Search Results

The accurate prediction of fuel sprays is critical to engine combustion and emissions simulations. A fine computational mesh is often required to better resolve fuel spray dynamics and vaporization. However, computations with a fine mesh require extensive computer time. This study developed a methodology that uses a locally refined mesh in the spray region. Such adaptive mesh refinement will enable greater resolution of the liquid-gas interaction while incurring only a small increase in the total number of computational cells. The present study uses an h-refinement adaptive method. A face-based approach is used for the inter-level boundary conditions. The prolongation and restriction procedure preserves conservation of properties in performing grid refinement/coarsening. The refinement criterion is based on the mass of spray liquid and fuel vapor in each cell. The efficiency and accuracy of the present adaptive mesh refinement scheme is demonstrated.

A general combustion model, in the context of large eddy simulations, was developed to simulate the full range of combustion in conventional diesel-type and HCCI-type diesels. The combustion model consisted of a Chemkin sub-model and an Extended Flamelet Time Scale (EFTS) sub-model. Specifically, Chemkin was used to simulate auto-ignition process. In the post-ignition phase, the combustion model was switched to EFTS. In the EFTS sub-model, combustion was assumed to be a combination of two elementary combustion modes: homogeneous combustion and flamelet combustion. The combustion index acted as a weighting factor blending the contributions from these two modes. The Chemkin sub-model neglected the subgrid scale turbulence-chemistry interactions whereas the EFTS model took them into account through a presumed PDF approach. The model was used to simulate an early injection mode of a Cummins DI diesel engine and a mode of a Caterpillar DI diesel engine.

Parallel computing schemes were developed to enhance the computational efficiency of engine spray simulations with adaptive mesh refinement (AMR). Spray simulations have been shown to be grid dependent and thus fine mesh is often used to improve solution accuracy. In this study, dynamic mesh refinement adaptive to spray region was developed and parallelized in KIVA-4. The change of cell and node numbers and the local characteristics in the dynamic mesh refinement posed difficulties in developing efficient parallel computing schemes to achieve low communication overhead and good load balance. The present strategy executed AMR on one processor with data scattering among processors following the adaptation, and performed AMR every ten computational timesteps for enhanced parallel performance. The re-initialization was required and performed at the minimized cost.

A particle swarm optimization (PSO) algorithm was implemented with engine testing in order to accelerate the engine development process. The PSO algorithm is a stochastic, population-based evolutionary optimization algorithm. In this study, PSO was used to reduce exhaust emissions while maintaining high fuel efficiency. A merit function was defined to help reduce multiple emissions simultaneously. Engine operations using both single-injection and double-injection strategies were optimized. The present PSO algorithm was found to be very effective in finding the favorable operating conditions for low emissions. The optimization usually took 40-70 experimental runs to find the most favorable operating conditions under the constraints specified in the present testing. High EGR levels, small pilot amount, and late main injection were suggested by the PSO. Multiple emissions were reduced simultaneously without a compromise in the brake specific fuel consumption.

The combustion and emissions of a diesel/natural gas dual-fuel engine are studied. Available engine experimental data demonstrates that the dual-fuel configuration provides a potential alternative to diesel engine operation for reducing emissions. The experiments are compared to multi-dimensional model results. The computer code used is based on the KIVA-3V code and consists of updated sub-models to simulate more accurately the fuel spray atomization, auto-ignition, combustion and emissions processes. The model results show that dual-fuel engine combustion and emissions are well predicted by the present multi-dimensional model. Significant reduction in NOx emissions is observed in both the experiments and simulations when natural gas is substituted for diesel fuel. The HC emissions are under predicted by numerical model as the natural gas substitution is increased.

Multi-dimensional computational efforts using comprehensive and skeletal kinetics have been made to investigate the cool flame region in HCCI combustion. The work was done in parallel to an experimental study that showed the impact of the negative temperature coefficient and the cool flame on the start of combustion using different fuels, which is now the focus of the simulation work. Experiments in a single cylinder CFR research engine with n-butane and a primary reference fuel with an octane number of 70 (PRF 70) were modeled. A comparison of the pressure and heat release traces of the experimental and computational results shows the difficulties in predicting the heat release in the cool flame region. The behavior of the driving radicals for two-stage ignition is studied and is compared to the behavior for a single-ignition from the literature. Model results show that PRF 70 exhibits more pronounced cool flame heat release than n-butane.

Neural networks are useful tools for optimization studies since they are very fast, so that while capturing the accuracy of multi-dimensional CFD calculations or experimental data, they can be run numerous times as required by many optimization techniques. This paper describes how a set of neural networks trained on a multi-dimensional CFD code to predict pressure, temperature, heat flux, torque and emissions, have been used by a genetic algorithm in combination with a hill-climbing type algorithm to optimize operating parameters of a diesel engine over the entire speed-torque map of the engine. The optimized parameters are mass of fuel injected per cycle, shape of the injection profile for dual split injection, start of injection, EGR level and boost pressure. These have been optimized for minimum emissions. Another set of neural networks have been trained to predict the optimized parameters, based on the speed-torque point of the engine.

