Using Simulation Software in the Courtroom
Simulation software is increasingly used in diverse fields of science and engineering. What about in the field of law? A simulation is inherently an imitation or representation of reality, and in court, lawyers try to establish what did or did not happen in reality and why. Here, we explore the use of simulation software in the courtroom and, specifically, the rules by which judges (in the United States) determine when the use of simulation software is allowed in a legal trial.
Types of Legal Cases in Which Simulation Software Is Used
There may be no limit to the types of cases in which simulation software can help prove what did or did not happen and why. Simulation software is perhaps best known in the legal world for its use in automobile accident cases, where specialized programs for these types of simulations have been in use for many years. However, simulation software has also been used for more diverse legal purposes, including:
- Illustrating how much steam was likely used by a utility customer (Commercial Union Ins. Co. v. Boston Edison Co.)
- Demonstrating vehicle dynamics and occupant kinematics in a product liability case involving a vehicle rollover (Harvey v. General Motors Corp.)
- Determining the likely height of the perpetrator accused of destroying property (Commonwealth v. Caruso)
To illustrate the power of simulation software in the legal arena, let’s look at the following example, in which Veryst Engineering, a COMSOL Certified Consultant, was hired to analyze the failure of a disk brake. Due to deterioration of the calipers controlling the arms that open and close the brake, the brake linings had worn prematurely. At issue was whether the caliper and brake linings deteriorated rapidly, such that the deterioration might not have been timely identified, or slowly, such that the problem should have been anticipated. Simulation software was used to analyze the effect of caliper failure on brake lining wear.
Testing the failed calipers after the incident at issue revealed them to have approximately twice the closing force of the newly manufactured product. The question then became whether this change in application force could explain the increased wear in the brake pads of the braking system.
Veryst implemented a wear model based on a modified version of Archard’s law:
ẇ = k(H,T)pNVT
where the rate of change in wear depth ẇ at any point is related to the normal contact pressure pN; magnitude of sliding velocity VT; and a constant k that is a function of the material hardness H, and temperature T.
Veryst implemented the above wear equation in COMSOL Multiphysics using boundary ordinary differential equations (ODEs) with the wear depth as the degree of freedom. The wear depth is then used as an offset between the contacting surfaces. Contact is enforced only when the penetration between the contact surfaces is equal to the wear depth. This wear algorithm is very efficient but is only suitable for cases where wear is significantly less than the width of the contact surface.
Veryst included heat generation at the brake pads, which was then coupled to the temperature-dependent wear rate. This multiphysics model is illustrated below, where contact pressure and wear are shown at a given time. The bottom image provides a view of the brake disk and pad temperature again at a given time. The model and physics shown below are from the actual case. Some of the details, such as the geometry and loading values, have been changed for confidentiality purposes.
Animation courtesy Veryst Engineering.
Animation courtesy Veryst Engineering.
Simulation software was key to determining whether the brake pads could have deteriorated rapidly given the unintended change in caliper force and given that examination of the actual brake linings revealed extensive damage. The image below shows the increase in brake pad temperature over time, which affects wear rate.
Image courtesy Veryst Engineering.
The results of time-dependent simulations under different loading conditions revealed how wear could develop over time, with a predicted abrasion profile consistent with that observed in the failed brake. Thus, the simulation ultimately indicated that a poorly performing caliper could rapidly produce high temperatures and lead to catastrophic and rapid brake pad wear.
Using simulation to support or refute theories as to what happened and why can be key to establishing critical legal questions.
2 Standards for Admissibility
Computer simulations are offered through the testimony of an expert witness, qualified in a particular field, who will use the simulation to support an expert opinion. In the United States, courts largely follow one of two tests for determining whether an expert may testify based on scientific or technical information: the Frye and Daubert tests.
The Frye Test
The Frye test is named after the 1923 case Frye v. United States. In this case, the appellate court affirmed the trial court’s decision to exclude Dr. William Moulton Marston, the inventor of the systolic blood pressure lie detector, from testifying on the results of the test that he had administered to defendant James Frye, on trial for murder. (Dr. Marston later went on to fame as the creator of the comic book Wonder Woman. See K.J. Weiss et al., “Frye’s Backstory: A Tale of Murder, a Retracted Confession, and Scientific Hubris”.)
The Frye test, under which Dr. Marston’s testimony was excluded, requires that scientific evidence has “gained general acceptance in the particular field in which it belongs” in order to be admissible in a courtroom. This restrictive test, perhaps due to its pedigree, reflects an effort to keep so-called “junk science” out of the courtroom. (See Peter W. Huber, Junk Science in the Courtroom.) Some state courts still follow the Frye test to this day.
