When I cook, it’s typically Indian cuisine and I use about 30 ingredients. I recently got into minimalist baking, and while it’s relatively straightforward to reduce a nine-ingredient macaroon to five, the question I struggled with was can you do that with Indian food with its huge ingredient list and still keep the complicated flavor profile intact? The answer is yes, and the key is to pick the right five ingredients out of the 30. The bigger the cut, the harder it is to identify what’s actually important.
Luckily, with mechanisms, CONVERGE CFD does that for you.
We took the Lawrence Livermore National Laboratory diesel combustion mechanism (2885 species) and reduced it three times with our mechanism reduction tool to produce a mechanism with 83 species. We then ran zero-dimensional ignition delay simulations under lean, stoichiometric, and rich conditions. The plots below show that while a reduction to 931 species does not significantly change the ignition delay time, the dramatic reduction to 83 species does.
I’ve previously talked about the value of the mechanism reduction tool and dynamic mechanism reduction. The cost that you must mitigate—whether you develop the mechanism, as Argonne National Laboratory did (54 species1) or automatically reduce it with our mechanism reduction tool or our Dynamic Mechanism Reduction capability—is that each time you reduce a mechanism, the odds are that you’ve introduced an error. With each reduction, the results of a simulation will drift further away from the results of a simulation with the original mechanism. On the other hand, a simulation with a small mechanism is much faster to run.
Enter mechanism tuning. The mechanism tune tool first analyzes sensitivity analysis associated with ignition delay or laminar flamespeed data to identify the reactions in the mechanism that most impact your chosen targets. If you don’t have ignition delay or laminar flamespeed sensitivity analysis data, mechanism tuning can obtain these data from the 0D and/or 1D solvers.
Almost 20 years ago, Convergent Science co-founder Kelly Senecal spearheaded the use of genetic algorithm (GA) optimization in the design of internal combustion engines. In its mechanism tune tool, CONVERGE 2.4 leverages the possibilities inherent in GA optimization to build a number of different mechanisms based on modified reaction A-factors for the most sensitive reactions in a mechanism. Since setting up a GA run can be tedious, the mechanism tune tool sets up all the input files for you so that you can simply run the GA.
In the plots above, we show the impact of mechanism tuning on ignition delay. We used the mechanism tune tool to optimize the 83-species reduced mechanism against the original LLNL diesel mechanism’s ignition delay data. We show data from two simulations run with mechanisms that have A-factors modified by the mechanism tune tool. You can see that the ignition delay from these optimized mechanisms matches the original LLNL value.
With the mechanism tune tool, you have the flexibility to optimize your mechanism to match ignition delay targets, or laminar flamespeed targets, or both. You can tune to, say, a set of laminar flamespeeds while simultaneously ensuring that the mechanism’s ignition delay isn’t altered. You can base this on three-dimensional simulations assuming you have the computational time to run the 3D simulations for mechanisms produced by GA to identify the best results (watch for an upcoming journal publication with this study).
If you have a large mechanism and need both efficient runs and robust results, give our mechanism tune tool a whirl. At the end of the day, the proof is in the pudding.