Neural networks have been used for engine computations in the recent past. One reason for using neural networks is to capture the accuracy of multi-dimensional CFD calculations or experimental data while saving computational time, so that system simulations can be performed within a reasonable time frame. This paper describes three methods to improve upon neural network predictions. Improvement is demonstrated for in-cylinder pressure predictions in particular. The first method incorporates a physical combustion model within the transfer function of the neural network, so that the network predictions incorporate physical relationships as well as mathematical models to fit the data. The second method shows how partitioning the data into different regimes based on different physical processes, and training different networks for different regimes, improves the accuracy of predictions.

One of the most critical elements in diesel engine design is the selection and matching of the fuel injection system. The injection largely controls the combustion process, and with it also a wide range of related issues, such as: fuel efficiency, emissions, startability, load acceptance (acceleration) and combustion noise. Simulation has been a valuable tool for the engine design engineer to predict and optimize key parameters of the fuel injection system. This is a problem that spans a number of subsystems. Historically, simulations of these subsystems (hydraulics, gas dynamics, engine performance and 3-D CFD cylinder modeling) have typically been done in isolation. Recently, a simulation tool has been developed, which models the different subsystems in an integrated manner. This simulation tool combines a 1-D simulation tool for modeling of hydraulic and gas dynamics systems, with 3-D CFD code for modeling the in-cylinder combustion and emissions.

A modeling study has been conducted in order to characterize the heat transfer in an automotive diesel exhaust system. The exhaust system model, focusing on 2 exhaust pipes, has been created using a transient 1-D engine flow network simulation program. Model results are in excellent agreement with experimental data gathered before commencement of the modeling study. Predicted pipe exit stream temperatures are generally within one percent of experimental values. Sensitivity analysis of the model was the major focus of this study. Four separate variables were chosen for the sensitivity analysis. These being the external convective heat transfer coefficient, external emissivity, mass flow rate of exhaust gases, and amplitude of incoming pressure fluctuations. These variables were independently studied to determine their contribution to changes in exhaust gas stream temperature and system heat flux. There are two primary benefits obtained from conducting this analysis.

Detailed chemical kinetics was implemented in the KIVA-3V multidimensional CFD code to study the combustion process in Homogeneous Charge Compression Ignition (HCCI) engines. The CHEMKIN code was implemented such that the chemistry and flow solutions were coupled. Detailed reaction mechanisms were used to simulate the fuel chemistry of ignition and combustion. Effects of turbulent mixing on the reaction rates were also considered. The model was validated using the experimental data from two modified heavy-duty diesel engines, including a Volvo engine and a Caterpillar engine operated at the HCCI mode. The results show that good levels of agreement were obtained using the present KIVA/CHEMKIN model for a wide range of engine conditions, including various fuels, injection systems, engine speeds, and EGR levels. Ignition timings were predicted well without the need to adjust any kinetic constants.

The present study investigated HCCI combustion in a heavy-duty diesel engine both experimentally and numerically. The engine was equipped with a hollow-cone pressure-swirl injector using gasoline direct injection. Characteristics of HCCI combustion were obtained by very early injection with a heated intake charge. Experimental results showed an increase in NOx emission and a decrease in UHC as the injection timing was retarded. It was also found that optimization can be achieved by controlling the intake temperature together with the start-of-injection timing. The experiments were modeled by using an engine CFD code with detailed chemistry. The CHEMKIN code was implemented into KIVA-3V such that the chemistry and flow solutions were coupled. The model predicted ignition timing, cylinder pressure, and heat release rates reasonably well. The NOx emissions were found to increase as the injection timing was retarded, in agreement with experimental results.

Artificial neural networks (ANN) have been recognized as universal approximators for nonlinear continuous functions and actively applied in engine research in recent years [1, 2, 3, 4, 5, 6, 7 and 8]. This paper describes the methodology and results of using the ANN to model a turbocharged DI diesel engine. The engine was simulated using the CFD code (KIVA-ERC) over a wide range of operating conditions, and numerical simulation results were used to train the ANN. An efficient data collection methodology using the Design of Experiments (DOE) techniques was developed to select the most characteristic engine operating conditions and hence the most informative data to train the ANN. This approach minimizes the time and cost of collecting training data from either computational or experimental resources. The trained ANN was then used to predict engine parameters such as cylinder pressure, cylinder temperature, NOx and soot emissions, and cylinder heat transfer.

