Search Results

A computational fluid dynamics (CFD) guided combustion system optimization was conducted for a heavy-duty compression-ignition engine with a gasoline-like fuel that has an anti-knock index (AKI) of 58. The primary goal was to design an optimized combustion system utilizing the high volatility and low sooting tendency of the fuel for improved fuel efficiency with minimal hardware modifications to the engine. The CFD model predictions were first validated against experimental results generated using the stock engine hardware. A comprehensive design of experiments (DoE) study was performed at different operating conditions on a world-leading supercomputer, MIRA at Argonne National Laboratory, to accelerate the development of an optimized fuel-efficiency focused design while maintaining the engine-out NOx and soot emissions levels of the baseline production engine.

A Machine Learning-Genetic Algorithm (ML-GA) approach was developed to virtually discover optimum designs using training data generated from multi-dimensional simulations. Machine learning (ML) presents a pathway to transform complex physical processes that occur in a combustion engine into compact informational processes. In the present work, a total of over 2000 sector-mesh computational fluid dynamics (CFD) simulations of a heavy-duty engine were performed. These were run concurrently on a supercomputer to reduce overall turnaround time. The engine being optimized was run on a low-octane (RON70) gasoline fuel under partially premixed compression ignition (PPCI) mode. A total of nine input parameters were varied, and the CFD simulation cases were generated by randomly sampling points from this nine-dimensional input space. These input parameters included fuel injection strategy, injector design, and various in-cylinder flow and thermodynamic conditions at intake valve closure (IVC).

Technical groups worldwide have been actively developing specifications and requirements for biodegradable hydraulic fluids for mobile applications. These groups have recognized that an industry-wide specification is necessary due to the increase in environmental awareness in the agriculture, construction, forestry, and mining industries, and to the increasing number of local regulations primarily throughout Europe. Caterpillar has responded to this need by publishing a requirement, Caterpillar BF-1, that may be used by Caterpillar dealers, customers, and industry to help select high-performance biodegradable hydraulic fluids. This requirement was written with the input of several organizations that are known to be involved with the development of similar types of specifications and requirements.

A previously developed CFD-based optimization tool is utilized to find optimal engine operating conditions with respect to fuel consumption and emissions. The optimization algorithm employed is based on the steepest descent method where an adaptive cost function is minimized along each line search using an effective backtracking strategy. The adaptive cost function is based on the penalty method, where the penalty coefficient is increased after every line search. The parameter space is normalized and, thus, the optimization occurs over the unit cube in higher-dimensional space. The application of this optimization tool is demonstrated for the Sulzer S20, a central-injection, non-road DI diesel engine. The optimization parameters are the start of injection of the two pulses of a split injection system, the duration of each pulse, the exhaust gas recirculation rate, the boost pressure and the compression ratio.

A global optimization method has been developed for an engine simulation code and utilized in the search of optimal fuel injection strategies. This method uses a Lagrange interpolation function which interpolates engine output data generated at the vertices and the intermediate points of the input parameters. This interpolation function is then used to find a global minimum over the entire parameter set, which in turn becomes the starting point of a CFD-based optimization. The CFD optimization is based on a steepest descent method with an adaptive cost function, where the line searches are performed with a fast-converging backtracking algorithm. The adaptive cost function is based on the penalty method, where the penalty coefficient is increased after every line search. The parameter space is normalized and, thus, the optimization occurs over the unit cube in higher-dimensional space.

Optimal fuel injection strategies are obtained with a micro-genetic algorithm and an adaptive gradient method for a nonroad, medium-speed DI diesel engine equipped with a multi-orifice, asynchronous fuel injection system. The gradient optimization utilizes a fast-converging backtracking algorithm and an adaptive cost function which is based on the penalty method, where the penalty coefficient is increased after every line search. The micro-genetic algorithm uses parameter combinations of the best two individuals in each generation until a local convergence is achieved, and then generates a random population to continue the global search. The optimizations have been performed for a two pulse fuel injection strategy where the optimization parameters are the injection timings and the nozzle orifice diameters.

