How computer modelling could save your life
2nd Jul 2013 | 11:00
And save car manufacturers a lot of time
Cars have been living in a virtual world for the past few decades. Computer simulations of crash tests, stylistic changes that help us visualise a new sports car and models that analyse how much fuel you will consume on your next road trip have helped speed up the prototyping process.
Yet recent advancements in data processing, storage architectures and even a new-found faith in the merits of virtual testing have helped car makers even more.
For some, building the car of tomorrow is getting easier, and the results are safer and more reliable.
GM virtually crash tests parts in a Chevy Impala
Computer models have existed for years that show how a vehicle will deform in an accident, what will happen to the occupants, and how the airbags will deploy. In recent years, GM and other car manufacturers have created even more sophisticated models.
Mick Scherba, senior manager of global computer animated engineering strategy at General Motors, says computer crash test models are improving quickly.
"We've been able to model everyone from infant to three up to 10, 50th percentile male, from front-side-angle, rear - all of those. We've developed sophisticated occupant models that have not just the size and geometry of occupants but also the internal organs. More recently, we've worked on computer models of an occupant that flexes like a real human. As the head hits the windshield, we'd get the load imparted and look at the internal organs," he says.
The idea is to understand very specific injuries and change the vehicle accordingly. One example he gave is a model that shows what happens when the hood of a car deploys in an accident. Of course, the auto industry is moving toward lighter, stronger vehicles such as fibre-reinforced composite materials. Each material deforms differently in a model.
"If you had a hood made out of traditional steel it deforms one way, but carbon-fibre deforms another way. You've got to have a model for each," Scherba says.
Multi discipline optimisation (MDO) is the latest frontier in computer modelling for crash tests. For a vehicle like the 2014 Chevrolet Impala, GM was able to model very specific scenarios, such as a side-impact at 30mph.
But it went much further: each specific part inside the car was modelled as well, such as a rocker inner assembly or a roof-side rail. The optimisation adjusts the gauges of each part, running thousands and thousands of models.
"You end up with a vehicle that has the lightest gauges possible in all of the structural members but still meets the structural requirements for all of those load cases," says Scherba.
Amazingly, the physical tests for these individual tests would take decades to build per car for each crash test scenario. "This looks at every load-carrying part and does take several days. We run one iteration of an MDO now over several weeks," he says.
Audi models the LED lights in an R8
Lighting design is one of the most interesting advancements in car technology. With a sleek sports car such as the Audi R8, your eye is immediately drawn to a bank of LED headlights, but they also play an important role in reducing accidents - say, warning other drivers around you about a crash, emanating brighter when you brake, or casting a wide glow over the road.
Johannes Scheuchenpflug, the coordinator of simulations in the Audi department handling light functions and innovations in Germany, says the Audi engineering team works closely with their design teams to run simulations of lights in new vehicles.
"We use CAD software such as Catia V5, and software to calculate and visualise optical surfaces, such as Optis Speos and Brandenburg LucidShape," he says. Interestingly, the models - run on a cluster of computers and graphics cards - must analyse everything from wind direction and velocity to how the parts inside the headlamp will deform in a crash.
"Safety of course is of crucial importance, to achieve good pedestrian protection," he says. "That requires placing and mounting all rigid parts inside the headlamp in a way that minimises the negative influence on a pedestrian in case of a crash. To accomplish this goal, structural analysis calculations are required, also using computer clusters."
Subtle changes in the computer models enable designers to see how the lighting will look in the final production car: the brightness levels, how much ambient light is created, how lights work when there is an oncoming car or an empty road.
Upcoming Audi vehicles will even use a technology called Matrix Beam that can automatically lower the beam of light so the road is illuminated but the oncoming driver is not blinded.
The result is an eye-catching final design that's fully optimised within a computer model for safety and aesthetics long before the vehicle ever reaches a physical production stage.
Toyota speeds up the Rav4 EV development
In developing any new vehicle, automobile manufacturers look at both the entire vehicle design progression and, at the same time, constantly monitor individual components. Each go hand-in-hand: an over-focus on the final outcome or the component design can lead to problems.
"Computer modelling enables innovations such as the use of new materials and processes by thinking on the vehicle level, doing component tests, then going back to the vehicle level," says Scott Miller, the general manager of vehicle performance development - safety and crashworthiness at the Toyota Technical Center in Michigan, USA.
For example, with the 2013 Rav4 EV, Toyota was able to simulate battery components within the vehicle long before they ever reached the physical prototype stage. Yet it could also model how a component would work in the entire vehicle design.
"In the case of the battery protection energy absorption systems, sub-system component models were built for various local phenomena to set deformations and load targets on a part-by-part basis," he says.
"The result was a nearly optimised section geometry and attachment scheme prior to any vehicle testing. The design from these iterations were then confirmed in full vehicle simulation with minor changes before building initial prototypes. The initial round of physical testing was very successful, with only minor follow-up items."
Miller says computer models had a dramatic effect. Prototyping usually takes about 10 weeks, and test preparation and test analysis for new designs usually each take about two weeks. With the Toyota Rav4 EV, the design feedback cycle went from a 14-week total down to just two weeks, which includes 2-3 days of prototyping, 2-3 days of test prep and one week of analysis.
The implication of all this is that as computer processing speeds improve and software designs become more realistic, the total development time for new cars will reduce even more. That means we should see fresher designs as opposed to sticking with staid vehicles that have worn out their welcome.