“What would happen if…?” is a question that humans have been wondering about various aspects of their lives since the dawn of time. But if you are a design engineer, you can’t afford to have that kind of uncertainty in your products. “What ifs” need to be addressed. And, thankfully, engineers have always had their ways of finding out. Simulation optimization is the way to go these days.
Simulation optimization is the process of finding the best input variable values from all possibilities, without explicitly evaluating each one.
Digital simulation: the powerhouse behind the third industrial revolution.
In the past, it was done with a combination of hand calculations and prototyping. In the present, engineers verify their design’s behaviour under various circumstances by digitally simulating it. Why the shift towards simulation? As we’ve discussed before, simulation enables the engineer to test their design in thousands of different scenarios. Then, using the gathered data they can come up with a design that meets the specs requested in an optimal way.
So engineers are optimizing their desings using simulation – simulation optimazation!
Doing that without a simulation tool would be so time consuming that it would be deemed impractical.
However, this isn’t the only benefit of simulation: in some cases, measuring the impact your design changes have on the product is hard and requires expensive equipment. For example, measuring electromagnetic phenomena (e.g. for checking the performance of a touch screen) would require a dedicated lab and experts to run it.
In this case optimization would require a lot of resources. Prototyping designs with small changes is straight up inefficient. Simulation optimization is not!
Even though simulations are truly life saving for engineers and offer unprecedented insight into their design’s behaviour, they also introduce new problems for them to face. Due to the complexity of the products nowadays, these problems are magnified and they need to be addressed. As a result, engineers must pay serious attention to simulation, otherwise the amount of problems they’ll face will be overwhelming. They need to use optimize the simulation practices they use during optimization simulation.
Hardware prototyping and testing requires special equipment.
Running a simulation: a step-by-step analysis of its pain points.
Simulation engineers face challenges throughout the stages of a simulation optimization. Simulation stages are: pre-processing, sending the model to the solver and post-processing.
Let’s examine the pain points of running a simulation, step by step, using the simulation of a touch sensor as an example.
The pre-processing phase is usually the one that takes the longest.
Sometimes, the CAD geometry needs to be recreated within the CAE tool, due to lack of compatibility. This leads to doing double work. Then geometry clean-up begins. Problematic geometries are hard to detect and refine. Also, oversimplifying the geometry can lead to incorrect results.
But this isn’t the only thing that can lead to incorrect results.
Mesh density is also crucial. Meshing is a fine art. Mesh too densely and your simulation time will skyrocket. Don’t mesh densely enough and it can lead to errors. There is one final step before sending the model to the solver: applying boundary conditions. It requires a certain level of expertise.
All these are extremely time consuming, even for seasoned simulation experts. And, to make matters worse, some of these steps will need to be repeated in each iteration of the simulation with different parameters. Iterations with slight tweaks are crucial for optimization simulation.
So, it makes sense that top-performers try to automate as many of these tasks as possible.
When they’re done with preprocessing, engineers send the model to the solver. This step poses new problems of its own.
Due to the complexity of the products, it is common that multiphysics analysis may be required. This means that for each type of analysis, a different solver may be used. On average, companies use 3.3 solvers! The model needs to be optimized for each solver. So, a simulation engineer can be called to prepare 3 separate models and send them to 3 different solvers, adding time, complexity and frustration to the process. And sometimes, when advanced analysis needs to be conducted, engineers may even be called to adjust the solver input file manually.
Simulation engineers face challenges throughout the stages of an optimization simulation
The main point of running a simulation is to get the data you need to make an informed decision. However, as we’ve seen so far, getting simulation results is much harder than it sounds. As a matter of fact, more than 50% of the time spent running a simulation is spent in trying to extract the results and not analyzing them. Data generation and analysis are performed on the final phase of a simulation, the post processing phase.
But even at this stage, engineers need to overcome some difficulties. The amount of time it takes the solver to send data can be long. This can be either due to poor performance of the solver, or due to the data set itself. One of the main pain points of the engineers is the amount of data they get from a simulation: it can be overwhelming. If too much data is received, then the engineer needs to skim through this staggering volume of information to find what they need and focus on that.
But it’s not just the amount of data, their format also has an impact. Often, due to their formatting they can be hard to read, or it can be hard to extract results for further calculations.
Again, top performers take steps to remedy the situation: they use tools to visualize the results or they automate the process of exporting the data in popular programs, such as MS Excel.
Lack of automation in various stages of a simulation can prove to be time consuming
Fieldscale SENSE: an automated electromagnetic simulation software solution for simulation optimization.
Fieldscale is no stranger to these problems. In fact, it was these problems that inspired and motivated the company’s founders to start the company. For us, running a simulation should be as straightforward and as intuitive as it gets. Those were the principals we kept in mind when designing our touch screen simulation tool, SENSE.
SENSE is an application specific simulation tool that can be used to simulate capacitive touch screens in minutes, with no expertise needed. SENSE automates most of the pain points of a simulation that were described above.
Here’s an example of how a touch screen is simulated using Fieldscale SENSE.
During the preprocessing phase, you can either choose to import your custom patterns as a .dxf file, or choose one of the popular patterns that SENSE offers (e.g. Manhattan or diamond) and create your touch sensor model within SENSE without using any hand-drawn geometry. Then, you can choose the stack-up of your liking (again, no need for any 3d modelling, SENSE takes care of that). Then, you can choose what you want to compute.
Geometry clean up? Boundary conditions? Meshing? That’s all automated! SENSE features the knowledge of an expert engineer, hidden behind a super intuitive GUI.
So, with a few clicks of a button, you’re ready to simulate a touch screen. In order to avoid a chaotic amount of data, SENSE lets you choose what you want to calculate: be it resistivity or capacitance, or a parametric analysis, SENSE enables you to only see what you want to see. But that’s not the only step we’ve taken to making the generated data easily digested and manipulated: you can see the results plotted or as a heatmap. Do you need to do further calculations with them? Simply export them to a .csv or .xls file and you’re ready to go!
And all of this is done without compromising simulation accuracy: SENSE’s under the hood powerful algorithms are hidden from the user behind it’s simple to use GUI. The state of the art algorithms deployed by the solver guarantee unparalleled simulation speed as well as verified accuracy.
SENSE features the knowledge of an expert engineer, hidden behind a super intuitive GUI.
The future of simulation: fit-for-purpose simulation technology.
Simulation should be just a tool used by engineers to make informed decisions and needs to be as automated as possible. But not only that: most simulation tools are general and require experienced analysts. The future of the simulation lies on its democratization: simpler, fit-for-purpose tools that work just for niche applications. That will give access to simulation optimization to more engineers.
If someone only simulates capacitive touch screens why should they have to go through a steep learning curve to be able to run the specific simulation they want?
The world is in shortage of engineers in general, even more so in simulation engineers. That’s why, we believe that the future of simulation lies in intuitive, easy to use, with a small learning curve software that non experts can use to get the data they need.
And also, that’s the reason why we created SENSE, a simulation tool that easily allows non experts to simulate capacitive touch sensors. Simulation optimization of a touch sensor has never been easier!
Do you want to experience the future of simulation? You can request a free trial here.
‘A third industry revolution‘, the Economist.
Addressing the bottlenecks of FEA simulation: Enabling Innovation by Getting Even More Value From CAE, Tech-Clarity, 2016.