Computer Simulations in the Courtroom: An Expert Perspective

Computer simulation can be a very powerful tool in a variety of fields. This includes accident reconstruction and biomechanics to fire investigation and product design. But computer simulations can also be full of unrealistic assumptions and invalid techniques. If an opposing expert relies on a simulation, how do you know if it’s reliable? Here, I’m discussing how you can parse valid, scientific techniques in computer simulations from junk science that should be excluded before trial.

Animation or Computer Simulation?

Determining whether the opposing expert has created an animation or a computer simulation may not be initially obvious. But this distinction is important for how the court will evaluate its acceptability.[1] An animation is usually a video demonstrating an expert’s opinion or a witness’s description of an incident. Therefore, an animation is just a demonstrative exhibit.

In contrast, a computer simulation involves calculations that ultimately form part of the basis for the expert’s opinion. A computer simulation may also produce a video with the results of the simulation. As such, it may require some investigation to determine if a video comes from a computer simulation or just an animation.

Example animation

An expert produces a video showing two vehicles approaching an intersection. The video is used to demonstrate how far from the intersection the vehicles were as they approached impact. This is an animation because it is only being used for demonstrative purposes.

Figure 1. Example animation of vehicles at an intersection.

Example Simulation

An expert produces a video of two cars crashing in an intersection. The cars impact each other and move to their respective points of rest. In this case, the video captures the physics of impact. Further, it shows the resulting damage and calculated movement post-impact. Therefore, this is a computer simulation.

Figure 2. LS-Dyna finite element model simulation of a vehicle to vehicle collision.

Acceptability of Simulations in Court

Determining if an animation is acceptable in court is relatively straightforward. The animation must be an accurate representation of the facts of the case without being overly prejudicial to the jury. A computer simulation is more complicated. Computer simulations should be broken into two parts. The first part includes the calculations and results of the simulation. The second part is the resulting video (if one exists). Once the computer simulation is broken into these two parts, the analysis can be evaluated like any other analysis using Rule 702,[2] Daubert,[3] or the applicable law in the jurisdiction.

Rule 702 states in part that an expert may testify if:

• The testimony is the product of reliable principles and methods
• The expert has reliably applied the principles and methods to the facts of the case

Daubert clarifies that the following factors can help determine whether the testimony is scientific:

• Whether the expert employs a theory or technique that is generally accepted in the scientific community
• Has it been subjected to peer review and publication
• Whether it can be and has been tested
• If the known or potential rate of error is acceptable
• Was the research conducted independent of the particular litigation or dependent on an intention to provide the proposed testimony

Accepted Scientific Methods: Simulation Validation

Evaluation of an expert’s validation is a key step in determining whether their simulation follows generally accepted scientific methods. Validation is defined as “the process of determining the degree to which a model is an accurate representation of the real world from the perspective of the intended uses of the model.” Therefore, “regardless of the use, confidence in computational simulations is only possible if the investigator has verified the mathematical foundation of the model and validated the results against sound experimental data.” [4]

In a New York Supreme Court case, the court excluded a computer simulation because the technique was not generally accepted by the scientific community.[5] Specifically, in the product liability action, the plaintiff’s expert had created an alternative design and demonstrated it using a computer simulation. Further, the expert did not perform any real-world testing on the computer simulation. The defense expert testified that using a computer simulation without validation is not an accepted scientific technique.

Understanding “Intended Use”

The “intended uses” part of the definition is also particularly important. Often an expert will use a commercially available computer model and say that “it has been validated.” However, has the model been validated for the intended use?

For example, I worked on a case where the opposing expert used a Madymo computer simulation to investigate the likelihood of a thoracic spine fracture. Madymo has a suite of models that have been validated to represent many aspects of the human body. These models are used to simulate human movement and injury risk. However, the risk of thoracic spine fracture was not one of them.[6] Therefore, the model was not validated for the expert’s intended use and was therefore not using a method accepted by the scientific community.

Figure 3. Images of Madymo human models (left) and dummy models (right).[7]

Reconstruction vs. Illustrating a General Principle

Computer simulations may support an expert’s opinion in multiple ways. Largely, these are either reconstructions or simulations illustrating a general principle. Either category can be evaluated in relation to Part (d) of Rule 702. This rule indicates that an expert must apply their method to accurately represent the facts of the case.

