<<BACK TO HOME

Replacing dummies with FE models of humans will help engineers prevent more injuries and improve comfort, but there’s a lot of research to do first. By Tristan Honeywill

A car hits a barrier, propelling the driver forward so his ribcage hits the steering wheel. The force of the impact pushes the ribs into his liver, which in turn bumps against his abdominal aorta. Working out whether this knock-on effect is likely to rupture the aorta is not usually left to engineers. Crash testing deals with dummies and acceleration values on key body parts – blood and guts aren’t in the equation.

Human body modelling could change this. As finite element (FE) models become more refined, it will become easier to mitigate serious injuries and the number of prototypes crashing into walls should also decrease. Pedestrian safety, ingress and egress ergonomics, seating and ride comfort will benefit too.

“Dummies give a rough idea of the risk of head injuries for regulatory or consumer tests, but can’t really predict injuries,” says Jean-Pierre Verriest, head of Inrets, the French national transport safety research laboratory. Verriest has led European research into a usable human model. “FE can analyse loads on the head and address mechanisms such as skull fractures and the hematomas and haemorrhages they cause in road accidents,” he says.

Building the models is not the work of an afternoon, however. Gathering data on the human body is hard. Experiments on live humans are limited to low-energy situations and yield information only on the body’s basic mechanics.

Geometric data for FE models can come from medical scans, but the mechanical properties of the materials and interfaces between organs are also needed. This leaves experiments on cadavers, but because the subjects tend to be elderly, it is necessary to make allowances for bone strength, blood pressure and circulation.

Toyota was responsible for the first usable human body model, Thums, commercially available via FE firm Dynamore. Thums represents a seated 50 percentile American adult male. It models the skeleton with bones consisting of solid and shell elements. Ligaments are modelled using shell or beam elements.

With around 60,000 nodes and 80,000 elements, Thums is on a par with a structural component in a car. It correlates well with front and side impact dummy tests, but the lung and heart are treated as one entity, as are the abdominal organs, so it’s not detailed enough to make complex decisions about injuries.

The European Humos model brings together more research, but is less robust. Last year’s Humos 2 corrected some geometry and mesh problems. It also produced a standing human, models of various sizes and postures, and a female version.

The model is available via three code editors: ESI of France, Altair of the US and Madymo of the Netherlands. VW, BMW, PSA, Renault and Mercedes-Benz are among the OEMs evaluating Humos, using it in parallel with dummy models to see what kind of advantages they can gain and giving feedback to developers.

Aprosys, another EU project will transfer Humos into a proper industrial tool with guidelines and templates. It will be ready in March 2008, says Verriest.

Efforts to coordinate international research are just beginning. A consortium of mainly US OEMs issued its first call for research proposals in March. The General Human Body Modelling Consortium comprises DaimlerChrysler, Ford, GM, Honda, Hyundai, Nissan, PSA, Renault, Toyota, and suppliers Takata and TRW.

“Working together won’t just speed up development and cut cost,” says Bing Deng, a researcher in GM’s Occupant and Pedestrian Safety Systems group. “We’ll prevent conflicting solutions, do more basic research on the models, improve their fidelity to real crashes and expand the range of collision and occupant conditions.” Modelling children is one of its more ambitious aims: the data is far harder to acquire than for adults.

“Ultimately, we’ll be able to perform certification to injury criteria using simulations rather than physical testing,” says Deng. Virtual type approval may be some time off, but wider use of human body models will also help develop more realistic crash dummies in the meantime.

The OEMs decided not to invite the code editors into the consortium because of intellectual property rights issues, but the software firms are focusing on other areas of human modelling. ESI is developing multi-body and FE human models to address vehicle comfort issues instead. Models such as Casimir can show the pressure on muscles and skeleton caused by different seating designs and materials.

“Dummies exist, but are complex to use,” says ESI scientific director Eberhard Haug. “Simulation lets you look at seating comfort and even optimise ride comfort. Driver fatigue is increasingly important, especially in the truck industry.”

Comfort CAE is still in its infancy, but Karl Siebertz, a biomechanics expert at Ford’s Aachen Research Centre in Germany, says the long-term value could be enormous.

“There’s still a long way to go until subjective evaluations can be substituted,” he says, “but CAE comfort evaluation is a key enabler for shorter product cycles and prototype reduction.”

Several things need to happen before it can play a full part in series development. Analysis of human motion has so far been confined to sports science and physiotherapy. It is unusual in the car industry. “Tracking a full body’s motion inside a car is far harder than just analysing a person’s regular gait,” says Siebertz.

Ergonomic engineering has the advantage that it often doesn’t require a detailed 1:1 model. Fracture modelling is not really relevant; normal muscle movement is needed. “Crash models would be far too slow to perform calculations,” says Siebertz.

Ford is working with the AnyBody CarDriver model, which includes some 600 muscles. Whereas crash models use forward dynamics to calculate the movement from a given force situation, AnyBody uses inverse dynamics to work out the forces required to perform tasks, such as operating the vehicle or ingress or egress. Ergonomics is increasingly important as populations age and arthritis starts to be part of the buying decision.

But it has its limitations. “Nobody can tell you how long it takes to get a back injury,” says Arne Kiis, AnyBody’s business development manager. “You must turn it into an ergonomic design problem. Instead of knowing whether 700N in your lumbar spine is a problem, you design a seat that lowers the forces in the joints.”

AnyBody aims to come up with guidelines and limits for different joints. But that’s 10 or 15 years away. The model’s maturity is akin to finite element analysis 20 years ago, says Kiis.

Next to building the “perfect” human model, one challenge will be the acceptance of human models by the industry. It may hesitate to use them for designing new cars, because legislation prescribes crash dummies in physical tests. “If not used, the models will not be developed fast enough to be good enough to be used,” says Haug. “But an OEM using the models, could have a marketing advantage. An intermediate step might be for EuroNCAP to award extra stars when an OEM uses human models.”

Standardisation will be important too. At present, there are different dummies for different sizes, impacts and markets.

Being able to have one scalable model for use in every situation would be an important step in achieving proper global vehicle development. ISO standards will be crucial for techniques and evaluation criteria.