Search Results

Model based computer-aided processes offer an economical and accelerated alternative to traditional build-and-test "Edisonian" approaches in engineering design. Typically, a CAE based design problem is formulated in two parts, viz. (1) the inverse design problem which involves identification of the appropriate geometry with desired properties, and (2) the forward problem which is the prediction of performance from the product geometry. Solution to the forward problem requires development of an accurate model correlated to physical data. This validated model could then be used for Virtual Verification of engineering systems efficiently and for solving the inverse problem. This paper demonstrates the rigorous process of model development, calibration, validation/verification, and use of the calibrated model in the design process with practical examples from automotive chassis and powertrain systems.

Virtual Verification (VV) of engineering designs is a critical enabler in the Product Development (PD) process to reduce the time-to-market in a cost efficient manner. Reliance on cost effective VV methods have significantly increased with increased pressure to meet customer expectations for new products at reduced PD budgets. Computer Aided Engineering (CAE) is one such VV method that affords an engineer to make decisions about the ability of the designs to meet the design criteria even before a prototype is built. The first step of the CAE process is meshing which is a time consuming, manual and laborious process. Also mesh development time and accuracy significantly varies with the (1) component (trim body, engine, suspension, brakes, etc.), (2) features predominantly occurring in the component (welds, ribs, fillets, etc.), meshing guidelines based on which the model needs to be developed (durability, safety, NVH, etc.), and the expertise of the meshing engineer involved.

Aftertreatment system design involves multiple tradeoffs between engine performance, fuel economy, regulatory emission levels, packaging, and cost. Selection of the best design solution (or “architecture”) is often based on an assumption that inherent catalyst activity is unaffected by location within the system. However, this study acknowledges that catalyst activity can be significantly impacted by location in the system as a result of varying thermal exposure, and this in turn can impact the selection of an optimum system architecture. Vehicle experiments with catalysts aged over a range of mild to moderate to severe thermal conditions that accurately reflect select locations on a vehicle were conducted on a chassis dynamometer. The vehicle test data indicated CO and NOx could be minimized with a catalyst placed in an intermediate location.

Engine operating strategies typically geared towards higher fuel economy and lower NOx widely affect exhaust composition and temperature. These exhaust variables critically drive the performance of After Treatment (AT) components, and hence should guide their screening and selection. Towards this end, the effect of H2 level in diesel exhaust on the performance of a Diesel Oxidation Catalyst (DOC) was studied using flow reactor experiments, vehicle emission measurements and mathematical models. Vehicle chassis dynamometer data showed that exhaust from light-duty and heavy-duty diesel trucks contained very little to almost no H2 (FTP average CO/H2 ∼ 40 to 70) as compared to that of a gasoline car exhaust (FTP average CO/H2 ∼ 3). Two identical flow reactor experiments, one with H2 (at CO/H2 ∼ 3) and another with no H2 in the feed were designed to screen DOCs under simulated feed gas conditions that mimicked these two extremes in the exhaust H2 levels.

A typical diesel After-Treatment (AT) system to meet 2010 North American regulations consists of a Diesel Oxidation Catalyst (DOC) for oxidizing Hydrocarbons (HCs) and Carbon Monoxide (CO), followed by a Selective Catalytic Reduction (SCR) catalyst for NOx reduction, and a diesel particulate filter. DOCs are also used to oxidize NO in the exhaust to NO2 that substantially improves low-temperature SCR NOx reduction performance. Vehicle and engine dynamometer data under typical operating conditions show (1) engine-out NO2/NOx can be as high as 40%, (2) DOC can reduce NO2 to NO in the presence of reductants, and (3) an aged DOC with a low CO/HC conversion can be a net consumer of NO2 over a wide range of temperatures and space velocities. Flow reactor experiments simulating diesel exhaust conditions show that DOC-out NO2 is independent of the inlet NO2/NOx in the feed.

Experiments were conducted to determine the effect of accelerated aging on LNT-SCR system performance. The systems were aged and evaluated on an engine dynamometer using a DW12 2.2L I4 Diesel. The aging consisted of three modes, including sulfur exposure, desulfation (deSOx), and DPF regeneration conditions. Two systems were aged using different desulfation protocols and then evaluated over steady-state cycling conditions. The sulfur release characteristics were measured. Overall system gross NOx conversion was robust to the aging protocol; however, there were significant differences in the individual component performance depending on the nature of the aging conditions. LNT NOx conversion was significantly lower for the case where maximum temperatures reached only 730°C, as opposed to the more severe aging where temperatures reached 800°C.

An easy-to-use implementation of a Differential Evolution Based Stochastic Optimizer (DEBSO) for nonlinear, multi-modal problems is presented. Using two case studies, we demonstrate that DEBSO is (1) more effective and (2) less sensitive to user defined initial guess values, in finding the global optimum, as compared to that of a gradient based deterministic optimizer. Results from using DEBSO for construction of empirical catalyst maps from pulsator data and estimation of parameters in a diesel oxidation catalyst model are also presented. The effectiveness and efficiency of DEBSO has been compared to other evolution-based optimizers in Appendix A.

This paper presents a hybrid approach for developing a robust model of a diesel oxidation catalyst (DOC). Information from multiple sources including detailed thermal balances, laboratory performance data, phenomenological description of adsorption and desorption in catalyst pores, and experience based correlations are seamlessly integrated using optimization and statistical tools to create an easy-to-use, computationally inexpensive predictive model. Light-off, Light-out, and fuel quench data from a diesel pulsator and engine dynamometer are used for model calibration. The calibrated model predicts cumulative HC and CO tailpipe vehicle emissions as well as DOC NOx outlet composition (NO vs. NO2).