Tuesday 22 October 2013

Negative Electricity Prices: Bleg

A couple of years ago, a visitor to our department here at Canterbury told me how sometimes in Texas, windfarms are able to generate so much electricity that the price of electricity goes negative. This recent article in the Economist (HT: Tyler at Marginal Revolution), gives a similar example from Europe. In both cases, I find this puzzling, and so I am seeking enlightenment from those who know the physics of electricity generation better than I do. To explain why it is a puzzle, let’s consider some examples of the economics of negative prices.

First imagine a pure-exchange world (i.e. one where commodities just exist rather than being created, so that economic activity consists of trade and consumption, not production). If all commodities are desired by all consumers, then competitive markets will result in all prices being positive, with prices reflecting the relative desirability and abundance of each good. If, in contrast, one of the commodities is not only not enjoyed but would be positively disliked by all consumers, the extent that its price would reflect that dislike would depend on whether there was “free disposal”, meaning whether the owner of the commodity could costlessly avoid consuming it. If there is free disposal, the price of the commodity would be zero. Without free disposal, the competitive price would be negative, again reflecting the relative (lack of) desirability.

A related situation arises when a firm produces a main product and an associated by-product. For instance, consider a motel that produces accommodation services during the peak holiday season. To provide this service, it has to incur the capital cost of building motel units that exist during the peak period, and then, as a by-product, these units exist during the off-peak times. If off-peak demand is low, some of these units might well be consistently vacant during the off-peak times, even if the price fell to zero. There again is an implicit assumption of free disposal here. If, for some strange reason, there was a requirement that motel units be occupied at all times to prevent depreciation of the capital stock, one could easily imagine the off-peak price going negative; that is, it could be worthwhile to motel owners to pay people to stay in their units during the off-peak times in order to ensure they were available for renting out at positive prices during the peak period. In effect, the opportunity cost of maintaining the units during the off-peak time would be negative, which could be reflected in the price. Again, the key assumption allowing negative prices is no-free-disposal. It does not make sense to see sellers choosing to sell something at a negative price if they could simply dispose of the good or service for free.

So now consider electricity. It is a key attribute of thermal power plants (particularly those using coal as the fuel source), that it is cheaper to keep the plant running 24/7 then to shut it down and heat it back up every day. This is just an on-peak/off-peak problem. Even if one only wanted to generate power during the peak periods each day, it would be cheaper to keep the plant running than to shut it down at off-peak times, giving a negative opportunity cost of generating power during those times. In the examples given above from Texas and Europe, wind or solar generation was able to meet regular demand at off-peak times, but shutdown costs made it economic for thermal stations to keep producing, sending prices negative. But, as we have seen, negative prices require an assumption of no-free-disposal.

My question then is: What is the physical or political constraint that implies an absence of free disposal in the electricity market? Why is it not possible to run a plant spinning the turbines, but simply not connect the station to the grid? In the case of Texas, I have heard that it is a purely political constraint: in a heavily regulated market, it is not palatable to have stations burning coal and not then produce any electricity. Is that all there is, or is there something about the physics of electricity generation that makes it imperative to force power onto the grid when a station is up and running? 

17 comments:

  1. At least with wind turbines the constraint is physical, if they're spinning they need to be attached to a load or they will fail. One possibility is building a massive resistance - paddle blades in water or whatever - with associated land, opportunity, and maintenance costs, and direct the output into that. For thermal plant you could dump the steam to atmosphere before it gets to the turbine, but I think it's treated water and dumping that's expensive, so the constraint is also physical - it needs to be connected to a full load

    Or you could deal with it via the market as you've described. I suppose the market is better than every company building a grid-sized bar heater at their plants.

    ReplyDelete
  2. Thanks for that: I would presume that with thermal plants, you could build it so that the steam could either pass through a turbine or be diverted through a different route, much like water in a hydro reservoir can be spilled, but that would require the negative prices being anticipated at the time the plant was built.


