W. Kendal Bushe

Predicting the ignition delay of turbulent methane jets using Conditional Source-term Estimation

A predictive simulation of the autoignition process of non-premixed methane in a turbulent jet configuration was performed. Closure for the chemical source-term was obtained using Conditional Source-term Estimation with Laminar Flamelet Decomposition (CSE-LFD). The ambient oxidizer conditions – the high pressure and moderate temperatures characteristic of compression ignition engines – were chosen with the intent to validate the combustion model used under engine-relevant conditions. Validation was obtained by comparison of the predicted ignition delay to experimental results obtained from a shock-tube facility at several initial temperatures. Overall, the combination of full chemistry that has been carefully tuned to predict autoignition of premixed methane–air mixtures under similar temperature/pressure conditions with the CSE-LFD model is able to successfully predict the autoignition delay time of methane–air jets well within the scatter in the experimental data.

Laminar flamelet decomposition for conditional source-term estimation

A new decomposition approach to conditional source-term estimation (CSE) is proposed and discussed. The new approach is tested in the a priori sense using direct numerical simulations (DNS). It is found that—where CSE had previously been found to provide closure for chemical source-terms with arbitrary chemistry in the large eddy simulation paradigm—it can provide this closure in the Reynolds averaged Navier–Stokes paradigm as well. Using the proposed decomposition improves the predictions of CSE considerably. Only the assumptions that gradients in conditional averages are small and that the probability density function of mixture fraction can be adequately approximated using a presumed functional form are needed. The computational cost of the new laminar flamelet decomposition approach to CSE is also substantially lower than that of the original approach.

Emissions variability in gaseous fuel direct injection compression ignition combustion

Measurements of ignition characteristics and emissions were made in a shock tube facility operating at engine-relevant conditions. Methane and methane/ethane fuels were injected down the centerline of the shock tube using an electronically controlled prototype gaseous fuel injector developed by Westport Innovations. Air was preheated and compressed using a reflected shock technique that produced run times of 4-5 ms. Particulate matter (PM) emissions were found to be highly intermittent. In only 6 out of 97 experiments was PM detected above background levels. In all of these 6 sooting experiments ignition kernels were located relatively close to the injector tip and ignition occurred prior to the end of fuel injection. Using the large orifice injector tip with pure methane fuel, PM was detected in 4 out of 28 experiments; using the small orifice with pure methane fuel, no PM was detected in any of 50 experiments. When 10% ethane was added to the fuel, PM was detected in 2 out of 20 experiments. Orifice size was found to significantly influence emissions of nitrogen oxides (NO x ). Emissions were reduced by over an order of magnitude when the orifice size was reduced by a factor of 4. Run-to-run variability in NO x emissions was relatively high at nominally constant operating conditions. NO x variability was associated with variability in ignition timing. At fixed operating conditions, NO x emissions were found to increase significantly with advancing ignition timing. Over the range of conditions studied, NO x emissions were found to be relatively insensitive to injection pressure ratio and injection duration.

A mesh partitioning algorithm for preserving spatial locality in arbitrary geometries

An algorithm for partitioning computational meshes is proposed.The Morton order space-filling curve is modified to achieve improved locality.A spatial locality metric is defined to compare results with existing approaches.Results indicate improved performance of the algorithm in complex geometries. A space-filling curve (SFC) is a proximity preserving linear mapping of any multi-dimensional space and is widely used as a clustering tool. Equi-sized partitioning of an SFC ignores the loss in clustering quality that occurs due to inaccuracies in the mapping. Often, this results in poor locality within partitions, especially for the conceptually simple, Morton order curves. We present a heuristic that improves partition locality in arbitrary geometries by slicing a Morton order curve at points where spatial locality is sacrificed. In addition, we develop algorithms that evenly distribute points to the extent possible while maintaining spatial locality. A metric is defined to estimate relative inter-partition contact as an indicator of communication in parallel computing architectures. Domain partitioning tests have been conducted on geometries relevant to turbulent reactive flow simulations. The results obtained highlight the performance of our method as an unsupervised and computationally inexpensive domain partitioning tool.

Autoignition and Emission Characteristics of Gaseous Fuel Direct Injection Compression Ignition Combustion

Heavy-duty natural gas engines offer air pollution and energy diversity benefits. However, current homogeneous-charge lean-burn engines suffer from impaired efficiency and high unburned fuel emissions. Natural gas direct-injection engines offer the potential of diesel-like efficiencies, but require further research. To improve understanding of the autoignition and emission characteristics of natural gas direct-injection compressionignition combustion, the effects of key operating parameters (including injection pressure, injection duration, and pre-combustion temperature) and gaseous fuel composition (including the effects of ethane, hydrogen and nitrogen addition) were studied. An experimental investigation was carried out on a shock tube facility. Ignition delay, ignition kernel location, and NOx emissions were measured. The results indicated that the addition of ethane to the fuel resulted in a decrease in ignition delay and a significant increase in NOx emissions. The addition of hydrogen to the fuel resulted in a decrease in ignition delay and a significant decrease in NOx emissions. Diluting the fuel with nitrogen resulted in an increase in ignition delay and a significant decrease in NOx emissions. Increasing pre-combustion temperature resulted in a significant reduction in ignition delay, and a significant increase in NOx emissions. Modest increase in injection pressure reduced the ignition delay; increasing injection pressure resulted in higher NOx emissions. The effects of ethane, hydrogen, and nitrogen addition on the ignition delay of methane were also successfully predicted by FlameMaster simulation. OH radical distribution in the flame was visualized utilizing Planar Laser Induced Fluorescence (PLIF). Single-shot OH-PLIF images revealed the stochastic nature of the autoignition process of non-premixed methane jets. Examination of the convergence of the ensemble-averaged OH-PLIF images showed that increasing the number of repeat experiments was the most effective way to achieve a more converged result. A combustion model, which incorporated the Conditional Source-term Estimation (CSE) method for the closure of the chemical source term and the Trajectory Generated Low-Dimensional Manifold (TGLDM) method for the reduction of detailed chemistry, was applied to predict the OH distribution in a combusting non-premixed methane jet. The model failed to predict the OH distribution as indicated by the ensemble-averaged OH-PLIF images, since it cannot account for fluctuations in either turbulence or chemistry.

Conditional Moment Closure Based on Two Conditioning Variables

A conditional moment closure approach for modelling turbulent combustion is proposed, based on two conditioning variables. The two conditioning variables used here are mixture fraction and a second conserved scalar a, which is initialized perpendicular to the mixture fraction, such that the two conditioning variables constitute a plane. With this we hope to capture important physical features of turbulent reacting flows which the single conditioning variable approach cannot, including local ignition, extinction and re-ignition due to small scale strain fluctuations. We propose to model the stress tensor with a stochastic process to account for chaotic turbulent fluctuations.