy, it costs less. Basically, this tells us “if costs change, prices change.”

The other set of sensitivities varies the plan, but keeps the assumptions constant. For example, six of the 17 sensitivities replace Chackachamna hydro with a different version of the Susitna project. They all show Susitna is more expensive. Which is obvious, because the way they got the first plan is by feeding it the default assumptions and asking the model to minimize costs. By definition, anything the model didn’t choose the first time is more expensive under that set of assumptions. This also doesn’t help anyone make decisions.

The only way to make choices based on sensitivities would be to vary both aspects at once -- assumptions and plans. One possibility is to allow the model to choose a new plan based on each different set of assumptions. I.e. “If gas costs are high, does a different plan make more sense?” This seems attractive on its face, and I think it’s what’s being proposed in the current planning process.

That won’t help either. The flaw here is that if you optimize for total cost across a set of different imagined futures, you turn the whole exercise into a contest of “which predictions are most believable?” And you’ll choose the same default assumptions -- after all, those were already the modelers’ best guess.

All of this would just spit out the same bad plan.

We need to model the future to make choices, but we need to do it in a way that recognizes that some of our assumptions are inevitably wrong -- and allows us to make choices that are robust across a wide set of assumptions and also across a varied set of values. That includes wholesale power cost, but also all the other pieces that make up “greatest value,” including stability, resilience and environmental impacts.

What now?

What I’m hoping for this time is something that helps us make good choices in a world that’s even more uncertain than it was in 2010.

We need to vary plans and assumptions independently, so we can compare multiple different plans across the same set of possible futures. To do that, we need some meaningfully different plans to start out with. What if we had made one plan that minimized overall costs, one that minimized emissions, one that minimized up-front capital costs, one that minimized cost volatility, one that focused on lots of small projects, one that focused on a diversity of resources, etc…? We also need an assumption space that is broad enough to encompass most possible futures, across all the relevant dimensions.

Maybe we’ll find plan A is cheapest under a certain future scenario, but super sensitive to changes in those assumptions, while plan B doesn’t look the best in any one of them, but is pretty good across the widest range. We’ll only know that if we set up the model to tell us.

I hope we will.

There’s been a lot of good discussion about what matters to different stakeholders, but not enough clarity, at least from my perspective, if what gets spit out at the end of this all will really let us evaluate those values.

The next public meeting is Tuesday February 17 at 1PM. The public can make comments at the beginning or end or submit them in writing. The draft working objectives document is here.

Last year, the legislature passed the SAVE Act (Saving Alaskans Money with Voluntary Community Energy, SB 152), which allows community solar farms to exist. But basically nothing about how they should exist was written into the bill -- it was all pushed to the RCA to create regulations.

Subscribe now

And hidden in those draft regulations, in the definition of a community energy facility, is this sentence:

The capacity of the community energy facility shall not exceed ten percent of the energy requirements of the subscribers.

This requirement is unmeasurable, and to the extent you can determine it, is far too tiny to be useful, especially for the low-income households that community solar is meant for.

The first problem is that the sentence doesn’t make sense. Capacity is the maximum amount of energy a power plant can produce at once; measured in kilowatts. Energy is how much energy you use over time; measured in kilowatt hours. This is undefinable, since those units can’t be compared directly.

However, it seems reasonable to assume that they mean the energy produced by the facility across the year shouldn’t be more than 10% of what the subscribers use in a year. This is also problematic to define. Energy production and use both vary. You can model energy production, but what about energy use? Do you use an individual consumer’s past energy use? Over what period? What if they’re a new customer? Or buy an electric car? Or their kids move out and go to college? Will a facility dip in and out of compliance as their consumers’ behavior changes?

Most importantly, 10% is tiny. For an average Railbelt household, 10% of energy consumption is equal to $10-$16 dollars on a monthly bill. If each person is only allowed to subscribe to enough panels to equal 10% of their energy, then all they can get is around $13 in credits. Since they will have to pay something for the subscription (this is how the solar panels get paid for), the most they could save would probably be a few dollars per month.

If a developer wanted to build a 1 megawatt solar farm, they’d have to find nearly 1500 households willing to subscribe, all of whom were willing to go through the trouble of signing up to see their bills basically unchanged. And there are costs to administer those subscriptions. Who would ever do that?

People with rooftop solar usually have systems around 7 times this size. Chugach’s community solar project (which got RCA approval before the community solar bill was passed) allows people to subscribe to enough panels to get 13 times more energy than this would allow them to.

Is there an alternative to selling a lot of mostly useless tiny subscriptions? The only one I can see is selling some extremely useless subscriptions to allow you to size up the other ones. The provision does say that it’s the aggregate 10% that’s important, not the individual 10%. So, I suppose you could beg a large consumer like a refinery or a mine to subscribe to a panel or two, which would instantly jump the allowed capacity to a level that would allow it to be useful for everyone else. But that seems crazy.

This provision is especially bad for the lower income people community solar is intended for. Since it costs money for a developer to administer a subscription, it makes sense to try and sign up the biggest consumers possible -- those where 10% of their energy use is a big number. Low income families often live in small spaces and use significantly less electricity than average.

The RCA should just delete that line. The regulations have other provisions that allow the utilities to set the maximum allowed capacity of community solar farms, and define that these facilities are supposed to produce within subscribers’ needs, and that excess unsubscribed power should be compensated at a utility’s avoided cost. Therefore, there are already plenty of guardrails to make sure this isn’t creating a backdoor way for developers to build utility scale power plants and get a rate intended for consumers.

Smaller Problems:

That one sentence is the only thing that renders the whole program useless, but there are other things I’d change. The utilities can set the maximum size of facilities, but there aren’t any limits on the number they can choose. It makes sense to let different sized utilities decide whether they want say, 1 or 2 megawatts as a limit, but it makes no sense to allow them to kill the program in their tariff by setting a 1 kilowatt size. Also, all the language about bill credits and excess power mirrors the current language on monthly net metering for rooftop solar. This makes sense now, but there is a bill that might change that to annual net metering. If that happens, the community solar rules should change to match, rather than having two different credit frameworks. It should be defined as working equivalently to net metering, so they can switch, or not, together.

To comment, go through this form, or email the RCA at rca.mail@alaska.gov, and reference docket number R-24-004.

Share

Subscribe now

The investment tax credit (ITC -- 48E) and the production tax credit (PTC -- 45Y) are the primary federal tax credits for renewable energy. While they’ve been around in some form for my entire life, Alaska hasn’t had much chance to take advantage of them, because the credits didn’t usually apply to our system of tiny publicly-owned utilities.

Now they do. The current version of these tax credits, as passed in the Inflation Reduction Act (IRA) in 2022, is vastly more beneficial to Alaska than any previous version has been. And these improved credits come at a time when grants are much less likely to be available. But just as Alaskan communities are gearing up to take advantage of the credits they finally qualify for, they’re poised to disappear.

The current House bill phases out the utility tax credits on an accelerated schedule, and requires a project to be fully operational before claiming them. Provisions on material sourcing are so complicated and onerous that no Alaskan entity can likely meet the tracking requirements. Residential solar credits end entirely at the end of the year. The practical effect is the same as eliminating the credits entirely.

Subscribe now

History of the investment and production tax credits

The investment tax credit has existed in some form since 1978. It gives a utility or developer a tax credit for a percentage of their investment in a renewable energy project. The production tax credit began in 1992. It gives credits out over time based on energy produced, on a cents per kilowatt hour basis.

These credits have been part of the backdrop of energy policy for more than 40 years, working alongside other supports for fossil fuels. They’ve also been subject to the winds of politics for all of that time. Nearly twenty different pieces of legislation extended or altered them. For most of their existence, these credits applied only to solar and wind energy. Until 2022, only taxpaying businesses could qualify.

Most utility-scale renewables in Alaska weren’t eligible in the past

Over 90% of Alaska’s existing renewable electricity comes from hydropower. Hydro wasn’t included in the credit system until 2005, and even afterwards it only received half the credits that wind did.

However, that didn’t actually matter here, because nearly all of our power projects were built by entities that don’t pay federal taxes. The vast majority of Alaska communities are served by non-profit electric cooperatives. Some major projects were built by the state. Neither the state nor the co-ops could benefit from these tax breaks. While there was always a possibility to set up a complicated financing structure with a for-profit company, Alaskan projects weren’t big enough to make that worthwhile.

The Inflation Reduction Act changed the credits in Alaska’s favor

The IRA made the credits ‘elective pay’ so non taxable entities could take a tax credit directly as cash. Immediately, this makes every owner of a power project eligible, including the state, cities, tribes, and the non-profit electric cooperatives that provide most of Alaska’s power.

It also made the credits technology-neutral. The Alaska Energy Authority can get the credits for its expansion of Bradley Lake, and diesel-dependent communities can get them for new hydro projects that have been long planned but difficult to fund. In some communities, particularly in Southeast Alaska, hydro may be the only non-diesel power option available.

And it made the credits bigger. The baseline credit level (after meeting prevailing wage and apprenticeship requirements) is 30%. And it also added various ways to stack 10% bonus credits on top of that baseline. The entire state of Alaska qualifies as an “energy community,” bumping the baseline up to 40%. Additionally, nearly all of our small communities would qualify for a 10% low income community bonus. While that only applies to projects less than 5MW, those are exactly what our small communities need.

Even the few for-profit independent power producers in the state that benefited from the old credits will benefit more from the new ones. Because they’re bigger, and because they’re extended. Alaska’s never been able to attract independent power producers from the lower 48. Our independent power producers are small home-grown enterprises often just getting off the ground. Our public utilities never qualified for the credits, and our independent power producers haven’t had time to grow to take advantage of them.

So now, our nascent independent power producer companies can still use the tax credits to lower the prices of the power they offer -- and they can lower them even more. Additionally, co-ops, municipalities, and the state can get the credits for their own projects. And these credits work for hydropower, geothermal, and any other non emitting technology.

What will we lose if they disappear?

Adding up potential credits (estimating capital costs and production) for Dixon Diversion hydro, Shovel Creek and Little Mount Susitna wind, and a Kenai Peninsula solar farm, the Railbelt alone stands to lose around $450 million dollars. Dixon Diversion is only potentially eligible because of the new elective pay provisions from the IRA -- but couldn’t possibly get built in time to use them. Likewise, HEA is pursuing a solar farm on its own, and needs elective pay to qualify.

It’s pretty obvious that removing a credit that makes energy cheaper will make energy more expensive. An annual study of US energy costs shows that over the last ten years, the credits have made solar and wind energy 15-20% cheaper on the high end of the cost range (where Alaska would fall). The current credits are more generous. Any project eligible for the production tax credit will cost 3 cents per kilowatt hour more without it.

If it gets built at all. Recently, Alaska’s largest solar developer pulled out of all their planned projects based on uncertainty over tax credits. Their Puppy Dog Lake project on the Kenai Peninsula would have been Alaska’s largest, and could have received over $20 million in tax credits to lower the power costs to customers.

The impacts of the tax credits on smaller utilities’ projects are harder to assess. While grants often cover a significant portion of the capital costs, the remaining matching costs can be a major burden, especially for projects that exceed the $2 million per project limit for state Renewable Energy Fund grants. The state is in fiscal crisis, and the federal government is freezing and clawing back any grants it can. If a rural utility can’t fund its matching costs today, it won’t likely get another chance.

The economics of rooftop solar get dramatically worse

Rooftop solar systems have been eligible for a tax credit since 2005. In 2010, Alaska passed a net metering law allowing people on the Railbelt to connect solar panels to the grid and get credit for that energy. Since then, around 3000 Railbelt homes and businesses have installed grid connected solar panels. Additionally, there are off grid houses and cabins all across the state, many of which have solar panels.

For homeowners putting up rooftop solar panels, the loss of the tax credits turns home solar from a decent investment in most places to something that won’t break even for many.

How does the House Budget kill the credits?

Residential solar ends this year

Homes wanting rooftop solar panels have to get them running by the end of the year to qualify for the 30% credit.

The accelerated phase-out ties eligibility to ‘in service’ date

The ITC and PTC credits phase out quickly. 80% of the baseline in 2029, 60% in 2030, 40% in 2031, and nothing after that. Historically, the eligibility date was determined by the start of construction. Now, it’s tied to the project being operational. Alaska sits at the far end of every supply chain; weather, funding, and logistics delays are normal. A halfway built project could lose the credit for factors outside the utility’s control

The “Prohibited foreign entity” requirements add likely impossible paperwork, even for projects that move fast.

The even bigger problem is the “Prohibited-foreign-entity” requirements. This prevents utilities from getting “material assistance” from countries our government deems problematic -- like China. But “material assistance” is defined so broadly that utilities and developers would have to prove that every wire, switch, and chip in a project contains no material that ever passed through or was processed by a prohibited country. Even huge companies will have trouble with that. Our small co-ops (and even the Railbelt ones are small) don’t have nearly the manpower or resources to have a hope of tracking that. There’s a one year grace period on this, but no more. But we could face claw-backs years later for an honest mistake.

Our expensive power will get more expensive.

Our power started out expensive and is already getting more expensive. Railbelt utilities’ rates have risen 10% in the past year and a half. Fuel costs are driving those up further. Rural rates have always been incredibly high. Losing the tax credits just as Alaska is finally able to take advantage of them will make everything more expensive. More importantly, it will make it a lot harder for anything to get off the ground. Capital costs have always been a huge barrier here, where our utilities are small, and everything is remote and expensive.

The uncertainty itself is an additional hurdle. Planning Alaska’s future grids, we are already trying to juggle unknown future fuel costs. The diesel that powers many of our communities has always been volatile, and the import replacement for Cook Inlet gas will probably be expensive, but also unknown, and likely volatile itself. Tariffs and the global economy add the same uncertainty for materials and supplies. Taking all of that, and then trying to sign a contract or make a construction plan without knowing whether you get a 40% tax credit or not just makes everything that much harder.

Since our utilities are small and mostly publicly-owned, getting something wrong (thinking a project can get built faster than it does, thinking a product is free from foreign material when it actually contains some), is a direct hit on the ratepayers.

These tax credits finally work for Alaska. Alaska should fight to keep them.

Share

Subscribe now

Subscribe now

The fund sets up a bureaucratic structure to protect ratepayers against a single source of potentially tiny rate increases, while ignoring all the other sources of much larger rate increases. The point of including it in the bill seems to be mostly that people are far more worried about the impact of solar panels than the numbers suggest, and an “I’m scared of solar panels” fund could cut off these arguments by forcing people to do actual math. Both the new bill and the existing law allow utilities to limit solar installations as needed to protect grid stability and reliability.

There are a number of key questions here, most of which were asked by the legislators in the bill’s April 8 hearing, but none of which the presenters from AEA gave much of an answer to.

How many people would this impact today? How would it change the incentives for future rooftop solar installations? How much revenue could be “lost” by utilities due to rooftop solar? How big would that fund need to be?

An important backdrop to any conversation about rooftop solar is that when people buy less power, per kilowatt hour rates increase. Solar panels are a particularly modern way of buying less power that has increased dramatically in some parts of the world (generally sunnier ones than Alaska), and set off debates about their impact, and whether electric rate structures are fair in the first place. I’ve got a whole other post about that, but the simple upshot is that this is far too tiny a rate impact to matter here.

Today I’m going to take my old analysis of how annual net metering would change the bills for people who have solar panels, and how more net metering under current law would impact overall electric rates, update both, and put them together to take a stab at those key questions.

(I’ll refer to net metering systems as solar systems throughout. The law allows other forms of energy, but it’s almost all solar. I’m also going to model everything based on residential customers and rates. Small business rates are generally similar, and industrial customers tend to pay less per kilowatt hour and would benefit less from net metering.)

Share

With existing installations, not much would change

Utilities buy around 15% of member-generated power at wholesale rates. Under the proposed law, they’d spend a total of $300,000 more to credit the same power at retail rates.

Currently, anyone who produces more power than they use in any given month sells the excess to their utility at a wholesale rate, for about a third of the retail value.

