Friday, March 1, 2019

Value lessons from Ontario electricity statistics

Data - and lots of it.

While I'll try to prevent this post from sliding into an abyss of Ontario electricity statistics, I'll be citing provincial data as the basis for discussion about public understanding being restricted by the presentation of data from official sources, and present new views of generally unreported data that would benefit literacy in valuations of electricity sources - an area where reckless ignorance blooms again and again.
but enough about academics, let's dive in!

The old standard of valuing generation sources is the Levelized Cost of Electricity (LCOE ). I was particularly pleased in late 2015 when a report from Ontario's Auditor General included a figure (5) indicating the cost and quantity of energy sources for 2014. I was more pleased a few years later when I received figures from a freedom of information (FOI) request with the same information for years 2007-2015. While I think the data is terrific, when I wrote about it I cautioned on presenting LCOE, "stressing these calculations deserve a big asterisk and lengthy footnote on the impacts of things such as curtailment and capacity payments." 

I have now done the data work required to add that lengthy footnote on curtailment and capacity payments.

I am certainly not unique in hoping for superior valuation tools to LCOE: the U.S. Energy Information Administration (EIA) has developed a  Levelized Avoided Cost of Electricity (LACE) metric, and the International Energy Agency (IEA) a Value-Adjusted Levelized Cost of Electricity (VALCOE - see pg 41). LACE is intended to value the cost (of alternatives) avoided by the generation, with the intent LACE > LCOE would signal a good project. VALCOE attempts to recognize the different capabilities of sources in providing firm capacity and flexibility. The specifics are less important than the principles: not all generation is of equal worth to systems that are intended to minimize loss-of-load situations. Concrete examples of LUEC's limitations will help conceptualize the issues that have people looking for better valuations.

Before I discuss my data I will note one other frequently cited source of Unit Cost in my province (I regard LUEC and LCOE as interchangeable terms): the Ontario regulator. Their most recent explanation of regulated rates includes a table (3) indicating hydro at 6.2 cents per kilowatt-hour, nuclear at 7.7, wind at 15.9, gas at 18.8 and solar at 51.3 c/kWh. I'll show replacing the OEB list's lower cost supply with gas is likely to lower total costs.

I have collected hourly data for the transmission-connected (Tx) generators reported by Ontario's system operator, including imports, and I've estimated (hourly) distribution-connected generation (Dx), curtailed supply, and contracted cost, either by unit generated (or curtailed), or by capacity required to be available. My base union query in working the data has over 16.4 million records. I will not repeat the word "estimate" in this post but simply note this one time it may be applied to everything (my work and the numbers from the IESO and OEB I cite).  

Here is my summary of annual generation and costs from 2008-2018:

ALL GENERATION AND COSTS (tables in Google Sheets)

For the years I can compare to the FOI data noted earlier, I'm very comfortable with the differences. There are costs in the IESO data I haven't attempted to replicate (including the one-time payment for parts for the cancelled/relocated Oakville Generating Station, the one-time costs from a court awarding former generators a significant sum, the costs of buying out non-utility generator contracts), and there's a benefit of rebates in 20009 I have also ignored. My figures for 2016-2018 total to less than my analysis of total market value and global adjustment figures indicate they should: I can isolate the difference to the broad "IESO-OPA" pot of the global adjustment (but not Bruce Power), and that reporting also indicates I've underestimated the cost of wind and solar in recent years. I suspect other differences are in imports from Quebec as the system operator (IESO) contracted some of this at the rather dopey direction of the former government. Only the IESO (the system operator) can know more precise figures and perhaps external estimates will motivate them to publish an accounting. Data is never perfect, but the data I've created/collected is now good enough to move beyond the broad overview of generation and costs that has hidden important details on both generation and its costs.

The IESO frequently reports figures of interest primarily to their system of grid-connected generators (Tx) and wholesale consumers - and once a year provides an accounting of curtailed nuclear, wind and solar generation. Ontario ratepayers do pay for that curtailed supply. The IESO also provides infrequent low-detail summaries of distribution-connected generation (Dx). I've monitored that reporting, and contracted supply levels, in developing algorithms to estimate hourly generation from distributed generation and hourly curtailment by generator (including hydro-electric generating stations which the IESO does not report but the public generator, OPG, notes in its annual reporting).

The focus of the IESO's Tx generation in reporting makes sense in viewing total system procurement by output (including deemed output for curtailed supply), as the grid-connected generators to provide most supply:
BUT, it makes little sense from a cost analysis perspective, and no sense at all in evaluating changes to costs in the system over the past decade. The largely ignored elements of distributed generation and curtailment grew by about $2 billion from 2008-2016, and adding the oft-ignored capacity payments, and conservation spending, 70% of the increase in the cost of supply from 2008 to 2016 is attributable to non-Tx energy cost factors.
Graph (and previous one) can be viewed, with data in Google Sheet "Cost Components" tab
This displays the substantial share of costs people would be unaware of from most system operator (IESO) and regulator (OEB) data, but it mixes energy and non-energy costs. The distributed (Dx) supply costs demonstrated here are significant, largely driven by a growth in solar, which is hidden by the lack of reporting on distributed supply. To demonstrate the disconnect in common reporting I'll note the IESO's Year Energy Output by Fuel Type shows 0.6 TWh of solar in 2018 while its reporting on the global adjustment showed $1.5721 billion (14% of the total) attributed to solar. Solar in Ontario was contracted at high prices, but not nearly so high as the $2,620/MWh implied by the selected IESO's reporting available for 2018.

Better reporting on distribution-connected (Dx) supply is a necessity to understand the growth in costs apparent in the first data table shown above - which is one reason I estimate it.

