|Published on:||June 29th, 2021|
|With:||Eric Miller & Tom Chadwick|
In this episode your host and Co-Founder of PADT, Eric Miller is joined by Tom Chadwick, Senior CFD Engineer at PADT to discuss what’s new regarding heat transfer and flow in the latest Ansys release.
If you have any questions, comments, or would like to suggest a topic for the next episode, shoot us an email at firstname.lastname@example.org we would love to hear from you!
ANSYS Discovery is a wonderful tool for fast and first look structural, fluid flow, and thermal simulations. Discovery gives us the ability to modify geometry very quickly within the interface and to add or remove features to view realtime simulation reactions. This allows us to quickly iterate, explore design changes, and better understand interaction between our design and the environment. Today we are going to be investigating the pressure profile on the exterior of a travel trailer being pulled behind a truck down the highway.
In this analysis we are using a 2018 Chevy Silverado 2500 model pulled from GrabCAD with a generic 25 foot travel trailer. Based on described experience we’ve noticed that the roof and sidewalls of a moderately sized travel trailer seem to bow outwards at highway speeds near the front of the trailer.
The first thing to do is to put this model in an enclosure and prescribe a flow condition at the inlet and a pressure boundary at the outlet. Modeling the truck/trailer combo at a speed of ~55 mph (25 m/s) confirms that there is suction (negative pressure) present in these key areas:
While the actual roof and sidewall separating could be attributed to poor manufacturing processes we wonder if there could be a design change to minimize negative pressure/suction. One idea would be to incorporate some sort of turbulator to break up the laminar flow. I’ve seen turbulator tape in a zig-zag pattern used in aviation for this specific purpose so we’ll try recreating the travel trailer equivalent and see how it goes.
I started with a zig-zag pattern about 4″ tall on top of the trailer to see if I could “pop the bubble”
This did have the intended consequence and it was curious to see how much impact the turbulator on top of the trailer had on the negative pressure at the sidewall of the trailer:
The next thing I wanted to try was moving the turbulator forward or backward to see the effects. Moving the turbulator towards the aft of the vehicle has limited effects but moving it to different locations within the suction “bubble” seems to effect our -500 Pascal isosurface:
This would seem to indicate the presence of a “sweet spot” for turbulator location that merits further research in either the “Analyze” mode within Ansys Discovery or within Ansys Fluent.
Before I hang up my coat I’d like to investigate one alternate design that I’ve seen more often in automotive applications. I’m going to try adding vertical pillars and see how that goes:
We can easily change the height and position of the pillars to see the resultant effects on the pressure isosurface. The pillars also have a significant effect on the suction bubble but I notice that it has less effect on the suction on the sides of the trailer.
Using Discovery we can quickly and easily iterate on designs, get a first-view of the physics, and determine which change or design merits further investigation. In this analysis we can see that there is most definitely a suction profile at the front of a generic travel trailer. If the suction proves damaging we can see that there are several design changes which will help to mitigate this effect.
For more information on ANSYS Discovery please reach out to email@example.com.
If you would like to play with the models themselves, you can download a zip file with all three models here.
This post was created based on the expert advice of PADT CFD engineer and Project Lead, Nathan Huber.
Simulating the behavior of liquids and gases has become a standard part of product development in products where fluid behavior plays an important role. Here at PADT, we have been using Computational Fluid Dynamics, or CFD, for years to model everything from combustion in turbine engines to cooling of electronics, to golf balls. With that experience, our estimates for a given project have become reasonably accurate.
However, we can only estimate accurately if we have complete and accurate information on what you need simulated and what you hope to gain from the simulation. To help everyone arrive at more accurate cost and schedule estimates, even if you are planning a project internally, we offer the following list of five questions we always ask:
Before we can do anything, we need to have an agreement in place that clearly defines how both sides handle proprietary information. When we have tried holding meetings to gather information for a quote before an NDA is in place, we almost always waste time. There is just too much that is proprietary in most products.