GVEA only reports purchased power, not power fed into the grid, so I assumed the ratio of purchased to net-metered power was equal to the other utilities.

Not much power is sold that way. Utilities buy back only 15% of the power produced by members’ solar systems. Around a third is net metered; credited against a customer’s bill at the retail rate.

A little over half the power doesn’t show up in utility calculations at all. We know roughly how much solar systems produce, but the utilities report receiving much less. Why is that? Most of this power is likely self-consumed; made by the solar panel and used by the house before it ever gets to the meter. Some of it probably really doesn’t exist (solar panels producing less than they theoretically should). And some power may have flowed out to the grid at a time scale shorter than the utility is measuring. For example, if a house was a net importer of energy for a given 15 minute increment (or whatever scale they’re measuring), the utility can’t tell if they were actually exporting for 5 of those minutes, but that export was smaller than what they imported in the other 10 minutes.

If net metering becomes annual, that ‘purchased by utility’ power would turn into credits rolled over to a different month, where they could be credited at retail rate instead. (Assuming any excess production in summer months is balanced by low production in winter - almost certainly true for households with solar.)

The increased costs to the utilities would add up to around $300,000 per year across the Railbelt, saving 3000 net metering customers an average of $100 each. Railbelt utilities wouldn’t notice the less than 0.1% loss to their annual revenue, whether or not the legislature appropriated the money to pay them back for it.

Annual net metering would encourage people to install larger solar systems, and make solar a better investment

The best size to build, under annual net metering, is a system that produces the same amount of power you use in a year.

These graphs show what the bill savings would look like for each utility’s average customer with solar panels, now and under the annual net metering proposal.

Monthly residential consumption from 2022 sales, Q2 2025 electric rates, solar capacity based on real-world data where available, PV watts modeling otherwise. HEA has a ‘system delivery charge’ for low-consumption customers.

Of course, bill savings are only half the equation. One way to try and determine whether it makes sense to install solar panels is to look at them like an investment, and calculate an ‘internal rate of return.’ Doing this requires a whole lot of assumptions. So rather than looking at the exact numbers in my graphs, it’s better to look at the pattern. Your specific rate of return will be shifted higher or lower based on individual circumstances. Factors that impact electric rates or panel costs will shift them all higher or lower together. But the way the values change with system size and annual net metering will stay the same.

Solar panels are always a better deal in places with higher rates. Annual net metering makes the math much more attractive, especially for larger systems.

Assumptions: Tax credits continue at 30%, rates increase 3% per year, initial costs based on system size here, monthly energy use equal to average residential use for that utility in 2022.

Net metering rules allow systems up to 25KW, much bigger than I’ve shown here. Will people just build bigger and bigger solar systems with annual net metering? No, because the credits expire after a year, and there doesn’t appear to be a mechanism in the bill to credit that excess annual power at wholesale rates. Anything you produce over your own annual needs becomes literally worthless. (This is likely why the bill allows people to elect to stay under the current structure -- someone who wants to produce and sell lots of power at the wholesale rate can still do it).

Rooftop solar produces less than half a percent of Railbelt energy today.

Right now, there’s around 17,500 kilowatts of net metered solar on the Railbelt, producing around 0.35% of our annual energy. In recent years, homes and businesses on the Railbelt have installed around 2,000 kilowatts of solar each year. At that rate, it will take around 15 more years to get to a point where net metering produces 1% of our energy.

However you measure it, this tiny amount of energy equates to a tiny revenue loss for the utilities.

The bill allows a utility to “seek recovery of revenue losses attributable to the entity's net metering service,” through a system where the Regulatory Commission of Alaska would set up rules for how to count the losses, the legislature would put money (or choose not to) into a fund to pay back those losses, which would be administered by the Alaska Energy Authority.

The structure is reminiscent of the Power Cost Equalization fund, but the similarity ends there. Power Cost Equalization is “equalization.” The state built cost saving hydropower projects for the Railbelt, Kodiak and Southeast, and set up an interest-earning fund to try and give similar benefits to small communities. Power Cost Equalization is also very important to the people receiving it. It lowers residential rates on the order of 20 or 30 cents per kilowatt hour. In contrast, with changes to net metering rules, there’s no particular reason the state owes a subsidy to Railbelt electric customers. The theoretical rate increases due to other people’s solar panels are irrelevantly small, and dwarfed by all the other rate increases the state won’t help you with.

The point of including it in the bill seems to be mostly that people are far more worried about the impact of solar panels than the numbers suggest, and an “I’m scared of solar panels” fund could cut off these arguments by forcing people to do actual math.

So, how big would the “I’m scared of solar panels” fund need to be?

It depends on how much solar is installed, and how ‘revenue loss’ is counted. There are three obvious ways I can think of to count revenue loss.

Difference vs Current Law: The revenue loss is the difference between the current monthly net metering framework and the proposed annual net metering framework.

Power ‘Fed Into the Grid’: Here, we take not just the power the utility is already buying at wholesale rates, but all the power they receive from customers. This would be a way of paying back the current cost of net metering, rather than only the new costs from letting the credits carry over month to month. It’s the difference between the retail rate a utility pays for the customers’ power and what the utility would have spent to produce the same power.

Self-consumed Power That People Might Have Bought: In this case, you calculate how much power all the rooftop solar panels should produce, and assume that it would have otherwise been bought from the utility. This assumption is probably wrong, partly because some systems might under produce, and mostly because energy production and consumption aren’t independent.

The one time HEA actually measured before/after electricity use for net metering customers, they found that the average customer who’d installed solar panels reduced their purchases only 485 kilowatt hours from April - June (comparing 2018 to 2020). A 4 kilowatt solar system (closer to the average in 2019) should produce about three times that much. Therefore, those customers must have been actually consuming substantially more, on average, after they bought their solar panels. It’s hard to know exactly why. Sometimes people purchase solar panels precisely to go with other electricity-intensive purchases, like an electric car or heat pump. Sometimes people use more energy once they have solar panels because it feels free.

So the theoretical generation is an overestimate, but it’s what I modeled below.

If installations increase a lot, the production becomes meaningful, but revenue losses are still very small on the Railbelt scale

In the ‘current installation rate’ scenario I assumed that the percentage of net metering and overproduction equals that of today. In the increased installation scenarios, I assumed only 1/3 of power was self-consumed, with 1/3 each over produced or consumed within the month, to account for larger installations being incentivized.

If solar installation rates increase dramatically with annual net metering, the power produced starts to become a lot more meaningful. If ten times as much got installed every year; 20 megawatts instead of 2 megawatts; those solar panels would be producing nearly 5% of our power by 2035, saving over a billion cubic feet of gas (southern utilities) plus millions of gallons of oil (GVEA) per year.

Theoretical revenue losses would also increase, to over $30 million dollars per year with the most dramatic increase in installations and the most generous way of counting losses. This sounds big, but the rate increase you’d be preventing is still well under a cent per kilowatt hour.

It’s very weird for the state to pay utilities because a law encourages conservation

The most extreme bar in my graphs, looking at the impact of home solar based on “Power People Might Have Bought”, is often how net metering is considered. But it’s bizarre it is to pay utilities if we create a policy that leads to people buying less of their energy, and we don’t do it in any other case. For instance, when the state ran the home energy rebate program from 2008-2018, they put a substantial amount of state money into reducing energy use in homes. Afterwards, the state bragged about the $261 million people saved in home energy bills. These savings were considered a great success, and no one ever considered paying utilities to offset any increased costs for non-participating houses.

Forcing utilities to actually do the math might get us out of a cycle of unreasonable fear of solar panels, allowing them to focus on bigger things.

While philosophically the fund doesn’t make much sense to me, maybe it’s a great idea, since it will force utilities to spell out actual economic impacts of net metering under some sort of formula. That would be a marked improvement over the current state where utilities are spending time, energy, and RCA filings painting rooftop solar as a bogeyman when they should be worrying about bigger things.

The person presenting the bill at the April 8 hearing mentioned that utilities, particularly MEA, were already worried about the rate impacts from net metering.

So, if this passes, MEA could go to the legislature and say, hey, under annual net metering, the rooftop solar already on our grid is costing the other customers around $60,000 to $390,000 per year (depending on which method is used to count it). If you allocate money to the fund, we can lower rates a tiny fraction of 1 cent per kilowatt hour.

If I were a legislator, I’d focus on other issues that can have vastly larger impacts on household finances. But the option is there for them to decide otherwise.

Conclusion

In conclusion, making net metering annual would increase the economic benefits of rooftop solar, and incentivize people to install larger systems. It would save some people a lot on their energy bills, and give the solar industry more business. It’s impossible to know how much difference that incentive would make to overall installation rates. If it makes a big difference, the amount of power produced and fuel saved by these solar systems could rise from tiny to meaningful, but would still be small.

The fund to pay back utilities for revenue they might lose through these net metering systems seems like an awkward bit of bureaucracy to solve a non-existent problem, protecting customers from miniscule potential rate increases that are dwarfed by the actual rate increases we don’t protect them from. But if it forces people to actually do the math, the “I’m scared of solar panels” fund might stop us from arguing about the theoretical impacts of rooftop solar so we can focus on the much less theoretical impacts of the Cook Inlet gas crisis.

Share

Subscribe now

What does it do? Can we meet the goals? What happens if we don’t meet the goals? What will this save or cost customers?

It requires the Railbelt to reach 40% renewable energy in 2030, and 55% in 2035, with fines for non compliance. The 2030 goal could be met by already-proposed projects without new transmission. The worst case scenario, where no one even attempts to comply and no new projects are built, could raise bills by 5.5% in 2030, and 8.5% in 2035. If utilities do comply, costs might be about the same, or bills could be lowered by as much as 10%, depending on currently unknowable economic details. Renewable energy projects have their big costs up front, so these new projects will provide some price stability and a hedge against volatile future fuel prices.

Enacting the RPS requirements could attract competition to our tiny energy market, providing pressure on both renewable energy and gas prices. Implementing them would also be a substantial step toward addressing the natural gas crisis.

Subscribe now

Over the next 10 years, HB 153 would bring the Railbelt from 15% renewable to around half renewable.

Currently, the Railbelt gets around 15% of its energy from renewable sources. This bill requires the Railbelt to get 40% of it’s energy from renewable sources by the end of 2030, rising to 55% by the end of 2035. While the last RPS bill set both a long term ambitious goal (80% by 2040), and a short-term smaller goal (25% by 2027), this bill chops off both those ends, focusing on medium-sized goals over a 5-10 year time frame.

Renewable energy is defined in a generally conventional way, including hydro, solar, wind, geothermal, tidal, and biomass. This is in contrast to a previous Clean Energy Standard that attempted to define all our coal plants as clean.

The energy generally has to be physically produced on the Railbelt, but utilities are also allowed to meet the requirements with credits bought from rural Alaska communities eligible for Power Cost Equalization (PCE). There are extra incentives for large wind projects that provide energy to multiple utilities, and for distributed generation (such as rooftop solar). These incentives take the form of multipliers, where the energy produced is counted as more than it actually is for the purpose of meeting the standard. Distributed generation gets a 2x multiplier, while large shared wind projects get a 1.25x multiplier. The biggest in-the-pipeline renewable projects on the Railbelt are wind projects that would qualify for this multiplier.

The 2030 goal could be met by already-proposed projects. And we’d only be 2% short of the 2035 goal

While there are many possible ways to meet the standard, I’ve analyzed a set of in-the-pipeline projects with publicly available production numbers and timelines, where utilities have already modeled integrating the power into their systems.

By far the largest chunk here is Alaska Renewables’ two wind projects at Shovel Creek and Little Mount Susitna. They are still pursuing both projects, and say that their projects do not depend on the continuation of federal tax credits. Importantly, they also don’t require any new transmission, according to an extensive integration study commissioned by the Railbelt utilities. The Railbelt has often been described as a “long extension cord,” and while new transmission would have benefits, there is a lot that can be built before that happens. This scenario would require the utilities to work together using “economic dispatch,” which means they’d run the cheapest power plants first, regardless of where they are and who owns them. This pair of projects, on their own, more than meet the 2030 goal.

The other projects I’ve included here are the state’s Dixon Diversion project (which would add more water to the Bradley Lake hydro facility), and a project modeled on the Puppy Dog Lake solar project. Though the Puppy Dog Lake contract was recently canceled, it’s possible that that project could go forward in a different fashion. Even if not, the production numbers in the contract are a reasonable estimate of what a similar solar project might produce, and HEA already modeled its integration into their system. Since both of these are on the southern end of the Railbelt, where transmission is more constrained (the bottleneck is South to North), it’s possible that adding these to the wind would require additional transmission. But they aren’t required to meet the 2030 goal.

I also modeled that all existing renewable projects continue to produce at recent average levels, and that distributed solar increases at a recent average rate. Potential renewable energy credits purchased from PCE-eligible communities aren’t included in the graph, but could cover around 1-2% of Railbelt energy. For simplicity, I’ve assumed flat electricity loads into the future. This matches recent history, but of course, the future might be different. If loads rise, more renewable energy will be required to hit the goals, but the increased loads will drive rates down (through spreading out the non-fuel fixed costs of the system).

In the worst case, the Renewable Portfolio Standard could raise costs 5.5-8.5%

Let’s do the worst case first. This is really a straw man, but is useful to get a sense of scale. Potential costs to ratepayers are driven by the fines the bill imposes for non-compliance. So to get the worst-case scenario, we need to assume that all possible renewable energy projects are more expensive than the cost of existing energy plus the proposed fines, and therefore absolutely no new renewable projects are built in the next ten years. We also need to assume that the utilities are pretty much ignoring the law, and that none of them manage to get any fines waived for “good faith efforts” to comply, factors outside of their control, or signing contracts for facilities that aren’t built yet.

When I wrote about the last Renewable Portfolio Standard proposal, I said that the fines that would be imposed on the utilities for not complying were so small as to be meaningless. That is no longer true in this version. First, the fines have been raised from $20 per megawatt hour to $45. More importantly, they’ve been adjusted for inflation (any dollar number in a piece of legislation should be adjusted for inflation!).

I assume here that non-fuel costs rise exactly with inflation, and fuel costs rise 1% faster than inflation. Gas costs will likely rise more, making the percentage impact of any fine smaller.

In this “do nothing” scenario, fines would kick in at around $50 million total across the Railbelt in 2031. That represents around 5.5% of total Railbelt revenue. So, it could raise bills about that amount over what they would otherwise have been. If everyone continued to do nothing for the following five years, that would reach 8.5%. This is meaningful, but not earth-shattering.

In my opinion, meaningful but not earth-shattering is probably a good place to be for fines that are intended to drive action. Without meaningful incentives, nothing tends to happen. This can easily be seen in the 50% renewable target the state was supposed to meet this year (set by the Palin administration), as well as voluntary goals set and missed by individual Railbelt utilities. On the other hand, while customers are responsible for their utilities’ decisions (these are all democratically-run cooperatives), it’s nice if individual customers aren’t entirely screwed over if their utilities make poor choices. It’s a balance.

A more reasonable bad case might be kind of a wash

It seems quite unlikely that all the Railbelt utilities would neither build/purchase any renewable energy nor get any waivers. But what if they do, and renewables are expensive? A lot of factors impact renewable energy prices, including interest rates and federal tax credits. Tax credits, as they exist today, could provide a 40% tax break to these projects. But that might be zero tomorrow. Our existing energy also has super uncertain costs, as we move to imported gas from one or two unbuilt import facilities, with unknown costs, and a supply price driven by global markets.

From that impossible mess, I’ll try and pull what numbers I can. The economics of Railbelt power plants are public data, so it’s easy to substitute higher gas prices into existing generation cost data.

Current gas generation costs are around 7.8 cents per kilowatt hour (total cost of power as it leaves the power plant, for all gas plants on the Railbelt in 2023). This will rise as gas from old Hilcorp contracts and Chugach Electric’s gas field is replaced by newer more expensive contracts and expensive imports.

Future gas generation costs might range from 11 cents per kilowatt hour (most efficient gas plant with Furie’s $12.30/Mcf gas) to 16 cents per kilowatt hour (less efficient plant with $16 gas). Oil generation costs and Healy 2 coal generation costs are already in that range or higher.

