Scottish Government

4 Electricity Demand

Electricity cannot, at present, be economically stored in large volumes and has therefore to be generated as it is required. Short-term demand forecasts allow scheduling of dispatchable power stations on a minute-by-minute basis. The forecasts are based on records of historic load demands and trends (including transmission and distribution losses), weather forecasts, timings of major events (including television programmes), scheduled requirements at individual generating stations (for example maintenance activities), regional imports and exports through interconnectors and any load-control strategies that may be in place. Accurate long-term estimates of both power and energy requirements are crucial to effective power system planning and operation.

4.1 Scottish System Demand

There is a continuous increase in the demand for electricity, although in the UK and many industrialised countries, the rate of increase has declined in recent years (Weedy & Cory, 1998). The seven year transmission statements of Scottish Power ( SP 2001 - SP 2004) and of Scottish and Southern Energy ( SSE 2002 - SP 2004) list historical and predicted system demands, both in terms of system maximum demand (in MW) and of energy demand (in TWh). These are shown graphically in Figure 4.1.

Figure 4.1 System demand in the Scottish Power and Scottish and Southern Energy areas.
(a) System maximum demand; (b) System energy demand.

The apparent decrease of SP's energy demand is attributed to the increased usage of CHP plants, which locally reduce demand levels and, consequently, the usage of the transmission system. Nevertheless, it is anticipated by SP and SSE that the overall customer demand will rise in the future as indicated by the dashed lines. SSE specifies an underlying growth pattern between 0.5% and 4% per annum. For the development of a scenario for the year 2020, an annual growth rate of one-percent was assumed. This led to increases in energy consumption in the SP area from approximately 24 TWh in year 2003-2004 to 28.4 TWh in year 2020-2021. In the SSE area, the corresponding increase was from approximately 8.4 TWh to 10 TWh. Measures to reduce demand were beyond the scope of the study.

By definition, the system maximum-demand occurs once per year. In 2003/2004 in the SP area the maximum-demand of 4,227 MW was recorded in the evening of 20th December 2003. In the SSE area the corresponding figure was 1,694 MW on 28th January 2004. The minimum demand for both areas is of the order of 36% of peak demand. The load factor in 2003-2004 was 64.2% for SP and 56.8% for SSE.

Figure 4.2 Annual load profiles for Scottish Power and Scottish and Southern Energy.
Weekly minima and maxima are shown; the 2001/2002 SSE values were predicted.

Figure 4.3 Daily load profiles for SP (top four curves) and SSE (lower four curves).

Each figure shows profiles corresponding to days of maximum demand (weekday), average winter demand, average summer demand and minimum demand (weekend).

Demand varies seasonally. In Scotland, the profile is highest in winter due to heating and lighting requirements. Figure 4.2 shows weekly minimum and maximum values within the areas of the two companies. The SSE values for 2001 and 2002 were predicted using the technique described below.

The daily demand profile also varies across the seasons as indicated in Figure 4.3. For each year, profiles are displayed for the days of maximum and minimum demand as well as for typical winter and summer days.

4.2 Demand Modelling

A grid supply point ( GSP) is an electrical sub-station which connects a distribution system to a transmission system. It was assumed that all of the renewable-energy generators considered in the context of the study are connected directly to GSPs, and not to some part of the distribution system as embedded generation. The GSPs conveniently provide a finite number of nodes for the assessment of both the geographically-specific demands for electricity and the renewable energy supplied within specific areas.

Time-series of load-demands at each GSP are not available in the public domain. SP publishes measured 'maximum transformer loadings' (in MVA) and power factors for its GSPs, along with predicted values for future years. As there is embedded generation within certain parts of the distribution network, the figures cannot be strictly interpreted as demand. SSE publishes winter loads at each GSP which are useful for power-flow applications. In order to take load diversity into account, these '100% demands' at each GSP are not the measured peak demand but a scaled value. If all of the grid supply-point demand figures are added up, the sum will then just reach or slightly exceed the system maximum demand in the area.

The demand modelling within the study would ideally have used time-series of hourly demand for every GSP in Scotland for the years 2001 through 2003. However, this amount of data would have been excessively large for convenient handling and the electricity utilities are understandably reluctant to release it. It would also have needed exhaustive scanning for errors. Comparison of known individual GSP demand profiles suggested that time-series data from one GSP cannot be applied directly to another GSP if the only correlating information is the maximum demand. The method used in the study was therefore to appropriately scale the overall system demand pattern to each individual GSP. For this purpose, half-hourly system demand data was obtained from SP and SSE. The SP data corresponds to the period 1st April 2002 through 31st March 2004 and the SSE data to the period 1st January through 31st December 2003.

For the 'missing' SSE data corresponding to the period 1st January 2001 through 31st December 2002, the hourly demand was estimated as follows:

• Monthly demand patterns: Hourly averages were calculated from the 2003 data for each calendar month for each power company, separately for weekdays and for weekends (including bank holidays), giving a total of two sets of twenty four patterns.
• Daily minimum and maximum: The weekly demand minima and maxima published annually by SP were interpolated to provide daily values by assuming that the minima occurred on Sundays and the maxima in the middle of the weeks. The SSE values for 2001 and 2002 were based on the calculated ratio of the SP and SSE demands for 2003.
• Daily demand curves: The monthly demand patterns were assumed to represent the 15th day of each month. For other days, weighted average curves were calculated (separately for weekdays and weekends). The curves were time-shifted so that on weekdays the daily maximum was reached once and on weekends or holidays the minimum was reached.

As a quality check, the daily demand curve predictions for SP and SSE were calculated for the complete three year period 2001 through 2003. This allowed comparison with measured values where they were available. Table 4.1 lists the total energy consumption in the SP area for two complete business years. The predicted values match well both with the published and the measured figures.

Table 4.1 Model results for the annual energy consumption in the SP area.

As a further comparison, Figure 4.4 shows measured and predicted demand patterns for four-day, Friday through Monday periods during the winter and summer of 2003.

Figure 4.4 Measured and predicted hourly system demand for SP and SSE.
(a) Winter: Friday 17th to Monday 20th January 2003;
(b) Summer: Friday 18th July to Monday 21st July 2003.

The demand values finally used in the study were scaled up from the hourly values in relation to the anticipated load growth corresponding to the year 2020, assuming an annual growth in energy demand of one-percent. This process yielded scaling factors of 1.208, 1.196 and 1.184 respectively for the load data from 2001, 2002 and 2003. Scaling the time-series to 2020 predicted a peak power demand of 7.29 GW and an average energy demand of 41 TWh.