| With the continuous development of internal combustion engine technology,performance simulation has ushered in new challenges in different application fields.As an important part of the performance simulation calculation of internal combustion engines,intake and exhaust flows have an important influence on the prediction accuracy and calculation efficiency of performance simulation software.With the increase of engine enhancement degree,higher flow rates and energy are generated in the intake and exhaust systems,especially the gradient of exhaust pressure wave variation,and the propagation process of pressure waves in the pipeline is more complicated.To accurately predict the energy loss and conversion process in the intake and exhaust system,the intake and exhaust system model needs to have a higher accuracy.On the other hand,in the design and verification of the internal combustion engine control system,it is necessary to adjust the control strategy in real time according to the working conditions,and the simulation model should be able to meet the real-time requirements.Since the existing models are developed for specific application scenarios,considering both the calculation accuracy and calculation efficiency is difficult;therefore,achieving a balance between the calculation accuracy and calculation efficiency in the simulation process of the intake and exhaust system has always been an urgent problem in the field of internal combustion engine performance simulation.Considering this,this study investigates the calculation method of variable-scale intake and exhausts flow and achieves a balance between simulation accuracy and efficiency according to different applications.In order to improve the calculation accuracy of intake and exhaust flows on the basis of satisfying certain computational efficiency,a pseudo one-dimensional flow(POD)model was established by combining traditional volumetric method with one-dimensional flow.Firstly,the pipeline was divided into several volumes.The continuity equation and energy equation were established within the volume,and the velocity at the interface between the volumes was solved.Then,based on the characteristic line theory and the assumption of linearization of boundary conditions,the variation of middle point velocity in the pipeline was deduced,and the POD calculation method of middle point velocity in the intake pipeline was established.Then,the impedance volume was established at the interface of pipeline volume,and the POD calculation method of exhaust pipeline was established considering the influence of wall friction and heat transfer on the flow.Finally,single-cylinder engine test was carried out to verify the prediction accuracy of POD model.The results show that the prediction accuracy of POD intake model is relatively higher,with the maximum and average error of 4.96% and 2.17%,and the exhaust error is slightly higher,with the maximum and average error of 8.1% and 2.76%,which can meet the performance simulation calculation with low demand for the prediction accuracy of pressure wave shape.The POD model has a high computational efficiency and can meet the real-time requirements in the speed range of 900?1400 r/min.To improve the computational efficiency of one-dimensional unsteady flow numerical simulation of intake and exhaust systems,based on the uniform-grid Total Variation Diminishing(TVD)format,the non-uniform-grid TVD iterative format is established.The density gradient variation parameter is used as the monitoring value.By introducing the mesh smoothness index,the adaptive operation of mesh encryption and merging is realized,and the variable-scale numerical simulation method is established.Calculation examples of different shock tubes were built to study the computational accuracy and efficiency of different solution formats.The results show that for uniform-grid TVD format,the more the number of grids,the higher the prediction accuracy,but the CPU computational consumption showed a linear growth trend.Under the same grid number,using a nonuniform grid can greatly improve the simulation accuracy on the basis of consuming the same CPU time.When a non-uniform grid is used,compared with the encryption grid scheme,on the basis of satisfying the simulation precision,the calculation time can be greatly reduced.The variable-scale numerical simulation method can reduce the data storage capacity,reduce the CPU consumption time,and improve the computational efficiency while ensuring that the numerical simulation accuracy is not reduced.To achieve the applicability of the manifold branch model to different branch numbers and meet the needs of different pipeline flow models,based on different theoretical assumptions,three theoretical models for multi-branch manifold boundary calculation are established,which are the multi-branch isobaric boundary model,multi-branch momentum conservation boundary model,and multi-branch pressure loss coefficient boundary model.Then the pressure loss coefficient of steady flow was measured,and the pressure loss database was established.Finally,with the experimental results as the benchmark,calculation examples were built to compare and analyze the prediction accuracy and calculation efficiency of different models.The results show that the isobaric model has the highest computational efficiency and can predict the pressure trend,but its prediction accuracy is lower.The momentum conservation boundary model considers the momentum loss along the flow path,and the model prediction accuracy is higher,but the calculation efficiency is lower.The pressure loss model can predict the pressure with the same precision as the momentum conservation boundary model,and the computational efficiency is greatly improved.To verify the accuracy of the intake and exhaust pipeline and manifold simulation model,other components and boundary models needed for the whole machine simulation were established,and a turbocharged diesel engine test bench was built.The model and test comparison verification research were carried out.The results show that the prediction results of the pipeline model coupled with different boundary test examples are consistent with the prediction results of mature software.For different speeds of the HC4132 turbocharged diesel engine,the simulation model could accurately predict the variation trend of cylinder pressure,intake and exhaust pressure wave,and main performance parameters,and the average error was less than 4.65%.To further improve the adaptability and efficiency of the simulation model,the coupling of intake and exhaust fluid network at different scales and the mathematical model are studied.First,based on the detailed model of the HC4132,the effects of different pipe and manifold multi-branch models on the performance and simulation efficiency of the whole machine are compared and analyzed.Afterward,the fluid network of intake and exhaust system is divided into three scales.Then the effects of different scale fluid networks on the performance and simulation efficiency of the whole machine are studied.Finally,the switching of fluid networks of different scales is coupled with the switching of simulation models to achieve intelligent switching between different models of the whole engine simulation.The results show that when the pipeline adopts the pseudo one-dimensional model,the whole engine performance prediction accuracy is less than 6%,the prediction accuracy is lower than that of the TVD format,but the computational efficiency is 7.59 times higher than that of TVD format.When the manifold branch boundary model adopts the isobaric model,the performance prediction accuracy of the whole engine is about 5%,the prediction accuracy is lower than the pressure loss model,but the calculation efficiency is 1.5 times of the pressure loss model.The performance prediction errors of the whole engine model after equivalent gas valve instead of multi-valve and similar pipe equivalent merger are within 6%,and the computational efficiency is 1.45 and 2.44 times that of the detailed model,respectively.By combining different stages of fluid networks with the mathematical model,the prediction error of the scheme combined with the simplified structure and the simplified model is less than 6.45%,but the simulation real-time ratio of HC4132 diesel engine in the calibration speed condition is 0.98,which can meet the realtime requirements. |
PDF Full Text Request