Category: Simulation

COMSOL, the last major independent simulation vendor, continues to maintain a torrid pace of innovation, adding to its lead among all multiphysics simulation software suppliers with the release of COMSOL Multiphysics Version 6.3. This latest update introduces features that enhance modeling efficiency and simulation speed (GPU-accelerated acoustic computations), improve the user experience, plus add a ton of functionality, such as with the dedicated Electric Discharge Module.

Electric Discharge Module

New in COMSOL Version 6.3 is the Electric Discharge Module, which lets you analyze electric discharges, such as lightning, and arcs, such as arc welding. The analysis strings together a variety of physics and considers the complex behavior of various materials. (solid, liquids, gases in a material library), plus phase changes, chemical reactions, and short total time duration, which requires infinitesimal time steps as low as sub-nanoseconds to evaluate transient activity and boundary conditions near electrodes. The Electric Discharge Module can interface with electrical circuits to analyze reactions to electrical components using 2D or 3D field models or by importing data in SPICE format.

GPU-Accelerated Simulations

Version 6.3 introduces GPU acceleration for transient acoustics simulations, bringing a performance boost of up to 25 times compared to CPU-based computations. GPU acceleration is now available across key modules:

• Automotive and architectural acoustics.
• Consumer electronics

Geometry Preparation and Automation

Importing CAD models into COMSOL Multiphysics has been streamlined with advanced geometry preparation tools. Key features include:

• Automated Defeaturing: The software removes irrelevant details such as small fillets and holes, simplifying the model while preserving essential geometry.
• Error Correction: Automatic identification and fixing of geometric issues such as gaps, overlaps and misalignments.
• Simplified Meshing: Cleaned-up geometries lead to better-quality meshes, which improve simulation accuracy and stability.

Interactive Java Environment

The addition of an interactive Java environment in Version 6.3 provides advanced users with new opportunities for customization and automation. Highlights include:

• Dynamic Model Modifications: Users can programmatically alter models during simulations, creating adaptive workflows.
• Chatbot-Assisted Coding: A built-in chatbot tool simplifies programming in the COMSOL API by providing code snippets and answering user queries.

Application-Specific Improvements

The updates in Version 6.3 extend across multiple domains, enriching the simulation capabilities of the platform:

1. Fluid Flow Modeling

• Reynolds-Stress Turbulence Models: Introduced for more accurate simulations of anisotropic and separated flows, these models are essential for industrial processes involving heat exchangers, reactors and aerodynamics.
• Free Surface Flow Improvements: Enhanced tools for modeling interfaces between immiscible fluids, particularly in simulations involving mixing, sloshing and coating.

2. Electromagnetics Simulations

• Electrostatic Forces for MEMS: Improved methods for calculating forces and torques in MEMS devices result in better optimization of designs.
• Laminated Iron in Motors and Transformers: New tools for accurately modeling magnetic losses and thermal effects in laminated iron cores.
• Transmission Line Modeling: Enhanced RLGC parameter extraction for RF and microwave devices ensures better performance predictions.

3. Structural Mechanics

• Thin Structure Electromechanics: Enables precise modeling of piezoelectric and electrostatic effects in thin structures, such as sensors, actuators and energy harvesters.2
• Stress Recovery: Improved algorithms for stress recovery provide more detailed insights into structural integrity.

4. Wave Optics

• Periodic Structures Workflow: Simplifies simulations of photonic crystals, diffraction gratings and metasurfaces, speeding up workflows while maintaining accuracy.
• Nonlinear Optics: Improved modeling of nonlinear optical phenomena, essential for laser design and telecommunications.

5. Surrogate Modeling

• Efficient data sampling tools have been introduced for surrogate model generation, providing approximate results for complex simulations with reduced computational effort. This feature benefits optimization studies and parametric sweeps.

Enhanced Usability

Version 6.3 also emphasizes user accessibility and collaboration:

• Improved Model Manager: Expanded version control and tagging capabilities make it easier to organize, search and collaborate on simulation projects.
• Visualization Updates: New rendering options for clearer presentation of results, including dynamic visualizations for transient and multiphysics simulations.
• Documentation and Training: Updated resources, tutorials and examples have been added, facilitating faster onboarding for new users.