The Daubert Test
The Daubert test is named after the 1993 case Daubert v. Merrell Dow Pharm., Inc.. In this case, the United States Supreme Court rejected the Frye test for use in federal courts and instead required trial judges to screen scientific-based evidence by weighing various factors in order to determine whether the expert’s testimony is relevant to the facts and reliable. These factors include whether:
- The theory or technique can be or has been tested
- The theory has been subjected to peer review and publication
- The known or potential error rate is acceptable
- Standards controlling the theory or technique’s operations exist and are maintained
- The theory or technique has gained widespread acceptance
Subsequently, in the 1999 case of Kumho Tire Co. v. Carmichael, the United States Supreme Court made clear that the factors listed above are not exhaustive and may vary depending on the type of expert.
In addition, Rule 702 of the Federal Rules of Evidence requires that an expert’s testimony be based on sufficient facts or data and that the expert has reliably applied the principles and methods to the facts of the case.
Applying Admissibility Standards to Simulation Software
Next, we look at different contexts in which these standards have been applied to expert testimony based on the use of simulation software. Before examining the elements of these tests, it is worth noting that at least one court has held that the Daubert factors are to be applied to the overall simulation program and not to any models supplied with that program. (See In re Yamaha Motor Corp. Rhino ATV Prods. Liab. Litig.)
General or Widespread Acceptance
One influential Massachusetts case, Commercial Union Ins. Co. v. Boston Edison Co., explored what constitutes “general acceptance” of simulation software within a particular field for purposes of the Frye test. Commercial Union involved a dispute between a utility customer and the utility provider, in which the customer claimed the provider overcharged based on a faulty reading of the amount of steam used by the customer. The court ruled that an expert could testify regarding a computer simulation showing the amount of steam used based on heat transfer, building materials, operating characteristics of heat and air equipment, and weather conditions.
That court determined that the simulation program at issue met the “general acceptance” standard insofar as:
- The program had been used by engineers and HVAC design professionals to model energy consumption in more than 40,000 buildings
- The use of the program to predict a building’s energy consumption was one of the most common uses of the program, and its productions were sufficiently accurate that the state of California approved the use of the program to fulfill a required computer analysis of a proposed building’s energy consumption as a prerequisite to issuing a building permit
- The program was also used to recreate past energy consumption in buildings
One indicator of “general acceptance” is use of the program at issue in other litigation. For example, in State v. Phillips, the Washington Court of Appeals found a valid scientific basis for expert testimony based on a specialized simulation program for automobile crashes based, in part, on the fact that the expert not only testified regarding the use of simulation programs in the accident-reconstruction field generally, but also had testified using the program at issue over a thousand times and testified regarding the advantages of the program at issue over others used in the field. Similarly, in Kudlacek v. Fiat SpA, the court found general acceptance to be satisfied where the program was used by the largest accident reconstruction company in the world and was “generally relied upon by experts in the reconstruction field for single-vehicle accident simulation.”
In contrast to the Frye test, which requires general acceptance in a relevant field, the Daubert test merely weighs against other factors whether the theory or technique has gained widespread acceptance. Nevertheless, the types of circumstances identified above in applying the Frye test should be useful in establishing that a simulation program has gained widespread acceptance under the Daubert standard.
Peer Review and Publication
One of the Daubert factors is whether the theory or technique at issue has been subject to peer review and publication. A case applying this factor to simulation software is Livingston v. Isuzu Motors, Ltd. There, the court allowed an expert to testify based on a simulation of an automobile collision. In addressing the peer review and publication factor, the court ruled that this factor was satisfied based on the fact that the expert had made several lectures and presentations concerning the methodology employed to members of the scientific community, including automobile manufacturers and engineers.
Testing of Theory
The Livingston case also addressed the Daubert factor inquiring into whether the theory has been tested. The court noted that the program at issue incorporated various laws of science, which are inherently tested, as well as case-specific data. As case-specific data was incorporated into the program, the court found that the complexity of the intermingling of this data precluded testing. The court further noted that an exact reenactment of the accident would not be practical or prudent. Thus, the court concluded that the theory behind the simulation had been tested to the extent feasible, which was sufficient to allow testimony based on the simulation.
Acceptability of Error Rate
Simulation is inherently an approximation of reality and not an exact replica of reality. This point is important in light of the Daubert factor asking whether any known or potential error rate is acceptable.
This issue was also explored in Livingston. There, the expert testified that the software program in question represented the best existing technology, which might be improved over time. The court ruled that the proper way to treat this factor, in light of this testimony, was not to preclude the testimony, but for the opponent of the evidence to cross-examine the witness to explore the possibilities and uncertainties regarding error so the jury could weigh such potential errors in assessing the strength of the expert’s opinion.
