Decentralized next cycle control of RCCI combustion

  • B.F.P.M.M. ten Doeschate

Student thesis: Master


High efficiency and low emissions have become essential for internal combustion engines, especially in heavy-duty applications. Heavy-duty engines are the work horse in transport, and they will remain the primary power source for the foreseeable future. To realize clean and efficient transport in the future, developing ultra-efficient combustion concepts that can utilize sustainable fuels has become an important challenge. Reactivity-Controlled Compression Ignition (RCCI) is a promising dualfuel combustion concept that has the potential for ultra-clean and highly-efficient engines. As this advanced concept is sensitive to variations in operating conditions, the development of closed-loop combustion control is essential to implement RCCI.
Cylinder pressure-based closed-loop combustion control (CPBC) offers new possibilities to improve control of the combustion process. In particular, next cycle control (NCC) enables cycle-by-cycle control actions by calculating the required control action for the next cycle based on the results from the previous combustion cycle. Because the combustion process is subject to various disturbances in the combustion conditions (e.g. ambient conditions and varying fuel properties), the main control challenge is to guarantee stable and safe combustion, and robust performance.
This thesis describes the design and implementation of a robust decentralized next cycle control system for control of RCCI combustion. Step response analysis and multi-sine frequency responses define the system’s dynamic behaviour, which is highly interactive and generally non-linear. This system analysis results in the determination of the dc gain matrix and the complex frequency responses.
The control configuration design involves a methodical screening of potential feedback variables and combinations of feedback variables, and RGA analysis to determine the optimal input–output pairings. RCCI combustion control is a complex multi-variable control problem. To realize multivariable control the system interactions are compensated using a pre-compensator (or decoupling matrix), in order to use decentralized feedback control. Two sets of feedback variables and two decoupling strategies are defined.
The first step in the controller design process is the identification of the decoupled system transfers of the defined input–output pairs. This is accomplished by estimating transfer functions (TFs) from the measured frequency responses. This results in 5 independent SISO TF models, for which 5 individual proportional-integral (PI) controllers are designed. The realized feedback controllers are subsequently implemented to control an RCCI engine simulation model developed by TNO Powertrains.
The reference tracking and disturbance rejection performance of the realized control system are assessed in two operating points, comparing the two feedback sets and the two decoupling strategies.
Also the effects on high-level performance objectives are discussed. Closed-loop control of the intake manifold temperature and the peak cylinder pressure rise rate is feasible, and results in stable and safe combustion. Decoupling at the closed-loop bandwidth frequency is the best strategy to maximize closed-loop performance.
Date of Award25 Mar 2022
Original languageEnglish
SupervisorBram de Jager (Supervisor 1) & Frank P.T. Willems (Supervisor 2)

Cite this