About COMSOL Multiphysics

COMSOL Multiphysics allows its users to model, analyze and optimize designs across a broad spectrum of industries. Its defining feature is the ability to couple multiple physical phenomena—mechanical, electrical, thermal, fluid and chemical processes—within a single model. This approach is not only convenient but accurate since there is no data lost when converting from the output of one solution to the input of another.

Core Components:

1. Model Builder: The central interface for constructing, solving and evaluating models.
2. Application Builder. COMSOL alone allows its users to create apps tailored to specific simulations so that more users can apply the COMSOL solver without needing to have and learn the complete COMSOL program.
3. Model Manager: Organizes and tracks versions, ensuring streamlined workflows and robust collaboration.

This software is widely used in various industries, including renewable energy and electronics, automotive and aerospace, and it is used to simulate just about anything, including wind turbines, battery design, and all sorts of cons

At the COMSOL Conference held in Boston recently, the turns to brain cancer. Not what you were expecting: a talk about cancer at a conference for simulation software users? Get used to it. It’s day 3. I have learned it is called COMSOL Multiphysics for a reason. The physics spans multiple disciplines, including medicine. Day 1, I learned how a medicine in a pre-filled syringe can jam and now this … hyperthermic treatment (HT) of brain tumors.

However, on the main stage is Dr. Dario Rodrigues, a medical researcher from the University of Maryland in Baltimore.

The good doctor has the audience riveted. He masterfully presents the problem (94 thousand new cases of brain cancer every year, with 28% of them malignant). For the worst cancers, such as glioblastoma, which has only a 15% chance of a good outcome, Dr. Rodrigues offers his favored treatment: hyperthermia – turning up the temperature on the tumors – and describes devices that will kill cancer cells safely (without damaging healthy cells).

How Does Hyperthermia Work?

Hyperthermia is derived from the Greek terms υπέρ/ypér, meaning “above, “more,” or “over,” and θερμός/thermós meaning “hot.” Hyperthermia therapy (HT) is most often used to describe the therapeutic technique of induced temperature increase through thermal energy delivery. – National Library of Medicine

Hyperthermia creates a hotspot in the brain that is 44 to 44° C for varying periods. Microwaves, RF, lasers can supply the energy…it doesn’t matter. The important thing is that the beams focus the energy on a small volume, so it only affects where it is aimed and not damage healthy tissues nearby.

The physics of energy penetration means that hyperthermic treatments are most common close to the surface, like skin cancer. Several clinics have opened to do this. However, the University of Maryland, where Dr. Rodrigues works, is able to work in deep-seated tumors, such as cervical, prostate, bladder and rectal cancer.

Hyperthermia can be used in conjunction with other treatments: surgery, immunotherapy, chemotherapy and radiation. In fact, one study found that if hyperthermia is used with radiation therapy, patients can have up to 5x better outcomes, which doctors define as being 2 years cancer-free.

Combined with surgery, hyperthermia will have an even better outcome – so long as the surgery is done immediately. Don’t wait a month, he says.

The surgeon “scoops out the tumor,” then the patient has to wait 30 days for hyperthermia treatment. Surgery will get most of the cancer cells but leave nearby cells unharmed. Thirty days late those cells have moved on — as much as centimeters away, explains the doctor.

Microwaves, which stimulate the water molecule, causing the lone oxygen atom to flip back and forth, generating heat, are perhaps the best energy source to use. HT microwaves are either 915 Mhz or 433 Mhz, the lower wavelength for going deeper under the surface.

The deadliest for form of brain cancer, glioblastoma, was removed successfully with hyperthermia 31% of the time, compared to a mere 15% of the time with surgery, according to a study, says the doctor.

COMSOL to the Rescue

The SAR radiation model is created with COMSOL Multiphysics and models the skin, fat, skull, spinal fluid and brain tissue.“ The interaction of fluid, solids, and electromagnetics make this truly multiphysics. Therefore, Dr. Rodrigues finds COMSOL indispensable to his research.

Dr. Rodrigues uses COMSOL to model the brain undergoing HT to predict the efficacy of HT devices under development. By varying the number of amplifiers, antennas and placement on a device worn around the head can be simulated. The cancer-killing hotspot shows up in red. He also likes that COMSOL’s visualization will let him customize the color mapping.

“Won’t it also help with how hyperthermia could increase the permeability of the blood-brain barrier?” asks an audience member.

