It is widely considered that the adoption of iterative software engineering methodologies and in particular continuous testing helps ensure high quality software and reduce bugs. The successful application of continuous testing however rests on the assumptions that testing is cheap, fast and easily repeatable. Software development for control systems in the chemical production domain generally cannot satisfy that constraint as evaluating the correctness of a recipe program requires its execution on a live production environment which can take multiple days to complete, usually comes at great expense in raw materials and can sometimes create a reliance on safety systems to manage risk. As a result testing in the chemical domain becomes a bottleneck that prevents true iterative cycles taking place. This in turn leads to a linear waterfall-like process with all its inherent problems and limitations.

To help resolve this problem, we propose a generic simulation framework, based on a domain model of core components of chemical productions plants. This simulation can be used in place of the live plant during a first phase of testing. Only once an engineer is satisfied that the software is performing as expected on the simulation, will that live plant hardware need to be involved. This will help greatly in reducing the bottleneck in the testing phase by allowing this to be quick and automated while reducing the risk and cost involved.