The wind integration study I referenced earlier says that the 300 megawatts of wind would pencil out (saving consumers money) if it cost between $97 and $126 million each year in total, equivalent 8.2 to 10.3 cents per kilowatt hour. [edit — I had higher numbers here earlier (9.7-12.6), as I accidentally wrote total costs in place of costs/MWh] Since the developer says they aren’t relying on tax credits, presumably they can hit this range without them.

The developer of the canceled Puppy Dog Lake solar contract mentioned tax credit uncertainty in their news articles. Their contract price was 7.4 cents per kilowatt hour. Taking away an assumed 40% tax credit would give you a price of 12.3 cents.

So one plausible scenario is that we meet the goals and all the new energy, gas or renewable, costs around 12 or 13 cents per kilowatt hour (wholesale).

In a good case, the Renewable Portfolio Standard could lower costs around 10%

If you assume that tax credits continue, reaching the RPS goals could save a significant amount of money.

Last time there was a Renewable Portfolio Standard bill for the Railbelt, the National Renewable Energy Laboratory did a study on the potential economics. Their assumptions included renewable energy tax credits, and imported gas costs. While the study looked at a higher renewable target (80% by 2040), the earlier years in that study look a lot like this current RPS.

The targets from this current RPS bill line up pretty well with the generation mix around 2029-2030 in the least cost scenario from that study. Cost savings over a no new renewables scenario were around $90-$100 million in that time frame, which would be about 9-10% of total Railbelt Revenue.

There are non-price benefits to adding more renewables

Other than straight up energy price impacts, what would this do? On the minus side, integrating variable renewable energy is more complicated than turning on a gas plant. There’s no re-inventing the wheel needed, since this has been done all over the world (including in some rural Alaskan villages), but it takes careful modeling. On the plus side, renewable energy projects will provide price stability and a hedge against volatile future fuel prices. Also on the plus side, any renewable energy we add to the grid saves Cook Inlet gas. This is important, because while there are now two preliminary plans for import facilities, nothing is a done deal yet, and shortfalls are still quite possible. Even without actual shortfalls, any gas saved will displace imports. Renewable energy also has vastly lower carbon dioxide emissions and emissions of other pollutants, which is important for mitigating climate change, and for human and environmental health.

If this is a good idea, why would we need to legislate it?

In a perfect world, you might imagine that all utilities always make the best decisions for their members on their own, and no legislation or regulation is ever beneficial. But, well, the world is imperfect. Doing things is always harder than kicking the can down the road, even when those things are beneficial or necessary. The Cook Inlet gas situation is a great example of this. Fuel prices are an especially easy place to kick the can down the road, since these enter our power bills in automatically-calculated ‘cost of power adjustments’ that don’t require anyone to actually vote to raise rates.

Mandatory goals can push us into action, add certainty, and lower prices through competition.

Certainty makes things easier. To reach these goals, utilities likely need to negotiate complicated contracts with each other, and with independent power producers. On the Railbelt, we have four major utilities with four different management teams and four different boards of directors, adding up to dozens of individual humans with different goals and viewpoints and relationships. If they have to be on the same page, that simplifies negotiations.

The standard itself will likely lower costs by increasing competition. Right now, we have just a couple of local companies that can build major energy projects. So when the only solar company has to cancel its plans, there’s no one else to turn to. A standard could attract more renewable energy companies to compete with each other, knowing that there’s a guaranteed demand. And because renewable electricity also competes with gas electricity, it can provide pressure on local gas prices as well.

Share

Subscribe now

Subscribe now

On the Railbelt, solar panels on houses and businesses produce a tiny fraction of our power today (0.3%), but still more than utility-scale solar (0.25%). That number has grown rapidly in recent years, and a number of policy and regulatory changes are poised to expand or contract it. A community solar bill was signed into law this summer,  but its rate structure hasn’t been defined yet. Rules that would make home solar more favorable (annual net metering), or less favorable (net billing), have cropped up in past legislation, and are likely to do so again. Rules in other states have been changing. The Regulatory Commission of Alaska has recently proposed rules that would raise the “cap” on Railbelt net metering to 20% of demand (about 2% of energy generated).

If we get more of these ‘distributed energy’ systems, what happens?

At the numbers we’re talking about (possibly reaching up to 3.7% of energy sales with net metering maxed out, and 100MW of community solar), the physical impacts of that solar energy on the system are pretty small. Mostly what would happen is that we’d burn less fuel, and people would buy less power. 

Buying less power is the heart of the controversy. Your electric bills depend, in part, on everything your neighbors do. Installing solar panels, insulation, or a hot tub. Buying an electric car, or becoming a snowbird. If your neighbors conserve power, and you don’t -- your bills will go up. In most of the country, individual households buy significantly less power than they used to. In a few states, distributed solar systems are a significant reason for that, but not in Alaska. It’s probably not a surprise that Alaska is firmly in the bottom half of the nation for distributed solar. 

But conservation has played a huge role here. Since 2010, Railbelt per household electric use has dropped 15%. Individual consumers have saved over 50 billion cubic feet of Cook Inlet gas through electricity conservation, and utilities have saved over 50 billion cubic feet through power plant upgrades and pooling. This is despite the addition of more than twenty thousand new electric customers. All those savings add up to around four years of electric utility gas use. Without them, we’d likely be importing gas already.

Here are a few facts about household Railbelt electric bills that might surprise you.

• Alaskans use less electricity per household than all but four other states. 

• Our bills are lower than the US average -- $1555 per year vs. $1623 per year.

• They are the same now as they were in 2010 (after adjusting for inflation).

One could take this as a reason to focus less on electric rates and more on heat bills, but somehow we don’t, and people can’t stop talking about electric rates. So here you go.

Distributed solar rate impacts would be tiny

If we want to know how more distributed solar might impact other people’s energy bills, we can do two things. 1. Look back at what did happen when we saw a much larger drop in electric use, and 2. Look forward to what could happen if we add solar systems and keep everything else the same. Luckily we can do this, because all the Railbelt utilities are cooperatives. We have public data on fuel costs, other costs, electric use, and money collected from customers, and can model what might happen if any of these things were different.

By conserving power, consumers have kept average annual bills pretty constant. However, if you didn’t get in on those efficiency improvements, and still use the same amount of power as you did in 2010, your annual costs are $275 higher. How does this compare to the potential impacts of distributed solar? If we assume rooftop solar systems grow to fill the RCA’s proposed cap, average bills go up around 60 cents per month. Community solar doesn’t have a rate structure to model, so for a placeholder, let’s say it works like adding 100MW more rooftop solar, roughly doubling the proposed cap. Together, that would add up to a $1.40 increase on a monthly bill, or around $17 per year. 

Everything else will change bills a lot more

Everything else that will happen in the time it might take to reach that level of distributed solar will have much larger impacts. All the distributed solar I modeled could raise electric rates around half a percent, spread over however many years it would take for it to be installed. Over the past year, Railbelt electric rates have changed an average 3.5% every quarter, sometimes rising as much as 23%, or dropping more than 9%. This is mostly due to shifts in fuel and purchased power costs. The Cook Inlet gas crisis will drive this further, possibly raising rates around 10 to 17% depending on the costs of outside gas.

Fuel savings benefits outweigh costs to other members 

If distributed solar doesn’t make much difference to rates, does it matter for fuel savings? On a percentage basis, if you add 3% solar energy, you’ll save roughly 3% of the fuel. Added up, that’s around 950,000 Mcf of gas (from the gas dependent utilities: CEA, MEA, and HEA), and 3.3 million gallons of oil (from GVEA) per year, similar to the potential fuel savings from the Dixon Diversion project. It doesn’t solve climate change or the gas crisis, but it’s a significant piece. 

It’s especially relevant because it’s additive to all the other pieces, and can happen more quickly. It doesn’t require either utilities or the state to spend money, negotiate contracts, or build anything, and therefore it can happen beside all those larger efforts. 

Utilities should focus on bigger things

Utilities are likely to provide much of the testimony on any regulations or bills involving distributed energy. This is silly. On an individual consumer scale, installing distributed solar has large financial and environmental impacts. On a utility scale, it doesn’t. 

Utilities on the Railbelt are all cooperatives, and work for their members. Allowing interested members to connect solar systems reduces fuel use and emissions a modest amount and empowers those members to lower their bills. But utilities can do vastly more to reduce fuel use and costs for all their members with utility-scale renewable projects. 

HEA just signed a contract to buy power from a 30MW solar farm that will more than double all the solar on the Railbelt in a single shot. The other Railbelt utilities are in negotiations with wind power developers for much larger projects. Utilities have yet to nail down where their natural gas is going to come from. These are the sorts of decisions that make a difference for the members as a whole.

Is it fair?

A big argument against net metering is that while the environmental benefits of burning less fuel are enjoyed by the entire world (climate changing emissions), and entire community (local air pollution), rich people receive all the economic benefits. High-income people are more able to afford solar panels. The small size of the rate impact renders that irrelevant here, but additionally, half of the distributed solar I modeled is community solar (which should be accessible to low income people) and the state also has a grant to help low-income households install solar. 

But for some people, it’s the principle that matters. The fact that one person’s actions affect another person’s electric rates seems fundamentally unfair. This argument has three major problems:

• If raising your neighbors' rates with your solar panels is unfair, so is raising them by buying a better refrigerator, or lowering them by buying an electric car.

• Rewriting rates in such a way that your power use doesn’t affect your neighbors rates would mean raising costs dramatically for anyone that doesn’t use much electricity -- which would cause vastly more harm to low income people than more solar panels ever could.

• Electric rates are designed to charge consumers based on their responsibility for past infrastructure decisions, not current marginal costs.

You might think you can calculate how much it costs the utility to serve a house with a solar system on it, see whether those houses are paying more or less than that number, and adjust accordingly. It turns out you can’t possibly do that. And to explain, I have to back up and explain electricity rates more generally. 

Fair electric rates are a fiction

An electric utility has a bunch of costs, and a bunch of customers to divide them between. Sometimes it’s easy. Add up the cost of fuel, divide it by the number of kilowatt hours produced and charge each person that much per kilowatt hour. That becomes part of the “Cost of Power Adjustment” on an electric bill. Everything else is much less straightforward.  

What is each customer’s fair share of power plant debt, lineman salaries, new transformers or regulatory compliance? 

There are standard ways this is usually done, based on a cost allocation manual from 1992, that involves dividing customers and costs into categories, then deciding how much responsibility each category of customers bears for each category of costs.

Those standards, baked into RCA regulations, don’t fit how our actual costs work. For instance, a large portion of total costs, including most of the costs of the power plants themselves, gets classified as “demand.” Which means that we look at the hours of the year when people used the most electricity (cold days in winter), figure out who used the most at those times (houses), and then assign them a higher portion of power plant costs. This is based on the theory that the power plant got built because the old power plants couldn’t provide enough energy to meet the highest demand.

But that’s not true. The natural gas plants that provide the majority of the power on the Railbelt today were built to be more efficient than the old ones because gas prices were going up, even though demand was going down, and there was no problem delivering enough electricity. But the fact that residential customers used more power in some of the coldest hours a decade ago is part of the reason for their rates today.

That doesn’t seem fair. It seems even less fair for a snowbird or a fish processing plant that never used power in the winter in the first place. Even when the division of costs isn’t obviously wrong, it’s still contentious (this RCA order is 89 pages long, and only summarizes all the division arguments in the Chugach rate case). And even the categories themselves are pretty arbitrary. Why should a house pay a different rate than a coffee shop, while an apartment in a dense city core pays the same rate as a house at the end of a long rural road lined with beetle-killed spruce trees?

The principle is that the “cost causer should be the cost payer.” Which means “we divide things into broad and fuzzy categories based on a decades old manual in order to figure out whose fault it was that the utility had to buy that thing in the first place” not “we charge you based on how much it costs to bring electricity to your house.”

The logic of per-unit rates collides with the day-to-day reality of fixed costs

Almost everything in your electric bill changes based on the amount of electricity you use. On the other hand, most of an electric utility’s budget stays the same.

From the “cost causer should be the cost payer” principle, dividing costs by the amount of power used makes a lot of sense. It’s supposed be divided based on the reason things were built, and things get built based on the amount of power they need to deliver. If my house used as much as a refinery, we’d have built a bigger power plant and bigger transmission lines. If the refinery used as much as my house, we’d have built everything smaller. 

But on a day-to-day basis, most of these costs don’t change if people change their power use. Fuel costs do, but the infrastructure is already built, the people are already hired, and the debt is already incurred. 

Another way to think about it is that everyone should pay the same amount for those fixed costs. No one advocates for that on an absolute level (that a house should pay the same fixed costs as a hospital), but some people do within a category (all houses should pay the same fixed costs). But this is just as arbitrary and no more fair than the way we do it now.

To explain, let’s take the example of my property. My utility is HEA. In 2023, the average HEA residential customer paid just over $1700. $500 of that paid for fuel and Bradley lake hydropower. The other $1200 went towards fixed costs.

My house used a little less power than average, so I paid $1100 in fixed costs. My sister in law has a house on the same property that also used a little less power than average. She paid $950 in fixed costs. If you take that $1200 number as the correct one, we’re both being subsidized -- to the tune of $350 total per year. What if I connected her house to my meter? Suddenly we’d turn into a single customer paying $1750 for fixed costs. It would appear that my family was overpaying -- subsidizing all the other customers $550 per year. But we’d actually be paying $300 less in electric bills. The actual costs to HEA of serving my property would be the same. 

Confused yet? 

Here’s some more example HEA customers, with graphs showing how much they paid over a year in fixed costs, power costs, and per kilowatt hour.

Dark blue bars are fixed costs, light blue are fuel and purchased power cost.

Customer I paid the least fixed costs - only $550. Under the “$1200 is correct” framework, they’re being heavily subsidized. On the other hand, with HEA’s mandatory minimum payments, they’re paying over $1 per kilowatt hour for their power, much more than in a lot of rural villages. If every customer were a customer I, we’d have built a very different power system in the first place. Customer I is paying a lot for the debt and depreciation for our new natural gas plants, but isn’t benefiting from it much.

If customer I had to pay $1200 in fixed costs, they would pay $2.23 per kilowatt hour. With such a small use, and such a high price, they’d probably disconnect entirely -- and the utility would collect less money overall. Customer I could, in fact, argue that almost any amount they pay for fixed costs is actually a benefit to the utility, since the utility would otherwise get nothing from them. Maybe they could offer HEA $100 per year instead. That doesn’t sound fair, and we don’t allow it for houses, but this is exactly how rates work for large industrial customers.

The Marathon refinery gets its power from a special contract, for around 2 cents per kilowatt hour beyond the generation cost. The rationale, as written in the contract, is that since the refinery would otherwise generate its own power, as long as it pays HEA a little more than the cost of that power, it’s a benefit to everyone else.

The people contributing the least to fixed costs often aren’t the ones with solar panels. 

Who are customers A-J? Most are real houses on the HEA system that shared their bills with me. Some have solar panels, some have heat pumps, some have both or neither. Two are 2023 averages from RCA filings -- the average of residential customers with and without solar panels. One is a theoretical customer with average use and a 6 kilowatt solar system, which I used for my rate impact analyses at the beginning.

The person paying the most towards fixed costs actually has a very large solar system. The people paying the least don’t have solar panels at all. Small houses and apartments (often occupied by lower income people) often don’t use much power. And people that use more power are disproportionately likely to buy solar systems. Which is probably why the average house and average net metering house are so similar. 

It’s all about the fuel

What about duck curves? Peaker plants? If you read about net metering regulation debates in other places, there are all sorts of factors I haven’t mentioned at all. That’s because they aren’t currently relevant here. We are nowhere close to being Hawaii.

The proposed RCA regulations would allow net metering to gradually expand from tiny to small. This won’t be enough to produce more solar than we can use during summer days. On the other hand, distributed solar also won’t save us from turning on inefficient power plants at peak hours or allow us to build smaller powerlines. Our peak hours are in winter, and we don’t need to turn on other power plants to cover them. 

