This proposal integrates expertise in system dynamics and control with expertise in fluid mechanics and engine modeling to develop a systematic approach to modeling transient phenomena in compressible fluid flow systems. Such a methodology does not exist today. The successful outcome of this project will be the ability to systematically carry out model order reduction for distributed parameter systems that describe the flow of compressible fluids, in analogy with existing methods available, for example, for mechanical systems. The ability to develop physically consistent lumped-parameter models of compressible flow systems will enable improvements in the control of energy conversion systems such as internal combustion engines, turbomachinery, fuel cell systems, HVAC and refrigeration systems, leading to better performance, reduced costs and reduced environmental impact.
The automotive industry is striving to improve engine fuel economy and to reduce their emissions in view of increasingly stringent government mandates and market demand for vehicles with near-zero exhaust emissions and significantly improved fuel economy. It is well understood that such gains can only be achieved if improvements in engine design can be matched by the ability to closely control engine efficiency and emissions. The present project will lead to better fuel economy and emissions control, but will also significant shorten the development of new, more efficient and cleaner engines. Today, this process is very much ad-hoc and it requires significant amounts of experimental calibration. The outcome of the proposed project is a systematic and widely applicable methodology that will shorten development time and that will lead to robust engine control designs that will also be easily portable among engine platforms. The project involves significant participation of Honor undergraduate students from under-represented groups, and an international collaboration with the University of Stuttgart, Germany.
With the advancements in design, the dynamics of compressible flows in air or gas delivery systems have become so complex that advanced optimization and control design methods are nowadays needed to fully exploit the potential of the existing technologies. Remarkable examples are the internal combustion engines for passenger cars and trucks, where solutions such as downsizing, boosting, and flexible valve actuation have contributed to a dramatic improvement of fuel economy in the past few years. The automotive industry is currently seeking to considerably improve the modeling tools employed for the engine air path system control design process, as the control-oriented models currently in use are limited in their accuracy and predictive ability, and rely significantly on calibration. As engine systems become increasingly complex, the ability to predict engine flows through low-order, physically based system dynamics models is therefore of critical importance to the development of model-based control algorithms that can be easily adapted to different engine platforms. This project integrated expertise in system dynamics and control with expertise in fluid mechanics and engine modeling to develop a novel approach to modeling transient phenomena in compressible fluid flow systems. The intellectual merit of the project was in the creation of model order reduction techniques for nonlinear distributed parameter systems describing the flow of compressible fluids. This project fills a historic gap between the traditional approaches to model such systems, based either on high-fidelity numerical simulation methods originating from partial differential equations (computationally expensive and not suitable for control and estimation), and low-fidelity, control-oriented heuristic methods based on lumped parameter approximations (calibration-intensive and usually inaccurate). The idea of applying model order reduction to nonlinear partial differential equations leads to the advantage of retaining much of the fidelity with the physics of the system, while guaranteeing low calibration effort. The project was executed by focusing on the following priorities: Formalize theoretical approach for model-order reduction methodology applied to 1D hyperbolic Partial Differential Equations (PDEs) for compressible flows (nonlinear Euler equations); Formalization of solution approach based on definition of Spatial Basis Functions; Verification of approach through application to theoretical problems and comparison with analytical solutions (linear and nonlinear wave propagation in straight pipes, shock tube); Application to internal combustion engines simulation and benchmarking against numerical solution methods and experimental data; Disseminate results of research to Industry; Develop education programs on methods for control-oriented thermo-fluid modeling. The objectives of the project have been met in full, demonstrating a systematic methodology to transfer the detailed understanding of gas flow processes from well-developed 1-D physics-based numerical simulation tools to control-oriented models. This approach was demonstrated to replace the current processes in use, which are very much ad-hoc and require significant amounts of experimental calibration. As part of the broader impact, three B.S. Honor students and two Ph.D. students graduated as a result of this program (one is now a research engineer in a major automotive company, the other is continuing towards an academic career). The outcome of the research was shared with the scientific community through numerous papers, presentations and seminars, and was transferred to the automotive industry. In particular, Ford and Chrysler have valued the results of this project, and are currently providing support and funding to research on model-based control design techniques using the methodology developed in this project. The project has strengthened collaborations with foreign institutions, namely the University of Stuttgart Automotive Research Institute (FKFS - Germany) and the Politecnico of Milano (Italy). The material developed in this project contributed in strengthening the material taught in undergraduate and graduate courses offered in the Mechanical Engineering program, namely ME-7281 (Powertrain Dynamics – Rizzoni), ME-7440 (Internal Combustion Engine Modeling – Guezennec) and ME-5339 (Simulation Techniques for Dynamic Systems Analysis and Design – Canova).
