FOR 2401:  Optimization-Based Multiscale Control of Low-Temperature Combustion Engines

• Subject Area: Thermal Engineering/Process Engineering ; Computer Science, Systems and Electrical Engineering
• Term: since 2016
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 277012063

Cycle-based control is the main research approach currently under investigation for controlling low-temperature combustion (LTC). This control, however, only allows the stabilization of the LTC in a very limited operation range. With a cycle-based control, only cycle-integral system dynamics and disturbance variables can be controlled. The chemical-physical processes, which occur on in-cycle time scales, are relevant for the stability and emission generation of the LTC. They cannot be influenced by a cycle based control. Consequently, the research unit is investigating multi-scale control in order to be able to take the smaller, in-cycle time scales into account. It is expected that the stability can be improved, the operation range can be extended significantly, the efficiency can be increased and the pollutant emissions can be reduced. Multi-scale control is an innovative and novel approach to solve this problem. In the research unit the LTC processes PCCI and GCAI are studied. A controller architecture consisting of a combination of cycle-to-cycle control and an in-cycle controller is developed. To account for the complex nonlinear multivariable system dynamics, optimization-based methods based on models of the process are developed and applied. For this purpose, numerical methods for iterative learning nonlinear model-based predictive control (IL-NMPC) developed for cyclic processes are used as a basis. Specifically for the processes, an analysis and evaluation of possible manipulated and controlled variables as well as the allocation of these variables to the different time scales is done. In addition, suitable formulations of the optimization task are also investigated. The controller-internal models are developed for both the GCAI and the PCCI process. Derived from experiments on singlecylinder test benches, physical models and quantitative descriptions of the processes are created. Furthermore, kinetic descriptions of the combustion processes, which are obtained by using surrogate fuels on optimized test laboratories such as reactors and flames, are emphasized. Furthermore, the integration of ion current as a novel sensor concept in the control system is considered. The developed descriptions are transferred into a structure that can be effectively used for realtime control. Furthermore, physically motivated grey box models are to be created for this purpose. Based on physical and system theoretical observations, models for disturbance variables will be developed. Finally, the resulting control algorithms will be validated on engine test benches. Evaluation criteria are the combustion stability, the coverable operation range in which stable operation and transient load profiles can be realized, and the potential for emission reduction and efficiency increase.

• DFG Programme: Research Units
• International Connection: Switzerland

Projects

• Characterization of the low temperature chemistry of large hydrocarbons for model development in motor control (Applicant Kasper, Tina )
• Control of quasi-homogeneous combustion in diesel engines using fully-variable injection (Applicants Beeckmann, Joachim ;  Pitsch, Heinz )
• Coordination Funds (Applicant Andert, Jakob )
• Detailed chemistry investigation for model development for engine control (Applicant Kohse-Höinghaus, Katharina )
• Multi Scale Control of PCCI Combustion (Applicant Onder, Christopher )
• Multiscale-control of the low-temperature combustion process GCAI (Applicant Abel, Dirk )
• Parallel multi-level learning and optimization algorithms for control of cyclic processes on embedded systems (Applicant Diehl, Moritz )
• Stabilization of the GCAI combustion process by in-cycle correlations (Applicants Andert, Jakob ;  Pischinger, Stefan )

• Spokesperson: Professor Dr.-Ing. Jakob Andert, since 12/2019