I played around with Flow Illustrator. I can't figure out if I'm using it wrong or if it is really inaccurate. Here is a 256x128 mask of Phil's template.
The flow field is not hugging curvature at all. The tail is shedding a KV street.
Your RE is off the chart, Dr C himself sez to use between 100 & 1000. Here is a copied quote from the "More Info" tab on his website.
"Very crudely, for bluff bodies (like cars, or any body the flow past which is separated) effective Re, if based on the body largest cross-section dimension, is usually between 100 and 1000. Note that the laminar, numerical, and turbulent viscosities differ not only in magnitude but in many other respects, so that one can never be accurately modelled by the other."
So Re set to 2.5 million is a bit outside this.
Also, I have noticed a huge difference in how shapes perform in "free air" and "ground effect". The ideal template in free air acts quite different than when close to the ground.
Also keep in mind, this flow illustrator is just kind of a toy, it is very difficult to set it up accurately to use as an actual predictor of real world flow depiction.
I set Dt=.2 it simply speeds things up a bit, and time of more than 15 Sec does not accomplish much.
Hope this helps. Don't forget to check out the "More Info" tab on the Flow Illustrator site.
I played around with Flow Illustrator. I can't figure out if I'm using it wrong or if it is really inaccurate. Here is a 256x128 mask of Phil's template.
The flow field is not hugging curvature at all. The tail is shedding a KV street.
Bob,I think that Flow Illustrator represents only the centerline flow in 2-D flow and is not sophisticated enough to represent the aft-body of any 3-D structure.It's why I never gravitated towards these CFD programs.
Hucho is quite explicit about the limited usefulness of CFD,especially with respect to the wake.
Current CFD technology IS sophisticated enough to accurately model wake flow,however I believe it is still relegated only to big-budget super-computers.
The template is derived from empirical wind tunnel studies of actual 3-D structures,with appropriate Reynolds numbers.
Anything constructed to its architecture should be separation-free in ground proximity.
The 2.5:1 ground-reflection finess ratio,when cut for vehicle ground clearance produces the 5.0:1 automotive fineness ration Hucho suggests as a 'minimum' for drag,and just squeeks by with respect for Mair's limit of 22-degrees max tangent angle for separation-free aft-body flow.
The ground clearance 'cut line' actually ends at about 80 % of the reflected image.This 20% of tail would demonstrate phantom flow,and from fuselage research can be cut away with no appreciable increase in form drag.
Steeper aft-bodies violate Mair's 22-degree limit and suffer separation.
Longer aft-bodies suffer skin friction increase due to the greater wetted area,although I admit that with respect to vehicles this is almost laughable.
Thanks for chiming in, Phil. My goal was to validate the simulation against a problem with known result. Flow should hug the template until very late. (It must separate eventually which is probably why late tail chopping does little harm IMHO.)
I first ran a cylinder, hoping to see some regime change as a function of Re but no dice. That's when I switched to foils before giving up. (To its credit a pure white image produced a clean flow field.)
Unless parameters can be tuned by someone more talented than yours truly, I'm inclined agree with Chernyshenko that these (finite difference) simulations are a toy for art not a tool.
Thanks for chiming in, Phil. My goal was to validate the simulation against a problem with known result. Flow should hug the template until very late. (It must separate eventually which is probably why late tail chopping does little harm IMHO.)
I first ran a cylinder, hoping to see some regime change as a function of Re but no dice. That's when I switched to foils before giving up. (To its credit a pure white image produced a clean flow field.)
Unless parameters can be tuned by someone more talented than yours truly, I'm inclined agree with Chernyshenko that these (finite difference) simulations are a toy for art not a tool.
Bob,her are some hard data that may shed some light on the CFD model:
*the original Jaray/Klemperer 'pumpkin seed' wind tunnel model of 1922 registered Cd 0.13.The model is virtually the template out to 65.5 % aft-body,then it takes a dive below 22-degrees,ending at 75 % aft-body.
* Kamm tested a simplistic version of Jaray's pumpkin seed in 1935.This model had zero tumblehome and zero body radiusing at edges.This model had Cd 0.21.
* Kamm's K-form model is the same length as the Jaray but his roofline departs the template at around 17 % of aft-body,arcing down to 25 degrees at rear terminus.This model also scores Cd 0.21.
*Walter Lay's model of 1933 follows the template to 88 % aft-body.This model,while having no tumblehome,does have plan taper beginning at the max camber point.The plan taper boat tail angle is a constant 12.5 degrees all the way to the tip.This model scores Cd 0.12.
--------------------------------------------------------------------------
I'm not sure about all the parameters Flow Illustrator will accept,but if there was a way to include any of these features,then we'd have hard numbers with which to compare the results.
Bob,her are some hard data that may shed some light on the CFD model:
*the original Jaray/Klemperer 'pumpkin seed' wind tunnel model of 1922 registered Cd 0.13.The model is virtually the template out to 65.5 % aft-body,then it takes a dive below 22-degrees,ending at 75 % aft-body.
* Kamm tested a simplistic version of Jaray's pumpkin seed in 1935.This model had zero tumblehome and zero body radiusing at edges.This model had Cd 0.21.
* Kamm's K-form model is the same length as the Jaray but his roofline departs the template at around 17 % of aft-body,arcing down to 25 degrees at rear terminus.This model also scores Cd 0.21.
*Walter Lay's model of 1933 follows the template to 88 % aft-body.This model,while having no tumblehome,does have plan taper beginning at the max camber point.The plan taper boat tail angle is a constant 12.5 degrees all the way to the tip.This model scores Cd 0.12.
--------------------------------------------------------------------------
I'm not sure about all the parameters Flow Illustrator will accept,but if there was a way to include any of these features,then we'd have hard numbers with which to compare the results.
Phil, at present with only 4 parameters I don't advise holding one's breath. Making the right friends in academia could be immensely valuable.
There is now a Flow Illustrator mailing list for a beta of a standalone version. If you have written to Dr. Chernyshenko about FI in the past please check your inbox for an invite and respond at least once to the list so they have an idea of actual interest.
I played around with Flow Illustrator. I can't figure out if I'm using it wrong or if it is really inaccurate. Here is a 256x128 mask of Phil's template.
The flow field is not hugging curvature at all. The tail is shedding a KV street.
I think the program is treating the 3-D,2.5:1 'Template' streamline body of revolution as a 2-D streamline section.I don't have Marchaj's table with me but memory( what a joke!) tells me that for best 2-D flow you need 3.92:1 streamline section.That should kill the Karman vortex street your seeing behind the 'Template.'
If you'll go down the home page here at the Aero Forum to the Tread: new mercedes b class 0.24 Cd ,go to page-3,#34(permalink) and click on "Aerodynamik Report" you'll see a 3-D flow image of the VW 1-Liter Car.
If you'll do a Template comparison you will see a very close match and where the VW goes hypo-Template you'll see a corresponding thinning of the smoke as the boundary layer is being compromised.
If you do a 3.92:1 section on Flow-Illustrator you should see something similar.
Or put just the upper half of your image in there and see what that does.
Today
Other popular topics in this forum...