In the case of reconstruction, an expert must appropriately support all of their simulation modeling choices as representative of the subject case. This can be very difficult to accomplish as almost all simulations must make assumptions or simplifications relative to reality. For example, consider a simulation of an automobile accident. This simulation demonstrates the effectiveness of a seat belt in injury prevention. A reconstruction would need accurate vehicle dynamics of the subject accident, plus accurate body weight and dimensions of the occupant. In addition, it would require accurate properties for the seat belt, steering wheel, instrument panel, etc.

Such a reconstruction is likely very difficult to match all the facts of the case with accuracy and precision. Therefore, it could be subject to attack for any aspects of the simulation that do not match the subject case. On the other hand, a simulation illustrating the general principle of seat belt effectiveness in a similar collision may not be subject to the same expectations. However, if the simulations are not close enough to the subject case, they may still be excluded. The simulation’s presentation to the jury is an important piece of the admissibility evaluation.

Figure 4. LS-Dyna simulation of a frontal impact with a seatbelt and airbag.[8]

Process for Evaluating a Computer Simulation

So you’ve determined that an opposing expert has used a computer simulation as opposed to animation. First, identify what type of software was used. There are numerous types of computer simulations an expert can perform. A few examples are finite element analysis, computational fluid dynamics, multi-body physics, and modeling chemical and biological processes (Table 1).

Each type of computer simulation has many commercial and custom software packages associated with them. Once you identify the type of software and any commercial version information, find an expert in that method to evaluate the simulation.

Common Computer Simulation Software Types and Commercial Packages

Accident Reconstruction and Biomechanics

• PC-Crash
• HVE (Human Vehicle Environment)
• ARAS 360
• Madymo
• ATB (Articulated Total Body)
• GEOBOD (Generator of Body Data)
• FE-BIO
• AnyBody
• OpenSim

Finite Element Analysis (FEA)

• LS-Dyna
• Abaqus
• RADIOSS
• Solidworks
• Pam-Crash
• Ansys

Computational Fluid Dynamics (CFD)

• Ansys Fluent
• OpenFOAM
• LS-Dyna
• Abaqus
• FLACS

Fire, Smoke, and Evacuation Simulation

• FDS (Fire Dynamics Simulator)
• PHAST (Process Hazard Analysis Software)
• PyroSim
• Pathfinder

The Importance of Software Specifications

An expert in the specific type of software is necessary for several reasons. First, most computer simulation files are specific to the software. This means they may not be possible to open without specialized software. Second, experts do not always readily produce all files associated with simulations. For example, in a recent case, we requested disclosure of all input and output files associated with the computer simulations. The expert disclosed only 3 of the 35 files produced by the simulation software. In a deposition, the expert claimed to have deleted all the other files.

Without an expert, the attorneys would not have known the disclosure was incomplete or how to address the insufficient disclosure. Be prepared to file motions to get complete disclosure. Remember that a complicated set of calculations are all that makes up any computer simulation. You would not let an expert testify to the results of calculations without disclosing what equations they used and what numbers went into the equations. Disclosing all input and output files for the computer simulation is the only way to get that information.[9]

If an opposing expert relied on a computer simulation, it’s important to evaluate this. It’s critical to determine if the expert used reliable methods that appropriately represent the facts of the case. If not, the simulation may be excluded. However, evaluating the simulation is not a trivial endeavor and will require an expert with specialized knowledge of the simulation software used as well as careful disclosure requests to get all information related to the simulation.

Works Cited

[1] Webster, Victoria, and Fred E. Bourn III. “The Use of Computer-Generated Animations and Simulations at Trial.” Def. Counsel J. 83 (2016): 439.

[2] Federal Rules of Evidence, Rule 702, December 1, 2014.

[3] Daubert v. Merrell Dow Pharmaceuticals, Inc., 43 F. 3d 1311 – Court of Appeals, 9th Circuit 1995

[4] Henninger, Heath B., et al. “Validation of computational models in biomechanics.” Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in Medicine 224.7 (2010): 801-812.

[5] Lascano v. Lee Trucking, 2007 N.Y. Misc. LEXIS 6872.

[6] MADYMO Human Body Models Manual, Release 7.5, June 2013. P29-30.

[7] https://tass.plm.automation.siemens.com/madymo

[8] https://cae.jsol.co.jp/en/product/structure/thums/cases/caseA01/

[9] Aldous, Alan. “Disclosure of expert computer simulations.” Computer/LJ 8 (1987): 51.