Preparing for a resilient energy future: learning from Texas’ rolling blackouts

18 Feb 2021

Snow in Texas

Hitting the headlines this week has been the 4 million electricity customers across Texas undergoing “rolling blackouts”, a deliberate rationing of access to electricity while production capacity is much reduced. Around 45 GW of generation is said to have been out of action, of which 30 GW represents thermal plant using heat from nuclear power or from burning gas or coal to make steam to turn the turbines. These capacities can be compared with an expected peak electricity demand for an extreme winter event in Texas of 67 GW (Britain’s typical winter peak demand is around 60 GW).

Different parties with various agendas have sought to blame, in turn, wind turbines, batteries, unprecedented electrical heat demand and a broken electricity market. Yet the core cause appears to be simple: a significant proportion of the power generators on which the State relies – predominantly gas-fired – has not been sufficiently ‘weatherised’ to be able to keep operating through the severe cold conditions which have beset a large part of the US. However, it’s also unclear whether enough gas would have been available to supply the gas-fired generators had they been in service, with daily gas production in Texas having been reported to have fallen by between 30 and 50% compared with early February.

It is notable that only Texas – which operates an independent electricity system  – has seen the weather have a major impact on electricity supply despite the storms affecting many states across the Mid-west. The difference is that while, for example, Minnesota would normally expect these conditions on an annual basis and prepare accordingly, Texas has far more moderate climate expectations and despite federal regulators pushing for increased investment in reserve and weatherising following similar events in 2011, this does not appear to have happened. The State is only weakly interconnected with the neighbouring North American transmission grids – ostensibly for political reasons – further minimising their options in managing disturbances. It also lacks any market mechanisms directly targeted at ensuring any particular level of ‘generation adequacy’. However, the system operator in Texas, ERCOT, re-assured the North American Electric Reliability Corporation in its 2020/21 Winter Assessment that it “expects to have sufficient resources under scenarios that assume low wind output as well as extreme peak load conditions with an associated increase in unit outages and derates [sic] due to weather-related natural gas supply disruptions”.

Making sure you can meet demand

In Great Britain, we have an electricity Capacity Market designed to ensure a certain maximum ‘Loss of Load Expectation’ (LOLE). LOLE is the amount of time we might expect there to be insufficient generation available to meet total demand. In Britain, the target is for this to be, on average, no more than 3 hours per year. In practice, it’s unlikely that any of those hours would see the disconnection of any demand that didn’t have a contract to be disconnected. If possible, the Electricity System Operator (ESO) would increase imports (or reduce and reverse exports) on the interconnectors or reduce supply voltages first.

The authors of this blog can’t remember a time when simple inadequacy of generation margins – as distinct from large, almost simultaneous ‘loss of infeed’ fault events or network faults – caused interruptions to demand. The vast majority of demand interruptions in Britain are caused by distribution network faults. Moreover, a shortage of generation relative to demand is very unlikely to cause a whole system blackout (something that has never happened in Britain) – the ESO should be able to see a tightening of margins develop over time and take steps to manage a power shortage, such as the “rolling blackouts” applied in Texas.

Even if the timing of it has been notified in advance and it lasts no more than a few hours, a disconnection of electricity supply when it’s your only source of heat in very cold weather will have a major impact. Weather-related network faults will be even worse as they may take a few days to put right, especially if a repair team’s access is impaired.

‘Generation adequacy’ and system-wide metrics such as LOLE are used to guide investment in generation capacity and, increasingly, energy storage and flexible demand. The estimate of LOLE for a given background of generation and demand uses probabilistic methods to account for variations in demand and the availability of generation, affected by both forced outages and variations in weather. When aiming at quite a high long-term average security of supply, the calculation will be sensitive to the ‘tails’ of random distributions. These are quite difficult to know with confidence as they depend on data gathered over many years. Furthermore, some of the biggest impacts come from ‘common mode’ events that affect many generators simultaneously, such as problems with particular designs of generation plant or the gas supply, or caused by extreme weather events that are credible but for which there might have been few if any, observations.

A particular ‘adequacy’ or ‘capacity margin’ does not guarantee that all demand will always be met. It is not designed to, and is just one part of a jigsaw that also includes a need for clear signals to owners of generation to make it available at key times, and industry policies on how to use it, plus sufficient network capacity to allow power to flow from generators to loads. In both investment planning and operational timescales the probabilities of random disturbances are balanced against the costs of being able to prevent those disturbances from causing interruptions to demand. However, measures to ensure resilience also include facilities to contain the impact of a disturbance and to recover from it.

New challenges

A number of complex factors are making assessments of costs and benefits of different resilience measures more challenging:

The impacts of extreme weather events are difficult to predict but likely to be severe – we can expect to see more and longer periods of wet and windy weather, as well as water shortages from extended periods without rain. Floods are likely to increase in frequency and impact. And while these key events are perhaps going to be infrequent, as with the case of Texas they could be protracted and highly damaging to an extent which is perhaps disguised by simply assessing average annual values.

Looking back, we can point to events such as the 2006 and 2019 European heat waves which bore significant excess mortality as perhaps becoming part of the ‘new normal’ even in the UK. Similarly, the North American cold wave of 2013/4 which caused large-scale blackouts across the North-Eastern US and parts of Canada and which in turn led to significant investment by generators in the affected areas to protect against the cold. Some scientists are suggesting that global warming is disturbing the polar vortex in complex ways that contribute to spells of colder weather further south.

Be prepared

We are already building today much of the energy infrastructure that will be in place in 2050. We need to be sure that we are building into that system the reliability and resilience that will be needed then, recognising both the greater value electricity is likely to have to end consumers and the increasing ways in which it might be disrupted, rather than assuming that normal practice will be sufficient.

The economic rationale which seemingly led to Texas failing to ‘weatherise’ its generators might only appear irrational with the benefit of hindsight. In looking ahead to potential extreme events that are by nature difficult to predict, it could be argued that the precautionary principle prevails and it would be better to be prepared, to invest and not need it, than to not invest and be found wanting. We might also look to Texas as a warning against isolationism and leaving preparedness for relatively rare conditions entirely to market signals, demonstrating perhaps that a wholesale market for energy alone will not protect the consumer against infrequent extreme events, and that sharing of resources can significantly contribute to mutual energy resilience.

Graeme Hawker is a post-doctoral researcher at the University of Strathclyde and a contributor to the UKERC themes on Energy Infrastructure Transitions and Local and Regional Energy Systems.

Keith Bell holds the Scottish Power Chair in Smart Grids at the University of Strathclyde and is a co-Director of UKERC where he leads the theme on Energy Infrastructure Transitions.