Reaction Design announced the release of its FORTÉ CFD package in time for the 2011 SAE World Congress, slated for April 12-14 in Detroit. It is designed for 3-D modeling of fuel effects in internal-combustion engines. Incorporating advanced fuel spray models and high-fidelity fuel chemistry models, the package boasts accurate results in hours—one to two orders of magnitude faster than current CFD tools, according to the company.
“Chemistry simulations, embedded in an overall solution that uses the Navier-Stokes equations [for fluid flow], is a big bottleneck in developing truly predictive engine simulations,” said Bernie Rosenthal, CEO for Reaction Design.
The FORTÉ CFD package builds on Reaction Design’s existing chemistry simulation tools: CHEMKIN and CHEMKIN-CFD/API. The company developed FORTÉ CFD in response to issues with current CFD tools for simulating engine combustion.
“Developers felt they spent too much time generating CFD meshes, compute times were prohibitively long, and in some cases emissions models gave wrong results and could not predict trends accurately,” Rosenthal said.
To compensate for accuracy shortfalls, engineers needed to calibrate models to match real data. “They literally needed to build [their experimental] engine before they could build the model of how that engine performed,” Rosenthal said. This is because many current combustion models do not predict critical combustion behaviors such as ignition, flame propagation, and emissions, according to the company.
They are constrained by computational limits to using too few chemical species.
“Typically, they use about a hundred or fewer, which can be as much as five times fewer than are required to capture accurate trends,” said Rosenthal.
He noted that new technical challenges are stressing simulation capabilities even further. New engine designs using high-efficiency, low-emissions designs are emerging and may need better modeling of chemical kinetics. These include dual-fuel engines, staged spray injections, homogeneous-charge compression ignition, or premixed charge compression ignition.
There are also new fuel additives and customized biofuels. “We felt that effective simulation must be efficient enough to run on 8 to 32 standard CPUs in less than 15 hours,” Rosenthal said. “Engineers typically don’t have access to supercomputer farms. However, they need answers overnight to keep their projects running.”
Reliability of the runs is vital as well. This means runs have to be completed unattended without crashing or failing to converge to a solution.
To increase the computational speed of these chemistry equations, engineers at Reaction Design developed and combined three different approaches. First, advanced numerical algorithms scaled back the computational complexity of the chemistry problem from N3 to N, where N is the modeled number of chemical species. The new meshing algorithm coupled with reformulated numerical algorithms probably contributes to this. The mesh appears based on boundary-fitted coordinate algorithms and captures fine detail near important parts of the model, such as the intake port.
Second, it judges on the fly. Calculations of different species will actually affect the parameters measured in that slice of time. If they have no effect, then FORTÉ does not calculate them.
The third method takes advantage of the chemical similarity of groups of cells, or clusters. It calculates such clusters only once.
Together, these new techniques reduce simulation run times by up to two orders of magnitude, according to the company.
“We can easily compute fuel models with over 400 species—a level of detail most accurate fuel models need,” said Rosenthal. Years of research by the Model Fuels Consortium validated the mechanisms for real fuel chemistry in FORTÉ CFD, according to the company.
Spray modeling is also addressed. Through a joint venture between Reaction Design and Wisconsin Engine Research Consultants, FORTÉ CFD integrates unique “grid independent” spray models that reduce the amount of calibration required for predicting engine performance, according to the company. These high-fidelity spray models also allow real fuel analysis by matching multicomponent physical properties with a multicomponent chemistry model.
“This removes the burden from the designer to build a mesh of any particular size,” said Rosenthal.
If this new model delivers on the speed as advertised, it offers engine developers a tool that will extend rather than depend on experiment. It allows engineers to focus on other problems that combine CFD and combustion simulation, such as turbulence models, according to Rosenthal. He also pointed out that the company customized FORTÉ CFD for engine designers, with guided setup, run, and visualization screens.
A further challenge is making sense of all the data. “Engineers need visualization tools to compare simulation results to experimental data,” Rosenthal said. “With existing tools, there was not an easy way [to export or] import that data to compare predictions to experimental data.”