Better models need better data

4 mins read

Tom Shelley explains how advances in physical understanding are making the modelling of mainstream products much more realistic

Better understanding of what occurs during deformation and failure processes, and how a materials history affects its mechanical properties, can improve the design of products if the information is built into advanced finite element analysis software. Advances in computing hardware means that models are no longer limited by number crunching process power, but by a lack of correct data and understanding of physical assumptions on which they are based. Getting the understanding right, gets the modelling results right, which as well as improving designs for aerospace and automotive applications, also advances the design of relatively humble consumer products. These were the conclusions drawn from the Virtual Product Development Conference held by MSC Software in Munich earlier this year. As well as the vendors explaining the latest enhancements to software, it provided an opportunity for users to explain some of the things they had done, which were in some cases a revelation to the vendors. Dr Reza Sadeghi, chief technical officer at MSC looks at problems that are not immediately solvable by the standard commercial software. He takes the view that users should come forward with problems and if MSC knows a researcher who can provide a solution, and if the application shows commercial potential, he would consider developing a new function for its standard products. Dr Martin Stumvoll of the Austrian company, Blum, makes kitchen equipment. He says his main concern was modelling hinges for cabinets. He says he had great difficulty in modelling the contacts. "That at the moment is my biggest problem," he says. "The software is hard to use. You need technical support and engineering understanding of what is going on to get accurate results." He used a number of different packages including MD Nastran, Marc and Adams and was just introducing SimXpert. He added: "The bottom line benefit is longer lasting hinges. Development time is shorter, but they still don't get cheaper." In response to this problem Dr Sadeghi replies: "There has recently been a lot of work on improving the modelling of assembly connectors, especially spot and seam welds, adhesive joints, and contacts involving large rotations." Other advances he mentioned include the ability to model flexible body contacts, such as a tyre contact on the road and multi modal optimisations that take into account the effects of possible variation in material properties and dimensions. Another capability that the team is proud of is the ability to model liquid sloshing around inside a container, which is of considerable interest to designers of marine and road tankers, especially those used by the military where it is required to be able to drive over rough ground at speed. Dr Manfred Achenbach, manager of analytical service/FEA for Parker Hannifin in Germany, says he was concerned with the design of piston seals for Bosch electro pneumatic hammers and dampers. Friction effects are significant, especially when they affect the motion of flying pistons inside the hammers. Stiction is well known, but the behaviour of the rubber seals is also affected by speed of deformation and what has previously happened to them. Parker Hannifin engineers have come up with a new 'Friction law' that takes these effects into account, and taking advantage of the ability to insert user subroutines into Marc, Parker has been able to greatly improve its modelling and the performance of its hammers and dampers. MSC says that discovering new laws that more accurately represent the behaviour of materials was hard to do, but as many of our readers will be aware, these may be quite easily revealed by experiments by trained mechanical engineers. The methods to discover them are fairly standard, and where relationships are not obvious, computer based statistical methods are nowadays readily available to uncover them. Another example of where models have been improved by better physical understanding and representation comes from Javier Romo, a project engineer with the Cidaut at Valledolid, based in Spain. His task was to model springback in seam welded and bent stainless steel tubes for car exhausts. This is important if they are to be formed into components that will readily fit the cars they are to be attached to. Forming parameters are normally established by trial and error. He found that the trick to getting the modelling right was to take into account changes in Young's Modulus caused by plastic deformation. If these are not taken into account, errors are around 4.2%, and if they are, these reduce to 1.4%, which is enough to make the difference between parts that fit and parts that do not. Thorough understanding is definitely required for modelling to be successfully, outside relatively simple tasks. Carlos Amezqueta, a mechanical engineer at the Spanish Department of Wind Energy at the National Renewable Energy Center (CENER), had the task of modelling the adhesive bonds between the shells that together make up wind turbine blades. He explained that if one modelled the bonds in a simple manner, refining the mesh produces maximum stresses that, "seem to get higher and higher as the mesh size is refined. The way to obtain sensible results is to concentrate on the elastic energy released when a defect grows since it is reasonable to assume that in a very large bonded area, there are always going to be some defects." The approach he used is called Virtual Crack Closure Technique (VCCT), which is based on in house developed code that interfaces with Patran and MSC Nastran. The modelling method has been validated by comparing model predictions and real world tests with a bonded metal composite sub component, and tests performed on more complex structures within the FP6 Upwind project. Dr Sadeghi had earlier revealed that conceptual fracture mechanics is already in some of the company's mainstream commercial products. He mentioned growing cracks along element edges in bonded joints, and concepts based on Genoa micromechanics in MD Nastran. He says: "I am proud of what the team has done, but there is a lot more to do."