Mats Larsson

Saab Automobile

I started my education back I 1971 at the Royal Institute of Technology in Stockholm. I received a M.Sc in mechanical engineering in 1978. I continued as a Phd-student at Solid mechanics and Strength of Materials at RIT for 3 years. An offer was made by the Aeronautical Research Institute in Sweden so I quit school, there I worked with damage tolerance assessment applied to a current Swedish fighter aircraft to be implemented on a new design. In 1983 another offer was made from Saab Cars, today Saab Automobile in 1983. I moved from Stockholm to a town called Trollhättan (50000 inhabitants) where the (most) Saab cars are designed and produced. I started there with structural analysis for all possible problems.

In 1985 we started with crashworthiness analysis, I worked with this topic for 5 years and then I moved to manufacturing Engineering, stamping department. There I started with stamping simulation work in 1990. For some years it was much work done to establish the working methods and so forth. The big change came when all parts where drawn in CAD and then stamping simulation became a standard tool in our development process.

From my start at Saab until 6-7 years ago I was also very involved with computer hardware ad software, what type of computers and capacity we needed for all type of finite element work made for the car program development.  Discussions on what workstations we should use and a Cray J90 and two SGI Origin 2000 were installed during my active time on these issues.

In the beginning of 2005 the Swedish government issued a new research program called MERA, Manufacturing Engineering Research Area. Two friends working at Volvo and myself made plans for a Finite Element Stamping Simulation project, we got some funding. In this project there are five working packages and one of them is about stochastic analysis.

Going back a little in time some of us at Saab realised the need to include natural variation in our processes and when we found out that we some computer power to start with stochastic analysis with FEM we made acquaintance with some software companies as well as established some cooperation within GM this was a log time ago, my memory fails me I would say it was around 2000 or so. Then some activities where made I several fields and I started to think how this should be applied to sheet metal stamping simulation. (I am still thinking on some issues, we can certainly discuss this further.)

We tried to install commercial software for stochastic analysis package to our main production stamping simulation software. Due to some problems with software/hardware this project never went into production. Then I made an internal proposal on this item and received some funding and together with IVF a Swedish Research Institute for Production. My needs and ideas became Mathlab-programs at IVF, these programs were userunfriendly.

With this development as input to the MERA-program we continued our efforts to develop options that we need and today we have some nice features and still learning, have ideas on further development. Still our software is userunfriendly, it is a research base and our aim is to be able to explain to software companies was we want and what we feel is important. To be able to do this we feel that we need to do this development work otherwise we are not able to evaluate if we are on the right track.  Our work has resulted in one paper accepted for publication in the "Journal of Structural and Multidisciplinary Optimization” and a presentation at the International Deep Drawing Research Group meeting recently in Hungary, 2007.

My formal background in mathematical statistics is not deep enough to follow all detailed discussions. Though my understanding and learning by doing has compensated that to some extent. I think I have a decent understanding on how we must take uncertainty and complexity into our (finite element) predictions and hopefully I can contribute in this field with some advice and thoughts.