PowerGen optimizes power plant performance using NAG’s Algorithms

The combustion process within a coal-fired power plant is complex and subject to a number of variable parameters that can affect performance. These include the distribution of coal in the boiler, the amount and distribution of the air in the boiler and, sometimes the particle size of the ground coal, which can be sourced from several different coal mines.

There are constraints, however, on which parameters can be adjusted to achieve optimization. Fuel flow, for example, needs to be maintained in order to keep the amount of power generated constant. If this constraint is removed, the model would simply shut the plant down – great for reducing NOx emissions but not appropriate for generating power or profitability.

There are also constraints on steam temperatures which can limit the range of operation and also constraints on engineering the upper and lower boundaries of the control variables.

With such a complex process, in which both dependent and independent variables can play an important part in overall performance, it is clear that optimization represents a genuine challenge. And, it is a challenge that needs to be tackled, if the balance between commercial and environmental targets is to be achieved.

There was really only one effective way forward – the development of a software model of the combustion process in the boiler of the coal-fired power plant that can account for all these variables.

Dr Mayes explained: “With constantly changing environmental legislation and commercial pressures, there is a need for continuous optmization of the power station boiler and traditional combustion testing only provided a snap-shot. It was also too expensive to frequently run tests on the actual plant – it simply wasn’t feasible. So, we’ve developed mathematical models to enable the process to be optimized based on certain criteria – a general concept that can be applied across many different industrial systems. The key feature is the need for a model of the process that allows us to use optimization techniques to find better solutions.”

In writing and developing the model, PowerGen turned to NAG for mathematical analysis software that could be integrated within its specialist application.

Dr Mayes explained: “With so many variables and parameters that can influence performance, a large degree of number crunching is required within our model. Whilst we concentrated on developing the specialist software for our application, we used the NAG Fortran Library to carry out mathematical manipulation of data within the model. After all, there is little point in reinventing the wheel. We looked at what was available and NAG offered precisely what we required.“

“Not only that, NAG has a reputation for high quality, tried and tested software that is well documented and supported by first class technical help, if required. In fact, whenever a mathematical operation of any complexity is required, we check to see whether there are NAG routines that can do the job.”