Dr. Rodrigues: Someone on the staff is looking into that.

“COMSOL is wonderful for these studies,” says the doctor.

Despite all attempts to popularize it, CAE remains in the domain of specialists. CAD companies, in an effort to expand their business, have acquired CAE companies and tried to simplify analysis so even designers can use it, but beyond the most straightforward stress analysis, CAE is attempted only by a small portion of design engineers and a smaller portion of designers.

Let’s face it: simulation is hard. It’s hard enough when the simulation involves only one discipline but when a simulation spans multiple disciplines, it becomes near impossible. An engineer can get bogged down in unfamiliar theory. With time pressures, they could make wrong assumptions, miss important details, get results that are way off — and never know it.

What is a product engineer, an expert in one discipline but a babe lost in the woods in every other, to do when faced with a challenging analysis that involves multiple disciplines?

They can call in the cavalry.

Cue Veryst

Holding a class at the recently concluded COMOL 2024 Conference in Boston titled “Solving Challenging Multiphysics Problems” is Veryst. This consulting firm has built a business in guiding simulation users out of the fine mesh (literally) they often find themselves when trying to use COMSOL Multiphysics and other simulation programs.

One such challenge presented: a prefilled syringe that clogs.

This would baffle a mechanical engineer finding nothing in the geometry of the syringe to cause such behavior. But Joseph Barakat, PhD, a chemical engineer at Veryst, notices that the needle clogged only when the plunger was pressed quickly. The cause: particles in the fluid that can’t get out of their own way, and get stuck. That happens with a non-Newtonian fluid, he says.

I flashback to Bird’s custard powder, always in the cupboard of our London house when I was young, and the mysterious way it resisted the spoon at times. It was not explained by a mechanical engineering education, in which fluids were always fluid — not somehow turning solid.

Barakat is on the verge of stealing the show. A PhD in chemical engineering from Stanford, sealed with a dissertation on “Microhydrodynamics of Vesicles in Channel Flows,” Barakat is nevertheless exceedingly polite and personable. Rheology is his specialty, and he is able to explain the intricacies of it in a way the typical (non-chemical) engineer can understand.

“It’s the same phenomenon you encounter with corn starch,” he says of the clogged syringe problem.

After the class, I sought out Veryst to thank them for solving a childhood mystery. There’s Barakat humbly pointing out that such knowledge is standard to his discipline. He has respect for all engineering disciplines, though he can’t resist sharing a joke about 3 of them with different perspective on fluids.

An aerospace engineer is bragging about designing Mach 10 aircraft.

A mechanical engineer thinks Mach 1 is plenty

A chemical engineer says “What’s Mach?”

It turns out these chemical engineers have been having a lot of fun with non-Newtonian fluids. Their favorite demo: filling a bowl with a water/corn starch mix and slapping at it towards the unsuspecting and amusing themselves at their overreaction. Haha, the mix clogs and won’t splash. Haha. That oobleck*. So much fun.

We Meet the Team

I had the pleasure of meeting Matthew Hancock, a partner at Veryst, and every bit a gentleman. Veryst (pronounced VER ist) is a made-up name, he says. We were called something else but that domain was taken.

I’m having a blast learning how analysis can be used to solve problems in multiple fields — in no small part due to a Veryst’s presentation. I came to the conference to learn about COMSOL’s ability to handle multiphysics. I left impressed by Veryst, which has among its staff a profound knowledge of various disciplines and the skills to use COMSOL Multiphysics.

These guys are super impressive. The typical engineer considers themselves smarter than the average Joe with a bachelors in engineering, but Veryst, based in Needham, MA, is packed with PhDs, many of them earned at Eastern elite schools, such as MIT and Harvard.

Six of the posters stem from Veryst’s cadre of engineering expertis. Is the COMSOL conference at risk of becoming a Veryst conference? A branding problem for COMSOL, perhaps, but the quality of posters would not suffer.

AltaSim

Also with a booth and posters was Veryst’s main competitor, AltaSim. AltaSim has fewer posters but a commanding position at the exhibit entryway with a booth stocked with brochures and manned by a director of business development and a VP of sales. Clearly, these guys are good at marketing.