Distributed solar probably saves more fuel than I’ve calculated here. There is line loss getting grid power to where it’s used, and power plants themselves use fuel. And all the saved fuel could be more expensive. I modeled the Cook Inlet gas conservation benefit based on $12/Mcf gas, but imports could cost much more, especially if we don’t get an import facility built soon enough.

In Conclusion

There is no ‘right’ or ‘fair’ answer on distributed solar policy or electric rate structures in general. All we can do is look at the potential impacts of any policy change. From where we’re at on the Railbelt today, allowing net metering to grow is a no regret move that could have a small benefit on overall fuel burning and the Cook Inlet gas crisis.

The proposed RCA rule changes to increase the net metering ‘cap’ to 20% of demand (around 2% of energy) simply requires the utilities to allow solar systems to continue to connect in the future as they have in the past, up to a level where they will remain irrelevant to other consumers. By itself, that doesn’t do much.

I’ve already shown that rate impacts to other customers would be tiny if we had that much net metering. Additionally, of all the Railbelt utilities, only the smallest (HEA) is close to its current cap, and no utility has ever actually cut off connections when the cap is reached (MEA spent years allowing connections above the cap before raising it). So the cap has had no impact whatsoever on net metering buildout so far. Changing the cap does nothing to change the incentives, which aren’t all that great for many consumers. It also doesn’t change the barrier of high up-front costs.

Policies that change net metering incentives could do a lot more. Annual net metering would allow people to get retail-rate credits for extra power they produce in summer that currently only gets wholesale rate credits, likely driving more installations. On the other hand, a change to ‘net billing’ would get rid of retail-rate credits entirely and give people only wholesale credits, likely shutting down the industry.

Community solar policy is more complicated, and deserves its own post. It’s more accessible to consumers than net metering, and while the impacts I’ve modeled here (as if it was equivalent to net metering) are a reasonable placeholder, implementation will matter a lot. It could end up being a better deal for consumers than net metering, or worse, and we also need to consider the incentives for the solar developers.

Anything that addresses distributed solar specifically is limited in its utility-scale impact, because it targets such a small portion of our grid. There are other ways to change rates that would have more dramatic impacts. Sometimes, utilities charge higher per KWh rates once consumption gets beyond a certain level. Renewable Energy Alaska Project proposed this as a gas-saving conservation measure in the Chugach rate case. Power Cost Equalization does this as well, with a cutoff on eligible kilowatt hours in a month. This incentivizes conservation, but disincentivizes electric cars and heat pumps, which also save fuel and energy. Other utilities do exactly the opposite, charging lower per KWh rates at higher use levels. MEA does this. This could make electric cars and heat pumps more attractive, but disincentivizes conservation. 

You can comment on the RCA regulations until November 12.

Subscribe now

Sources, assumptions, and methods

• All utility level data from RCA filings and the EIA

• Actual household bills are based on the most recent full year available, and are a very non-random sample of people who shared them with me, all served by HEA.

• Rate impacts from solar systems were calculated separately for each utility, and weighted averages used for the Railbelt-wide number.

• Net metering systems were modeled by taking average monthly use for a residential consumer for each utility, and combining that with monthly home solar capacity factors based on actual data for HEA and CEA systems, and combined with PV watts estimates to extrapolate to MEA and GVEA territories. 

• For Railbelt-wide rate impacts, I assumed all net metering (and all community solar) was made up of 6KW systems (the average size) installed on an average use home. Electric bills for these houses were calculated based on Q3 2024 rates for each utility. Real net net metering systems are probably more often installed on homes with above average use.

• To determine the rate impact, I subtracted COPA payments from the annual bill, to determine the net metering customer’s contribution to fixed costs. The difference between this number and the average household’s fixed cost contribution was considered to be a revenue loss to the utility. For gas-dependent utilities (all except GVEA), I assumed that each KWh of solar power acted to ‘save’ cheap Cook Inlet gas for future use, and that the cost difference between producing a KWh at a utility’s current gas costs and at $12/Mcf is a revenue benefit to the utility. Rate impacts were determined by applying these revenue impacts on a per KWh basis to the utility’s entire retail sales. 

• I did not model reduced losses from distributed generation, but Railbelt utilities generally report losses of around 4.5-6%.

Subscribe now

Utilities think it’ll take longer to construct an import facility than they did originally. If Cook Inlet gas production declines as predicted, we’ll need to import stopgap gas in little tanks on container ships --  at more than triple current prices.

Enstar can’t get anyone to promise to sell them new gas that supposedly is available in Cook Inlet. Whether or not they have to draw down their stored gas substantially in the next year or two might be a useful signal of how bad this is going to be.

We haven’t actually done anything to address the crisis yet. Lots of folks have studied and planned things, but there haven’t been any final investment decisions on any major renewable projects, gas developments or gas import facilities. The longer this remains true, the bigger the supply gap could be.

Even with all of that, consumer impacts are probably not as scary as that seems. Current contracts act as a partial buffer through 2032, even if we have to mix in some triple price gas. Especially if we work to conserve gas, and actually invest in solutions. Business and industry will be proportionally more affected than households.

Gas crisis timeline - Who said what when

“Gap begins” when predicted Cook Inlet supply is less than demand -- if a range was given, I chose the middle. “Earliest Imports ready” is the fastest possible timeline for a large scale LNG import facility of any kind. “Import price” is the price of that gas (including import costs), and “Stopgap price” is the price of importing gas through small ISO tanks on container ships before the existence of a large scale import facility.

Reasons for Enstar to worry

Import facilities will take longer than anticipated to come online: Last summer, Enstar and the electric utilities published two separate reports that estimated gas imports could be ready as early as 2027, one year ahead of their projected supply gap. Since then, Enstar has told both the legislature and the media that their consultants now say it can’t happen before 2030, a timeline that will continue to slip until they make a final investment decision on an import facility.

Enstar needs more gas than it used to: Enstar has started supplying gas to electric utilities. At the end of 2022, Enstar contracted to provide 1 Bcf (billion cubic feet) per year of gas to GVEA for two years. When HEA’s Hilcorp contract expired this spring, Enstar committed to sell 3-4.5 Bcf to HEA for a year. This means that Enstar is projecting to need 37 Bcf of gas for the upcoming year, significantly above the recent average of 33 Bcf per year. 

Enstar’s contracts don’t cover all its needs: This isn’t actually new. Every year, Enstar files a “Gas Cost Adjustment” with the RCA, which both sets its customers’ gas price for the following year, and predicts where that gas will come from. Enstar generally predicts that around 10% of its needs will come from “undetermined supply” or from its gas storage reserves. In practice, both Hilcorp and smaller producers have supplied more gas than predicted each year, and Enstar hasn’t had to draw down those reserves (gas storage is used heavily for seasonal demand fluctuations, but hasn’t been a contributor to overall supply). Now, higher volumes make that gap larger -- around 14% of expected gas needs.

Based on gas cost adjustments and gas cost balance reports filed with the RCA. 

This is likely why Enstar is raising the alarm rather than the electric utilities. Chugach and MEA have all their gas covered under contract or self production until  2028, and GVEA doesn’t need gas to generate power (though they’d prefer to buy gas power). Electric utilities also have the ability to reduce gas use through adding renewable energy, while a gas utility cannot.

Suppliers may not fill those gaps as they have in the past: Relatively little gas is supplied by small (non-Hilcorp) producers, but what there is is poised to shrink. Last summer, Vision and Enstar renegotiated their contract for smaller volumes based on poor production. Enstar was waiting on drilling results from Furie/Hex at the end of 2023 to determine future gas amounts, but now says they anticipate no gas from Furie/Hex after March 2025. Furie will not drill new wells this season.

Hilcorp’s signals have been more mixed. They’ve clearly said that their entire well drilling program is intended to meet existing contracts (see graph below). They also agreed to sell Enstar 3.5 Bcf more for one year to cover most of HEA’s demand (Enstar agreed to cover the rest through their own existing contracts). That agreement only increased the price a little, but was full of provisions to mitigate supply risk -- requiring Enstar to use some of its stored gas before buying the extra ‘call’ gas it usually buys in winter, and exempting Hilcorp from penalties on some amount of delivery failure.

From Hilcorp’s presentation to the legislature February 2024

Normal variation could make things better or worse: Gas use varies quite a bit year to year, around 11%. In cold years, we use more gas for heat. On the electricity side, we use substantially more gas when GVEA is able to purchase power from southern utilities, and a little bit more in dry years when there’s less energy available from hydropower.   

2024-2026: Will producers sell gas beyond their firm contracts, or will we draw down gas storage?   

The state’s forecast modeled gas pool decline. The dropoff is steep, but not as steep as the dropoff in firm gas contracts. This means there’s a substantial amount of gas that is theoretically economic to produce in Cook Inlet, but no one has promised to do it -- represented by the green bars in the graph below.

From the June 2023 BRG report by Enstar and Electric utilities.

The space between the bottom of the green bar and the purple ‘medium demand’ line was pretty small in 2023 -- and was indeed supplied from Cook Inlet. But that uncontracted wedge gets bigger quickly. And while Cook Inlet producers may still be able to supply that gas, they certainly aren’t willing to promise it.

Producers might sell gas to Enstar on a non-firm basis: Gas producers are probably motivated to be conservative with their firm contract promises. If they do produce gas beyond contract minimums, they might sell it to Enstar for somewhere around current prices, as has occurred in the past.

Or we might use gas we’ve already stored: If producers can’t produce more, we’ll have to turn to storage. Since the CINGSA gas storage facility was created, 9.7 Bcf more has been put in than taken out, and it also contains extra gas that was discovered when it was built, for a total of 13.5 Bcf working gas there today. Additionally, Hilcorp has its own storage, which had around 26 Bcf stored as of February.

Neither of those outcomes is a crisis. But how much we need to draw down gas stores may be a sign of what comes next.

2027-2029: Renewables, new gas, or super-expensive stopgap imports?

This time frame seems like the biggest wild card. Ordinary variation in demand and drilling success might mean that we have a major supply gap in this time frame, or none at all. And while it’s too soon for large-scale imports, this is enough time to make significant changes to gas demand, storage, supply, or all three. 

Renewables could push demand down -- if we sign contracts soon: In my last post about Cook Inlet gas, Alan and I analyzed how proposed renewable energy projects and ambitious efficiency measures could delay the supply gap. 

The recently-passed HB 307 improved the prospects for all these renewable projects (assuming it gets signed by the Governor), by reforming the property tax structure for IPPs, and removing wheeling fees. But we’re midway through 2024, and no power purchase contracts have been signed yet. It seems likely that some timelines will shift later, and it remains possible that none of these projects will happen at all.

Utility demand of 65 BCF from the mid-case of the 2023 working group report (p 12-14), production data from DNR’s truncated 2022 forecast.

Gas storage might help a little: Chugach is planning to rely on new gas storage by 2027 or 2028, allowing it to save some of the near term production from its Beluga River gas field for later years. Chugach often sells gas energy to GVEA, so if they stored more gas it might push up GVEA electricity prices even as it delays their own shortfall. Cingsa also has expansion plans, and the legislature just passed a bill setting up a framework for regulation of third party gas storage. This is important for a future import scenario, and for reliability on super cold days, but doesn’t help overall supply issues much. Hilcorp has storage already, and small producers seem unlikely to end up with so much gas they can’t sell it.

New gas wells show diminishing returns, but without them, production would likely crater: Overall, the Cook Inlet data shows that well drilling hasn’t slowed down, but we get a lot less gas out of each well. This makes it hard to get more by drilling more. Nearly all of the recent wells have been drilled by Hilcorp, which has said they don’t expect to produce more than their current contracts require, and are drilling to offset the natural declines. So if Hilcorp backs off drilling, they would likely produce a lot less -- as they threatened to do when the legislature suggested making them subject to corporate income tax.

Cook Inlet well data from the state’s AOGCC data miner - gas development wells only

From the state’s AOGCC data miner

Small producers might produce more gas than expected -- if they get financing and/or royalty relief: In contrast to Hilcorp, two of the small gas producers do say they could potentially increase gas production between now and 2030.

Bluecrest has gas reserves that require a new offshore platform, and a $400 million dollar investment they haven’t been able to secure. This gas would cover around a quarter of current Railbelt demand for 13 years -- not enough to change the long term picture, but enough to buy time. However, they say the earliest it could happen is 30-40 months after financing -- 2027 if that happens today. AIDEA now has the authority to lend money for Cook Inlet gas development from a Cook Inlet Reserve based lending account (from HB 50). But making use of that would require a future legislature to fund the account with a rather large amount of money -- not guaranteed, and at least a year away. Bluecrest is currently in forbearance with its existing AIDEA loan.

Furie/Hex also says they could produce more gas from their Kitchen Lights Unit, possibly ramping up to 12 Bcf/yr (17% of current demand) by 2028. But they framed this to the legislature as contingent on using the Inlet’s single jack up rig, which isn’t certain or quick (they’re still working on permits, and Hilcorp already has plans to use the rig in 2024). They also framed it as contingent on financing, and contingent on royalty relief (which they haven’t gotten), and recently confirmed they will not drill this summer. I’m also not sure what portion of this gas is already included in the state’s Cook Inlet forecast, since it’s a currently producing pool. 

We could import super-expensive gas in little tanks on container ships -- and we probably will: One thing that’s gone little noticed in Enstar’s dire warnings is what happens if we have a shortfall and no import facility. In that case, Enstar has said we’d need to import gas in ISO tanks instead (shipping-container sized tanks, stacked up on container ships), and that it would cost $25 or $30 per Mcf -- over triple current prices. 

In the mid-range predictions -- this happens. They all show a supply gap at least a couple years before an import facility. And developers’ timelines for new renewable energy or new gas are more likely to be optimistic than pessimistic. This will cost the Railbelt $150 million to $600 million extra per year (beyond the current costs of gas) until large scale imports come online, with costs ramping up as the gap grows.

Triple prices sounds big and scary, but around half the Railbelt’s gas will still be under contract at prices close to today’s through 2032. And a good chunk more theoretically will come from Cook Inlet. For example, the utility mid-case forecast for 2029 shows around half the gas from existing contracts, and only around a quarter coming from outside Cook Inlet. Triple price stopgap gas could make up a huge portion of the total cost in this scenario, but existing contracts buffer their impact, leaving the overall price within the range predicted for large scale imports.

Current contracts include CEA’s owned gas from Beluga River. Prices modeled in today’s dollars at $8/Mcf for gas under contract, $10/Mcf for new Cook Inlet gas, and $27.50 for stopgap ISO tank gas (halfway between Enstar’s two quoted estimates).

I’ll cover the cost impacts of those large scale imports in the next section. If we don’t get imports online as predicted, and end up with an extended period where we’re burning triple-price stopgap gas to heat our homes, using diesel power, etc… Then the extra costs could add up to billions.

2030 and Beyond: LNG imports -- when will we get them, and what will we pay?

Why isn’t Enstar pulling the trigger on an import facility right now? There’s a lot of uncertainty in the forecasts, but once you start getting to this time frame, they all collapse into “not enough”. If you add Bluecrest and Furie’s optimistic forecasts to aggressive renewable development, we still run short of local gas in the early 2030s. This is basically inescapable because most gas is used for heat. We can, and should, electrify heat, but there isn’t a way to step up both renewable buildout and heat electrification quickly enough. 

Enstar recently filed an application to extend their service territory to encompass a potential import site. But while the urgency of their warnings has increased, they’ve moved more slowly on an investment decision -- originally planned for the end of last year. This is the piece of the Cook Inlet gas puzzle that confuses me the most. Stepping up warnings and delaying action looks like a move to try and get someone else to solve the problem -- but they haven’t clearly asked anyone else to solve it. Most of what they said to the legislature was along the lines of ‘do what the gas producers want’ -- but what the gas producers say they want might change the numbers a little, but won’t solve the long term decline. 