Increasingly, the exchanges of data in complex ECM (Electronic Control Module) systems rely on multiple communication networks across various physical and network layers. This has greatly increased system flexibility and provided an excellent medium to create well-defined exchangeable interfaces between components; however this added flexibility comes with increased network complexity. A system-level approach allows for the optimization of data exchange and network configuration as well as the development of a comprehensive network failure strategy. Many current ECM systems utilize complex multi-network communication strategies to exchange and control data to components. Recently, Caterpillar implemented an HIL (Hardware-In-the-Loop) test system that provides an approach for developing and testing a comprehensive ECM network strategy.

Faced with competitive environments, pressure to lower development costs and aggressive timelines engineers are not only increasingly adopting numerical simulation techniques but are also embracing design optimization schemes to augment their efforts. These techniques not only provide more understanding of the trade-offs but are also capable of proactively guiding the decision making process. However, design optimization and exploration tools have struggled to find complete acceptance and are typically underutilized in many applications; especially in situations where the algorithms have to compete with existing swift decision making processes. In this paper we demonstrate how the type of setup and algorithmic choice can have an influence and make optimization more lucrative in a new product development atmosphere. We also present some results from a design exploration activity, involving linkage and structural development, of an earth moving machine application.

Today's engine and combustion process development is closely related to the intake port layout. Combustion, performance and emissions are coupled to the intensity of turbulence, the quality of mixture formation and the distribution of residual gas, all of which depend on the in-cylinder charge motion, which is mainly determined by the intake port and cylinder head design. Additionally, an increasing level of volumetric efficiency is demanded for a high power output. Most optimization efforts on typical homogeneous charge spark ignition (HCSI) engines have been at low loads because that is all that is required for a vehicle to make it through the FTP cycle. However, due to pumping losses, this is where such engines are least efficient, so it would be good to find strategies to allow the engine to operate at higher loads.

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.

A study of the combined use of split injections, EGR, and flexible boosting was conducted. Statistical optimization of the engine operating parameters was accomplished using a new response surface method. The objective of the study was to demonstrate the emissions and fuel consumption capabilities of a state-of-the-art heavy -duty diesel engine when using split injections, EGR, and flexible boosting over a wide range of engine operating conditions. Previous studies have indicated that multiple injections with EGR can provide substantial simultaneous reductions in emissions of particulate and NOx from heavy-duty diesel engines, but careful optimization of the operating parameters is necessary in order to receive the full benefit of these combustion control techniques. Similarly, boost has been shown to be an important parameter to optimize. During the experiments, an instrumented single-cylinder heavy -duty diesel engine was used.

The NOx reduction performance under lean conditions over γ-alumina was evaluated using a micro-reactor system and a non-thermal plasma-equipped bench test system. Various alumina samples were obtained from alumina manufacturers to assess commercial alumina materials. In addition, γ-alumina samples were synthesized at Caterpillar with a sol-gel technique in order to control alumina properties. The deNOx performances of the alumina samples were compared. The alumina samples were characterized with analytical techniques such as inductively coupled plasma (ICP) emission spectroscopy, temperature programmed desorption (TPD) and surface area measurements (BET) to understand physical and chemical properties. The information derived from these techniques was correlated with the NOx reduction performance to identify key parameters of γ-alumina for optimizing materials for lean-NOx and plasma assisted catalysis.

Most of the history of systems engineering has been focused on processes for engineering a single complex system. However, most large enterprises design, manufacture, operate, sell, or support not one product but multiple product lines of related but varying systems. They seek to optimize time to market, costs of development and production, leverage of intellectual assets, best use of talented human resources, overall competitiveness, overall profitability and productivity. Optimizing globally across multiple product lines does not follow from treating each system family member as an independently engineered system or product. Traditional systems engineering principles can be generalized to apply to families. This article includes a multi-year case study of the actual use of a generic model-based systems engineering methodology for families, Systematica™, across the embedded electronic systems products of one of the world's largest manufacturers of heavy equipment.

A state-of-the-art spray modeling methodology, recently applied to RANS simulations, is presented for LES calculations. Key features of the methodology, such as Adaptive Mesh Refinement (AMR), advanced liquid-gas momentum coupling, and improved distribution of the liquid phase, are described. The ability of this approach to use cell sizes much smaller than the nozzle diameter is demonstrated. Grid convergence of key parameters is verified for non-evaporating and evaporating spray cases using cell sizes down to 1/32 mm. It is shown that for global quantities such as spray penetration, comparing a single LES simulation to experimental data is reasonable, however for local quantities the average of many simulated injections is necessary. Grid settings are recommended that optimize the accuracy/runtime tradeoff for LES-based spray simulations.

