“Simulation powers innovation,” stated Uday Korde, Engineering Group Manager, Powertrain CAE Methods for General Motors, speaking at the recent Dassault Systèmes Regional User Meeting for users of SIMULIA engineering CAE tools. He noted that over the last ten years, GM Powertrain has dramatically changed how it employs CAE. Where once using CAE was a forensic afterthought, it is now forefront in their design cycle.
Much of this is forced by the modern challenges of powertrains. To meet stringent regulations on fuel economy and emissions, engineers are turning to more efficient combustion, various forms of boost, cylinder deactivation, thermal management, stop/start, advanced transmissions, and aftertreatment—plus varying degrees of electrification. With such complexity comes the difficulty in managing all design variables by laboriously building numerous prototypes. “It's just impossible to play with hardware and come up with innovative designs,” he said.
Today’s success, foundations of process
Nothing describes success as well as numbers. Korde reports that the number of analysis outputs from GM’s powertrain CAE is growing more than 10% per year. He described a disciplined process behind this success, featuring a predetermined catalog of standardized CAE work flows. “Inputs and outputs are well defined, along with the toolset we used and how we use it,” he said. “Before any program is launched, we actually put a catalog together, identifying the risk items and plan the work ahead of time,” he elaborated.
A key outcome of standardizing work is it enables process automation. He provided a number of examples at the user’s meeting, featuring simulation tools from Dassault Systèmes as well as others, such as GT Power from Gamma Technologies. He showed some examples of process automation illustrating the power of ‘analysis led design.’
Using automated analysis, an advanced intake port was designed that maximized the tumble and swirl of air going into the cylinder as well as the mixture preparation inside it. “This is something we could never have been able to do without automation,” he said.
As a result, over the last decade he reported that this process has significantly reduced the amount of expensive prototype hardware built for development. This decrease dovetails with the growth in high-performance computing infrastructure available to calculate all of that analysis (see figure).
This is not to say GM has eliminated any hardware testing or plans to in the future. “As we assess the risk of new technology, we do recognize that we need hardware—to make sure that we are not overconfident,” admitted Korde. However, CAE tools combined with mathematical optimization and automation means GM can explore many more potential designs. “It is just impossible to play with hardware [alone] and come up with innovative designs,” he said.
Despite the gains they have made in efficiently designing 21st Century powertrains, there is more potential as software gets better and computers get more powerful. He listed a number of specific improvements for which they will next concentrate on development.
First is a process to sort concepts. “We want to be able to analyze quickly and explore different options. Not necessarily at this stage to be able to exactly correlate to physical test, but enough that we can distinguish a good design from a bad design,” he said.
Second, is a process to design to target by using simulation to avoid over-design, saving weight and expense. He claims the default today is to over design, which is a form of waste.
The third is creating robust designs, which are designs that are less sensitive to noise or uncontrolled inputs—a challenging topic, but vital for future competitiveness and cost. In particular, he points to the use of Design for Six Sigma techniques used in conjunction with simulation for best effect.
The fourth is simultaneous co-simulation. Many CAE simulations are built on different disciplines such as heat transfer (thermodynamics) and stress analysis (solid mechanics). However, heat-induced stress is a major concern in engine blocks, for example. In the real world, heat and stress occur simultaneously. For best results, they should be calculated together as well, but this requires specific and new techniques for combining them. This is in tandem with future development of large-system simulations, moving from simulating a single cylinder to an engine, to a powertrain, to an entire automobile.
“This is all part of evidence-based decision making that creates business value,” Korde explained. “The biggest effort has been to educate the non-CAE engineering community. Whenever we have demonstrated how well simulation correlates with physical data, the next step is realizing how much easier it is to create a design.”