What will imported gas cost? The first utility reports on the gas crisis last summer put the costs of imported gas at around $13 per Mcf, compared to the current average of around $8. More recently, Enstar has used higher numbers -- $15 or $16 per Mcf. Pacific LNG prices haven’t changed substantially since last summer. Possibly the numbers are higher because the consultants who said import infrastructure couldn’t happen until 2030 also said that infrastructure would cost more. Possibly no one knows what prices will be and folks are just pulling numbers from a large hat of possibilities in every interview.

For households, electric rates probably won’t change much, but heating costs for Enstar customers will: I analyzed the impact of gas costs and renewables on both electric and heat rates as part of my post on natural gas vs. heat pump costs earlier this year. It matters less than most people suspect. Gas costs make up around 20% of total electric costs on the Railbelt, for around $170 million per year. Doubling that would matter -- but it wouldn’t come close to doubling electric bills. 

Gas efficiency and non-gas components of the rate (including other fuels) are held constant, and renewable energy is modeled at $0.08/Kwh.

On the other hand, 75% of a residential gas heat bill from Enstar is the gas cost. Doubling the gas cost increases the bill by 75%. Yet it still remains well below the cost of heating oil. Nearly 40% of Railbelt households don’t have gas heat, and outside the Railbelt, almost no one does (and their oil is usually a lot more expensive).

See detailed assumptions and versions that include heat pumps in this post

Gas costs are a bigger deal for commercial and industrial users: For both electricity and gas heat, fuel costs are a larger portion of your bill the more energy you use. Fuel costs are captured in a per Kwh “Cost of Power Adjustment”  (COPA) for electric customers, and a per ccf “Gas Cost Adjustment” for gas customers. Those costs are the same for every customer. Meanwhile, the rest of the per-unit charges are much smaller for large users. For example, the COPA is less than a third of an electric bill for an HEA residential customer, but 85% of an electric bill for the Marathon oil refinery. 

Chicken and Egg -- are all the options delaying all the other options?

In their presentations to the legislature, gas producers mentioned that possible competition with gas imports, a pipeline, or renewables makes their investment more uncertain. But that could go in all directions. Why invest in in imports if local gas might become available after all? Why invest in renewables if utilities might sign long term gas contracts that don’t leave room for them? The risk, of course, is that we don’t invest in anything at all, ride the downward slope of Cook inlet production, and find ourselves in a few years with nothing but $25-$30 per Mcf ISO tanks of gas, and all the solutions years farther off than they would be if we pursued them today. I’d like to think we’re not going to do that, but I’m not so sure.

In Conclusion

Someone should probably do something. Specifically, Enstar should make a final investment decision on an import facility, all the Railbelt utilities should sign contracts for large scale solar and wind facilities, and the rest of us should make our buildings more efficient. We said the exact same thing before, but I’ll repeat it until we do it. Doing it sooner minimizes the likelihood and quantity of triple price stopgap gas imports. And if we do end up with more Cook Inlet gas than expected, those actions will still set us up well for the future. It’s theoretically possible that an import facility could be rendered useless by a combination of amazing short term gas production and a fast buildout of the North Slope pipeline that nearly everyone’s given up on. It’s also possible that if you leap off a cliff, someone will toss you a parachute before you hit the ground.

In the meantime, I’ll be watching what happens with Enstar’s gas supply in the next couple years to see how hard that landing might be.

Subscribe now

Subscribe now

Right now, utilities generally own and pay for for the transmission lines in their territories. If other utilities move energy across those lines, they pay the owner a “wheeling fee” based on the amount of energy transferred. These proposals would change that to a system where all the costs are pooled, and each utility pays based on their proportion of the total electric load. 

So I did the math to see what impact that would have on the various Railbelt utilities. The short answer is that transmission is a tiny piece of current costs, and changing how we allocate them would only shift things a little bit -- no utility would see more than a 1% total change in revenue. 

So why does it matter? 

The costs to operate a transmission line don’t increase if you move more power across it, but wheeling fees are charged based on the amount of power transmitted. They create a strong disincentive for a utility to buy power from another utility’s territory, even if that would otherwise be the cheapest and most efficient option.

That doesn’t matter much under our current system. Almost all of the energy transfer that happens is either required by law or saves so much money that it almost doesn’t matter what the fees are. MEA and Chugach are required to pool their generation, and don’t pay wheeling fees for power transferred back and forth in that pool. GVEA and MEA both pay wheeling fees to Chugach to receive their share of power from Bradley Lake, but since Bradley Lake produces some of the cheapest power on the Railbelt, it’s well worth it. Bradley Lake also has special agreements that exempt it from wheeling fees in HEA territory -- any other power moving off the Kenai Peninsula could be subject to two sets of fees. Similarly, GVEA’s oil-fired generation is so expensive that gas power from the south is worth buying, even with wheeling fees.

Wheeling fees could have a big impact on the price of renewable power 

Wheeling fees discourage all sharing of energy between regions. This matters for renewables particularly, because renewables are the only energy we’re likely to want to share. There isn’t enough extra gas to ship more gas power north, and there isn’t much to be gained sending gas power back and forth from the Kenai Peninsula to Anchorage. Wheeling fees would probably increase the price of new renewable energy by over 10% for each utility territory it passes through. Wheeling already adds more than that to the price of Bradley power for MEA and GVEA. It costs literally zero extra dollars to ship this new power over existing lines (and we can transfer a lot more power over our existing lines if we have reason to do it). It’s just a fee.

A utility buying energy from elsewhere still pays just as much to maintain its own transmission system as it did before -- plus some extra to other utilities. Therefore, there’s an incentive for utilities to want any new generation in their own territory, even if there are better resources in other areas, if a single large renewable project would be cheaper than a bunch of smaller ones, or if buying power from multiple different renewable projects in different places would help smooth out the spikes and valleys in generation.

In 2021, Railbelt utilities paid around $32 million total in transmission costs, 3.6% of their total revenues. This includes operations, maintenance, and depreciation. Depreciation is how the initial cost of a piece of infrastructure is spread out over time. That figure also includes the costs utilities pay to run the Alaska Intertie running between Southcentral and Fairbanks, which is owned by the state. It doesn’t include the costs of the transmission infrastructure that’s part of the Bradley project, but that’s small, and is already paid for in proportion to each utility’s load. In total transmission costs add up to around three quarters of a cent per Kwh sold. Individual utilities costs range between 0.55¢ - 0.85¢ per Kwh.

In contrast, when utilities charge wheeling fees, they charge more -- more than the costs of building, operating, and maintaining the infrastructure. Wheeling isn’t charged on a simple per Kwh basis, and varies depending on where the electricity is coming from and going. But in 2021, Chugach charged between 0.88¢ and 1.2¢ per Kwh for wheeling on its system -- substantially more than overall transmission costs. Near term solar and wind power is likely to cost around 7.0¢ to 8.5¢ per Kwh. If that power had to move through more than one utility, the cost would be at least 20% higher by the time it reached the other end of the system -- possibly rendering it uneconomic. Pancaking wheeling rates are parasitic on the system, letting utilities profit off each other based on geography, or worse, preventing good projects from getting built at all.

Getting rid of wheeling fees wouldn’t be a major harm to any utility. My graphs show how those costs are distributed now, and how that would change if they were instead allocated by ‘load ratio share’ (each utility’s percentage of total power sales), similar to what the two bills propose. The numbers are small, and the changes are also small. MEA would pay around $1.4 million more , and GVEA would pay around $1.5 million less, but the total change would be less than 1% of total costs for each utility. 

Additionally, the Railbelt is poised to build a lot of new transmission in the coming years. HEA is the only utility with a large-scale battery right now. Its depreciation costs aren’t captured in the 2021 data I show, but it increased HEA’s transmission costs 1.8 times between 2021 and 2022. In the next few years, all the utilities will have batteries. A federal grant will build a new transmission line across Cook Inlet if the state provides matching funds. Changing how these costs are allocated will be easier while the numbers are small.

The two transmission bills put different groups in charge of future transmission

Both (SB 217/HB 307) and (SB 257) would likely allocate transmission costs in similar ways. 

SB 217 uses  load ratio share (what I used in my analysis) as the basic idea, but leaves the details to the RCA, which it directs to “gradually transition” to a cost recovery mechanism which “must take into account each load-serving entity's load in comparison to the total load on the integrated transmission system.”  SB 257 uses similar language, but adds another possible way to calculate the proportions, based on maximum rather than annual energy flows: “The cost recovery methodology for the backbone transmission system adopted by the transmission organization must pool backbone transmission system costs and allocate those costs through certificated load-serving entities on a coincident peak or load ratio share basis, or a combination of both, as approved by the commission.” SB 217 includes all old transmission in the costs, but only includes new lines if they connect more than one new power project. SB 257 requires the organization it creates to determine which parts of the system are the “backbone” and which serve only one utility. Neither specifies whether depreciation is considered in transmission costs, but since depreciation represents the costs of actually building those lines, I think it should be.

But they differ dramatically in the ownership, management, and planning of the system as a whole

SB 217 leaves ownership and management of existing transmission infrastructure the same as today. Future planning is left under control of the Electric Reliability Organization (ERO, also RRC). The ERO was mandated by a bill a few years ago, and is currently only partly stood up. It includes both utilities and non-utilities in its governing structure, and is supposed to create both reliability standards, and an “integrated resource plan” for the whole Railbelt that includes both transmission and generation. The ERO doesn’t own anything, so the utilities and/or the state would still build and operate the transmission, but that transmission would have to fit within the integrated resource plan, either by being listed in the plan or getting pre-approval from the ERO.

SB 257 would create a new transmission organization, the “Railbelt Transmission Organization,” within the Alaska Energy Authority. This organization would actually own and manage all the transmission -- the utilities would be required to transfer all their transmission lines and batteries. The new organization would also be responsible for the grid-wide planning that currently is supposed to be the ERO’s job. It would be governed by the utilities, AEA, and one member from the ERO (which would still exist, but have fewer responsibilities). The main controversy over this proposal is the fact that this organization would be much more utility-controlled, without the consumer, independent power producer and independent board members the ERO has.

Both bills also do some entirely unrelated, and generally positive, things. SB 217 changes the tax structure for Independent Power Producers in a way that would put them on the same footing as utilities, substantially decreasing the price of renewable energy. SB 257 ensures that commissioners on the RCA meet stricter qualifications.

In conclusion: 

Whichever organization ends up doing the Railbelt planning, we should fix the problem of wheeling fees soon, rather than waiting years for that plan to get made. No one loses anything significant if we reallocate current costs slightly. But independent power producers are negotiating renewable power contracts with utilities right now. These projects are the best thing we can do to reduce gas usage in the near term, as well as the cheapest power for the Railbelt in the longer term. Pancaking wheeling rates might significantly shrink the renewable projects they decide to build. If that happens, we all lose.

Subscribe now

Instead, it’s an extremely expensive power source that tries to justify itself by ignoring the costs of transmission and the West Susitna Access Road and comparing its cost to inaccurate and inflated prices for natural gas power. Also it fails to compare to cheaper renewable options, and uses overly optimistic assumptions for both coal plant costs and experimental carbon capture technology. Rosy presentations with bad math have led the Mat Su borough assembly to pass a resolution stating “Coal-fired power generation with CCUS presents a compelling alternative to imported LNG by providing affordable, reliable, clean electricity to the Railbelt grid at substantially lower than current costs” -- when nothing could be further from the truth.

Subscribe now

There are a lot of things left out of those rosy presentations. Not just roads and transmission. Most plans to build carbon capture technology on power plants have failed before they ever started running. There are only four currently operating in the world, and they have been plagued by cost overruns, and operational problems that cause them to capture far less CO2 than initially planned. All things that we ought to think about before diving into a $4 billion publicly-funded project that can’t possibly be built quickly enough to impact our near term gas shortfall.

So I decided to do a better analysis.

According to this feasibility study’s  assumptions, federal tax credits subsidize the larger version of the coal plant enough to make it only a little more expensive than existing  solar and gas generation, and slightly cheaper than the 76% renewable scenario in the recent National Renewable Energy Laboratory (NREL) report. According to more plausible (but still optimistic) assumptions, this plant will end up more expensive than gas power from imported LNG and far more expensive than the high renewable scenario, with carbon capture that’s likely to cost more than the tax credits will pay for.

What’s wrong with those assumptions?

Roads and powerlines are not free: They ignore all the costs of the road and transmission needed to actually connect this plant, 75 miles from the current electric grid. “The West Susitna Access Road is assumed to be funded by DOT and/or AIDEA.” And: “Power customers are assumed to permit and install their own power lines independent of this project. Such costs and permitting are not addressed in this study.” In contrast, renewable energy developers include the costs of power lines and substations necessary to connect to the grid in their power price. NREL includes those costs in its study. Ultimately, the consumers bear that cost.

They assume carbon capture and sequestration will work. Sequestering CO2 in a depleted oil or gas field is a relatively figured out problem. Capturing it from the exhaust of a power plant is not. Most proposed power plant carbon capture projects failed before they ever got started, the most egregious running up more than 7.5 billion dollars  in costs (more than triple the initial estimate) along the way. There are only four such plants operating in the world. Two are recent pilot projects in China, without public data. The two in North America have a terrible track record. The Petra Nova plant in Texas aimed to capture 33% of emissions and actually got 27%, while the Boundary Dam plant in Canada aimed for 90% and averaged less than 60%, meeting its daily capture goals for only 3 days in the first 3.5 years of operations. Project proponents claim CCS at the proposed Susitna plant will operate at a never-before-seen success level, capturing 95% of emissions. If this plan runs into similar problems as other CCS operations, the federal tax credits won’t be enough to pay back the costs. 

The fuel and power plant costs are wrong: The report not only fails to compare the coal plant costs to renewable power costs, it fails to compare them to power plant level costs at all. The only comparison they make is to a Chugach retail industrial rate that they pegged at 18.8 cents per Kwh, over twice actual power costs from current gas plants. For the coal plant, they assume capital costs vastly cheaper than listed in other Railbelt studies, coal prices significantly less than the Healy plants pay, and a capacity factor higher than any fossil plant on the Railbelt today. (see detailed assumptions at the end)

Adjusting any one of these assumptions to something more plausible bumps up the coal plant price to well above the 76% renewable scenario. Adjusting all of them bumps it well above even the costs of running gas plants on imported LNG, double the proponents’ estimate, and on par with the most expensive plants we use today. There are also some optimistic assumptions I didn’t change. I adjusted the capture success rate of the carbon capture plant, but left its costs as stated -- accuracy on those costs seems extremely unlikely for a technology with only two examples, neither recent. Experimental technology is rarely under budget. I also kept their assumption that the federal tax credits will last for the entire 30 years of the project, rather than the 12 years that is in current law. 

Why are we talking about this at all?

My analysis above frames the question based on Railbelt power consumers. And from that perspective, it makes no sense to even consider this. If the question is “How should we power the Railbelt?”, it’s clear that even our not-so-good options (lots of imported LNG) are better than this, and our good options (mostly renewables) are vastly superior.

But the proponents of this plan may actually be answering a couple of very different questions: “How can we support a specific set of potential mines?” and “How do we take advantage of generous federal subsidies for carbon capture?”

The power needs of the proposed Donlin mine are mentioned many times in the report. Donlin is unlikely to be able to buy gas from Cook Inlet to supply its power. 

Additionally, Flatlands’ energy’s coal lease is useless without a customer for the coal, such as this power plant. Coal hasn’t been exported from Alaska since 2016, and the Usibelli mine supplies all the existing coal plants. Other potential mines are one of the big drivers for the West Susitna Access Road. Cheaper solutions to the Railbelt’s power needs don’t do anything for these mine claim holders. But they would benefit if the public built them some infrastructure.

This plan is being promoted by the state’s CCUS working group. With half the members from industry, and the rest from governments and the university, it’s mission is to “accelerate commercial carbon capture, use, and storage (CCUS) projects within the state,” by means of finding a project that could potentially be funded by the new $85/ton carbon capture and storage tax credits. Having dismissed all of Alaska’s existing coal plants for not being in good carbon storage locations, and our gas plants for their too expensive fuel, it seems they felt their only option was to put a new coal plant somewhere close to storage. There is no “commercial carbon capture” for power plants.  

