By Fanny Griesmer, chief operating officer, COMSOL, Inc.
Computer modeling and simulation has been used in engineering for many decades. At this point, anyone working in R&D is likely to have either directly used simulation software or indirectly used the results generated by someone else’s model. But in business and in life, “the best laid plans of mice and men can still go wrong.” A model is only as useful as it is realistic, and sometimes the spec changes at a pace that is difficult to keep up with.
Modeling and Simulation Is Great, But…
One of my favorite parts about working at a multiphysics software company is getting to see up close all of the clever and innovative ways our customers use simulation to move the world forward. There was the loudspeaker engineer who talked about turning an idea in their head into a viable product that passed both the technical spec and looked good, and they credited simulation for turbocharging their design iteration process. Another time, I spoke with someone who used our software for automating their process of designing boat landings for offshore wind turbines by creating their own library of parts, combining their learned experience with structural analysis. Someone else invited me into their impressive test lab where they showed off how they run experiments to generate material data, which they later used in their true-to-life computer models.
The benefits of getting a preview of the real-world outcome before you commit to a project plan or design transcend industry and product offerings. There are countless examples of how modeling and simulation speeds up innovation and reduces overall costs. That said, using simulation in the way it’s largely been done over the past 30 years has required specific expertise and training on how to use the software of choice. So while companies that use it have a lot to gain, the total gain is still limited by the number of employees who have learned the necessary skills to build computational models. But that doesn’t need to be the case.
Taking Simulation to Greater Heights Through Custom Apps
Take a company that produces speakers for luxury cars, for instance. Today’s consumers expect a lot more from their ride than safely going from point A to B, sound quality being one of those things. By using multiphysics modeling, it’s much faster and easier to visualize how different designs will sound in specific car cabins before going to production. You can make a perfect replica of a particular car design to test your loudspeakers in, but if that car is still in the process of being designed, your replica can become outdated very quickly. The way sound bounces off the interior might go from acoustic bliss to tinny tunes if the car door design changes or the customer chooses a different type of trim. Do you really want to go back and forth updating and rerunning the full-fledged model every time? And what about the lag between learning about the changes from the team designing the interior to actually being able to update your model to reflect those changes? In the case of one supplier of audio technology for car manufacturers, the team worked in different time zones and the lag was a real pain point. They needed something better.
What did they do? They built their own custom simulation apps based on the full-fledged model. Instead of constantly going back and updating large models to account for interior design changes, their global and cross-functional team could enter the changes into input fields in a custom user interface — built by them in-house, exactly to suit their own needs. Since the app is powered by their own underlying acoustics model, they could then quickly and easily visualize how their loudspeakers would sound inside the car environment, design changes and all.
An example of a custom app for designing a heated car seat, where the user inputs relevant data into restricted fields and the results combine the inputs with an underlying computational model.
Now, in this case, the apps were built by and for R&D teams to improve their own work. While this benefited the company and the team, it’s still “just” another example of using modeling and simulation for R&D. Apps have the potential to break far beyond the traditional simulation software user groups and we’ve already started seeing real examples of that.
Making Decisions in the Field, Factory, and Lab
Let’s imagine a construction company. Building more leads to more revenue, but hiring enough contractors for the job and motivating them to work fast does not guarantee a larger profit. The world at large is powered by laws of physics and chemical reactions that intertwine and affect each other. The same is true for concrete, which is easily affected by temperature in the air and soil during the curing phase as well as internal temperatures based on chemical reactions when water and cement are combined. Choices made at the construction site determine how fast the concrete will harden and ultimately how strong and durable it will be going forward. Without predictive analysis, picking the best concrete mix and deciding when to remove the supporting framework involves mostly intuition and guessing. Multiphysics simulation would give you a more accurate estimate of how fast the concrete will cure, especially if you can incorporate information such as what part of the building is being cast, what material surrounds the concrete, and what the weather conditions are like on site, both now and in the forecast. This type of information can really only be gathered in the moment, out on the construction site. It’s not practical or realistic to send a simulation engineer out with the construction crew nor is it realistic to teach the crew how to use simulation software. But it is possible to have a simulation engineer build a custom app for the onsite crew to use. You can have the best concrete in the world, but if it doesn’t set right, the project cannot be deemed successful. Contractors make in-the-moment decisions that will eventually determine whether a project is on track or delayed, profitable or not. Simulation apps would allow them to test their choices virtually before picking the best mix and curing time based on both science and their local onsite conditions. This is not a fictional example, by the way: one of the world’s largest suppliers of cement, aggregates, and precast concrete rolled out an app for this use a couple of years ago, and they are only continuing to expand on their use of simulation apps today.
A simulation app for determining concrete mix and casting timeline based on onsite conditions.
Next, let’s consider a company focused on manufacturing. In that case, weather is not as big of a concern, because an indoor environment can be more tightly controlled. However, there are still many uncertainties at play that can impact production outcomes and if you can predict them in advance, the business will be better off. Let’s take an additive manufacturing factory producing parts via metal powder bed fusion as an example. Back at the office, simulation engineers can optimize the designs ahead of production, but the end result might still not match the model if the facility conditions are not ideal at the time of production. Heat and humidity inside the facility can cause the metal powder to oxidize and pick up moisture while in storage, and this will alter how it flows, melts, picks up electric charges, and solidifies. Furthermore, the powder is flammable and toxic, even more so when it dries out. In other words, measuring and managing humidity levels in the factory impacts both product quality and worker safety. One such company modeled their own factory and built simulation apps around it to monitor and predict factory conditions based on variables such as outside climate, how many machines are running, and how machines are positioned. Their staff can then use the apps on the spot to figure out how to adjust ventilation and production schedules to create the conditions they need for the best production results.
Now, if you are running direct experiments in a lab or using test rigs, you can, of course, see exactly what the real outcome is based on carefully selected inputs and a controlled setup. By coupling experimental testing with simulation, though, you can improve understanding and make faster predictions using your lab-generated results. For example, if you’re researching thermal elastohydrodynamic lubrication of gear contacts, you might learn through observation that a diamond-like carbon coating on the gears’ surface improves their efficiency, but that only shows you what happens, not why. In this case, having a simulation app in the lab would allow you to easily input the details of your actual setup and get a multiphysics simulation of how the heat flows inside the system. A research team that did exactly this, understood from the model that the efficiency improvement stemmed from the fact that the coating traps heat in the contact, which lowers the lubricant’s viscosity and thereby decreases friction. They would not have known this using only the naked eye.
Simulation can be used as an effective decision-making tool in the office, field, factory, and lab. When organizations build and distribute their own custom apps, everyone in the workforce will be able to make decisions based on forecasts that account for real-world complexities and the underlying laws of physics — without having to first learn how to use simulation software or take up a lot of someone else’s time. The world is ever changing and simulation apps help companies and teams of all kinds keep pace.
If you’d like to learn more about simulation apps, here’s a suggested resource: https://www.comsol.com/benefits/simulation-apps
Fanny Griesmer is the chief operating officer of COMSOL, Inc., which develops, markets, and sells the COMSOL Multiphysics® simulation software.
