Development of an Intelligent Monitoring and Control System for a Heterogeneous Numerical Propulsion System Simulation

1. Introduction

Designing and implementing new propulsion technologies can be an expensive and time-consuming process. The Numerical Propulsion System Simulation project, sponsored by NASA Lewis Research Center, is bringing new computer simulation techniques and parallel hardware to bear on this problem [2, 3]. Specifically, it is fostering the development of parallel simulations to improve both the execution time and accuracy of the simulations. A simulation executive will be developed that will support complete engine simulations made up from improved component simulations. Research on the simulation executive includes developing the monitoring and control techniques needed to manage the simulation, and exploring the use of expert system techniques to assist the user in controlling the simulation.

Several key issues must be addressed in the design of the propulsion system simulation. One is the integration of simulation codes at different levels of fidelity. Low-fidelity (less detail) modelling requires empirical data that are not available at the preliminary design stage. On the other hand, high-fidelity (more detail) modelling overcomes this limitation, but at a substantial computational cost. Zooming allows selected components to be modelled in detail and integrated into a low-fidelity engine simulation. Additionally, during a low-fidelity simulation, zooming provides a means of selectively examining in detail the physical processes within components of the engine.

A second issue is the use of a monitoring and control system. A monitoring tool will allow the user to observe the progress of the simulation through displays of its key parameters. An expert system can further improve the simulation by continuously monitoring and actively steering the simulation. This requires support in two areas: The first area is the collection of knowledge and the formulation of rules that govern the design and operation of jet engines. The second area is the integration of expert system software into the simulation executive to assist the user in executing the simulation.

A third issue is heterogeneity. The engine component codes and the expert system take advantage of a variety of vector and parallel platforms, and employ a variety of programming models and languages. An interconnection system allows components to execute on the most appropriate platform with minimum effort on the part of the user and the scientific programmer. The user should not see individual simulations that execute in isolation, but rather a single integrated simulation.

This paper describes a prototype simulation executive designed to address all three issues: zooming, a monitoring and control system, and heterogeneity support. The prototype employs a one-dimensional model of a complete engine. In this model, the operational characteristics of the individual system components are supplied in the form of performance maps that are constructed from experimental data. To provide descriptions of the physical processes occurring in an engine component beyond that supplied by a performance map, a higher fidelity component simulation is used. The simulation executive uses a monitoring tool that provides information about the high-fidelity component simulation to the user and the expert system. Based on this information, the expert system provides warnings and errors to the user and will be able to actively steer the engine simulation. Heterogeneity is addressed in the simulation executive through an interconnection system that provides the software framework to connect the various tasks.

Section two describes an engine model that demonstrates zooming on the fan component of NASA's Energy Efficient Engine [5]. Section three describes the design of a monitoring tool and expert system that assists the user in executing the simulation and will be used to explore techniques for intelligent control. The section also describes how the interconnection system integrates the parts to create the simulation executive. The system is currently being implemented, and the last section gives a snapshot of the current status of this project, along with some of its future directions.

Engine Simulation: Intelligent Monitoring and Control

[Next] [Previous] [Top]

Generated with CERN WebMaker