What about the climate impacts?

The project proponents tout that a coal plant with 95% carbon capture would produce less than half the emissions of a natural gas power plant, at 0.2 metric tons per MWh as opposed to 0.44 (the current Railbelt total for all generation is around 0.5 metric tons/MWh). This is theoretically true, if such a plant existed. But neither of the two existing coal plants with CCS has come even close to capturing 95% of its CO2. One never aimed to capture more than a small fraction. The Boundary Dam plant that originally set its sights on 90% capture later changed its target to 65%. The plant’s recent emissions are 0.4 metric tons/MWh, and their reports have drawn the line of success as staying below the threshold to trigger a carbon tax. In contrast, the 76% renewable scenario would produce around 0.1 metric tons/MWh, less than half of that theoretical coal plant, and far less than any actual coal plant.. 

Of course, it’s true that any new technology starts out expensive and not working well. Some ideas will always fail, and some, over time, will improve. So there is a place for trying out CCS technology -- and that place is where there are already emissions. If you try this on an existing coal plant, and fail, at least the problem hasn’t gotten any worse. If you build a new one, you’re creating a brand new long lasting climate problem in order to maybe, partially, if you’re lucky, solve it. 

The report throws out a few ridiculous calculations to supposedly make the plant climate neutral. One involves offsetting the climate impacts of vast fields of imaginary coal-heated greenhouses by piping in CO2 and waste heat rather than heating them with coal directly. Certainly one could reduce emissions by not burning coal in your greenhouses, but no such greenhouses exist so it’s hard to see how one would achieve an offset by reducing their emissions. Another involves burning beetle-killed spruce trees so they won’t slow down the growth of future forests, but there is no evidence (even in the citation provided in the report) that trees somehow grow faster if you cut down the dead trees next to them. 

In Conclusion: 

This idea has received far more press than it deserves, and is distracting from sensible near-term efforts to address the gas crisis, cheaper longer term solutions like the 76% renewable scenario, and current legislation to smooth the way, such as the bill addressing wheeling fees and tax structures for independent power producers. But even if it doesn’t make sense, it could still get built. Or at least cost a lot of money along the way to being abandoned (like many CCS proposals).

And if it does, one of the most worrying aspects about this expensive coal plan is that the costs and risks will fall on the public. Unlike renewable power projects, there are no independent power producer companies coming in wanting to finance and build this. There won’t be, because any such company would have to account for those cost risks, fund the infrastructure, and offer the utilities a firm, competitive, long-term price for the power. If this is built, it will be on our dime (the state and/or publicly-owned utilities). Right now, the university is asking for a $2.2 million grant match to fund the next piece of the study. It’s small, compared to the roughly $4 billion total cost, but it’s a small step in the wrong direction. There may be some stakeholders who benefit from this idea, but it’s not the Railbelt ratepayers.

Subscribe now

Detailed Assumptions:

Road: West Susitna Access road costs from AIDEA. The Middle Susitna Skwentna option at $453 million is the only one that reaches the proposed mine/power plant area. I used half that cost, as the full proposed length goes farther than needed to reach the plant.

Transmission: Transmission costs of $17.7/KW/mile from the NREL study, doubled. 300MW is a large enough portion of the Railbelt load that it will need redundancy, to account for the contingency that a single line will trip and cut out half the Railbelt’s power. Similarly, the Susitna Watana hydro project modeled a triple circuit for that reason.

Coal plant

Size: The feasibility study highlights two proposed coal plant sizes: 400MW gross (300MW net with CCS) and 100MW gross (75MW net with CCS). While transmission is substantially cheaper for the smaller plant (and wouldn’t need a double circuit), everything else is more expensive, and it is less economic than the larger plant in both the authors’ and my analysis, so I don’t show it here. 

Capital costs: The feasibility study uses capital costs of $5,572.5/Kw for the 400MW coal plant, while NREL lists $7,799/Kw, and the 2010 Black and Veatch Railbelt Integrated Resource Plan lists $10,109/Kw (when adjusted from 2009 to 2024 dollars). UAF’s recent coal plant using similar technology was constructed for $15,000/Kw. I used $7,000/Kw, in my analysis, which is still much more optimistic than all the other estimates.

Capacity factor: The study states that the new coal plant would have a capacity factor of 85%, which is much higher than any fossil plant on the Railbelt today (topping out in the high 60s/low 70s). I adjusted this to 75%, which is still higher than any current fossil plant on the Railbelt.

Fuel: They give coal costs at $3.50/MMBtu. I used coal costs equal to 2022 coal costs at the Healy plants of $4.98/MMBtu. It seems unlikely a new mine will be substantially cheaper than an existing mine, mining what the authors list as a similar coal deposit. 

CCS: I used the overall CCS costs as stated, but adjusted the amount captured to 72% of their stated target, based on the average of the success (relative to targets) of Petra Nova and Boundary Dam, assuming most of the cost goes into building the facility, pipeline, and wells. This leads to a cost of $89.30 per Mt captured, slightly more than the $85/Mt tax credit. I did not adjust the capital cost of the CCS facility, even though any experimental facility is likely to come in over budget. I also kept the assumption that the tax credits, currently scheduled to expire in 12 years, will extend for the entire 30 years of the project.

Gas Plants:

Fuel: Current gas costs are taken from actual values from RCA filings for each power plant in 2022 -- the total amount paid at each plant for gas, including gas under Hilcorp contract and gas owned by Chugach. Imported LNG is modeled at $14/Mcf, with a gas plant efficiency of 8.2 Mcf/MWh, equal to overall efficiency of all gas plants in 2022. The NREL study lists gas costs at $12.1/Mcf, rising to $13.2/Mcf by 2040. I used the higher number as a more conservative estimate, to show how the coal plant compares to gas on the more expensive end of existing predictions.

Non Fuel: Non-fuel costs are derived by summing the “total production costs,” minus the fuel cost, for every gas plant on the Railbelt in 2022, from RCA filings. This number is divided by the total gas plant net MWh for 2022. This includes both regularly used and rarely used plants, and both fixed and variable production costs.

76% Renewable scenario: Total costs and fuel costs for the 2040 reference scenario of 76% renewables are from the graph on p 35 of the NREL report, divided by the total generation of 5941 GWh from  p38.

Subscribe now

The increased price of imported LNG is not, in itself, a crisis (more on that in a later post). Actually running out of gas -- being unable to provide enough heat and electricity to the Railbelt without other options in place -- would be a crisis. 

So we decided to analyze what we see as one of the biggest missing pieces of the conversation.

How much, and how fast, can we reduce consumption?

Consumption hasn’t been flat in the past: Historically, every official report on Cook Inlet gas shows future consumption flat, and every report has been wrong. In 2011 the state predicted 90 BCF of use every year, by 2018 it was 80 BCF, and in the 2022 report report, it was 70 BCF. Much of that was due to more efficient gas-fired power plants. There’s little opportunity left to improve our gas plants, but a lot of opportunity to use them less.

Consumption doesn’t have to be flat in the future: Using numbers from the utility working group 2023 report, we modeled known near-term renewable projects and an increase in home efficiency. With those straightforward measures, we can reduce gas use below the production estimates for several years. These gas savings (in the form of stored gas or unproduced gas) can be used to offset production declines in later years. 

The utility analysis shows a gas supply “gap” of 8 BCF in 2028. In our analysis, this gap is pushed back by nearly two years. 

Utility demand of 65 BCF from the mid-case of the 2023 working group report (p 12-14), production data from DNR’s truncated 2022 forecast.

How did we come up with that?

Current demand can be broken into three main categories; industrial use, heating, and electricity generation.

Industrial: Most of the industrial use is oil and gas field operations. Those have remained relatively constant, even as production has dropped dramatically (slide 10 - from DNR), and may be a sort of unavoidable ‘parasitic load’ on the system. Other industrial users include the Tesoro refinery. We didn’t model any decrease in industrial use.

Heat: Efficiency measures are effective. Overall, Enstar’s residential customers use 10.5% less gas today (weather normalized) than they did in 2007, before the Home Energy Rebate and Weatherization programs. Participants in the Home Energy Rebate program reduced their gas consumption by 26%. Those programs reached fewer than 20% of Enstar’s customers, suggesting there are significant savings still possible. 

However, heat consumption has to be reduced at the level of each individual building, through weatherization, furnace and appliance upgrades, behavioral changes, or switching to a different heating source. That makes it hard to do quickly. The number of inefficient heating systems and homes is smaller now than in 2007. Additionally, the economics still favor switching to natural gas for homes heating with oil or propane (as I showed in this post). 

Putting all of this together, we decided to model a 15% total decrease in gas for building heat, ramped in over 10 years. That will likely require incentives, but not necessarily from the state. IRA funds will provide rebates up to $8000 per home for efficiency upgrades, tiered based on energy savings and income.

Electricity: This is the easiest sector to reduce quickly, displacing a portion of gas-powered electricity with renewable electricity. There are a number of nearly shovel-ready wind and solar projects being pursued on the Railbelt today. Over the longer term, NREL’s recently released report shows that a 76% renewable system is the cheapest way to power the Railbelt, substantially cheaper than keeping our current generation mix.

Between now and 2030, we modeled only those renewable projects currently in advanced stages of negotiation with utilities or planning by utilities, using their most recently-reported completion dates from news articles or utility filings and reports. Houston solar (already operational), Kenai solar and landfill gas (2026), Shovel Creek wind (2027), Little Mount Susitna wind (2027-8), and Great Lands solar (2028 - Chugach electric personal communication). Home net metering was assumed to increase at the same rate as recent years, around 2.5MW per year.  Annual generation assumptions can be found in this post. Shovel Creek wind was presumed to primarily displace oil generation, while all other renewables were assumed to displace gas, at the current Railbelt efficiency adjusted for part-load and operating reserve fuel use impacts.

Beyond 2030, we ramped renewable energy linearly to the 55% by 2035 renewable target set in the proposed RPS bill. 

Previously stored gas can fill the remaining gap: Our analysis shows a supply gap of 0.3 BCF in 2029, and 7.5 BCF in 2030, potentially before imports could arrive. However, Cook Inlet has a small ‘savings account’ of sorts in the form of stored gas, which has been accumulated over the past couple decades, largely by Hilcorp (slide 6). The current amount stored is equal to about half a year of current demand. 

Without the savings we modeled, the cumulative supply gap would be 43 BCF by 2030, more than is in storage. With them, our cumulative cap could be filled by less than a quarter of the currently stored gas.

What about subsidizing gas? Last time we tried, we spent around $1.5 billion dollars in cash to Cook Inlet producers. In some years, those subsidies were more than the entire cost of gas for Enstar and all the electric utilities put together. A decade later, we find ourselves in the same place.

So should we spend $10.7 billion instead? The Alaska Gasline Development Corporation has been pushing their North Slope pipeline for years, with no luck so far getting the investment needed to pay for it. Now they are back to promoting a “phase 1” in-state pipeline for $10.7 billion dollars, which could theoretically, if we start now, beat LNG by a few months. And while they hope it will be an export project eventually, that hasn’t panned out so far. “In-state” means that one way or another, Alaskans are paying for it. 

Alternately, $10.7 billion could pay all gas and electric bills for all Railbelt households for 18 years.

Or give every household in Alaska over $40,000 for weatherization and efficiency.

Or build enough wind and solar farms to produce more than twice as much power as the Railbelt currently uses.

HB 222 directs the state to actually invest the permanent fund in a gas pipeline

What about royalty relief? There are two proposals to cut royalties on new gas (HB 276 and 223), and one to give tax credits for jack up rigs drilling in the inlet. These aren’t refundable, so they are less egregious than the cash credit subsidies of the past, and far less so than subsidizing a gas pipeline. But they’re unlikely to lead to more gas production in the short term, nor to stop the need for imports in the medium term, so they don’t address the main problem.

Exploration will be as expensive, risky and time-consuming as it ever was. Royalty relief only kicks in once gas is already being produced from these new areas, so will probably come after 2030, if at all. 

It could allow someone to sell gas that otherwise would be too expensive for the market to bear. We already have lots of unused flexibility in prices. Cook Inlet buyers, worried about the costs and availability of imports, would likely sign contracts for new gas at prices much higher than today. But those contracts aren’t being offered. Price does not appear to to be a limiting factor in current investment. 

In Conclusion: Don’t panic, build an import facility, move forward with renewable projects already in the pipeline, and make our homes and businesses more efficient.  Over the longer term, gas consumption can continue to ramp down. And if we avoid locking ourselves into long term contracts or expensive subsidized fossil fuel infrastructure, imports will allow us the flexibility and time to develop local renewable energy for both electricity and heat.

Subscribe now

It takes a fair amount of work to puzzle out what it would do. Or why. Did someone let Usibelli coal mine write themselves into a clean energy bill? Or did they throw it together at the last minute without really reading what they wrote?

Subscribe now

Before the bill’s first hearing tomorrow, I thought it was important to call this out. Defining coal as clean is the most egregious thing in this bill. As far as the rest goes, the incentives in it are probably too small to be meaningful, and strangely limited to exclude projects by independent power producers. It also appears to direct the RCA to come up with an entire Railbelt transmission plan, insists that Copper Valley be connected to the Railbelt, and sets the RCA to create a new net metering rate that may entirely replace the current system

Clean Energy Definitions

Here’s what “clean energy” means, according to HB 368.

“Sec. 42.05.935. Definitions. In AS 42.05.900 - 42.05.935,  

(1) "clean energy" means electrical energy that 

(A) when generated by a load-serving entity, does not release  carbon dioxide or releases carbon dioxide in an amount that is offset by the amount of carbon dioxide the load-serving entity absorbs or removes from the  atmosphere;

(B) is generated from coal with a sulfur content of one percent or less by weight;

(C) is generated from renewable energy resources; or 

(D) is generated from nuclear energy; “

This list defines coal as clean. It produces the nonsensical result that energy produced by low-sulfur coal is considered clean by default, even without carbon capture, while natural gas power (which produces vastly lower sulfur dioxide pollution and lower CO2 pollution by default) is not. I can’t see how that makes sense to anyone.

The only coal mine in Alaska is Usibelli, which produces coal with less than 1% sulfur by default -- therefore all coal counts as clean.

What else is in the bill?

Targets:

The bill sets “targets” for clean energy of 35% by end 2036, and 60% by end 2051. While there’s a whole lot of text about what counts for compliance with and exceptions from these targets, their only real effect is to set eligibility for the bill’s tax credits. (Actually, I can’t find anything that says a utility that fails to meet the targets is ineligible, only that one that has a waiver or has opted out of them is ineligible). 

Also, while the targets have years attached, they’re actually pinned to completion of an undefined transmission upgrade plan that the RCA has to create. 

“The commission shall adopt a minimum standard for electric power transmission lines sufficient to ensure seamless end-to-end electrical energy transmission.”

That sounds more like the RCA creating a transmission plan rather than a transmission standard, and is referred to as “the upgrade” thereafter. 

Regardless of the year listed in the bill, the first ‘target’ comes into effect 10 years after “the upgrade” is completed. And while there’s nothing much defined about that upgrade plan, it must be assuming that the plan includes extension of the transmission system to Copper Valley, because of this part: “60 percent of sales must be from clean energy within 25 years after the upgrade is complete, or when electric power transmission lines connect the interconnected electric energy transmission network in the Railbelt to the service area of the Copper Valley Electric Association, whichever is later.” This despite the fact that Copper Valley isn’t currently part of the Railbelt, so its connection can’t possibly be required to “ensure seamless end-to-end electrical energy transmission.” Apparently, the RCA is also going to define those ends, and someone wants to make sure they include Copper Valley.

Clean Energy Tax Credits:

Depending on that vague transmission plan, the “targets” may or may not ever be triggered to potentially render a utility ineligible for the credits. Certainly all utilities would be eligible for them between now and 2036. What would the credits do?

Credits are a 0.2 cents/kWh tax credit for the first 10 years of newly-built “clean energy” projects. That amount is the same as the proposed RPS penalties I discussed in the last post -- quite tiny compared to utilities’ overall costs. So even if they do incentivize coal power, they don’t incentivize anything very much.

