Abstract:
It is important to develop methods that are capable of successfully determining plant performance. The method used should be based on the ability to determine the performance of each of the various unit operations within the plant. This in turn will assist with the correct decision as to which unit in the plant should be improved first. The performance of the various units can be accumulated to give a representation of the performance of the entire plant. A plant-wide performance monitoring method has been developed to do just this. Originally it was developed for a specific unit operation. It has now been verified that this method is applicable to different unit operations. The method employed to determine this plant-wide performance is by evaluating how close the plant is to its inherent optimum. Where applicable, this inherent optimum can also be replaced with a user specified optimum. When an optimum is specified there is a possibility of oscillations around this “optimum” and the effects of this on the performance number are eliminated to give a more general plant-wide performance number for each unit operation. In addition to the “optimum” value selection the addition of performance weights to specific focus areas (utility usage or product quality) in the performance calculation will also improve the comparative nature of the plant-wide index for different unit operations. The scope of this investigation is limited to the experimental test rigs that were available in the Process Control Laboratory at the University of Pretoria. The methods that were used to determine the single loop performance of each of the different control loops are: <ul> <li>Minimum variance</li> <li>Generalised minimum variance</li> <li>Integral of the Absolute Error (IAE)</li> <li>Integral of the Square Error (ISE)</li> </ul> The single loop performance methods are required to determine how effectively the plant-wide performance index evaluates the plant, since these are existing means of determining how well a plant is operating, but these become impractical due to excessive amounts of information needing evaluation.
