Utilities Forecasting

This contract, which was originally intended to be for 3 months, eventually extended to nearly 4 years. A major electricity supplier needed an algorithm for forecasting electricity consumption by Commercial and Industrial customers. I designed and developed an original forecasting algorithm, which used a combination of analysis of past history, data smoothing techniques, temperature compensation, holiday and special event compensation and a self-optimiser.

I coded these in VBA to drive a spreadsheet test-harness. The algorithm was then converted to VB6 in a database and is now the core consumption forecasting tool used by the company.

The cost saving achieved by the improvement in forecasting is in excess of £1million per year.

As part of the same contract I developed algorithms for:

• Forecasting the consumption of their portfolio over the Christmas and New Year period.
• Forecasting the Triads, the peaks of electricity consumption, to enable customers to minimise connection charges.
• Estimating the cost that each customer causes as a result of the difficulty of forecasting their consumption profile.
• Estimating the costs associated with Triad avoidance for their largest industrial customer.

home