They are no slouches in COMSOL, either. Established in 2002, AltaSim, headquarters outside Columbus, Ohio, claims expertise in COMSOL, completing over 850 projects spanning many of COMSOL’s physics. They will take on the actual simulation, train users in how to do it themselves, or, if the company does a particular analysis over and over again, make the customer an app for it.

_________

*From a Dr. Zeuss book titled Bartholomew and the Oobleck, where a strange green substance falls from the sky. Oobleck is a non-Newtonian fluid, meaning it behaves like both a liquid and a solid depending on the pressure applied to it. It’s made by mixing cornstarch and water, typically in a ratio of about two parts cornstarch to 1 part water. When you apply force to oobleck—like squeezing it or hitting it quickly—it acts like a solid, resisting the pressure. However, if you touch it gently or let it sit, it flows like a liquid. It is a school science experiment every chemical engineer remembers.

The COMSOL conference returns to Boston after a 5-year hiatus. Thanks, COVID.

Once again, I am delighted to be able to cover good products and companies with EngTechnica. Here, we have COMSOL, arguably the first and best multiphysics simulation software. Ansys may claim to have multiphysics. Siemens, too. But both have bolted on the physics they didn’t initially have. For example, Ansys, a structural FEA program from the start, began calling themselves a multiphysics solution after acquiring Fluent, a CFD program. Sure, it was FEA + CFD. But COMSOL was multiphysics from the start. It’s in the name: COMSOL Multiphysics. And they’re still ahead of the pack.

Most of the presentations posted at the conference rely on COMSOL to go across the engineering disciplines, such as one on EV batteries, that involve electronics, chemistry, thermal and mechanical engineering.

COMSOL also distinguishes itself by letting you create apps. Here’s how it works. Suppose you have an oft-occurring situation that demands simulation. Say someone in the field needs to determine how long a truckload of apples can stay fresh. Stay with me here; this actually happened. The apple deterioration can be simulated with COMSOL Multiphysics, believe it or not. But where the simulation needs to happen, at the apple cart, so to speak, it is not practical to have a simulation-savvy engineer be present. Wouldn’t it be nice to have an app for this, to punch in the number of apples and where the apples are to be delivered and have the app do all the rest, have it sense prevailing conditions along the route and give you the number of days you’ve got. All the calculations are done in COMSOL Multiphysics, all of them under the hood. This is the Coldivate app, available in the Apple Store and for Android. Sure, it’s a niche, and there’s no telling how popular this is. Currently, there are no reviews on the Apple Store. But the point is…

You can package up a simulation for any case with an app that can be helpful for anyone, anywhere. COMSOL is to be commended for this. No COMSOL license is required to run the app. COMSOL gets no revenue from this. Their play is like Adobe with PDF. It costs money to create an Acrobat file, but everyone can read Acrobat files for free.

It’s genius. And not just because I attribute genius to those who agree with me.

Andy Fine has been in the design and engineering software world for quite some time. He has learned a lot. Now, he wants to impart that knowledge to startups. He’s no VC shark, looking for a chunk of equity and taking up a board seat forever. Startups can retain Andy for as long as they wish. On retainer, Andy will take the time to study the startup and their business plan and, among other things, suggest a go-to-market strategy. Perhaps most importantly, he can also help startups get funding, which he has done on two occasions.

The wide world of “Software for Hardware” is Andy’s backyard. He has worked for several of the companies above, dealt with many more and appears to be familiar with them all.

Born in the UK and now outside Denver, Andy is the founder of Fine Physics. He has an extensive background in FEA and CFD, using it early on while earning a Masters in Aeronautical Engineering from Cranfield University in the UK and logo on top of a bachelor’s in aerospace engineering from the University of the West of England.

After a stint as a hardcore user, Andy quickly got into sales. His resume reads like a Who’s Who of simulation, including Ansy and Altair – and we’re only on the A’s.

Andy now seeks to apply what he has learned to help young companies, some with founders who may need to learn how to spell FEA or CFD, into how they can do more with simulation software and get to a better-finished product with fewer prototypes and testing.

With the surge of AI, there could be no better time to use simulation without knowing its intricacies. For those who insist on learning how, Andy is confident he can demystify the usually very daunting software.

I asked about what the AI role should be in simulation. To the two items on my AI-for-simulation wishlist (A, converging quickly to a solution and B, smart refinement of a mesh), Andy added another:

AI should be able to suggest shapes based on the stated design intent. The shapes would have been predetermined to be valid. Then, the simulation can analyze those shapes in all the details necessary and at the time available. A few good shapes are better than thousands of shapes, such as generated by generative design programs.

