Multi-input–multi-output optimization of reactivity-controlled compression-ignition combustion in a heavy-duty diesel engine running on natural gas/diesel fuel. (March 2020)

• Record Type: Journal Article
• Authors: Ebrahimi, Mojtaba; Najafi, Mohammad; Jazayeri, Seyed Ali
• Is Part Of: International journal of engine research. Volume 21:Number 3(2020)
• Page Start: 470
• Page End: 483
• Publication Date: 2020-03
• Subjects: Reactivity-controlled compression ignition -- combustion -- diesel engine -- natural gas -- emission -- design of experiment -- factorial method -- artificial neural network -- multi-input–multi-output optimization -- genetic algorithm -- particle swarm optimization algorithm; Engines -- Periodicals; 629.25
• DOI: 10.1177/1468087419832085
• Languages: English
• ISSNs: 1468-0874

Abstract

The aim of this study is to implement the multi-input–multi-output optimization of reactivity-controlled compression-ignition combustion in a heavy-duty diesel engine running on natural gas and diesel fuel. A single-cylinder heavy-duty diesel engine with a modified bathtub piston bowl profile is set on operation at 9.4 bar indicated mean effective pressure and running at a fixed engine speed of 1300 r/min. A certain amount of diesel fuel mass per cycle is fed into the engine at a fixed equivalence ratio without any exhaust gas recirculation. The optimization targets include reduction in engine emissions as much as possible, avoiding diesel knock occurrence, and achieving low temperature combustion concept with the least or no engine power losses. To implement the optimization, the effects of three control factors on the engine performance are assessed by the design of experiment concept—fractional factorial method. These selected control factors are intake temperature and intake pressure (both at intake valve closing) and the diesel fuel start of injection timing. Some randomized treatment combinations of chosen levels from the three selected control factors are employed to simulate reactivity-controlled compression-ignition combustion. Based on the engine's responses derived from the simulation, reactivity-controlled compression-ignition combustion's mathematical model is identified directly using an artificial neural network. Next, an optimization process is conducted using two different optimization algorithms, namely, genetic algorithm and particle swarm optimization algorithm. For assessing and validating the obtained optimal results, the obtained data are used to simulate reactivity-controlled compression-ignition combustion as the engine input factors. The results show that the proposed artificial neural network design is effectively capable of identifying reactivity-controlled compression-ignition combustion's mathematical model. Also, by optimizing reactivity-controlled compression-ignition combustion through different optimization algorithms, the optimal range of the engine operation at 9.4 bar indicated mean effective pressure is well estimated and extended.