Forecasting electricity demand in the time of Covid

The Low Carbon Contracts Company (LCCC) was forced to rapidly re-evaluate its approach to forecasting electricity demand during the lockdown. Forecasting analyst, Andreas Zimmermann, writes for Utility Week about how LCCC responded and the trends he is predicting going into 2021.

With the Covid-19 pandemic taking the world by surprise, sudden and unprecedented changes came into our lives.

Most countries entered a strict lockdown, with non-essential businesses closing their doors, in line with the government guidance. To minimise the spread of the virus, people had to adapt to working from home, when possible, or even being furloughed or made redundant.

With travelling and social interactions severely limited, major sectors, including transportation, manufacturing and commercial, faced disruptions, leading to a significant drop in electricity demand. While the total daily eligible demand saw variances of up to 20 per cent compared to normal levels, there was also a shift in demand patterns throughout the day, given by the change in people’s behaviour.

Besides the absolute reduction in demand over an entire day, another significant change was in the demand pattern. With people working predominantly from home, the electricity demand in office buildings was reduced throughout the workday, with the largest demand gap seen between 8 am and 3 pm. Towards the end of the workday, the demand was picking up, narrowing the gap between the actual and the forecast and accentuating the “Duck shape”. The lowest demand gap was recorded over-night, typically between 11 pm and 5 am.

As lockdown was eased, some ongoing effects of the pandemic were reflected in the demand outturn value and shape, however this time with a lower magnitude. Compared to April and May, when the percentage demand variance was in the double digits, the period between June and September brought a gradual recovery, up to around 2-3 per cent lower than normal levels.

This recovery, driven by a sustained reduction in the number of infection cases, as well as series of economic stimuli implemented by the government, has exceeded expectations.

Demand forecasting in an unusual situation

Electricity is one of the most important commodities of the modern world, being at the foundation of our everyday lives. On one hand, electricity demand is the primary source of revenue for suppliers, thus producing an accurate and timely forecast is extremely important for budgeting and planning purposes. On the other hand, electricity demand is also used as the main charging mechanism for various levies used to subsidise renewable generation, driving the progress towards net zero by 2050.

The key role of the LCCC, as the Contracts for Difference (CfD) counterparty, is to ensure that the costs of CfD generation are successfully covered by the Interim Levy payments, collected from electricity suppliers. The Interim Levy Rate (ILR) is set a quarter in advance and is calculated on the basis of the forecast Eligible Demand, defined as the Gross Demand less the demand from Energy Intensive Industries (EII).

The Total Reserve Amount (TRA) is set so that, together with the ILR income, there is a 95 per cent probability of covering all the CfD payments in the quarterly obligation period. The relationship between Gross Demand and National Demand is illustrated in the diagram below.

At the end of March, as the pandemic was unfolding, it became clear that its effects would be felt throughout Q2 and perhaps beyond, significantly reducing ILR income. As the ILR for Q2 2020 had already been set at the end of Q4 2019, prior to Covid-19 appearing in the UK, we needed to decide whether an In-Period Adjustment (IPA) of the ILR was required. Given that any IPA requires us to give suppliers 30 days’ notice, we had to act quickly and estimate the demand reduction throughout the quarter, with very limited data and no previous similar circumstance.

Consequently, we immediately started an enhanced demand monitoring process, the first step of which consisted of a top-down analysis of demand, based on the sectoral split published in the Digest of UK Energy Statistics.

Considering the annual electricity demand profile, we estimated the weekly contribution to electricity demand from each individual sector. By applying demand adjustment factors for each sector, based on up-to-date relevant news articles and previous economic downturn in 2008/9, we estimated the likely demand reduction for the entire quarter.

Over the first couple of weeks in April, as more demand out-turn data became available, we were able to refine our initial estimate. To do so, we looked at the demand variance from different day types: weekdays, Saturdays and Sundays. It then became clear that there was a much larger variance during weekdays, compared to weekends, which in hindsight was obvious, as people were working from home and offices were closed.

Moreover, given that demand has a downward trend between March and June, we introduced a seasonality adjustment and arrived at our narrow range of 11.8-13 per cent demand reduction for Q2 2020, compared to the levels forecast in December 2019.

Given the general uncertainty, we decided to take 13 per cent as our central assumption for demand reduction, implemented in the weekly tracking for senior management.  Using an automated tool created internally for the purpose, we generated daily demand out-turn reports to check whether our assumptions were still valid and, if not, to enable us to make quick, informed decisions.

Reducing the impact on suppliers

From the start, we engaged with BEIS and Ofgem to highlight the need for an IPA, with the aim of supporting a joined-up and coordinated response from across government. In parallel, we ensured that commercially relevant information was communicated promptly and accurately to suppliers, while also understanding the likely impacts of Covid-19 on their cashflows.

Given the urgency of the situation, we increased the frequency of our cashflow projections from weekly to daily, as every extra settlement day would provide insightful information and would help in contouring the demand trend.

Upon realising there was a real chance of LCCC not collecting sufficient funds to cover the CfD generation cost until the end of Q2 2020, we had to act quickly and quantify the additional funds required, while also developing a funding solution which minimised the impact on suppliers.

The first step we took was to review the relevant regulations in detail, with the invaluable support from our legal team. Coming to the conclusion that the TRA can be used towards covering the payments, we still had a gap to fill for ensuring a 95 per cent probability of being able to cover CfD payments in Q2 – this was quantified at £43 million. In parallel, BEIS, with the rest of central government, was developing a coordinated public response aimed at ameliorating the impact on business, including suppliers.

While clearly necessary at the time, BEIS was very keen on avoiding the additional £43 million levy increase in order to protect suppliers, and thus offered LCCC an emergency loan of up to £100 million to avoid the need for an IPA and full drawdown of the TRA. After careful consideration, including whether this could be perceived as limiting LCCC’s carefully guarded independence, this was taken up by the LCCC board.

In the end, the LCCC board decided that in these extreme circumstances, the overriding concern had to be supporting suppliers and their customers through these extraordinarily difficult times.

Aftermath

April and May were a torrent of activity, with all this analysis, decisions on the course of action and the relevant regulations written and consulted on in record time. After this, the pressure on our staff started to ease off a bit, but only a bit. With the Q4 2020 forecasts needing to be produced by the end of June, we focused on incorporating lessons learnt through the previous quarter so as to balance the risk of residual low demand with the gradual opening up of the UK post-lockdown.

We settled on a 5 per cent reduction from normal demand, much smaller than the 11.65 per cent out-turn demand reduction in Q2 (just outside of our original 11.8 – 13 per cent range). Another significant challenge was establishing a central assumption for the demand recovery rate in the subsequent quarters. With Q3 2020 levy rate already set in March, just before the lockdown, there was no adjustment to the normal forecast demand, which increased the probability of an undercollection.

Our enhanced demand monitoring, along with top-down sectoral analysis extended over the following quarters, has brought confidence that we would be able to act quickly and make a timely in-period adjustment, no matter what the situation.

For the extended forecasts, we have assumed a gradual demand recovery, with 3 per cent reduction in Q1 2021, followed by a residual 2 per cent reduction for the following three quarters.

As for Q4 2020 demand outturn, the proof of the (Christmas) pudding will be in the eating. In addition to Covid-19, significant risks remain around the shape that Brexit will take and any consequent impact on electricity demand.

Suffice to say, our monitoring activities from the height of the lockdown are being maintained. We are also engaging with a range of stakeholders to gather views on possible demand outturns and refine our forecasts.

If, after reading this, you would like to get in touch to discuss further, our (virtual) door is always open – we look forward to hearing from you.