We also need to know the physical geometry of your system. That is why we ask for an accurate and complete CAD model. We take some time to poke through the files in our software to make sure we can use the geometry, it is accurate, and it has the level of detail required for CFD. Basically, we check to see if we can pull a fluid domain from your CAD models. Remember, we are not simulating the solid part of your product; we are modeling the inverse and therefore need to pull a negative volume from your geometry.
Now that the geometric domain is understood, we need to know what is inside that domain, and what is acting upon it. We will ask you for boundary conditions, and for the material properties of the fluid or fluids you are asking us to model. The complexity, time variation, and severity of the loads drive the difficulty of setting up and running the simulation. And the material properties can also impact the sophistication of the model as well as its robustness. Both, therefore, have a significant impact on cost.
When a simulation finishes, it can be post-processed to get a vast array of plots, figures, animations, pretty pictures, etc. Those take time to create, so we need to know what you want to see. Also, we set up some post-processing parameters before we start the simulation.
The whole point of doing a CFD simulation is to study the behavior of your system. We need to know what behavior you need to understand so we can make sure that the simulation we propose answers your questions and guides you in your design process.
We hope you find this review useful when you are planning your internal CFD project as well as those you outsource. And speaking of outsourcing, please consider PADT as your resources for any future simulation projects of any type, not just CFD. Now, you already know what questions we will ask.
Products such as ANSYS Fluent, CFX, and Ensight work together in a constantly improving tool kit that is developed to provide ease of use improvements for engineers simulating fluid flows and the impact those flows have on physical models.
Fluids simulation users will find that ANSYS 2019 R3 includes many enhancements that further simplify the user experience and broaden use to new applications. The new Fluent experience has been improved so you can enjoy more CFD in less time, with less training.
Join PADT’s Simulation Support and Application Engineer, Sina Ghods, for a look at what is new and improved for fluids simulation tools in ANSYS 2019 R3. This presentation includes updates regarding:
If this is your first time registering for one of our Bright Talk webinars, simply click the link and fill out the attached form. We promise that the information you provide will only be shared with those promoting the event (PADT).
You will only have to do this once! For all future webinars, you can simply click the link, add the reminder to your calendar and you’re good to go!
Below you can find some of the most frequently asked questions from this webinar, along with our answers.
Q: What computer and license type are you actually using for running the simulation? Ram, cores, CPU, etc.
A: 16 GB RAM Intel i7, NVIDIA Quadro K1100M
Q: Is the solver running a transient solution in the background while you are post processing and haven’t paused it, or is it marching towards better convergence of a steady state solution?
A: The solver is running a transient solution the entire time. The only time it is not doing this is if I pause the simulation.
Q: Does this require 3D models to be built within Discovery Live, or it can import outside 3D models (Solidworks etc) and still allows the user to add/modify geometry objects on the fly within Discovery Live?
A: Yes, you can import any CAD native format or neutral format in Discovery Live. You can also create Geometry from scratch in Discovery Live using the SpaceClaim tools.
Q: It looks like when you change the “speed – fidelity” setting, the solver resets; while when you change other settings (like domain geometry), the solution continues from last step. What’s being changed when you change the “speed-fidelity” setting?
A: When you change the Speed-Fidelity setting, the computational grid of the domain is coarsened or refined. It is also re-partitioned/parallelized on the GPU cores.
Q: Are there predictive features of simulation, obtaining design suggestions from the end results? Are there inputs for end results to accomplish reverse engineering?
A: Yes, in a sense. Because Discovery Live provides literally “real-time” feedback on design changes, you have design suggestions based on the end results. For reverse engineering in Discovery Live: there are not inputs for end results, but the part you’d like to reverse engineer can be imported (or created) in Discovery Live, and then analyzed in the range of operating conditions to understand it’s performance.
Q: Do you have any control of the mesh? How boundary/layer effects can be checked?
A: Control of the mesh available via the Speed-Fidelity slider bar. Boundary layer effects could be checked using surface integrals of force, but again recall that if you’re trying to finely resolve details such as boundary layer physics, you may need to consider the next level of accuracy which would be Discovery AIM or Fluent.