As automotive engine designers strive to meet tightened regulations for both fuel economy and emissions, they are increasingly using simulation to understand the subtle impacts of changes in the many elements of burning fossil fuels. Even though modeling and simulation tools still have some gaps, they are letting engineers try many innovative techniques without spending money on prototypes.
As virtual design technologies and techniques get better, engine developers can change far more parameters as the search for ways to meet conflicting rules for fuel consumption and emissions. More powerful computers and improved software also let design engineers try out innovative solutions that might not have been viable a few years ago.
“In the past, a new engine was often based on the last model,” said Mike Anderson, Director of Engine Development and Validation for General Motors. “Now we start with combustion, doing everything with modeling. We can look at great ideas we might not have explored before.”
Simulation makes it possible to test new control concepts before engines are available. That lets developers expand their experiments, altering and testing engine configurations and fuel distribution scenarios without the typical constraints of physical systems. These simulations are used extensively throughout the development cycle of engine control software to validate strategy, calibrations, and diagnostics.
“The usage of simulation starts early, on the developer’s PC, with Model-In-the-Loop (MIL) testing, where algorithms in pure model form are developed and validated against the engine model,” said Santhosh Jogi, Engineering Director for dSpace Inc. “This leads into Software-In-the-Loop (SIL) simulation. These steps address validating functional behavior of software independently from hardware.”
As engines become more complex, it is far harder to optimize performance. Software, electronic controls, and mechanical components all have many parameters that must be understood on their own.
Once this knowledge is gained, engineers must see how the many design elements interact when they are combined into a full system. Tool suppliers and users agree that as the number of variables goes up, the interplay between components and systems increases.
“When you add widgets like cam phasing, you have bigger control issues, so you need to simulate the engine plus the controls,” said Pete Maloney, Principal Consulting Engineer for The MathWorks. “If you do one without the other, it’s not that beneficial.”
Engine designers can often get results from many simulations in less time than it would take to run a single real-world test on a prototype engine. This speed-up is even more impressive when the time needed to build a physical prototype is factored in.
Simulations are also becoming more accessible, no longer requiring experts. Improvements in software and human-machine interfaces mean that the tests can now be run by a range of engineers. These improvements are augmented by advances in computing power that make it possible to run tests that used to take days or weeks in an afternoon or overnight.
“With simulation, you can push a button and in a matter of hours you know whether you’ve got better performance,” said Radu Theyyunni, Engineering Group Manager of Systems Design and CFD/Performance Analysis at GM.
The rise in computing power lets developers combine the many elements of an engine, even extending to the full vehicle. But even the fastest computer slows down as the size of the simulation rises. Many observers say that it is often wise to run quick tests of the components by themselves rather than to combine, say, the full engine and transmission.
“One trap people fall into is wanting to over-simulate and put everything in every test,” said Greg Brown, Product Manager at PTC. “If you manage the number of abstractions, you can do better simulations in less time.”
Though hardware and software improvements let engineers do all sorts of simulations before prototypes are built, today’s tools still leave engineers wanting more. That’s especially true on the emissions side, where analysis tools have taken more time to evolve.
Nonetheless, simulations still provide plenty of help. Even tools that are a bit flawed help engineers determine which design path they should take as they strive to reduce emissions.
“The models act like a compass, guiding people in the right direction but not telling them exactly where we are,” The MathWorks’ Maloney said. “The emissions models are generally not very good; we don’t have good enough models for the chemical processes. We know what’s going on outside the cylinder, but it’s not as clear inside the cylinder.”
The tools inside the cylinder are better, but they are also still evolving. GM’s Anderson noted that there are still differences between what is provided by simulations and what is gathered during dynamometer tests.
Fine points such as fuel burning rates and the center of the explosion are still fuzzy. “The data doesn’t necessarily match what’s on the dyno, but simulations provide a lot of data that we can’t get with a dyno,” Anderson said.
Though engineers always want more from development tools, they also realize that every time tests are run, real-world data can be used to improve models. The benefits of modeling and simulation are so significant that these tools are being used throughout the auto industry.
“Migrating to simulations lets us find solutions in a way that’s cheaper, faster, and better. We’re able to do analysis in far less time, getting to the right answer the first time so there’s a lot less rework later on,” Anderson said. “We can put the money into dynos and people instead of building equipment that would end up on the scrap heap.”