The federal production tax credit gives between 2.75 and 3.35 cents/kWh (though not for coal power!). So, more than ten times as much (and inflation-adjusted, unlike the proposed credits in this Alaska bill).

One of the facility qualifications is “is owned by the load-serving entity.” So none of the large renewable projects being discussed by Railbelt utilities right now would qualify (all are being built by independent power producers).

These tax credits can’t be paid as cash, but only against tax liability. Utilities on the Railbelt are co-ops, but they do pay a Gross Revenue Tax of 1/2 mill per kWh, which is paid out to boroughs and cities where the power is sold. The yearly total of these taxes is around $2 million across the whole Railbelt, which utilities could entirely offset by producing 1/4 of their power in a credit-eligible way.

Net Metering:

The bill instructs the RCA to come up with a new buyback rate for net metering systems. Currently, exported power is credited at the retail rate up to the user’s consumption within a given month, and at the calculated “small facility purchase power rate” (generally similar to the cost-of-power-adjustment, around 5-10 cents per kWh depending on the utility and quarterly power costs) for excess production. It’s unclear to me whether the new rate would only replace that excess production purchase rate, or whether it would replace net metering entirely.  

That depends on whether “electric energy supplied to a load-serving entity” counts everything that hits the meter or not. I’d tend to read it as replacing net metering entirely.

“The commission shall by regulation establish a method to determine annually the amount of a reasonable seasonal and time variant export credit rate for electric energy supplied to a load-serving entity by a customer's distributed energy system. In determining the export credit rate, the commission may consider any relevant factors, including avoided costs of load-serving entities.” 

PCE:

The bill changes PCE rules to exempt sales of recovered heat from the PCE calculations. I don’t know enough to say whether this is a good thing. Seems like it might be, but also seems like it’s easily missed hidden in this bill, so I wanted to mention it.

What next?

If someone writes a clean energy standard bill to define clean energy in a sensible way and attempt to drive it forward with meaningful incentives, it would be interesting to have that conversation. In the meantime, I’m stuck talking about how much of a mess things are.

Subscribe now

• Cold Day Deliverability: This is the ability of the natural gas infrastructure to deliver enough gas during a cold day in the winter when demand for gas is highest due to high heating needs.

• Sufficient Annual Production:  The total production of natural gas across the year needs to be sufficient to meet the total need for gas.  While we do have a substantial amount of stored natural gas, a deficit in annual production will quickly draw down those natural gas stores.

As I discuss below, the Cold Day Deliverability problem can be solved with substantial certainty and only moderate amounts of money. The much more serious problem is maintaining sufficient annual natural gas production or importation to meet the annual gas demand; I plan to discuss this topic in a future post.

Cold Day Natural Gas Deliverability

On an average day during the year, Cook Inlet natural gas users need about 190 million cubic feet (MMcfd) of gas.  However, on the coldest day designed for (a bit colder than January 31 of this year), users would normally demand about 390 million cubic feet of gas, slightly more than double the average day (see Table 4 of the Berkeley Research Group report on Cook Inlet Gas). For about 10 years, gas wells in Cook Inlet have not had the ability to ramp up to this high level of demand, so natural gas storage has been utilized to supplement the production from the gas wells.  When the wells are producing more than the need for gas in the summer, gas is stored into underground storage facilities (old natural gas reservoirs).  In the winter when demand is higher than production, gas is withdrawn from storage to supplement production from the gas wells.

During the recent cold weather event on January 31, it is important to know that there was plenty of natural gas volume in storage facilities to meet the gas demand.  The CINGSA storage facility alone had 6,730 million cubic feet of gas actually in storage, which is sufficient to supply the natural gas production deficit under design day conditions for 33 days.  In addition, Hilcorp has substantially more natural gas storage than CINGSA. Although I could not find data on the current volume of gas stored in Hilcorp storage, during mid-February of 2022, Hilcorp had a total of 15,300 million cubic feet in storage (Luke Saugier presentation to House Energy Committee on 2/21/2023).  So, we likely had enough gas in storage on January 31 of this year to supply the production deficit for multiple months of design-day cold weather. The problem on January 31 was not a problem of having enough gas in storage.

The ostensible problem on January 31 was the ability to supply natural gas from storage at a fast enough rate. Removing gas from underground storage involves using wells to extract the gas, compressors to provide sufficient flow and dehydrators to remove moisture from the gas. CINGSA, the regulated gas storage facility owned primarily by ENSTAR-related companies, had been experiencing problems with two of its wells.  CINGSA was designed to supply 150 MMcfd of deliverability, but because of sand intrusion in two of its wells, CINGSA showed its available deliverability to be only 97 MMcfd on January 31. ENSTAR is the largest user of CINGSA and this loss of deliverability from CINGSA put ENSTAR in a pinch.  ENSTAR has firm commitments for substantial deliverability directly from the gas producers, primarily Hilcorp, but not enough to make up for the lost CINGSA production.  Railbelt electric utilities helped out by reducing their gas demand through stopping sales of gas electricity to Golden Valley Electric (who replaced the electricity with oil generation) and by ramping up hydroelectric production.  Hilcorp came up with additional gas beyond their contractual commitments to ENSTAR.  So, heat and power in Southcentral Alaska continued to be delivered.

How close were we to a customer-impacting shortage?  This analysis I did looks at the total amount of deliverability in Cook Inlet (including CINGSA at its crippled 97 MMcfd level) and compares it to an estimate of January 31 demand.  That analysis shows that supply of deliverability exceeded demand by 24% assuming that all of the non-CINGSA sources of deliverability were fully available and not experiencing problems. My attempts to contact Hilcorp about their actual deliverability on January 31 were not successful.  This is an important analysis, similar to comparing available electrical generation capacity against peak electrical demand. We should have better publicly available data on the natural gas peak-day situation.

The analysis I did also estimated the temporary help that Railbelt electric utilities could provide by reducing their demand for gas.  A portion of these actions were implemented on January 31, but a number were still available if the situation got worse.  Specifically, the Matanuska Electric Association Eklutna Generating station can be switched from natural gas to oil fuel.  If that plant were run at full capacity using oil, it would have relieved about 9% of January 31 gas demand.  There was the ability to further ramp up hydroelectric production, saving about 4% of gas demand. Importing 30 MW of power from Golden Valley Electric across the Northern Intertie would have relieved another 1.6% of demand.  So, those potential contributions are significant and were not needed on January 31.

While the data available to me is not conclusive, let’s assume that we cut it too close on January 31.  How can we fix this problem?  Again, there is plenty of gas in storage; the problem is getting it out of storage at a higher rate.  ENSTAR is undertaking an expansion project at their CINGSA facility that will increase deliverability by 65 MMcfd, which represents 17% of peak design-day needs of Cook Inlet and represents 22% of ENSTAR’s peak needs.  So, this is a substantial increase in deliverability. The project also adds 2 BCF of storage volume to the existing 11 BCF storage capacity of CINGSA.  This very noticeable improvement in peak day withdrawal capacity comes with a price tag of $72 million.  To put $72 million in context, remember that ENSTAR was recently purchased for $800 million. Hilcorp has stated in their presentations to the Legislature that they invest hundreds of million of dollars per year into Cook Inlet gas facilities. The solutions to our annual gas production needs identified in the Berkeley Research Group report vary in capital cost from $570 million to $9 billion.  So, I would characterize $72 million as a very moderate investment.

My conclusion is that the cold-day deliverability problem can be solved with near certainty and with a moderate amount of investment. I do wonder why the solution hasn’t come earlier. I had much more concern about this problem when I first read the interview with John Sims, the ENSTAR president, in the Northern Journal and heard Mr. Sims’ presentation to the Legislature.  In my opinion, those presentations were more alarming than justified after a deeper dive into the cold-day issue.

In a future post I plan to address the more serious problem of adjusting to dwindling annual production of Cook Inlet natural gas. In my opinion, the solution to that problem will be more wind/solar electrical production, increased energy efficiency, and, unfortunately, importation of Liquid Natural Gas.

Subscribe now

Chugach Electric’s board just passed a resolution in support of an RPS, right after Golden Valley Electric published a statement opposing it. Homer Electric and Matanuska Electric have not put forth public statements on the bill, but Homer Electric recently repealed their voluntary renewable energy goal, and Matanuska Electric has never had a meaningful renewable energy or emissions goal.

SB 101 proposes a standard of 25% renewables by 2027, 55% by 2035, and 80% by 2040, with fines of $20 per MWh (2 cents per KWh) if the targets aren’t met.

Golden Valley’s statement specifically calls out the penalties as one of their objections. Last year’s bill hasn’t gotten much discussion yet, but utilities’ comments on Dunleavy’s version also revolved heavily around the potential rate impacts of the penalties. It’s true that penalties would be paid by co-op members — there’s no one else to pay them. It’s also true that the penalties proposed are a miniscule fraction of the total utility costs.

Even if utilities make no attempt whatsoever to meet the standards, the penalties increase total costs by less than 1% until 2036, and less than 5% at their maximum.

This makes sense if you think about it for a minute. Fines are 2 cents per Kwh, and in early years, would only be leveled on 10% of Railbelt electricity. Two tenths of a cent per Kwh is well within the range of quarter-to-quarter fluctuations in the cost-of-power-adjustment that most customers don’t notice.

And in the future, while 2 cent per KWh fines are fixed in law, inflation continues to apply to everything else, rendering the fines a much smaller percentage of the total. As the targets increase, those fines go up, but add less than 5% at their maximum in 2040. In its current form the bill forgives the first year of non-compliance, reducing the impact of fines even further.

In this analysis, I used a 2% escalation for general costs, a 3% escalation for fuel costs, and flat energy consumption. The Cook Inlet natural gas crisis means that fuel costs in this no-new-renewables case will likely increase a lot more than that — pushing total costs higher, and making the fines appear even more irrelevant. Additionally, a power purchase agreement signed and approved within a given year can count as part of a utility’s total even if the actual energy doesn’t come until two years later, giving plenty of wiggle room for the earliest 2027 deadline.

Utilities will save costs by pursuing renewables rather than relying on expensive imported gas. But if they choose not to, they can afford the fines. If they do try to transition, and miss a few deadlines along the way, the fines are even smaller. 

There’s also a debate about over whether or not fines can be paid from member rates. This graph assumes they can be, comparing the fines to total revenue. If the fines have to be paid out of margins, that blue area gets a whole lot smaller. To cover a fine with margins, a utility would likely take the money from past members rather than current ones -- deciding not to pay back capital credits to past members. Ultimately it is members’ money either way, so I think it’s better to include it in rates. Current members have the ability to complain to the RCA or vote out their utility boards if they don’t like what’s happening, which past members can’t do. Also, since the penalties are small, ratepayer ire at having any penalties at all included in their bill is probably more of an incentive to meet the goals than the fines alone could be.

You can have a renewable energy goal without penalties, of course. We’ve had one since 2010, which says Alaska will reach 50% renewable energy by next year. In the 14 years since the goal was passed, the Railbelt has added around 2% wind energy, and less than 0.5% solar, to reach a total of around 15% renewable energy (the state total is higher, around 1/3, largely due to hydropower in Southeast which existed long before the goal was passed). 

That’s where a penalty-free goal will get you.

We shouldn’t need penalties to incentivize renewable development. Economic incentives point that direction already. But that should make us even less scared of the penalties in an RPS. 

Subscribe now

Subscribe now

Most of the Railbelt’s electricity comes from natural gas, and with the Cook Inlet gas crisis, that’s most of what I’ve been talking about. But 12% comes from coal, all of it through Golden Valley Electric Association (GVEA), which doesn’t have access to gas directly. While gas power has decreased over the last decade, coal power production has actually increased.

All data in this post comes from public EIA, EPA, and RCA filings.

Despite providing a small fraction of total Railbelt electricity, coal plants are responsible for over a third of direct CO2 emissions. 

Two of the coal plants are in Fairbanks (Aurora and a small one at the University), and provide local heat as well as power. For the analysis shown above, combined heat and power plants are considered to have lower emissions per unit of electricity than power-only plants. Burning coal produces the same amount of CO2, but some of that is attributed to the heat production, which would otherwise have been produced by burning something. Therefore, the power-only coal plants, Healy 1 and Healy 2, have the highest emissions per kWh on the Railbelt.

The discussion about closing coal plants started with Healy 1. It’s an old plant (1967), and needed an expensive pollution reduction system (selective catalytic reduction system, or SCR) to meet EPA requirements. GVEA decided to spend $30 million on upgrading that system to avoid having to shut down the plant.

Healy 2 already has an SCR, and faced no such immediate requirement. But Healy 2 is unreliable and expensive, and GVEA decided to close it instead. Between 2018 and 2021, Healy 1 had a reliability of 91%, whereas the reliability of Healy 2 was only 59%.

This makes Healy 2 one of the most expensive plants on the Railbelt to run. The coal that fuels it is in line with current natural gas costs, but the rest of the operating costs are vastly higher.

Healy 2 is one of those quintessential Alaska boondoggles. It was built in 1998 as a $300 million demonstration project by the Department of Energy’s Clean Coal technology program (which kicked in $120 million), AIDEA ($150 million), the Alaska legislature ($25 million), Usibelli coal mine and GVEA ($10 million together). It used an experimental new technology to burn waste coal that never really worked. The plant never successfully passed its testing phase and shut down by 2000. GVEA and AIDEA were in a legal dispute over the plant, which GVEA ended up agreeing to buy in 2009. HEA was originally going to be part of that deal, but after public outcry and an overturning of the board, they pulled out. GVEA finalized the deal in 2013, but didn’t get Healy 2 into regular commercial operation until 2018.

GVEA’s self-imposed deadline to close Healy 2 at the end of 2024 is approaching quickly, and I don’t know if they are on track to do it or not. One of the elements of that plan involved contracting for a large amount of wind power (which is in progress). Another element involved buying more gas power from southern utilities (which depends on uncertain gas availability). But if they do close Healy 2, the legislature shouldn’t stand in the way. 

Subscribe now

Subscribe now

What is community solar? It’s a model that sits in between home solar systems and utility-scale solar farms. 

• Home solar systems serve a single house, and much of the power they produce is used to offset power consumption, credited at retail rates.

• Utility scale solar farms provide bulk wholesale power for a utility to distribute to all its customers. In Alaska, the largest such farm is RIPP’s Houston solar farm, which produces power equivalent to the consumption of around 1500 houses, and sells it to MEA at a rate below the cost of wholesale gas power. Other active proposals on the Railbelt would be much larger. 

• Community solar farms sell subscriptions to individuals, designed to imitate the effects of net metering for people who can’t install their own solar panels. 

In an ideal world, community solar would 

• Meaningfully increase renewable energy generation.

• Provide renewable energy that’s additional to and/or independent of utility efforts. 

• Provide access for lower income people and renters who can’t install solar panels. 

Chugach’s proposed plan does none of those things. 

Per kilowatt hour, the project is very expensive for subscribers

Chugach’s tariff proposes an initial subscription cost of $22.27 per panel per month, each of which is expected to produce 400 kWh of electricity per year. This results in a cost of $0.67 per kWh for the solar power. An average residential customer seeking to offset their entire consumption this way would roughly triple their electric bill. 

Chugach expects to get a federal tax credit of 30% for their project, and will pass that to the subscribers. They also propose a declining rate for their subscription. Including both of those factors leads to a project lifetime ‘best case’ cost of $0.45 per kWh. A Chugach staff presentation last summer contemplated a somewhat cheaper price than they eventually put in the tariff — but even those preliminary predictions were quite expensive, around $0.37 per KWh.

The best case is nearly double Chugach’s current rates of $0.21/KWh. 

Why is that? It’s not because solar power is inherently expensive.

Solar is actually quite cheap. Using the “Cost of Power Adjustment” as a rough approximation for wholesale power costs, Chugach’s current power costs (mostly gas power) are exactly the same as in the most recent Railbelt solar contract. And while those costs are likely to go up as gas costs rise, the solar contract price only increases a little.

