Goal-Directed Grid-Enabled Computing for Legacy Simulations

We describe a middleware framework conceived to enhance the effectiveness and efficiency of existing simulation applications by providing three capabilities: (1) access to grid-based and cloud-based execution, (2) access to advanced Design of Experiments (DOE) methodologies such as simulation-based optimization, and (3) access to robust data processing and visualization. The framework has been applied to a variety of simulations in both commercial and open source programming languages employing both discrete and continuous modeling formalisms. A key design objective is to minimize the workload necessary to adapt a simulation for use with the framework. User experience to date reveals that the learning curve for the framework is reasonable, but further automation of key tasks would enhance the framework´s utility.