A Simple Reason Why Net Zero Is Impossible

I took a look at this issue at hourly resolution for the lower 48 states for 2022. My procedure was slightly different. I looked at the hourly generation from wind and solar, and the hourly demand. Starting from a solar only assumption, I calculated the hourly surpluses and deficits against demand that would feed from and to storage. I looked at two different storage scenarios: a “battery” or “pumped storage” storage system with a round trip efficiency of 75%, and a Power-to-Gas solution with a much lower round trip efficiency of 36%. The total volume of surpluses had to exceed the deficits to cover the round trip losses, and generation capacity was calculated to deliver the same volume in storage at the end of the year as we started with.

I then looked at the trade-off from cutting solar progressively, and adding in wind to replace it while maintaining the same constraint on storage that generation must be sufficient to replenish storage over the year to its starting level, finishing with an all wind, zero solar solution. Of course initial storage has to be sufficient to ensure that the amount in store never falls below zero over the year. The time of minimum storage is at the end of summer before demand tails off in the shoulder fall/autumn months.

The idea was to look at how the storage requirement varied with changing relative proportions of wind and solar for a wind and solar only grid. I also looked at the maximum total hourly production for each scenario to see what grid capacity would be needed to transmit it to demand and to or from storage. These trade-offs are summarised in this chart:

There is a surprisingly broad range of mix of wind and solar with a relatively constant storage capacity required of around 300TWh, but extremes of depending on one or the other are to be avoided, because storage requirements escalate sharply. That is slightly higher than your figure of 250TWh because I included no nuclear or other generation, but we are in the same ballpark. Here, storage capacity is defined as redeliverable storage after the round trip loss. So if you are using gas as an intermediate store, you have to boost the amount of gas in storage to cover the losses due to the inefficiency of gas generation, which would set the gas storage at ~500TWh if we assume a 60% generating efficiency. Perhaps surprisingly the redeliverable storage requirement differs little between the very different round trip efficiencies over most of the range of wind to solar ratios: it is the extra capacity for both generation and grid transmission required to cover the round trip losses that takes the strain. Grid capacity required tends to be lower at higher relative amounts of wind capacity because solar is much more peaky in output.

In reality we really need to start by looking at the most difficult years, with low capcaity factors for wind and/or solar generation. If we can’t keep the lights on for these we don’t have a viable system. The corollary of course is that there will inevitably be years of high renewables output, but whether it is worth storing the surplus against future extended Dunkelflaute some years ahead is likely to produce the answer no. So that means that the surplus would be curtailed at least once the storage is full. There will be ongoing self-discharge losses from any kind of storage to further dampen the economics of inter year storage.

Your paper makes many excellent points about why the simple copper plate (any generating surplus can fill any demand anywhere) analysis for sample years is inadequate, and it certainly makes the point that storage is infeasibly expensive. There are two ways out of this conundrum. One is to invest in overgeneration, and the other is to cheat by using backup generation which would inevitably be fossil fuelled. Another cheat method favoured by those playing with the numbers in green think tanks and consultancies is to assume that it’s OK to have unsatisfied demand imposed via “demand side response”. Here the assumptions tend to be hidden from public view: my guess is that they are assuming you can cut large swathes of heating demand for days on end in a lengthy winter cold snap without provoking riots at the next election or killing too many people.

The reality is that long before we reach the point at which renewables are close to numerically generating enough over a year to meet demand we will be seeing rising periods of surplus generation, starting mainly in the lowest demand hours or locally around solar peaks when air conditioning demand is not too high – a sunny day in May for example. The duration of these surpluses will increase as more generation is added, and the size of the surplus will increase in the lower demand hours. The result is that we will have a small number of hours with a very large surplus, tailing down to many hours with smaller surpluses. That of itself creates an economic problem, since it will not be economic to transmit and store very occasional large surpluses, so much of that output will have to be spilled/curtailed. Already curtailment starts to eat into renewables economics. Initially, it rises roughly quadratically as the number of hours and the size of surpluses increase. Later, as the marginal useful generation from each capacity increment drops because it is simply adding to surpluses at times of reasonable output, but doing very little to solve Dunkelflaute deficits, curtailment asymptotically approaches a very high percentage of incremental output. The effective marginal cost of useful renewables generation can become a large multiple of any simplistic levellised cost calculation.

However, there is a trade-off with reduced storage requirement as extra generation is added, and since storage is much more expensive than extra generation, the optimal solutions will have significant over-generation. There is no single real optimum, because of the unpredictable variations in demand and generation. The best that can be done is to head in roughly the right direction so that there is a broad range where costs do not vary too much. The exploration of this range of sensitivities needs to start with the difficult years that impose larger storage requirements and lower renewables output. It is necessary to look at averaged annualised costs for the various elements. That means allowing for asset lifetimes and using an annualised capital charge basis for the asset itself, plus maintenance and other input costs. That applies to grids, wind and solar generation and storage assets. Of course, once you start allowing backup generation into the system you will likely canibalise the whole renewables setup. But it’s probably best to compare with a non-renewables optimised grid as a separate exercise to evaluate the extra costs.

I may have time to explore the rudiments of the over-generation tradeoff in a further post before the discussion here tails off. An introduction to the behaviour can be seen in my simpler analysis of wind and solar on Thursday Island.

https://euanmearns.com/wind-and-solar-on-thursday-island/