Virtual Product Development (VPD) is the scientific process of developing and analyzing digital prototypes in a completely 3D virtual environment utilizing virtual product design and simulation programs. VPD assists greatly in the innovation process by more accurately predicting product performance in virtual testing environments ultimately minimizing time to market, design failures, and product development costs.
VPD means many different things to different companies. To some companies, VPD may mean implementation of a computer-aided design (CAD) in their design groups. To others, VPD may mean implementation of a web-based collaborative development that brings together designers, customers, and value chain partners. While these initiatives are definitely necessary in today’s competitive marketplace, VPD in the biomechanics world is a much more powerful process that is required to drive innovation and minimize time to market, potential design failure, and product development costs.
Implementation of VPD in biomechanical applications is still somewhat lagging, given the obvious quantifiable benefits that it provides. The main reason for this is that many are unfamiliar with how easy it is to implement into today’s modern R&D environment. That includes many biomechanical engineers. This is mainly due to the fact that many are still biomechanical modeling and simulation tools that seem too complex and intuitive in nature to be the centerpiece of any VPD program. In many cases, they are somewhat correct in their assumption. The problem is they are using old, classical methods of developing dynamic equations. That is not to say they are incorrect by any means; rather, it is just that they are extremely complex models that require a lot of intuition on top of a lengthy debugging process that require years of advanced dynamics modeling background to be successful in any 3D simulation that has more than a few degrees of freedom (DOF). Plus these models are typically not very scalable and transferable to other modeling and simulation activities. Too much time is spent on the programming of the dynamic equations of motion rather than on any meaningful analysis with the simulation model. So it is easy to see why VPD has not taken off in biomechanics-based R&D programs using these old, classical dynamics techniques.
However, the necessary toolsets and software programs are readily available. The problem is that many are unfamiliar with or unwilling to learn new modeling techniques, as it took 5+ years to become advanced at the previous methodology they used. But with the increased powers in personal computing capabilities and with many commercially available programs available that automate the process of deriving and programming the dynamic equations of motion, it has never been easier to implement VPD programs in any size R&D group. And the cost to do this is minimal compared to the savings it will ultimately yield in VPD programs. Using Kane’s Method and one of the associated symbolic computer tools available, just about anybody can derive the equations of motion and create a bug-free computer simulation code for complex 3D systems having up to 30 degrees of freedom in a few days work.
Unfortunately, too many biomechanical engineers believe that motion capture studies provide the ultimate tool for studying what happens in high-speed movement activities. Motion capture studies are very powerful tools for observing high-speed motion activities that are too fast for the human eye. And the amount of quantifiable data that these studies provide is extremely valuable. I am not trying to diminish the power of a well designed and implemented motion capture study. My only concern is that too many biomechanical engineers view motion capture studies as the end all system technology for advanced biomechanical evaluation studies, and concentrate all of their activities on the setup, data collection, post-processing, and evaluation reams of data in search of the answer to any number of very important biomechanical questions.
In reality, motion capture studies provide potential answers to the effect of the problem, but really don’t address the cause of the problem. A well designed motion capture study can potentially investigate the cause through a properly designed design of experiments (DOE) study. However, in human testing programs it is extremely difficult to control even a few kinematic variables in a motion capture study. By the time you collect all of the data from a number of different test subjects, you are left with terabytes of data to analyze, without really knowing what was controlled in the test. And that is because the inverse dynamics process is counter-intuitive to truly understanding the cause of the change by the human body. That is because in an inverse dynamics study,measured motion capture data is used to estimate the joint torques/forces that are necessary to produce the recorded motions. So this type of study is really examining the effect of the change by the human body.
So how can one study the cause of a change by the human body? That is where forward dynamics modeling comes into play. Quite honestly, most biomechanical engineers are really not all that familiar with forward dynamics modeling. They typically get through the inverse dynamics phase and stop there as the terabytes of data they can analyze can cause data overload and paralysis by analysis. But the inverse dynamics process should really just be a precursor for the much more powerful forward dynamics process.
What is the difference between the 2 steps? Forward dynamics uses joint torques/forces to predict resultant motions. So essentially the output from the inverse dynamics process is used as the input for the forward dynamics process. Why is that more powerful? The reason why this is such a powerful predictive tool is that the software logic mimics the manner in which the human body actually functions. In real life, neural inputs give rise to muscle forces that generate joint and ground reaction forces that together drive the dynamic movement of the human body. This order is maintained in the forward dynamics or simulation paradigm. And that is why forward dynamics modeling is such a powerful predictive tool.
It is important that one really understand these differences to truly see the benefits of forward dynamic simulation routines. In the inverse dynamics study, motion capture data is the input data and calculated joint torques/forces are the outputs. So in a motion capture study, the investigator is typically trying to look for changes in the input (measured motion data) to try and determine changes in a movement exercise. Again, human movement comes after the neural input, and after the muscular activation, and after the joint torque which causes the bone of interest to move relative to another bone. An extremely difficult process to try and control kinematic variables in order to understand a cause and effect relationship. This type of study requires the human subject to perform condition A and then condition B, both subjected to the motion capture protocol. Dependent upon the number of trials, fatigue may set in and extreme care must be taken during data collection to control the kinematic variable(s) of interest.
