Power and prediction

Caught in the icy grip of winter, the last thing South Africans want is load shedding – but Eskom can announce load shedding at any time, throwing lives into disarray.

One way to prevent or minimise the risk of load shedding is to have a proper and adequate schedule for the maintenance of power generating units in place, says Dr Jancke Eygelaar, a doctoratal graduate in industrial engineering at Stellenbosch University.

According Dr Eygelaar, the ability of a power utility like Eskom to satisfy energy demand can be inuenced signicantly by unexpected breakdowns of power generating units whose reliability (whether they may fail or not) may be compromised, given how long the utility has been in operation.

The elective scheduling of this planned maintenance of power generating units is a considerable challenge for any power utility.

“In most cases, unexpected failures are also much more expensive to repair than taking planned preventative maintenance action. Maintenance of ageing power generating units is, however, often neglected due to high energy demand and low system capacity, as seen in the case of Eskom.”

Dr Eygelaar says his study took into account the demand of a system, as well as other factors, and aimed to schedule preventative maintenance on generators which are operating at a low reliability. These generators have a high probability of failing which may cause load shedding to be implemented as the demand cannot be met.

As part of his doctoral research, Dr Eygelaar put forward two new scheduling criteria that could help optimise the maintenance scheduling. The first involves minimising the probability that a power generating unit will fail, while the second involves maximising expected energy production during a scheduling window.

He analysed the effectiveness of these two criteria by applying them to a real-world case study based on the Eskom power system. “Both of these criteria can improve the reliability of a power system. By taking into account the reliability of the generators when scheduling preventative maintenance, load shedding and power outages can possible be avoided completely.”

In addition to the scheduling criteria, Dr Eygelaar developed a computerised decision support system that can provide an operations manager with good planned maintenance schedules for power generating units based on the two scheduling criteria. A decision support system helps the user to make informative decisions and to identify and solve complex problems.

“The decision support system will provide employees with information regarding available capacity and the demand of the system over the scheduling window, as well as information on the manpower required to perform maintenance over the duration of the maintenance window,” Dr Eygelaar says.