MMM...Aerodynamics...
Registered User
Joined: Apr 2003
Posts: 22,888
Likes: 3
From: Covington WA, USA
Originally Posted by slimjim8201,Oct 6 2005, 11:03 AM
For the top open case:
Velocity = 120 mph
Drag = 412.62 lbf
Downforce = 21.58 lbf
Surprised that it generated downforce. I imagine this will go away and become lift at even higher speeds.
Velocity = 120 mph
Drag = 412.62 lbf
Downforce = 21.58 lbf
Surprised that it generated downforce. I imagine this will go away and become lift at even higher speeds.
Registered User
Joined: Apr 2003
Posts: 22,888
Likes: 3
From: Covington WA, USA
Originally Posted by DrCloud,Oct 6 2005, 05:16 PM
Given the speed you're using, I'll assume that this is a compressible flow simulation
I assume it's inviscid.
The model and grid doesn't look detailed enough that it matters much what kind of turbulence and separation models are being used.
Thread Starter
Registered User
Joined: Jul 2005
Posts: 1,670
Likes: 0
From: Gie
Originally Posted by DrCloud,Oct 6 2005, 08:16 PM
Although this comment may clarify things somewhat, it may not. The two remarks in previous posts about the physical models (the paper one and the diecast one) suggest that some people think you're doing this in a wind tunnel with some kind of nifty visualization. (It's not, of course, this is all computers.)
Also, the 45 minute run-time is machine dependent: what're you using?
Not to knock your company's product, but canned CFD packages are always compromises, because they're designed to simulate anything. Nonetheless, this one looks pretty good, in the context of the rather coarse mesh. Given the speed you're using, I'll assume that this is a compressible flow simulation; further, the "iteration" note on the pix suggests a non-linear solver of some kind (for the pressure field, I'd guess). This, of course is more accurate than simple predictor-corrector time stepping.
If you're going to pursue refinements of this (and more power to you if you do!), I'd suggest curvature in the windshield. That, I think, will change the top-down streamlines dramatically. HPH
Also, the 45 minute run-time is machine dependent: what're you using?
Not to knock your company's product, but canned CFD packages are always compromises, because they're designed to simulate anything. Nonetheless, this one looks pretty good, in the context of the rather coarse mesh. Given the speed you're using, I'll assume that this is a compressible flow simulation; further, the "iteration" note on the pix suggests a non-linear solver of some kind (for the pressure field, I'd guess). This, of course is more accurate than simple predictor-corrector time stepping.
If you're going to pursue refinements of this (and more power to you if you do!), I'd suggest curvature in the windshield. That, I think, will change the top-down streamlines dramatically. HPH
4 GB Ram
Top of the line without being 64 Bit
www.cfdesign.com
We are the leading Upfront CFD company in the world. Feel free to read about us on the website.
Some interesting customers include
-Trek (Lance Armstrong's bikes and helmets)
-NASA
-Richard Childress Racing (NASCAR)
-Bugatti
I will have to check but I think we may be over 1000 customers strong now. I know for a fact that we have sold well over 1000 seats of software. Perhaps 1500 now.
I ran this analysis incompressible. 120 mph is only around mach 0.2. Compressible analysis are only valid for external flows above mach 0.3.
I am unsure about your linear solver question. I do know that we have over 20 solvers that can be used. The software picks the best solver for the analysis as it progresses. It will even change from solver to solver mid-analysis if it will produce a better result. The turbulence model I used was the industry standard workhorse, k-epsilon.
I can assure you. CFdesign is anything but a canned CFD package. We have been putting canned CFD packages out of business for over ten years now. We are even starting to pull customers from FLUENT
Registered User
Joined: Apr 2003
Posts: 22,888
Likes: 3
From: Covington WA, USA
Originally Posted by slimjim8201,Oct 6 2005, 05:38 PM
Compressible analysis are only valid for external flows above mach 0.3.
Thread Starter
Registered User
Joined: Jul 2005
Posts: 1,670
Likes: 0
From: Gie
Originally Posted by mikegarrison,Oct 6 2005, 08:38 PM
120 mph? That seems pretty incompressible to me, actually. Or at least, it seems like the benefits from using compressible flow would be small compared to the computational costs.
I assume it's inviscid.
The model and grid doesn't look detailed enough that it matters much what kind of turbulence and separation models are being used.
I assume it's inviscid.
The model and grid doesn't look detailed enough that it matters much what kind of turbulence and separation models are being used.
Tomorrow I will throw up some results from an analysis that is running overnight with approx 500,000 elements. Top down with a generic rear wing. Should be interesting.
You seem somewhat familiar with CFD. What packages have you used in the past?
Thread Starter
Registered User
Joined: Jul 2005
Posts: 1,670
Likes: 0
From: Gie
Originally Posted by mikegarrison,Oct 6 2005, 08:44 PM
Surely they are valid at low speeds, right? The question is just whether the compressibility changes the answer enough to be worth the computation.
Registered User
Joined: Apr 2003
Posts: 22,888
Likes: 3
From: Covington WA, USA
Originally Posted by slimjim8201,Oct 6 2005, 05:47 PM
You seem somewhat familiar with CFD. What packages have you used in the past?
Registered User
Joined: Jun 2005
Posts: 3,077
Likes: 0
From: EstesPark/BocaRaton
There's no point in debating the fine points of CFD here -- but note my screen name, and believe me that it's necessary to use the compressible Navier-Stokes equations for simulations of thunderstorms and various other atmospheric phenomena, if you want to capture all the important interactions (in which, of course, Mach number is negligible). Incompressible, inviscid flow would work if the medium were water, though. HPH
Registered User
Joined: Apr 2003
Posts: 22,888
Likes: 3
From: Covington WA, USA
[QUOTE=DrCloud,Oct 6 2005, 06:03 PM] There's no point in debating the fine points of CFD here -- but note my screen name, and believe me that it's necessary to use the compressible Navier-Stokes equations for
Thread Starter
Registered User
Joined: Jul 2005
Posts: 1,670
Likes: 0
From: Gie
Originally Posted by mikegarrison,Oct 6 2005, 09:15 PM
Yeah, but for the purpose of measuring force on solid bodies in a uniform flowfield, the 0.3M rule does usually suffice. Inviscid, incompressible calculations give a pretty good answer, and are much less costly than full-up 3-D compressible N-S. It all depends on what you are trying to do. Realize that you are trying to see what is happening inside the fluid, but pretty much slimjim just cares about what is happening at the surface of the car -- the details of what is going on inside the fluid only matter for how they influence the surface solution.
[edit: BTW, it is the peak speed of the flow that is the important parameter here, not the speed of the uniform flowfield. So if the high-velocity areas of the flow get into the compressible speed range, it doesn't matter that the overall flow velocity is only 120.]
[edit: BTW, it is the peak speed of the flow that is the important parameter here, not the speed of the uniform flowfield. So if the high-velocity areas of the flow get into the compressible speed range, it doesn't matter that the overall flow velocity is only 120.]
