lunduniversity.lu.se

Denna sida på svenska This page in English

Modelling language creator

2014-07-08

We are encouraged to think outside the box. Yet it can take time before the surrounding world is ready to take on innovative ideas. Hilding Elmqvist, LTH alumnus and creator of a programming language now used by many industrial companies all over the world, knows this only too well.

He first presented his ideas in 1978, when he publicly defended his doctoral thesis on a new way of simulating mechanical and physical processes. Those ideas formed the basis of a new programming language which, in 1996, became a language standard known as Modelica. But for many years, his thesis lay more or less forgotten in a desk drawer.

The particularity of the language idea resides in its construction, which enables it to manage many complex calculations, thereby providing a more complete picture of the engine, the car, the power plant or whatever object you want to simulate. Large numbers of calculations require a lot of computer power and memory, which is one explanation for the fact that it took almost two decades before the ideas caught on. Another, perhaps more important explanation, is that the methods of the time for calculating dynamic systems were quite different to what they are now.

In a simplified sense, classic automatic control engineering consists of calculating input and output signals in a block diagram. This means that any single adjustment will entail several subsequent changes.

‟The structure is excellently suited to control systems but becomes unwieldy for physical and mechanical systems,” asserts Hilding Elmqvist.

In his doctoral thesis, he argued that the focus should be on the relationship between variables instead. That would solve the adjustment problems, among others.

‟In that case it is sufficient to specify that the rotation angle of a motor shaft is the same as the rotation angle of the connected load, for example. It is not necessary to specify which of them affects the other.”

The differences between the two principles is crucial. Hilding Elmqvist remembers very well when lightning struck and he had his ‟Eureka!” moment.

‟It was Maundy Thursday in 1976. We were on our way up to Årröd, quite far into the forest in northern Skåne, where we had rented a cabin. I did not usually bring my books with me on holidays, but on that occasion I did. I just read and worked. My wife mostly took care of the children that time”, says Hilding Elmqvist.

After Easter, the doctoral thesis took a new direction. Until then, similarly to Elmqvist’s previous degree project, it had been an attempt to develop computer simulation programmes on the basis of the prevailing input and output signal idea which had come to nothing.

After publicly defending his thesis in 1978, Hilding Elmqvist wrote two conference papers, but was mostly received with a shrug. He therefore put his ideas aside, did some research and worked as a project manager for SattControl in Malmö.

Autumn 1991 saw a crucial development. Hilding Elmqvist was contacted by a former colleague from ETH in Zürich, François Cellier, who was about to write a book about modelling.

‟I was completely overwhelmed when I read the manuscript. He was using the principles and examples from my thesis! At the same time I realised how much easier it is to launch an idea if it has several proponents and a community is built up around it.”

François Cellier further developed the software and highlighted it in the book. Its publication became the starting shot for continued further development. In 1996, Hilding took the initiative to develop a common standard language, Modelica.

Since then, researchers, developers and other Modelica linguists meet regularly to discuss and develop the language. Hilding Elmqvist describes it as collaboration and competition combined. Recently, in mid-March, 400 experts from all over the world gathered at LTH for the tenth Modelica conference. The meetings can be seen as proof of the language’s international impact – and that the road from idea to innovation can be both long and winding.

Kristina Lindgärde

 

Facts: Hilding Elmqvist
Lives: in Lund, grew up in Urshult
Profession: Head of engineering in the large French group Dassault Systèmes, board member in Modelon
Education: Studied electronics (now electrical engineering) 1968–1972. Doctoral student in automatic control engineering 1972–1978. Postdoc at Stanford University 1978–1979.
Model: Karl-Johan Åström, professor emeritus in automatic control engineering. ”He was my supervisor and an important mentor for many years”.

 

Facts: The Modelica language 
The Modelica language enables computer simulations allowing the specification requirements for a new product or a product modification to be tested in advance, for example. This can concern the fuel consumption and energy output in a power plant, how a car behaves on a wet road surface with a specific depth of tyre tread, or how mobile a robot arm will be as a result of a change of material. Car manufacturer Toyota used the language to develop the Prius model, for example.

In the past, if a company decided not to build prototypes, it would have to make three separate sets of calculations for mechanics, electronic and chemistry. This is not very successful, besides costing time and money, as you want to find out what happens when you press the accelerator on a car while turning the steering wheel, for example.

In Lund there are two spin-off companies with links to Modelica: software company Dassault Systèmes (formerly Dynasim) and consultancy Modelon AB.