Removing curtailment and capacity payments from calculations can present a fairer look at values per unit/MWh. Curtailment is removed as it is required due to the system having too much supply, and at this point I won't assign blame for that to any one generation type, and capacity costs are removed as it is the system that requires the capacity - not solely the one generation type.

While removing curtailment costs does drop the levelized cost of wind well below the rates determined earlier (including the one noted by the regulator in their price plan reporting), the big reduction in removing systemic curtailment and capacity costs is in gas' LUEC (and less meaningfully in biofuel's). 

The distinction between capacity payments and energy payments isn't always obvious. I've estimated costs for generators paid to be available, not simply by the generation they produce: these are capacity payments. Included are reliability-must-run, contingency support,and net revenue requirement contracts: the latter primarily apply to natural gas plants that entered service after 2001, and the former to Lennox (oil/gas) and coal-fired generators. Coal is no longer an issue in Ontario, but it's history is demonstrative on how capacity payments come to exist. Coal-fired generating stations were owned and operated entirely by the public generator, OPG. In 2008 natural gas was expensive and coal-fired supply sold into the market at a price - so high that OPG was required to return profits to consumers in 2009 (due to a market power mitigation price ceiling), but with the collapse in pricing and demand by the end of 2009 OPG was receiving out-of-market support payments to keep the plants deemed to be required for the system to function.

Generators paid for capacity have little opportunity to profit from selling into the IESO's system (which they refer to it as a "market"). Removing the capacity payment portion of their revenue to show only what is paid for energy (generally the cost of fuel), shows the incremental cost to ratepayers when output from the generators is supplied. Over the past decade the average cost has been $37/MWh. That can be considered the incremental cost of the next megawatt-hour of supply from natural gas generators. Including the fixed capacity cost the average LUEC for the past decade's gas output would be $108/MWh. The variance is large because the capacity factors of this set of facilities is small: always below 20% and as low as 6% in 2017. One way plants have been categorized historically is as baseload (very high capacity factors), peaking (infrequent use), and intermediate. Most of the gas facilities in this group were designed as intermediate generators. If they ran at 45% capacity factors the annual all-in cost, including capacity payments, would have been between 5 and 7 cents/kWh most years, which is only a third of what the OEB shows as the levelized cost of gas-fueled generation.

There is a capacity element to levelized cost analysis, so the cost should not be excluded from gas generators altogether. In Ontario other contracts and regulated rates recover all cost through a single price for capacity and energy. Contracted, or regulated, rates for nuclear and hydro cover both capacity and energy elements. An ignored challenge in Ontario, as elsewhere, is valuing generators where the capacity has a value that is low - meaning little of the capacity can be relied on to be available as demand requires. In Ontario the sources with little capacity value are wind and solar.

Why should the lightly utilized gas generators be allocated the full cost of providing the necessary firm capacity to operate a system?

I'll return to that question after exploring a related one: why would the cost of curtailing supply due to an overall excess be attributed only to the the type of generation curtailed?

Curtailment estimates are difficult, and attaching costs to the curtailment nearly as difficult. There are annual figures for curtailment of nuclear, wind (from the IESO), and public hydro generators (from OPG). I am using my estimated figures which differ, but I don't consider the differences significant.

While OPG will eventually be paid for curtailed hydro, it will be via a variance account that builds over time to be eventually recovered through future rates. I've referred to the eventual costs as "Deferred $Ms (Hydro)" in the graphic, but this amount is not included in any other accounting shown in this post.

While removing capacity and curtailment costs is useful, there's not an easy way to redistribute the costs back in. The IEA's value-adjusted levelized cost (VALCOE) recognizes there is less of a value to sources that lack capacity value, and flexibility, but I'm not convinced they have meaningful measurement for those factors. The EIA's LACE/LCOE may not provide easy equations either, but it does provide a simple model for valuing very low capacity value resources in Ontario. The cost of the average contracted megawatt-hour of wind output (~$128/MWh in Ontario) can be evaluated against the avoided supply expense which is the cost of fueling the province's gas generators with capacity contracts (~$37/MWh). While it may be difficult to hear wind costs $91/MWh more than the alternative, wind enthusiasts might note this figure would be reduced by pricing emissions of greenhouse gases and potential penalizing local pollutants - but they'll be disappointed when I note the alternative would cost nothing when less wind would simply result in less curtailment.

There are general lessons that should be heeded beyond Ontario. 

While the value of capacity and flexibility are difficult to specify they're existence can be seen through analyzing market pricing. Even in Ontario's dysfunctional market, which recovers very little of the total cost of generation, looking at a value factor reflecting the average value of a generator as a percentage of the average value of all generation (see Hirth) one might conclude wind's eratic production pattern is not well matched to supply, nuclear's baseload profile is indifferent to supply, as is much of its hydro (which runs as baseload, and peaks during lower demand with the spring freshet), solar is not actually much above average where the priciest periods of the day are morning and early evening ramps in demand, gas is more valuable as it's only used to meet demand, and one could postulate Ontario coal's low load-point (flexibility) and ramping depth would make it the most valuable output.

I do not mean to imply many in Ontario miss coal's negative impacts on air quality and emissions. 

I do want to emphasize fixed-price contracts for low-value electricity don't incent improvements in the flexibility and capacity value that need to exist in the system: they incent only energy, and energy is only one component of value.

We don't know how to easily evaluate complex elements of supply including capacity and flexibility values, and we struggle with valuing emissions - but we should, by now, know a fixed-price must-take contract for a low-value source can't be compared, on LUEC, to higher value alternatives.

No comments:

Post a Comment