An investigation into the characteristics and optimisation of a high-pressure common rail injection system

  • Nao Hu

Student thesis: Doctoral Thesis


The primary aim of this research was to predict the emissions and fuel savings when replacing a mechanical fuel oil injection system with a high pressure common rail one. The work in this study consists of two parts. In the first part, a novel routine is proposed for the optimisation of electronic fuel injectors and their dynamic response, including the needle valve opening delay and the needle valve closing delay (The needle valve opening delay refers to the delay between the control signal being trigged and the needle valve being fully open; the needle valve closing delay refers to the delay between the control signal deactivating and the needle valve being fully closed).;Two injectors (Type-I and Type-II) were included; their one-dimensional (1D) models were built in AMESim software and validated respectively.;A parametric study on the Type-I fuel injector was conducted before the optimisation process in order to examine the effects of various parameters including the control piston diameter (CPD), control oil (i.e. the fuel oil used for control in a typical solenoid electronic fuel injector) inlet passage diameter (IPD) and control oil outlet passage diameter (OPD) on injection characteristics, i.e. injection rate, injection mass, needle valve lift and control chamber pressure.;Then, the optimisation of the injector dynamic response was investigated by the proposed routine in modeFRONTIER software. In detail, the routine included the following steps: First, a random sequence was adopted in the design of experiment (DOE) type. Then, an NSGA-II (Non-dominated Sorting Genetic Algorithm II) algorithm was selected. Next, a whole electronic fuel injector model was chosen, where the displacements of the needle valve were generated.;These data were first written into an input file, and to do this, appropriate writing and reading rules needed to be developed. The text file was read by the MATLAB code, where the control signal and needle valve displacement timings were calculated. The valve opening delay and the valve closing delay were thus obtained from these timings. Additionally, a constraint was set between the control oil inlet passage diameter and control oil outlet passage diameter in that the former should be smaller than the latter in each run.;The CPD, OPD and IPD were the three design parameters to be varied in the optimisation process of the Type-I fuel injector at a specific rail pressure. However, three more design parameters (the spring preload force (SPF), nozzle orifice number (NZN) and nozzle orifice diameter (NZD)) were involved in the optimisation process of the Type-II fuel injector under three different rail pressures (80 MPa, 120 MPa and 160 MPa).;The optimal design with the best trade-off between the valve opening delay and the valve closing delay of each fuel injector was singled out via a scattering chart. Results show that the optimum Type-I fuel injector achieved reductions of 40% and 25% of the baseline design on the valve opening delay and valve closing delay respectively. The optimal design of the Type-II fuel injector also achieved a huge reduction at all three rail pressures.;Specifically, the valve opening delay was reduced by 29.8%, 29.2% and 20.9%, and the valve closing delay was reduced by 25.6%, 24.5% and 30.1% at 80 MPa, 120 MPa and 160 MPa rail pressures respectively. RSM (response surface method) contour maps were used to study the interactions between design parameters. Results indicated that the CPD, IPD and OPD and their interactions are influential design parameters for the valve opening delay, while the IPD has a dominant effect on the valve closing delay.;. A large CPD together with a large IPD was found to increase the valve opening delay dramatically. Surprisingly, the effects of the spring preload force (SPF) on the valve closing delay are noticeable at low rail pressures. The valve closing delay decreases with an increase in the spring preload force. The feasibility and efficiency of the proposed routine was validated both on the Type-I fuel injector and the Type-II fuel injector. It was not only achieved great reductions on both the valve opening delay and the valve closing delay, but was also able to comprehensively disclose the effects and interactions of the design parameters on the injector dynamic response.;In the second part, Type-II fuel injector matches with the combustion chamber of a medium-speed marine diesel engine was conducted by a CFD model of the medium-speed marine diesel engine built in AVL FIRE software. The model was validated by using the cylinder pressures, rate of heat release (ROHR) and NOx emissions under four engine loads, i.e. L25 (25%), L50 (50%), L75 (75%) and L100 (100%) loads.;Seven engine design parameters, including four injection-related parameters (spray angle, nozzle protrusion length, injection timing and swirl ratio) and three combustion chamber geometry parameters (bowl diameter, centre crown height and toroidal radius), were examined by a parametric study and a multi-objective optimisation.;The parametric study was carried out to discover the sensitivity of design parameters on the objectives (nitrogen oxides (NOx) emissions, soot emissions and specific fuel oil consumption (SFOC)). In it, each engine design parameter was investigated independently under three engine loads (25% engine load was excluded due to the fact that it is very unstable operating condition). Results showed that the injection-related parameters were found to have much more influence than the combustion chamber geometries. In addition, the injection timing, one of the injection-related parameters, has the largest influence on the objectives.;In the optimisation study, two algorithms (the nonlinear programming by quadratic Lagrangian (NLPQL) algorithm and multi-objective genetic algorithm (MOGA)) were used in two stages, initially separately and then sequentially, for the first time in the engine optimisation domain. This study aims to reduce the NOx emission, soot emissions as well as to improve fuel economy. Detailed comparisons were made for NOx emissions, soot emissions and SFOC as well as the design parameters.;The optimisation study showed that the NLPQL algorithm failed to obtain optimal designs whilst the MOGA offered more feasible Pareto designs. Since the NLPQL algorithm is a local optimisation method, and as a result it is affected by the selection of appropriate initial conditions. The optimal design with the best trade-off between NOx and soot emissions obtained by the MOGA was set as the starting point of the NLPQL algorithm. In this case, a better design with lower NOx emissions and soot emissions was obtained.;The combustion processes of these optimal designs were also analysed and compared in detail. Late injection and small swirl ratio were reckoned to be the main reasons for reducing NOx emissions. In the end, contour maps generated by response surface method (RSM) were applied in order to gain a better understanding of the interaction and sensitivity of the design parameters on NOx emissions, soot emissions and SFOC.;Results indicated that NOx emissions and soot emissions can be greatly reduced by adopting a late injection, a low swirl and a large spray angle, but fuel economy was sacrificed.
Date of Award28 Sept 2017
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsUniversity of Strathclyde & Lloyds Register of Shipping
SupervisorPeilin Zhou (Supervisor) & David Clelland (Supervisor)

Cite this