The Impact of COVID-19 on the Electricity Network

Over the past couple of weeks, many of us have seen significant changes to our daily routines. The implications of these are far reaching, but my niche interest is in how the electricity network has been affected.

In the UK, we benefit from a relatively secure power system. Last August, parts of the nation experienced a blackout for the first time in recent history. Politicians, journalists, and the general public were eager for someone to blame for this (for the record, the fault was caused by a lighting strike, but if you’re desperate to point a finger it should be at the two power stations that failed). However, it also highlighted how we take for granted our secure supply of electricity, which is rare in many parts of the world.

A stable system requires there to be a constant balance of the supply and demand of electricity – in other words, at all times there has to be approximately the same amount of power going into and out of the grid. Given that many power stations take time to power on or off, this is a complex task requiring a mix of accurate forecasting and contingency planning. For 24 hours a day, several National Grid employees sit at control desks constantly monitoring the state of the system (when I visited two years ago, there were at least three people split between two locations). Therefore, it is not surprising that the lights have stayed on over the past week, but some things have changed.

Electricity use is a footprint of human behaviour, and so anytime there is a significant change in behaviour, we can expect a change in electricity use. Even without social distancing, the electricity demand changes throughout the day, and between days. Influencing factors include: the weather, the school calendar, and the TV schedule. The largest single uses of electricity in the UK are heating and lighting, so the largest energy demands are seen in the winter (when it is cold and dark). 

The graph below shows the demand profile for a typical week in March, with last week’s demand overlaid (please note the false origin). The ‘typical’ data was taken from the last 8 years, and the shaded area covers a 80% confidence interval. It is clear that overall energy demand has been lowered as a result of the lockdown; it is particularly interesting to look at the difference between Monday and Tuesday (when we moved from ‘work at home’ to ‘lock down’).

GB power demand last week compared to an average March week. Data source: [1]

This change can be further understood by breaking down the national electricity use into three categories: domestic, commercial, and industrial. Over simplifying slightly, domestic is what we use in our homes, commercial is used by shops/offices, and industrial is used in manufacturing. Domestic demand is the smallest of these, but is responsible for the evening peak, and therefore plays a disproportionately large role when it comes to maintaining the system.

It is likely that commercial and industrial load has fallen, due to shops and businesses being closed. Since the 23rd, Transport for London has been running a reduced tube and rail service, and this will also account for some of the drop. However, working and spending more time at home will increase domestic electricity demand. Octopus Energy have published a report in which they claim that their average customer increased their electricity demand by 4% during the first week of working from home. It should be noted that these consumers will include a number of retired or self employed people, who will not have seen as large a behaviour shift. This means that areas with a large number of office workers are likely to see a larger increase in network loading, which might necessitate upgrades – it will probably be months before we can know this for certain.

One of the benefits of a fall in demand, is that energy supply becomes a “buyers market”. Renewable energy such as solar and wind tend to be cheapest because they have “zero marginal cost” (basically, you don’t have to pay for fuel). Therefore, we can expect our use of high carbon fuels to drop. The graph below shows the percentage of demand that was met by “low carbon” fuels last week, and over a typical March. However, it should be noted that we have seen a general shift over the past few years to a lower carbon electricity mix, so even without the lock down we might have seen last week’s fuel fix outside of these bounds.

Low carbon fuel mix last week compared to an average March week. Data source: [1]

The laws of supply and demand say that a lower demand with the same supply should result in a lower cost. The graph below shows the average price paid by energy suppliers for electricity on each day. Note that the market operates in 30 minute periods, so there is price variation throughout the day which has not been shown. The price varies significantly day-to-day, but there does seem to have been a general reduction in cost.

The average price paid for electricity on the wholesale market. Data source: [2]

Those unfamiliar with wholesale prices might be surprised that average prices range from 1-3p/kWh, while consumers pay an average of 15p/kWh (most are currently on a flat rate). Infrastructure maintenance and profit margins account for a lot of this difference, however there are also costs arising from real-time balancing of the system’s supply and demand. Operator intervention is necessary because the electricity market mechanism will not necessarily arrive at a solution that can satisfy demand safely; an ex-colleague once explained real-time balancing to me as the process of engineers fixing the mess that the economists have made, but other opinions are available.

When a large percentage of demand is met by renewables the system frequency is more volatile, and so balancing can become more difficult (and expensive). If you are interested, National Grid have recently published this article describing the technical challenges they are facing in more detail. Additionally, when there is more uncertainty in demand (say, because the nation has just gone into lockdown) there is a higher chance that expensive reserve power will need to be utilised. Therefore, my expectation was the daily balancing costs would have increased over the past couple of weeks. However, the daily balancing costs (shown below) have not changed significantly.

The daily cost of balancing the system over per MWh of demand. Data source: [3]

Digging into the costs (which comprises a range of different balancing mechanisms) reveals that transmission constraint resolution heavily dominates the total cost. This is when the suppliers chosen by the market could cause sections of the network to become overloaded (e.g. because there are too many suppliers in one part of the country). This problem is not significantly affected by the changes to demand, which explains why the balancing costs do not seem to have been affected. In my view, these costs demonstrate a fundamental failing of the electricity market mechanism. However, given that I initially wanted this post to be impartial and factual, I’ll save this rant for another day.

To summarise, the electricity system is currently operating at very low costs. Unfortunately, consumers (who will see an increase in their domestic energy consumption) will see none of these savings. Maybe, after all of this is over, we could think about creating a fairer system?

[2] Elexon Portal SSP/SBP/NIV
[3] National Grid ESO, Daily Balancing Cost (Balancing Services Use of System)