A dual fuel engine simulation model was formulated and the combustion process of a diesel/natural gas dual fuel engine was studied using an updated KIVA-3V Computational Fluid Dynamic (CFD) code. The dual fuel engine ignition and combustion process is complicated since it includes diesel injection, atomization and ignition, superimposed with premixed natural gas combustion. However, understanding of the combustion process is critical for engine performance optimization. Starting from a previously validated Characteristic-Timescale diesel combustion model, a natural gas combustion model was implemented and added to simulate the ignition and combustion process in a dual fuel bus engine. Available engine test data were used for validation of both the diesel-only and the premixed spark-ignition operation regimes. A new formulation of the Characteristic-Timescale combustion model was then introduced to allow smooth transition between the combustion regimes.

The present study uses a numerical model to investigate the effects of flow turbulence on premixed iso-octane HCCI engine combustion. Different levels of in-cylinder turbulence are generated by using different piston geometries, namely a disc-shape versus a square-shape bowl. The numerical model is based on the KIVA code which is modified to use CHEMKIN as the chemistry solver. A detailed reaction mechanism is used to simulate the fuel chemistry. It is found that turbulence has significant effects on HCCI combustion. In the current engine setup, the main effect of turbulence is to affect the wall heat transfer, and hence to change the mixture temperature which, in turn, influences the ignition timing and combustion duration. The model also predicts that the combustion duration in the square bowl case is longer than that in the disc piston case which agrees with the measurements.

A level set G-equation model has been developed to model the combustion process in spark ignition engines. The spark ignition process was modeled using an improved version of the Discrete Particle Ignition Kernel (DPIK) model. The two models were implemented into the KIVA-3V code to simulate SI engine combustion under both premixed and direct injection conditions. In the ignition model, the ignition kernel growth is tracked by Lagrangian markers, and spark discharge energy and flow turbulence effects on the kernel growth are considered. Once the ignition kernel grows to a size where the turbulent flame is fully developed, the G-equation model is used to track the mean turbulent flame evolution. When combined with a characteristic time scale combustion model, the models were also used to simulate stratified combustion in DISI engines, where the triple flame structure must be considered.

DPF regenerations involve a trade-off between fuel economy and DPF durability. High temperature regenerations of DPFs have fewer fuel penalties but simultaneously tend to give higher substrate temperatures, which can reduce thermal reliability. In order to weaken the trade-off, the integrated system-level model [1,2,3,4] is used to conduct optimization studies and explore novel regeneration strategies for DPF aftertreatment devices. The integrated model developed in the Engine Research Center (ERC) includes sub-models for engines, emissions, aftertreatment devices and controllers. Based on the engine and regeneration fuel economy, multiple and single cycle regeneration tests are performed and analyzed. The optimal soot loadings to initiate and terminate regenerations are discussed. A pulsed regeneration strategy, which is characterized by injecting multiple pulses of fuel (upstream of a DOC) during regenerations, is investigated.

Large Eddy Simulation (LES) with a flamelet time scale combustion model is used to simulate diesel combustion. The flamelet time scale model uses a steady-state flamelet library for n-heptane indexed by mean mixture fraction, mixture fraction variance, and mean scalar dissipation rate. In the combustion model, reactions proceed towards the flamelet library solution at a time scale associated with the slowest reaction. This combination of a flamelet solution and a chemical time scale helps to account for unsteady mixing effects. The turbulent sub-grid stresses are simulated using a one-equation, non-viscosity LES model called the dynamic structure model. The model uses a tensor coefficient determined by the dynamic procedure and the subgrid kinetic energy. The model has been expanded to include scalar mixing and scalar dissipation. A new model for the conditional scalar dissipation has been developed to better predict local extinction.

To help account for fuel distribution during combustion in diesel engines, a fuel film model has been developed and implemented into the KIVA-II code [1]. Spray-wall interaction and spray-film interaction are also incorporated into the model. Modified wall functions for evaporating, wavy films are developed and tested. The model simulates thin fuel film flow on solid surfaces of arbitrary configuration. This is achieved by solving the continuity, momentum and energy equations for the two dimensional film that flows over a three dimensional surface. The major physical effects considered in the model include mass and momentum contributions to the film due to spray drop impingement, splashing effects, various shear forces, piston acceleration, dynamic pressure effects, and convective heat and mass transfer.

A computer model that is able to predict the occurrence of knock in spark ignition engines has been developed and implemented into the KIVA-3V code. Three major sub-models were used to simulate the overall process, namely the spark ignition model, combustion model, and end-gas auto-ignition models. The spark ignition and early flame development is modeled by a particle marker technique to locate the flame kernel. The characteristic-time combustion model is applied to simulate the propagation of the regular flame. The autoignition chemistry in the end-gas was modeled by a reduced chemical kinetics mechanism that is based on the Shell model. The present model was validated by simulating the experimental data in three different engines. The spark ignition and the combustion models were first validated by simulating a premixed Caterpillar engine that was converted to run on propane. Computed cylinder pressure agrees well with the experimental data.