    In the case of wind turbines, I would have thought it relatively easy to build a mechanism by which one could let the wind blow but not have the turbines spin, but the policy distortions in the market when it comes to renewable sources of generation are such that I imagine the windfarms are insulated from the negative prices.

    ReplyDelete
  3. In NZ electricity prices can go negative, but it is unusual. Different types of plant have different characteristics as you have identified. The primary constraints are whether there is sufficient demand and whether there is sufficient transmission and other ancillary services etc.


    In NZ (in a highly simplified sense) we have plant that must run, usually hydro that is subject to minimum river flow constraints. This however does not of itself require prices to be negative as almost all hydro plants have the ability to by-pass the plant. There is no value in putting water through a hydro plant and paying people to take it away, so first choice is store it, second choice spill but only to satisfy regulatory requirements. Even at minimum flow the hydro plant offer is unlikely to be zero or negative as the plant will have a positive cost from running, variable OPEX and market fees. In the NZ we have a must run dispatch auction as although the market can clear a negative price it cant solve for a negative offer. So the must run dispatch auction gives the "winner" the right to offer a zero.


    Wind is a bit different. It can be dispatched off, by effective allowing the blades to be feathered and then going to an effective open circuit situation. As such it does not have to run just because it is windy and at negative prices, again why would you!


    Thermal (coal, gas and overseas nuclear) can be a bit different, but again this can often be an economic not a physical limitation (although the engineers will deny this :) ). There are thermal cycles and costs associated with starting and stopping very large thermal plants. In the NZ market the respective generator needs to factor this into their offer prices. In addition certain fuel contracts such as take or pay gas, may limit flexibility. Having said that, a thermal fuel source is conceptually no different to a hydro reservoir and can be optimised given expected load, prices etc.


    Peaking gas turbines are designed for fast start and stop and have less thermal issues, but are also less fuel efficient.


    Geothermal is almost always run based load; again the primary restriction is thermal cycles and potential impacts on the steam field through time.


    So in a market it can be observed that different operators will structure their fuel and plant operations around their forecasts of demand and supply. As the demand and supply mix changes during the day, week, months and years, the effective operating profile of plant can change (generally except for geothermal) as participants seek to maximise some aspect (hopefully net cashflow) given overall market conditions, plant additions, withdrawals, maintenance, grid constraints etc etc.


    There can however be distortions or other incentives. Physical delivery type contracts can provide incentives to run even of supply could be sourced elsewhere. Note that this basically does not happen in NZ. In the USA with high levels of project finance being the norm, and often more physical type markets and contracts, plant often needs to run to get paid. If their contract price, less own operating costs and the "negative" price of being dispatched is still positive, that is a better outcome for them then not running.

    ReplyDelete
  4. I can't imagine that spot prices would go negative in NZ very often. Check out what a huge proportion of our electricity is generated by hydro (free disposal by using the spillway or sluice), wind (free disposal by blade feathering) and geothermal (free disposal by stopping injecting water into the wells): http://www.em6live.co.nz/Default.aspx

    ReplyDelete
  5. There is a must run auction that allows the winner/s to offer at zero. Geothermal injects used geothermal fluid (usually). To turn the plant down or off they just close the valves on the production wells. Approx 60% comes from hydro these days and spill is minimised (as it is free fuel, which of spilt is lost) unless affected by actual or expected floods or due to minimum flow requirements when demand is very low.

    ReplyDelete
  6. I had the impression that in addition to constraints at the generation end, there were also minimum loads that needed to be satisfied on the grid itself for stability. Is this correct?

    To get negative prices, you need assumptions about supply and demand. Some NZ users I thought had a pretty high short term elasticity of demand through inventory management (Fonterra/Methanex perhaps?), and in some cases it would be more efficient for the demand side to do make quick adjustments than the supply side.

    Pretty tough if demand side is dominated by residential users.