In the forward dynamics study, joint torques/forces are the inputs and predicted motions are the output. Sound familiar? Human movement comes after the neural input, and after the muscular activation, and after the joint torque which causes the resultant movement. That is why forward dynamics simulation routines are so powerful due to their predictive nature. I am not trying to diminish the advanced math and programming required to develop a forward dynamics model. But a well designed forward dynamics simulation routine can be used in a parametric design program to analyze any number of changes on the resultant motion. And this type of study can be run 24 hours a day on any number of computers simultaneously as it does not require a motion capture study to analyze the what-if scenarios thrown at it. Everything is done through virtual simulation. Questions can be studied that may be difficult or even impossible, costly, and/or unethical to answer through actual human subject testing.
A real life scenario would probably be very beneficial to prove my point. I worked at Callaway Golf for 5 years in a variety of R&D roles on the Product Player Matching initiative which included Player Profiling, Digital Human Modeling, and Club Fitting projects. The main testing tool used in the golf industry is robot testing. Different clubs are tested on robots at different swing speeds to measure their down-range performance. A standard testing platform used by golf OEMs following standard operating procedures (SOP) from the United States Golf Association (USGA).
The problem with robot testing is that no one really swings like that. I have 5 years of player profiling studies with instrumented clubs to prove that. I was tasked with developing a digital human swing model that could swing the same every time for any desired swing profile so that parametric changes in club properties could be examined by analyzing the resultant motion changes (speeds and orientations) of the club head at impact. How did I accomplish that? Through a well designed VPD biomechanical swing model.
Player profiling studies were used with instrumented clubs to measure and analyze swing profiles across all types of golfers – high swing speeds to low swing speeds, high closure rate swings to low closure rate swings, low handicappers to high handicappers. From these studies, swing profiles were created based on closure rates (how fast or slow the club head was closed during the downswing) and release rates (essentially how fast the swing speed was based on wrist release/hinge angles. From these profiles, characteristic dynamic swing models were created that best approximated these swing profiles. Motion capture studies were performed to ultimately create a forward dynamic swing model for that swing profile. The motion capture data was used in the inverse dynamics process to determine the joint torques/forces; the inverse dynamics outputs were then used as the inputs to the forward dynamics model to predict resultant motions. Was it successful? Correlation coefficients between 0.95 – 0.99 between measured joint motions and prediction joint motions for all 16 degrees of freedom and the resultant club motion.
Now that a dynamic swing model was created for each swing profile, what is the benefit? Using the desired swing profile, club properties can be parametrically analyzed to see the effect on resultant club head motion. If weight is moved in the clubhead to the rear, what is the resultant change of the club head at impact? Does it loft more and what happens to the swing speed? If weight is moved to the heel of the club, is the club more open or more closed at impact for different swing profiles? How does the weight of the shaft effect the club head orientations at impact for any swing profile? Using the exact same swing, what are the orientation changes in the club head at impact between a 3 wood and 3 iron? Even player effects can be analyzed. If we increase the torque of the torso relative to the hips (increased X-Factor), what happens to the swing speed at impact? All of these scenarios can be run on a computer without any human subject testing necessary. And it can be run all day long. That is the power of forward dynamics modeling.
Forward dynamic simulation routines can be used for a variety of biomechanics applications. They can be run to analyze sports equipment interaction – golf clubs, baseball bats, tennis rackets, or bikes. They can be run to analyze sports performance – the effect of increased torque in the hips/torso vs. additional torque at the wrists in the golf swing, the effect of sequential timing on shoulder and elbow loading during the violent overhand pitching motion. They can be run with and without medical device implants examining the effect of hip and knee implant designs or spinal fusion cages to examine altered kinematics or kinetics. Results from the forward dynamics routine can be used in synergistic finite element analysis (FEA) studies to examine stress values in adjacent bony tissue from a squatting motion or during a running motion. The options are endless…and extremely powerful.
Examples of VPD biomechanical analysis programs I have developed include:
- As part of an intelligent knee brace project for a start-up company, I was responsible for all proof of concept design programs integrated with VPD simulations to predict knee brace function and performance in a typical NFL game environment for both contact and non-contact ACL injury mechanisms prior to building a physical prototype.
- Developed and implemented a special purpose VPD design and analysis system for a kinematic based club head design program using actual measured player input data. VPD system allowed for database integration and selection of player kinematic data, CAD-based club head properties (geometric, material, and mass), mass and flexural properties of shafts from CAD output, and geometric and mass characteristics of the grip.
- Developed a lower body forward dynamics model to better understand the natural kinematics and kinetics of hip and knee joints and to simulate in vivo functional activities to evaluate the kinematic and kinetic performance of parametric implant designs in virtual product development (VPD) design analyses.
- Conducted comparative kinematic analyses of a human knee vs. a total knee replacement (TKR) system, specifically analyzing differences in deep knee flexion and resultant joint contact forces, using active muscle forces and soft tissue constraints to drive the simulation.
I have over 10 years of experience in developing VPD programs for sports and medical device products using physics-based engineering simulations of computer prototypes that predict product performance in an integrated virtual testing environment, resulting in increased product innovation and significant savings in product development costs and time. I have extensive experience with various 3D solid modeling programs and integration with VPD dynamic simulation routines for the sports equipment and medical device industry, specializing in spinal fusion devices and hip and knee implants.
I also have over 8 years of experience in the application of Six Sigma (DMAIC) methodologies for analysis and improvement of existing engineering design and development processes. I also have over 8 years experience in application of Design for Six Sigma (DfSS) approaches for design or redesign of new golf equipment and medical device products including advanced design tools such as Quality Functional Deployment (QFD), Failure Modes and Effects Analysis (FMEA), Design of Experiments (DOE), Analytical Hierarchy Process (AHP), and Voice of Customer (VOC).