Parallel three dimensional finite element analysis of dinosaur trackway formation

Fossilised footprints and trackways provide palaeontologists with information regarding dinosaur locomotion such as their gait, posture and speed. Current best practice is to interpret trackways as 2D surface features. Using computational geomechanics, this paper demonstrates that subsurface deformation can lead to transmitted or false footprints at different depths, whose size and shape relates to a distorted 3D pressure bulb. The results of parallel 3D finite element simulations are compared with a set of transmitted tracks owned by Amherst College, USA. At Amherst College, palaeontologists have peeled away each layer of solidified sediment to find a footprint in each one. Each of these footprints has a unique geometry, a different length and angle between digits. If each of these was found in isolation, they would be erroneously interpreted as coming from different species of dinosaur. Significantly, palaeontologists use a simple equation relating the length of the foot and the distance between two footprints (stride) to calculate the dinosaur's speed. As the footprint length changes with depth, so does the apparent stride. The consequence is a different speed for the trackway at each depth: a clear source of misinterpretation. To simulate the transmission of the footprint through the soil layers, an elasto-plastic soil model is used together with a fine resolution 3D mesh. The requirements of such a model have a marked impact on the computational cost and the authors describe how existing parallel libraries were extended to build a scalable footprint simulator. © 2006 Taylor & Francis Group.