What does Andy think about CAD companies trying to pass off every design tool as AI?

He laughs. “A lot of stuff is being placed under the AI umbrella.”

Also, Altair and SImScale are making smarter meshes, finer where they should be and coarse elsewhere.

Andy is working with a company that is using AI to create 2D drawings from 3D models. Its ability to infer tolerances may put it ahead of established CAD companies that are also making an attempt at automating dimensioning.

His other company on retainer is Key Ward in Berlin, which he helped secure €1 million. Key Ward specializes in extracting valuable insights from massive datasets, such as those created by simulation programs. Using low-code instead of SQL queries of programming, they hope to develop systems that allow more engineers to use simulation intelligently – as opposed to the perception of CAD companies just dumbing down the interface and giving it to untrained designers.

You can find out more about Andy Fine and connect with him on LinkedIn.

In what may be a significant leap forward for the visualization of computational design in the virtual world, nTop has announced a strategic collaboration with NVIDIA to integrate the latter’s AI-based OptiX ray tracing engine into its shape optimization program and NVIDIA’s Omniverse. The partnership comes with funding from NVentures, NVIDIA’s venture capital arm. The amount of funding was not disclosed.

While Optix ought to improve the look of the rendered parts produced and the speed of the renderings of parts made by nTop’s computational design products and assemblies, the bigger play is NVIDIA’s Omniverse, which allows designers and engineers to interact with the digital twins of the parts and assemblies, all showing up rendered in the Omniverse without having to have them remeshed.

Background

nTop, previously known as nTopology, is an independent vendor of shape optimization software, a design aid typically offered within a CAD program. nTop’s last independent competitor, Frustrum, was acquired by PTC in 2018.

Staying independent seems to have worked for nTop, letting its mastermind, CEO and founder, Bradley Rothenberg, steer it towards motivated customers that can clearly benefit from the company’s next-generation design/simulation technology — rather than have it included in general-purpose MCAD programs and try to convince users that they need it. nTop counts 400 companies as paying customers, including Lockheed Martin and Tesla.

The company received $65 million in funding in 2021, bringing its total funding to $135 million over the years.

NVIDIA, maker of GPUs, is currently on top of the AI tsunami. Demand for its GPUs, which are used in data centers, has made NVIDIA the darling of Wall Street, pushed its stock sky-high, and made it the most valuable company in the world. Unlike most chip manufacturers, NVIDIA creates software its chips can take advantage of, including OptiX, which uses AI to drastically cut the time to produce photorealistic images, improving movie and game scenes and scientific visualization.

How nTop Plans to Use OptiX

The primary application of OptiX in nTop’s platform will be to produce real-time visualization. The OptiX framework is designed to take full advantage of NVIDIA’s GPUs, enabling rapid rendering of parts, assemblies, and complex scenes. For nTop users, this means they can get instantaneous feedback on design changes. For example, every design iteration where a part’s shape changes, its designers will be able to see how these changes affect the model’s appearance in real time.

Tight integration between nTop and OptiX may eventually lead to the same parts and assemblies moving into the virtual reality of NVIDIA’s Omniverse, all without having to be rendered again.

Streamlined Workflow and Efficiency

The integration of OptiX into nTop’s platform is not just about adding new capabilities; it’s about making the computational design process more efficient. By marrying ray tracing to computational design with OptiX, the user avoids having to remesh and have it rendered again when it enters the virtual world. With the enhanced visualization and optimizations done once for the real world, entry into the virtual world could be seamless.

Conclusion

NVIDIA’s OptiX could bring beauty to nTop’s brawn. Generative design and computational design program have always been about how the parts perform, not about the way they look. While no engineer will admit they like how their products look, there is no doubt that a good-looking, realistic model adds to its sales appeal. Just as color changed our TV and movies with the realism the people back then thought they didn’t need.

Even if step improvements are gained at great cost (all those GPUs and server farms are not cheap), realism will always rachet up. And designers and engineers feel better about what they are designing. By offering accelerated simulation coupled with high-fidelity rendering, this collaboration could bring about a more intuitive and efficient design process on the desktop and in the virtual world.