$12B of US ratepayers' money wasted on a modeling mistake in PJM
newsletter.semianalysis.com
newsletter.semianalysis.com
And ISOs/RTOs/TSOs operate in complex political environments and have to ensure reliability in adverserial markets. They are often constraint in what they can/can't do.
What I will say is that their modelling should absolutely be open and transparent, and typically it's not. That would make necessary discussions around modelling assumptions, potential improvements and political constraints much much easier and more fruitful.
[1] That said, my intuition is that current outages are probably not due to supply insufficiency, but due to transmission system failures.
https://inthesetimes.com/article/the-excel-error-heard-round...
It also reminds me of the flaws in the London epidemiological model about how to respond to covid.
It just may be that these things should be reviewed a little closer by people who are obsessive about correctness
No amount of review is enough. As anyone who's worked in quant science will attest, a model that hasn't repeatedly made correct out-of-sample predictions is most likely worthless, as it's far far easier to overfit a model to data than to predict the future.
That one was fun. I got into an argument with a scientist who was (IIRC) working with or for the specific professor that made that mess, and he blamed the entire software engineering/computer science industry for making C++ too hard to use, and said it should never have been released if mistakes like this could be made. And then accused me of gatekeeping when I said perhaps they should get an expert to look over this stuff before submitting results to governments to form a policy basis.
The attitude seemed to be "I am a scientist and am clearly very smart, and therefore anything you do must be trivial in comparison. If I can't pick up in seconds what takes you years to learn and more years to perfect, you've clearly done it all wrong and it's your fault."
You know what I mean?
The wikipedia writeup is a lot kinder to it than anyone was at the time, it was generally considered poorly put together and full of errors that made reproducibility questionable if not impossible.
Please note I'm not a covid conspiracist and I don't have a strong anti-lockdown streak, I got vaccinated and stayed indoors when asked to. I do have a general interest in government policy being based on sound information though.
That was not clear in the article.
Whilst these sorts of analyses are informative, they lack answers to the who profits? question.
The intention of the payments is to increase revenue for that kind of power generation capability to encourage more such plants be constructed.
The argument in the article is dubious to me. Of course the higher price isn’t leading to more generation today, that’s not the point, the point is to reward developers that build and built capacity CA needs in winter. The disagreement then becomes which model is correct about how much capacity is actually needed.. but the fact that a tiny move in demand moves the price so substantially seems to me to undermine the entire premise of the blog post, clearly supply is severely constrained?
I’m not a quant, and I’ve worked energy trading desks long enough to know there is a lot I don’t understand.. but I don’t see how separating auctions by plant age does anything other than move numbers around while keeping the total bill the same. Plants still need the same lifetime revenue to make investment decisions pencil out; whether you front-load payments or spread them evenly, the total in current value needs to be the same.
So it can matter how we distribute that revenue as to whether or not the business responds in the desired way, eg, actually investing in new capacity by linking payments directly to new capacity.
2. If no new or existing market participants are willing to expand capacity, that would indicate that existing prices aren’t worth it for them to do so. Companies are willing to spend far more than the cost of building a new power plant for things like AI - and they’re often even building power generation on top of the AI build outs. So it has to be more than “It’s expensive”.
I know how I feel about this - I much prefer price signals here; it allows any developer that can meet the spec to build energy production - distributed decision making over centralized decree - and it allows the other side of the equation, consumers, to decide they'd rather not pay for this generation: Curtail demand rather than increase supply.
Like - maybe I don't want to pay energy prices this high, I can choose to insulate my house better or get a heat pump and thus reduce my electrical bill. The price signal lets everyone in the market choose how to act, rather than a central authority declaring we must build, say, more gas turbines and share the cost of that across rate payers.
But, there are people much, much smarter than I that completely disagree with this position. In the end I think the answer is how you feel about human nature, the capability and shortfalls of markets and the complexity of deciding how societies' resources should be allocated.
To have a market, some organization needs to administer it. PJM operates transmission and facilitates a regional wholesale electric market.
https://en.wikipedia.org/wiki/PJM_Interconnection
https://en.wikipedia.org/wiki/Regional_transmission_organiza...
That could have been prevented by simply not permitting connections in certain parts of the network. Yes, theoretically it could also have been prevented by forcing generators to directly bear re-dispatch costs which would have led developers to build a different set of generating assets. This would have been a different set of market mechanisms.
However, if generation decisions had been made centrally, they would have been made by the same system operators, regulators, and governments that mandated the construction of the current system and they would have therefore built the same system more or less. Do you really think that if e.g. Germany or the GB network had been centrally planned by their respective government mandated regulators and system operators, they wouldn't have put all those windfarms exactly where the market-led model we actually have has put them?
You don't seem to understand what's driving the so-called re-dispatch costs in Germany so let me explain: when there's power consumption in the South, it is paid at the market price, when the "supplier" is located in the North (typically wind) and there's not enough network capacity to transfer the power to the consumer, the transport authority sends an order to the supplier not to produce the electricity, and instead pays a thermal power plant in the South (at a higher price than market price, by definition) to provide it instead.
That's not a problem of physical realities driving constraints, that's the problem of a market that uses a very simplified model that simply can't deal with the complexity of the underlying phenomenon (we don't even have enough computing power to accurately model the electricity transport at national scale given how challenging the equations are, and even the lineralized version that are being used in practice are already very compute-intensive, it's entirely futile to expect to build a market that can efficiently reflect them).
The issue is that the physical reality of the network and the location of generators and loads is fixed at time of [re-]dispatch. Regardless of how you reconfigure the dispatch mechanism, there is energy that cannot be moved through the network from the wind in the North to the load centres in the South. The only way to avoid that physical reality is to either massively increase the N-S transmission capacity or to put generation capacity closer to load.
