Accident Simulation

While accident re-creation is on a physical prototype of model of the accident artefacts and events, accident simulation is a virtual reconstruction of the accident with a computer model or a physical small-scale model (or occasionally a large-scale version when the object itself is very small and you cannot see much in an actual scale re-creation).

This is also a godsend where physical re-creation would be too large, too expensive, too time-consuming, and/or too dangerous to people or property or environment.

Computer simulation of accidents

Today, most major accidents are simulated with computer models. The computer is used not only for very highly complicated mathematical analysis of accident parameters, but through computer graphics and animation, also for quite realistic visuals in still and video format.

Apart from impressing all viewers, the computer formulation also allows the analyst to 'play around' with the objects and their movements with very little effort beyond the first modelling, so that by repeated trials, one or a few scenarios which would have led to the accident can be quite reliably determined.

The computer applications will be discussed in detail in another paper.

Model Analysis of Complex Structures

Another alternative is some kind of model analysis in which a scaled version of the situation is subjected to the same event as in the accident.

Before computers became so common, so powerful and yet so inexpensive, model analysis flourished both as a science and as a practical procedure to solve complex engineering problems. Even now, except that people would not know or care for the power of model analysis, it can serve as an effective alternative to complex computer analysis, which even today depends much on the proper use of good software.

While doing my PhD in the USA, a Professor took our class to visit the well-known structural designer Ken R. White's office in Denver. Mr. White showed us some small-scale models of innovative structures he had built out of tooth-picks and drinking straws, and then loaded them and pushed them around until they failed, to get an idea of their failure modes and loadings.

This was before the days of finite elements and 3D computer graphics, but the idea was the same: You have no record of the accident itself. You develop a model, and experiment with various scenarios, until the outcome is closely similar to the actual accident. Even today this is the most common procedure. In many cases however, the creation and testing of the model, physical or digital, can get quite expensive.

For physical models in pre-computer days, scaling effects and material properties were handled by a science now relatively uncommon called 'Similitude'. I developed a lab-oriented course on ‘Similitude in Engineering’ during my tenure at Auburn University in Alabama, USA (1967-1975), and employed the method to good effect in my research and to guide student theses.

After the advent of computers, by the 1980s, we got the power to model huge structures in great detail and analyse them by computers, and then model analysis became less popular.