Mathematical Simulation as a Forensic Engineering Tool

Expert witnesses will provide attorneys with basic technical background regarding the types of mathematical simulation available for use in investigating cases and as courtroom evidence.† The presentation further reviews appropriate applications for each type of simulation, as well as the limitations of each type.† The presentation contains text, graphics, simulation output, animation output, and simulation data.

1. What is Simulation?
   1. Animation vs. simulation discussion (cartoon vs. physics)
      1. Input driven vs. output driven
   2. Classes of simulation codes
      1. Rigid body dynamics
      2. Finite element
      3. Hybrid (flexible bodies, rigid body & FE combined)
      4. Specific math models (e.g., vehicle dynamics programs, vehicle collision programs)
   3. Types of problems best addressed by rigid body dynamics simulators
      1. Large-scale motion, kinematics
      2. Whole body, object response
      3. No concern for local deformations or force concentrations
      4. ìFailureî of pieces generally not permitted
      5. Generally faster setup and run times, less computationally intensive
   4. Types of problems best addressed by finite element simulators
      1. Local deformations and force patterns of interest
      2. Quasistatic or small-scale motions
      3. Surface interfaces
      4. Fluid problems
      5. Generally longer setup and run times, highly computationally intensive
2. Why use simulations? What simulations can do for you.
   1. Visually appealing, scientific exhibit
      1. Viewable from any angle
      2. Playable at any speed
      3. Can be visually enhanced with background footage, human overlays, etc.
      4. "Clean" representation of potentially gory incidents
   2. Predict testing outcome or expand on testing analysis
      1. Use simulation to predict results of questionable tests
      2. Use simulation to conduct sensitivity study on several variables to better define desired test conditions
      3. Use single test for benchmark, then expand analysis with simulation for less cost than repetitive testing
      4. Use results from a standard (government) test with customized subject dimensions to get case-specific outcomes
      5. In many cases, simulation can be the only viable analysis alternative (e.g., environment is complex and cannot be recreated or does not exist; testing may cause damage to dummy)
      6. Examine "what if" scenarios
3. How to present simulations in your case.
   1. Thorough and understandable presentation of simulation process and distinction from animation
   2. Credible documentation on input information
      1. Present benchmarking test first, if available
   3. Carefully explained video output, with leading stills so jury gets oriented before viewing
   4. Relation of simulation output numbers to real world phenomenon or tolerances understandable by jury
   5. Witness prepared with sensitivity analysis and relevance of output data
4. How to address opposition simulations, common tomfoolery
   1. Must have input data in usable electronic format (e.g., CD, disk)
   2. Review data, run simulations on your platform to reproduce results
   3. Sanity check on all input numbers (e.g., no gravity)
   4. Tests for numerical stability
   5. Tests for sensitivity (e.g., simulation does something completely different when forklift weight changed by 2 pounds)
   6. Checks for non-physical constraints (e.g., foot tied to pedal with rope, enormous tensile forces developed; grip strength required vastly exceeds human ability; enormous torques applied to force vehicle to follow desired path; bodies pass through one another)
   7. Checks for differences between simulations (e.g., why was friction removed when the comparison was meant to be between the lap and shoulder restraints)
   8. Sanity check on output parameters (e.g., check for non-physical contacts, excessive forces not supported by injury or damage patterns)
5. Understanding the limitations of simulations
   1. Only as good as the input data
   2. Cannot provide more resolution than modeling assumptions (i.e., hand ellipsoids cannot provide individual finger grip forces)
   3. Dummy models are not always visually realistic
   4. Most human models assume passive subject in dynamic event
6. Examples and case studies (Supplemental to examples used throughout presentation)
   1. Pedestrian impact
   2. Booster seat simulation
   3. Train simulation
   4. Debunking of simulations (forklift, restraint geometry, dump truck)
7. Conclusions
   1. Powerful, scientific tools to demonstrate complex events
   2. Work optimally in conjunction with baseline test data
   3. Easily ìtweakedî by unscrupulous experts.† Professional help essential in uncovering these errors.
   4. Work best when the attorney and engineer both understand and concur on exactly what the simulation is demonstrating