Once these generators have been placed where they are and pending extremely challenging network expansion being realised, it will always be the case that some party or set of parties bears the cost of the constraint.
If you moved the ownership and control of the transmission network and all generators into a single operator that was responsible for optimising its operation in near-real-time and that bore all the costs within a single accounting perimeter, it would do more or less what the the existing market based mechanism does: turn off the wind generators and instruct the thermal plants in the South to run instead.
Yes, it's true that in that case there would not be a "market price" paid to wind turbines that actually cannot run, nor would there be a higher re-dispatch price paid to the thermal generator. However the total system cost would still be the same. Even though the wind turbines would "lose" money because they can't run and the thermal plant wouldn't receive its market price for relieving the congestion.
Of course you could argue that this efficient single operator simply wouldn't have built the wind assets in the first place in a location where they often couldn't run but that assumes that the same policy makers who created the current outcome wouldn't have engineered the exact same outcome through the direct control they presumably would have had of the integrated operator. I will note that many American utilities actually are vertically integrated like this (banned in the EU under the Third Energy Package) and they build in the exact same pattern and bear the constraint costs internally.
But it wouldn't count a profit in the North and build yet another pointless wind turbine in the North. Whereas the current system, because it pays the Northern producer the market rate, for electricity that isn't needed, shape the grid in a nonsensical way! That's the core of the criticism.
> Of course you could argue that this efficient single operator simply wouldn't have built the wind assets in the first place in a location where they often couldn't run but that assumes that the same policy makers who created the current outcome wouldn't have engineered the exact same outcome through the direct control they presumably would have had of the integrated operator.
There's literally no reason to assume they would have done the same choice given they would have responded to completely different incentives (minimizing the total cost, instead of individual actors maximizing their profit under the artificial price scheme).
They did so for political and ideological reasons that I think are pretty clear to anyone who follows European politics.
Had those same political stakeholders had more direct executive control of a single national electricity system, they would have done the same thing and for the same reasons.
Yes, of course a vertically integrated system optimising for efficiency would never have made these choices but the same political decision makers who shut down Germany's nuclear fleet would never have permitted such an integrated system to operate in such an efficient way if it meant not building wind assets.
We can see the impact of a different political direction in the French electricity system, which despite being structurally similar to the decision making of the German system is in practice much more centrally directed and which despite absurd attempts to sabotage it from within France and within the EC continues to operate in a much more optimal way.
It doesn't do this through formal centralisation and integration but by the careful design of grid codes and subsidy regimes to avoid these absurdities.
The “copper plate” rule appeared in 1998, long before the phaseout of nuclear and the move to wind power. Your argument completely confuses cause and effect: the reason why wind was pursued was because the market model made it look like it was the most economically efficient.
Thinking that someone made a plot to favor wind against more than a decade in advance by designing the market this way is laughable.
And by the way, I'm French and your point about my country couldn't be more inaccurate: the market rules are equally absurd, if not worse, the only reason the system works the way it does is that EDF is big enough (both in terms of electricity production and economic power) to be able to perform the central damage control. It severely affects their ability to make profits though (which is why the company was pulled back from the stock exchange).
And of course it's a market design issue, but as I said above, the problem is that you'll always face market design issues because the market designers face an impossible problem: you want to use a simple enough model and you're modeling an insanely complex problem for which any discrepancy between the model and reality will result in a catastrophic failure unless you have a central authority there to patch the holes.
At the end of the day, it's always going to be a centrally-managed system (because you cannot afford blackouts) but with the central planner merely swallowing the cost of the economic inefficiency rather than preventing it.
Suppose a new generator wants to know which is the best node to interconnect to. This should be obvious from nodal pricing, as you would simply check the node that has the highest INC redispatch/lowest congestion costs (the nodes in the south of Germany in your example). Similarly, the worst node to connect to will be the one with the highest DEC redispatch/highest congestion costs (the North). Instead, in the current german system, I would think you would need to do some contrived thing of figuring out which generators are getting the highest uplift payments, but this would be very imperfect as you would not know how sensitive a different node would be to the system redispatch (not to mention, this is assuming this information is even readily published).
There’s really no excuse to not adopt nodal pricing over whatever exists for a centrally planned grid operator. I can only really imagine it’s a political issue.
It's an adversarial setting, economic actors are incentivized to find any loophole and exploit them, so even if it doesn't need to be perfect it needs to be indistinguishable from perfection from the PoV of the market actors, otherwise the defects will be weaponized.
This is a huge "it depends", because we are then comparing a partly-political system with a fully-political system. The privatized model builds at the optimal locations for getting paid under the prevailing rules; the state one is much more prone to getting distracted by lobby groups. See the German over-dependence on coal and the UK conservative ban on onshore wind farms.
I will be interested to see how the deployment of affordable battery storage changes the economics. I've also seen a striking presentation about how much renewables the UK plans to deploy (roughly double!) by 2030. More on the strategy: https://www.gov.uk/government/publications/clean-power-2030-...
"Clean Power means that by 2030, Great Britain will generate enough clean power to meet our total annual electricity demand, backed up by unabated gas supply to be used only when essential."
(note details of wording)
This is anti-scientific woo. Both theoretically and empirically market-based electricity systems are much more efficient, that's why even China is adopting one: https://www.enerdata.net/publications/daily-energy-news/chin...
This is hilarious coming from someone defending a concept that assumes neither thermodynamics nor Maxwell's laws exist.
> Both theoretically
When making preposterous hypothesis you can prove literally anything to be efficient.
> and empirically
Ah yes, like the PJM and the European market are “empirically efficient”.