With stringent emission regulations, aftertreatment systems with a Diesel Particulate Filter (DPF) and a Selective Catalytic Reduction (SCR) are required for diesel engines to meet PM and NOx emissions. The adoption of aftertreatment increases the back pressure on a typical diesel engine and makes engine calibration a complicated process, requiring thousands of steady state testing points to optimize engine performance. When configuring an engine to meet Tier IV final emission regulations in the USA or corresponding Stage IV emission regulations in Europe, this high back pressure dramatically impacts transient performance. The peak NOx, smoke and exhaust temperature during a diesel engine transient cycle, such as the Non-Road Transient Cycle (NRTC) defined by the US Environmental Protection Agency (EPA), will in turn affect the performance of the aftertreatment system and the tailpipe emissions level.

With the rapid development of artificial intelligence, autonomous driving technology will finally reshape an automotive industry. Although fully autonomous cars are not commercially available to common consumers at this stage, partially autonomous vehicles, which are defined as level 2 and level 3 autonomous vehicles by SAE J3016 standard, are widely tested by automakers and researchers. A typical Human-Machine-Interface (HMI) for a vehicle takes a form to support a human domination role. Although modern driving assistance systems allow vehicles to take over control at certain scenarios, the typical human-machine-interface has not changed dramatically for a long time. With deep learning neural network technologies penetrating into automotive applications, multi-modal communications between a driver and a vehicle can be enabled by a cost-effective solution.

Caterpillar uses heat treatment to enhance the properties of a significant number of parts. Traditional heat treat process optimization is both time consuming and expensive when done by empirical methods. This paper describes a computer simulation of the heat treatment process, developed by Caterpillar, based upon finite element analysis. This approach combines thermal, microstructural, and stress analysis to accurately model material transformation during quenching. Examples are presented to illustrate the program.

The design of a pressure compensated hydraulic valve is optimized using CFD analysis. The valve is used in a hydraulic system to control implement movement. High flow rates through the valve resulted in unacceptably high pressure drops, leading to an effort to optimize the valve design. Redesign of the valve had to be achieved under the constraint of minimal manufacturing cost. The flow path of hydraulic oil through the valve, the spool design, and various components of the valve that caused the high pressure drops were targeted in this analysis. A commercially available CFD package was used for the 3D analysis. The hydraulic oil flow was assumed to be turbulent, isothermal and incompressible. The steady-state results were validated by comparison with experimental data.

A numerical simulation and optimization study was conducted for a medium speed direct injection diesel engine. The engine's operating characteristics were first matched to available experimental data to test the validity of the numerical model. The KIVA-3V ERC CFD code was then modified to allow independent spray events from two rows of nozzle holes. The angular alignment, nozzle hole size, and injection pressure of each set of nozzle holes were optimized using a micro-genetic algorithm. The design fitness criteria were based on a multi-variable merit function with inputs of emissions of soot, NOx, unburned hydrocarbons, and fuel consumption targets. Penalties to the merit function value were used to limit the maximum in-cylinder pressure and the burned gas temperature at exhaust valve opening. The optimization produced a 28.4% decrease in NOx and a 40% decrease in soot from the baseline case, while giving a 3.1% improvement in fuel economy.

Identifying edge cases for designed algorithms is critical for functional safety in autonomous driving deployment. In order to find the feasible boundary of designed algorithms, simulations are heavily used. However, simulations for autonomous driving validation are expensive due to the requirement of visual rendering, physical simulation, and AI agents. In this case, common sampling techniques, such as Monte Carlo Sampling, become computationally expensive due to their sample inefficiency. To improve sample efficiency and minimize the number of simulations, we propose a tailored active learning approach combining the Support Vector Machine (SVM) and the Gaussian Process Regressor (GPR). The SVM learns the feasible boundary iteratively with a new sampling point via active learning. Active Learning is achieved by using the information of the decision boundary of the current SVM and the uncertainty metric calculated by the GPR.