Reliable Application to the Facts of the Case
One of the criteria that must be satisfied under Rule 702 is whether the expert has reliably applied the principles and methods to the facts of the case. This issue recurs in cases involving simulation software to the extent that it is often necessary to make approximations or to use comparable inputs in undertaking a simulation.
Decisions in which experts have been allowed to testify based on simulations often stress the degree to which the expert’s model fits the facts of the case. For example, in Deffinbaugh v. Ohio Turnpike Comm’n, the appellate court affirmed a judgment in an automobile accident case where an expert had testified based on a computer simulation. In that case, the expert “utilized known facts such as the weight of the trailer, its physical dimensions, and the surface friction coefficient” in generating the simulation. Similarly, in Wheaton v. Bradford, the court ruled that an accident reconstruction expert could testify based on computer simulations where the expert used factual information and physical evidence, including measurements of the final resting positions of the vehicles taken by the police, his own measurements and observations from the accident scene, the physical properties of the vehicles at issue, and data from the control module sensor of the plaintiff’s vehicle.
Where the inputs to the simulation deviate from the facts at issue in the case, the expert must be prepared to address and account for those deviations. For example, in Hudson v. City of Chicago, an automobile accident case, the plaintiff’s expert testified based on a simulation using data pertaining to a Chevrolet Monte Carlo, when the actual vehicle involved in the collision was a Chevrolet Caprice. Because the expert testified that he had adjusted the characteristics of the car within the program to conform to the actual car involved, the court held that the expert’s testimony was permissible. Other courts have similarly ruled that questions about the accuracy of the inputs to the simulation are for the jury to weigh after cross-examination of the expert and are not grounds for excluding the expert’s testimony. (See In re Yamaha Motor Corp. Rhino ATV Prods. Liab. Litig.; Turner v. Williams.) However, where a simulation was based on inputs that were not themselves determined through scientific methodology consistent with the Daubert test, one court ruled that the expert would not be permitted to testify. (See Reali v. Mazda Motor of Am., Inc.)
The requirement that a simulation be reliably applied to the facts of the case does not mean that the simulation program must be created for the specific case. For example, in State v. Tollardo, the appellate court upheld the trial court’s admission of expert testimony that was based on the use of an off-the-shelf simulation program. However, where there is a choice among off-the-shelf programs, selecting a program with advantages for the type of simulation at issue may help ensure admissibility. For example, in State v. Phillips, the court upheld expert testimony based on an off-the-shelf simulation program for automobile accidents and cited, among other things, that the program at issue was three-dimensional, whereas any of the other available choices were only two-dimensional.
By contrast, if the program at issue has limitations that decrease the accuracy of a simulation of the phenomena at issue, expert testimony based on the use of that program is less likely to be admitted. For example, in State v. Sipin, the court overturned a criminal conviction for vehicular homicide where the prosecution had relied on the very same simulation program at issue in Phillips. The Sipin court found that the user manuals for the program described limitations on the program and model used that diminished its utility for the type of crash at issue.
Lessons for Effective Use of Simulation Software in the Courtroom
Simulation software can be useful for proving a broad range of disputed facts in court. Because of the widespread use of simulation software in industry and in academia over many years, there is no shortage of evidence to help satisfy the Frye or Daubert tests for admissibility. In establishing acceptance of a particular program, lawyers and expert witnesses should gather evidence of use of that program within the relevant industrial sector, which is often publicized by the software publisher and/or members of the industry. They should also look for use of such software in connection with standards or government regulation, use in other litigation, and use in academia (which may be highlighted in academic publications or even in published course descriptions).
Lawyers and experts may find evidence of peer review and publication in the form of conventional academic publication and academic coursework and research projects that are extensively designed around a particular simulation program. In addition, the software publisher may make available information on the types of verification and validation testing it performs.
Expert witnesses should be prepared to explain the ways in which simulations deviate from reality and why potential errors are acceptable. They should also be prepared to testify as to exactly how the facts of the case were incorporated into the simulation and be prepared to explain how deviations from the facts were compensated for. They should also be prepared to address any advantages the simulation program may have over other such programs for the particular application at issue.
Read about other legal considerations regarding simulation on the COMSOL Blog:
- Can Models Be Protected by Copyright Law?
- A Guide to Publishing and Licensing Simulation Applications
The views expressed in this blog post are offered for purposes of discussion only and should not be construed as legal advice or as the official position of COMSOL on any legal matter. Nor are the views expressed here intended as exhaustive coverage of the subjects addressed here. Each situation is different, and different circumstances may result in different legal outcomes.
- COMSOL Now
- Today in Science
- COMSOL Now
- Today in Science