Blog
Printer-friendly versionPDF version

Cool is a hard thing to define. It’s completely subjective. But you know it when you see it.

There are a lot of ways to present CAE/CFD data. Plots and tables are arguably what we make most of our decisions on. But, “Excel sheets…  aren’t everyone’s friend.1 Scenes then… you can put a lot of things into your scenes; results on the surfaces of the thing you’re simulating, streamlines that go in and around the thing you’re simulating… These visual abstractions are a deeply ingrained part of our engineering culture. But not everyone has a casual familiarity with this language. People with diverse levels of expertise have to make sense of these abstractions. And those who don’t, or no longer, speak the language daily and who typically have the least amount of time to assemble conclusions, also carry the heaviest decision-making obligations.

Maybe some of you are getting ahead of me here, recalling the phrase “Color For Directors,” a phrase I personally find to be demeaning to directors and dismissive of what we do. Cool pictures? Sure, but cool isn’t cool if it isn’t right. I submit that we have an ethical obligation to maintain the fidelity of our data,2 and taking it a step further, we rely on good fundamental data to make decisions. Now, data alone can’t capture an idea. 3 Effective visualizations (cool is implied here) can capture ideas, quickly and easily, inviting curiosity and engaging broader audiences. In STAR-CCM+® v12.02, you can create photorealistic images and animations, reducing the gap between the time needed for you to communicate your messages and the time needed for others to understand practical implications, quickly placing your information into their own knowledge frameworks.

Ray tracing, the underlying technology we are using, has been around for nearly 50 years.4 We’re used to seeing it in the movies we watch5 and the games we play.6 To give you a quick start into this realm of light and shadows, STAR-CCM+ provides basic and preset render materials (gold, chrome, rubber, water…) with control over light absorption, reflection, refraction and visual surface roughness. In the figure below, you can see a mixing impeller casting a shadow on the tank floor. At left, we use a simple render to show the reflection of the mixer shaft on the impeller blades. In the middle, we make a simple change where the impeller blades are scattering light instead of reflecting it. At right, we’ve made a transparent impeller blade. To gain a greater understanding take a look at the leftmost impeller blade in that third figure. Light passes through the blade and is refracted. On the bottom of that blade, the shadow from the tank floor is collected. This explains why the blade shadow and the lighter area corresponding to the hole in the blade are shifted on the impeller surface. Believe me, once you start working with these capabilities, you’ll be looking at the shadows and reflections in the real world around you in a whole new way.

Control of shadows is important since it has the potential to distort scalar displays – color has meaning here. Below, we see the same simulation results with and without ray tracing. The inset images at the bottom show just the CFD data. While there are some minor shading differences, the visual data integrity is largely preserved in the ray traced illustration: detail is added to the car body without losing fidelity in the CFD data.

We’ve also added support for environment maps which are essentially panoramic backgrounds. In the marine applications illustrated below, clouds from the environment map are reflected on the VOF wave surface. You might ask, is this a distortion of the data? I would argue no, the VOF wave surface isn’t artificially smoothed in any way and the shadows and reflections are providing valuable depth cues.


Using ray tracing is also a better way to show highlighting. In the side-by-side illustration below, we see much better context for the highlighted wheels in the ray traced image at left. The image at right, rendered using OpenGL, will be faster in terms of interaction. How much faster depends on your scene content, but keep in mind that you can control ray tracing quality (lower quality settings equate to faster interactive performance).

Finally, ray tracing is CPU-based, not GPU-based: it uses your computer’s processing power instead of your graphics card. It’s possible to do things with ray tracing that you just can’t do with OpenGL. With today’s CPU hardware, you don’t need your own personal render farm to make high impact, effect visualizations - that makes it practical. Another practical consideration is interactivity - doing all the advanced rendering passes can take time. To address this, progressive rendering is automatically invoked when you adjust your view in a scene. Rendering to the final quality level happens gradually, delivering a better user experience while ray tracing is ongoing. Ray tracing runs in parallel, scales well and you can specify the number of rendering compute threads. Below, we can see the scaling performance for an image containing volume rendering and custom rendering applied to the SR-71 surfaces. How well ray tracing scales will depend on what is being rendered. We’ve seen in general that as you increase your quality level, scaling performance tends to be higher.  

Speaking of volume rendering, because we can do this with ray tracing it is now possible to generate volume rendered pictures on systems where there are no graphics cards. If you’ve ever struggled with volume rendering on a cluster, this could be a solution for you.  

When is a picture really worth a thousand words? When it is effectively able to communicate ideas, when it can show why the best design is the best design and when it reaches everyone in the room with an economy of explanation. And, being cool doesn’t hurt either. Cool is truly cool when it is right.

  1. Data visualisation: Contributions to evidence-based decision-making. A SciDev.Net Learning Report (2016) Retrieved from https://social.shorthand.com/SciDevNet/3geA2Kw4B5c/data-visualisation-contributions-to-evidence-based-decision-making

  2. Stempeck, M., DataViz for good: How to ethically communicate data in a visual manner (2016), Microsoft. Retrieved from https://blogs.microsoft.com/newyork/2016/01/20/dataviz-for-good-how-to-ethically-communicate-data-in-a-visual-manner-rdfviz/

  3. Berinato, S., Visualizations That Really Work (2016), Harvard Business Review. Retrieved from https://hbr.org/2016/06/visualizations-that-really-work

  4. Appel, A. “Some Techniques for Shading Machine Renderings of Solids,” Spring Joint Computer Conference, (1968), 37-45.

  5. Christensen, P.H., Fong, J., Laur, D.M. and Batali, D., Ray Tracing for the Movie ‘Cars’, Proc IEEE Symp Interactive Ray Tracing, pp 1-6 (2006). Retrieved from http://www.seanet.com/~myandper/abstract/rt06.htm

  6. Pohl, D. Ray Tracing Egoshooters, Quake 3: Ray Traced (2004). Retrieved from http://www.q3rt.de/

Matthew Godo
STAR-CCM+ Product Manager
Stephen Ferguson
Marketing Director
James Clement
STAR-CCM+ Product Manager
Dr Mesh
Meshing Guru
Joel Davison
Lead Product Manager, STAR-CCM+
Ravindra Aglave
Director - Chemical Processing
Sabine Goodwin
Director, Product Marketing
Karin Frojd