Demand forecasting: multiple choice

Modelling a handful of future electricity demand scenarios is no longer sufficient. George Danner makes the case for simulation strategies that can model thousands of outcomes.

Electricity suppliers have seldom had much need for crystal balls. For decades, domestic electricity demand could be predicted with a high degree of certainty. Total household consumption would increase slowly but steadily, broadly mirroring the rate of growth of the general economy. There would be several small spikes in demand throughout the year – spells of cold weather, a rush for the kettle during half time of the FA Cup final, that kind of thing.

But the predictability of the past is no more. And just as the stable pattern of demand is changing, a raft of potentially game-changing green legislation is on the horizon. All of which makes the eternal challenge – matching supply to demand – harder than ever.

Understandably, in a sector that needs to forecast demand over the long periods of time associated with large capital projects, accurate demand predictions can provide a crucial advantage when it comes to devising strategy. Torus has just completed a project for a major electricity industry player, in which we helped build up a picture of what residential electricity use might look like in the short term (three years’ time) and the long term (ten years’ time).

Our tool of choice was cold, hard mathematics – not a crystal ball in sight. We calculated the likely impact of the following (both separately and in multiple, overlapping combinations): advances in renewable energy technology; the growing popularity of electric cars; the rising price of carbon-based fuels; the impact of present and predicted government policy on climate change; micro-combined heat and power systems; residential heat pumps; and the spread of solar panels. On the last point, for example, we calculated that the recent reduction in feed-in tariffs to 21p will result in four million British homes installing photovoltaic panels by the year 2020.

Such a simulation model is capable of considering thousands of plausible futures, even those that are unlikely, but which, if true, will have a meaningful impact. We produced models that map all the possible futures, so that the strategic decision-makers can spend their time thinking about action plans and contingencies, rather than fussing with the nuts and bolts of the data.

Intelligent utilities will embrace this new uncertainty and seek to exploit it for competitive advantage. They must do more than simply react to developments. They need tools to help detect even the possibility of periods of disruptive change so that they can prepare for them well in advance.

This can be done by embedding in their forecasting systems real-time feedback loops that can take a much broader view and avoid the danger of focusing on just one or two of the most likely long-term scenarios. They need to look to a future full of multiple eventualities, which nowadays can be done through simulation (see box).

Exploring multiple eventualities via simulation strategies is a lot less risky than going straight to a single trial in the real world. Many of the most progressive utility industry leaders are now using this technique to strengthen their strategic planning. While mathematical models alone cannot tell them what to do, they can equip senior managers with the knowledge and insight they need to make their most important decisions.

George Danner is a director at Torus Business Web

Simulation strategies

Using simulation, energy utilities can role-play thousands of plausible futures about what will happen to households’ demand for electricity over the next decade – and use the results to inform strategic actions that are both robust and relevant.

To succeed, a simulation strategy must provide the following:

· a simple visual representation of the results, and a clear user interface which keeps the complicated maths under the bonnet

· a formula which allows you to easily change the variables and see the long-term impact

·  a formula which can rapidly recompute the impact of changed variables

· an accurate representation of the company’s various possible futures, and a clear explanation of the combinations of conditions that will shape them

This article first appeared in Utility Week’s print edition of 6 April 2012.

Get Utility Week’s expert news and comment – unique and indispensible – direct to your desk. Sign up for a trial subscription here: http://bit.ly/zzxQxx