An Enhanced Iterative Process for Maintaining APC Applications
How can we generate data suitable for modeling while continuing to optimize the operation of the plant?
A well-engineered Aspen DMCplus application generally delivers substantial financial benefits via feed rate increases, better conversion, higher yield of valuable products, and reduced energy consumption. However, processes often change over time, requiring modification of the models. Why is this so problematic? Because the cost of modifying the model can be roughly 80% of the initial cost.
When a mismatch between the model and actual plant performance exists, we need to revise the model to eliminate the mismatch. The workflow is to collect current and representative process data. Then we must pre-process the data to remove bad values and other data anomalies. Next we produce a set of candidate models and evaluate the accuracy of those candidate models with due consideration of dynamics, gain ratios and Relative Gains (RGA numbers). We then adequately test the candidate models considering the current constraint sets. Finally, we configure and deploy the controller.
Most major commercial APC vendors offer tools to assist in these tasks. These tools have provided good value by reducing time and effort. But the potential benefits that can be achieved by making these tools work more synergistically are far greater than the benefits delivered by the individual tools.