POWER ALERT: AN INNOVATIVE SYSTEM TO CONTROL RESIDENTIAL LOADS UNDER PEAK CONDITIONS USING NATIONAL TV

摘要

The Western Cape Region of South Africa experienced severe power supply problems between February 2006 and August 2006. The reason for this was that Eskom who it the electricity supply utility, lost one of its Nuclear Supply Units. Forecasts predicted a winter shortfall of 300 MW for the region out of an average supply of about 4000 MW. It was clear that if something could not be done to reduce the demand, power outages would be experienced. The main concern was the evening peak between 18:00 and 21:00 due to the high evening load. Eskom sent out a request for proposals to energy solution providers and ESCOs to come up with solutions to solve the crisis. The problem was that there was not enough time to implement solutions like residential hot water control systems in the 625 000 households of the Western Cape. The total residential load was around 1 200 MW. Energy Cybernetics, an energy solution provider, came up with the solution to use the National TV Broadcaster to inform the public on the status of the electricity supply network between 18:00 and 21:00 on weekdays. Energy Cybernetics and Yardstick implemented the project for Eskom. It was called Power Alert. Power Alert made use of real-time total electrical load data obtained from Eskom's National Control Centre to identify when the electrical supply was under strain. Predictive modelling was also used to determine the total electrical demand for the Western Cape based on realtime data, historical data and predicted weather conditions. The real-time predictive modelling was then used to predict the demand of the Western Cape 30 minutes, 60 minutes and 24 hours in advance. An online computer system and application was also developed that connected the Power Alert Control Centre to the Broadcasters Control Centre. This provided the Power Alert Control Centre with the capability to change any message to be broadcasted within five minutes. The real time predictive modelling was used to predict the total electrical demand and consequently the expected level of strain.