Putting those two charts together, it’s clear that Chugach could buy wholesale solar power from an independent power provider, tack on the cost to distribute that power, and end up with something much cheaper than their community solar rate.

Solar isn’t expensive, but Chugach’s project is expensive. 

This Chugach project is actually much more expensive per unit than rooftop solar. Total listed project costs are $2.7 million for a 576 kW project. That’s more than $4,500 per kW. Current prices for a rooftop solar system in Anchorage are around $3,200 to $3,400 per kW. 

Why is Chugach’s project so expensive? I’ve already shown that the capital cost of the project is quite high relative to rooftop solar and IPP solar. A second part of the problem is how Chugach has chosen to structure their cost recovery. Here’s the issue in a nutshell, from an analysis by my blog partner Alan: Chugach plans to finance the project 100% with debt, making their interest payments quite high. Then they plan to collect a large margin component on top of this loan interest (amounting to 75% of interest payments). This margin component is not a cash expense and only serves to build up equity for the cooperative on the back of the community solar project.  On a present value basis, over-recovery of costs relative to actual cash expenses amounts to $1.1 million or 39% of the capital cost of the project.  Here is the link to Alan’s analysis.

The project is also so small as to be nearly meaningless: “generating electricity equivalent to an average of 0.02 percent of Chugach's retail energy sales annually.” Chugach is simultaneously pursuing utility scale wind and solar projects that are nearly 250 times the size of this community solar farm. Independent power producers are clearly able to build cost-effective solar in Alaska, and I’m sure they’d be happy to turn a 120 MW project into a 120.5 MW project if Chugach requested it.

Back to my three potential benefits of community solar… Does this:

• meaningfully increase renewable energy generation? -- No -- 0.02% isn’t meaningful.

• provide renewable energy that’s additional to and independent of utility efforts? -- No -- Chugach is spending money and effort on this plan that could be used for cheaper utility-scale solar, so it may lead to less solar overall.

• provide access to lower income people and renters who can’t install solar panels? -- No -- because it’s much more expensive than baseline power.

Chugach’s surveys show that its members want community solar. But I’m assuming that what they want is actually one or all of the benefits I listed above. 

Not only does this project not make sense, I don’t think that a community solar subscription program within a cooperative utility can ever make sense. Stay tuned for more on that in a later post.

This project is a distraction from the large utility-scale renewables that Chugach should be pursuing for all its members.

Subscribe now

With the impending Cook Inlet gas crisis pushing those prices up, will that still be true?

Subscribe now

I particularly wanted to know how those numbers worked out for air source heat pumps. Heat pumps (and direct electric heat) are the only easily available heating methods that don’t involve burning anything directly. Which means they don’t produce local pollution, and have the potential (depending on the electricity source) not to produce CO2 either.

I’ve been happily heating my place with a heat pump for a bit over a year now, which means I get lots of questions about how well it works (great), how much it costs (about $600/yr for my small place), and who installs them (harder question). 

My co-blogger Alan has developed an excellent heat pump calculator that lets you look at the economics for your specific house, including lots of characteristics of the building itself, different heat pump models, installation costs, and more. If you’re thinking about getting one, check it out!

For straight operating costs, the general rule-of-thumb answer is that Enstar gas is cheapest where it’s available, but heat pumps are basically always cheaper than propane, and often cheaper than heating oil, especially if your heating oil is expensive. 

I decided to analyze how that general answer might change with rising gas prices, renewable energy, and load growth. This is based on cost numbers for Wasilla as a representative Southcentral community -- the answer looks similar for most of the urban Railbelt except Fairbanks (where natural gas is less available for heat and electricity). Detailed assumptions and methods are at the end of the post.

Gas heat gets expensive if gas prices rise. But most of our electricity is produced from natural gas, so the heat pump also gets more expensive. It takes gas prices to triple before they get close, and at that point, oil heat is cheaper.

If more of the electricity is produced from renewables, the heat pump costs become less dependent on gas. The heat pump stays cheaper than oil heat under most scenarios, but more expensive than gas heat until gas prices reach around double those of today.  

Most of the heat pump operating costs are based on the fixed costs of operating the electric grid. These don’t change with gas costs or electric use, so if more electricity is sold, those go down. If electric use goes up dramatically (nearly double current use based on the 2050 “moderate load growth” scenario from this ACEP paper on beneficial electrification in the Railbelt), those fixed costs drop, and heat pumps become competitive with natural gas in the $12-$15/Mcf range that’s likely in the near term. 

Could that happen? The increased load in the beneficial electrification paper is mostly due to a projected increase in electric vehicles, rather than heat pumps. I don’t know whether that’s likely, but it could happen if electric vehicles save people money at current electricity and gasoline prices. It seems much less probable for people to increase load by doing something that’s currently more expensive than the alternative. That goes both for heat pumps and for the increase in industrial load that keeps getting promoted in the energy task force reports. (We can’t just “Be Iceland” and attract industry when industrial power is so much cheaper in so many other places -- unless we cancel out those gains with massive subsidies). 

People in real houses also have to factor in the capital costs and installation costs of a heat pump, the cost of backup heat if it’s needed, and any non-cost factors that matter to them.

When I got the heat pump at my place, I didn’t do any of these economic calculations. I didn’t even use Alan’s calculator. We were switching from wood heat, which had no direct monetary cost to us. Wood takes lots of labor, and most importantly to me, creates lots of pollution. Indoor air pollution, outdoor air pollution, and a great deal of climate impact. Burning wood sometimes gets treated as “carbon neutral,” but on the timescale of the dire consequences we’re trying to avert, it really isn’t. Wood has a worse carbon footprint than all these heat sources. My heat pump has the potential to be a cleaner and cleaner heat source if we clean up the grid. I’m a fairly low-income person, but to me, that’s definitely worth $600/year.

Subscribe now

Notes and assumptions

Costs are listed per mmBTU of energy delivered to the house, and don’t include installation costs for any heater. Assumed efficiencies are 80% for natural gas, oil, and propane furnaces, and a COP of 2.5 for the air source heat pump. Assumed prices are all based on Wasilla numbers, as a representative Southcentral location that avoids some of the complication of other areas (Anchorage has two different electric rates with different gas costs, Fairbanks doesn’t have Cook Inlet gas and would have less efficient heat pumps, Homer has an extra Enstar surcharge.) 

Current energy prices

• Heating oil = $3.64/gallon (from heat pump calculator). 

• Propane = $4.20/gallon (from heat pump calculator). 

• Natural gas = $1.15 per ccf (based on current base rate + gas cost adjustment (GCA) + monthly customer service charge normalized to average residential use). 

• Electricity = $0.2054/KWh (MEA Q4 2024 base rate + cost of power adjustment (COPA). Not including customer charge, since customers without gas heat are assumed to be without gas altogether, but customers without electric heat still have electricity)

Future gas prices:

The base rate and customer charge are assumed flat. The GCA was assumed to directly equal $/ccf in assumed future prices. This may be an underestimate, since the current GCA ($0.87/ccf) beats the current Cook Inlet prevailing value ($0.79/ccf).

Future electricity prices:

 Electricity base rates were assumed to be flat. The current COPA was split into a gas-dependent and gas-independent component (assumed to be flat). New renewable energy was priced at $0.08/KWh. Gas power efficiency was assumed at the 2022 central area plant efficiency of 8.09MCF/MWh. The gas-dependent component = 2022 fuel costs/ /total 2022 MWh (total generated+purchased - sold for resale). This component was varied based on hypothetical gas prices and renewable additions as shown below.

Base case: Gas price per KWh = $/MCF x MCF for native load in 2022 /total 2022 KWh 

Increased renewables cases: 

Gas price per KWh = $/MCF x (MCF reduced by defined percentage)/total 2022 KWh 

New renewable price per KWh = $0.08 x (KWh no longer met by gas)/total 2022 KWh

Renewable scenarios:

The 45% reduction is based on the lower bound of gas use for electricity presented in the 2023 joint utility report on Cook Inlet gas availability. The 75% reduction is based on 750TWh remaining natural gas generation from the 80% RPS standard in the initial NREL RPS report (range 600-900TWh depending on scenario). The 25% reduction is one I threw in to have a more moderate number.

Increased load scenario: This is based on the “moderate load growth” 2050 scenario from table 5 of this ACEP paper. Increased load was assumed to proportionately reduce the electricity base rate, as if the entire base rate is composed of fixed costs that do not increase with load growth.

Subscribe now

This gets brought up a lot in a couple of contexts.

1. Were the new power plants and upgrades in the 2010s a good investment in efficiency, or wasteful overbuilding?

2. How much fuel can be saved by “economic dispatch” -- always using the most efficient plant, regardless of which utility owns it? 

As the proud owner of a spreadsheet full of publicly available Railbelt utility data, I decided to do this analysis myself -- looking at the power production, gas use, and gas costs in the Railbelt’s central region from 2009-2022.

Over this time, there have been a series of efficiency initiatives in the Railbelt’s central region. First, there were a series of changes to the physical system. Chugach built the South Central Power Project (2013), MEA built its Eklutna gas plant (2015), and ML&P upgraded plant 2 to plant 2a (2017). Next, there were a series of changes in the way the utilities coordinated to use these plants. First there was a loose power pool between the three central utilities (2017). Then Chugach bought ML and P, merging three utilities into two (2020). Finally, Chugach and MEA were mandated to form an obligatory “tight” power pool (first full year in 2022). 

Generation from the new and upgraded plants more or less entirely replaced generation from the old plants by 2017. This had a substantial impact on efficiency -- improving it by 18%.

Any improvement after 2017 can likely be attributed to consolidation of utilities and better coordination between them. Since then, fuel efficiency has improved by an additional 6.5%, mostly in the first couple years of the loose power pool agreement. Efficiency improvements plateaued around 2021 with the completion of the Chugach and ML&P merger. 

Plant upgrades have saved an average of 4.6 Bcf of gas each year since 2013, and an average of $32.5 million on fuel costs each year (using Cook Inlet Prevailing Value gas prices). From 2013 through 2022, this adds up to 46.2 Bcf of gas, $325 million, and 2.5 million metric tons of CO2 emissions. Pooling and consolidation saved an additional 1.2 Bcf of gas and $9.4 million per year, for a total of 6 Bcf, $47 million, and 335,000 metric tons of CO2. 

News articles at the time estimated the three new plants cost around $1 billion in total (SPP, 2a, Eklutna). With that simplistic framing, about a third of that cost has been recovered in fuel savings. 

If it took 10 years to save a third of that cost, will it take 20 more years to get back where we started?

Probably not. If the plants hadn’t been replaced, presumably some fraction of that $1 billion would have been used to replace worn out components of the older plants. And as gas prices increase, each portion of gas saved is worth more money. Recent reports on the Cook Inlet gas crisis put future costs at around $12-$15 per Mcf. At those prices, the fuel savings from the newer plants nearly double -- to around $60-$70 million per year. On the flip side, that $1 billion number likely doesn’t include the ongoing interest payments for the loans used to build the plants.

What’s next?

By these metrics, our gas power will probably never get more efficient. It’ll likely get worse. But that’s actually a good thing. And the coordination between utilities will only get more critical. 

The graphs I’ve presented are looking at gas power only. We’ve likely achieved most of the savings we can in generating that power more efficiently. But we have a lot of opportunities left to use less gas power in the first place. 

Energy conservation has already saved a lot of gas. Even after accounting for a small shift from central region power production to Kenai Peninsula power production, gas power in the central region has declined by 18% -- an impact as large as the impact of all the power plant upgrades. 

The biggest opportunities for future savings are in renewable energy. If all of the central region’s gas power came from the current most efficient plant (not actually possible), we’d save the equivalent of 96,000 MWh per year. Just one of the proposed renewable projects in the region (Little Mount Susitna Wind)  would make more than four times that much energy.

If we replace a large amount of gas power with renewable energy, the remaining gas power will likely become less efficient. Gas plants run less efficiently at lower loads, so as we use them less, the fuel per MWh goes up for that remaining gas power, even as we save a ton of fuel overall. This makes the coordination of how we dispatch the plants especially important. For example, in a future scenario where we need half as much power from those three central gas plants, running them all at half the load will be much more wasteful than turning one or more of them off, and power pooling and economic dispatch will matter more than it does now.

Share

Subscribe now

Analysis details

All data on power plants is from public RCA filings, EIA data, and EPA filings. Gas prices are from the state’s Cook Inlet Prevailing Value numbers. I looked at central region gas plants only, in order to ignore the impact of transmission constraints. Some of this power is used outside the region, largely by GVEA. To calculate savings I looked simply at gas use per MWh and the prevailing Cook Inlet gas price for each year. Actual gas savings were probably less, since Chugach (and previously ML and P) gets a substantial amount of gas from its own Beluga River gas field at a lower price. Operation and Maintenance costs are ignored. All money values are nominal values.

Subscribe now

Currently, around 2,200 people on the Railbelt grid have solar panels on their houses. The total capacity (power generation potential) of these systems on Railbelt homes and businesses has quadrupled in the last five years, while remaining much lower than in the US as a whole. Across the US, 2.4% of residential customers have net metered solar systems (calculated from 2023 EIA net metering data), compared to just under 1% on the Railbelt.

These home solar systems work through monthly "net metering." Some electricity produced by the panels is immediately consumed by the house. When the house is producing more than it consumes, that electricity is fed back into the grid. When the house is producing less than it consumes, it gets energy from the grid. At the end of the month, those kilowatt hours in both directions are added up, and the customer is billed for the "net" amount they consumed. Some proposed bills in the legislature (including the RPS, a potential change to the community solar bill, and possibly others) have talked about changing that to annual net metering -- basically the same thing, but adding up once a year at the end of March. So I was curious to see what that might do.

The Railbelt has high electricity prices, and similar solar install costs to other states, which should make net metering a pretty good deal. But we also have less sun than a lot of places, and a lot less electricity use when it's sunny (not a lot of air conditioning, which dominates electricity use in warmer places). So most people with solar panels overproduce in the summer, and that overproduction gets credited at the lower “buy-back” rate (technically the Small Facility Purchased Power Rate) of 5.5 to 15 cents/KWh, rather than the higher retail rate of 18 to 25 cents/KWh. Additionally, HEA customers get charged for a minimum of 150KWh each month, regardless of use, as a "system delivery charge."

So I took the average monthly use for a residential customer at each utility, the average monthly solar production for a representative city in that utility's area, current electric rates, and did a little math to figure out what a solar system would do to that average home's bills -- now, and if net metering became annual.

Solar, right now, is the best deal for GVEA customers. They have high power consumption (just behind MEA), the highest power costs, and the highest buy-back rate. It’s worst for HEA customers, who have high rates, but the lowest consumption, and a “system delivery charge” that kicks in when monthly consumption drops below 150KWh. Interestingly, HEA customers are actually the most likely to have solar panels. Maybe other factors (permitting? environmental motivations?) are hindering or driving adoption in different areas. The smallest solar systems give the greatest savings per KW for a typical customer, but the average system size is significantly larger. It’s possible that people with solar use significantly more power than average, that the investment is still attractive at larger sizes and people put in the largest system they can afford, or that sizing is driven by other factors (like reaching net zero for a household).

Making net metering annual would reduce yearly bills almost $300 in the best case, and not much at all in others. The biggest change would be in Chugach territory, where current buy-back prices are lowest (because of the gas field Chugach owns). It would have the least impact on the HEA grid, where the system delivery charge would cancel out most of the gains.

Annual net metering would improve economics the most for customers installing solar systems with annual output matching annual consumption, which is around 6-7KW for an average user.

Would it make a difference? Anecdotally, people seem to care a lot about the payback economics of their solar panels (HEA got a lot of complaints when the system delivery charge hit net metering customers in 2020), but I’m not sure how much people are gaming out the system in detail, vs. installing what they can afford to hedge their bets against rising utility prices. Electricity prices are already set to rise based on inflation and interest rates, and are likely to rise substantially more with the impacts of the Cook Inlet gas crisis.

Data geeks can check out the data and assumptions

behind this on my Google sheet.

Subscribe now