    ReplyDelete
  7. I had not heard of a minimum load requirement, just that generation needs to match load very closely. Maybe The Other Neil, can clarify.


    The general assumption on demand (in any context) needed to get negative equilibrium prices is that demand doesn't tend to infinity as price tends to zero, which isn't a big stretch. Yes, I think there are some examples of individual users with high short-term elasticity of demand buying directly at wholesale prices, but the effective SR elasticity is zero for the majority of users as they are on fixed-price plans from their retailers. The whole question is moot in NZ thought, because of the fraction of our generation capacity that is hydro.

    ReplyDelete
  8. Thanks Neil. This is very helpful. BTW, are you the Neil who provided me with wholesale price and quantity data back in 2010 when I was writing a critique of the Wolak report?

    ReplyDelete
  9. These comments have all been very helpful. What I take from this is that there are genuine physical constraints preventing free disposal in thermal generation plants, but that these can be exacerbated by economic distortions in fuel contracts and the like. It is never likely to be an issue in New Zealand with so much of our capacity being hydro; I wonder if we would see negative prices in North American and Europe if there wasn't as much craziness surrounding renewable generation (in the form of both subsidies and regulatory interventions). Negative prices don't of themselves imply a badly designed market, but they may be a symptom of the design.

    ReplyDelete
  10. Don't recall, but I may have! Old age is getting to me :)

    ReplyDelete
  11. I don't see why wind farm advocates would brag about negative prices. Negative prices imply you are producing something that nobody wants. After all they are willing to pay you to stop doing it.

    ReplyDelete
  12. There's actually quite a lot of spilling done in the lower South Island. The Manapouri (owned by Meridian) and Clutha (owned by Contact Energy) systems have very little storage capacity due to tight resource consent constraints about minimum and maximum lake levels. Transmission grid capacity constraints come into it as well, obviously they'd prefer to be producing with that water.

    ReplyDelete
  13. I'm not sure that the advocates for wind and solar power brag about negative prices. But I certainly think it is true that they tend to think of sun and wind as being free (as well as clean) by looking only at the direct costs and not at the effect they have at the margin on total costs (and carbon emissions) in the system as a whole. And of course, the crazy political environment in the U.S. and Europe means that the system-wide costs of renewables is not necessarily communicated to the owners through negative profits.

    ReplyDelete
  14. There is a minimum stable load required on an electricity system. I would love to explain why, but my background is economics not engineering :) It is something to do with requirement for resistive load and inductance (I think).


    The minimum load in NZ is christmas day, overnight or following days. Even at this time the system is fine. although it can have spinning reserve issues, depending on which plant is running and whether the HVDC link is highly loaded. This is also the time a negative or zero prices are likely to be observed, especially if hydro lakes are high and inflows are high too.

    ReplyDelete
  15. using Meridian data covering the period May 2011 to Sept 2013, 2,444GWh was spilled. This was 11.8% of their total production. However, 70% of this was due to high flow requirements. The balance is dominated by regulatory and transmission constraints about equally.

    ReplyDelete
  16. Interesting. What's your source for that data? Not questioning it, would just be handy for future reference :)

    ReplyDelete
  17. When electrical demand and supply aren't closely matched then you get instability, shortened infrastructure life and quality problems. If there is too much demand then the frequency and voltage both drop, it's analogous to putting your car in too high a gear and watching the RPM drop. If there is too little demand then the frequency and voltage both rise, like putting your car in too low a gear. Grid management is done to err on the side of the former condition as it is cheaper and much easier to address; just generate more electricity. The latter condition is actually very difficult to control as operators can't easily increase demand.

    Electricity is meant to be delivered to the customer at 230V alternating at 50Hz, I can't remember the tolerances off the top of my head but they're quite tight. However the grid isn't a perfect conductor. Supply and demand aren't distributed homogenously across the country. So you can easily get the situation where there is too little demand in South Otago and too much in Auckland.

    ReplyDelete