197 Aerodynamics (CFD analysis)

tomislavp4

Paid Member
Hi guys. I'm running some CFD simulations on a model of a 197 in my spare time and wanted to open a thread so I can share the results with you. The goal is to see what aftermarket parts work and how well they work. Later on I will test out some custom stuff (wings, splitters etc.) as well. Now keep in mind that I am not a professional aerodynamicist so take this with a grain of salt.

The simulations are done on a half-a-car model with sealed engine bay and partially modeled underbody. Models of this kind are usually used by carmakers early on in the concept design stage for comparing different shapes. As these simulations are highly simplified, they tend to underestimate the drag coefficient and overestimate the lift coefficient.

Beow are pictures of the "stock" and the "cup" model used in the simulations and the results.
Stock model.jpg Cup model.jpg Results.png
The stock model has a drag coefficient of 0,365 and a lift coefficient of -0,261 which gives a lift to drag ratio of -0,715. The drag figure is in the normal range for hatchbacks. The lift value however is very optimistic. It means that the downforce produced by the car at top speed is 130 kg which is highly unlikely. As mentioned before, the simulations overestimate the lift values.

The Cup spoiler and splitter increase the drag coefficient by 19 percent but they also increase the lift coefficient by 54,8 percent. The lift to drag ratio goes up 30,1 percent so the car is more efficient at producing downforce with the Cup bits but it will have a lower top speed.

To see what is happening with the flow, I've attached pictures of the pressure and velocity on the symmetry plane as well as pressure on the models themselves. We can clearly see how the spoiler creates a high pressure area in front of it and how it alters the shape of the wake.
Symmetry pressure.jpg Symmetry velocity.jpg Car pressure front.jpg Car pressure rear.jpg

To better see the effects of the spoiler I have pictures of low pressure areas on the symmetry plane and the underbody of the models. The stock car has a region of lower pressure on the front of the underbody but the Cup car has greater area for the low pressure to act on. At the back of the car, it is clear that the low pressure zones are bigger on the Cup car so the aerodynamic balance should be shifted backwards.
Symmetry low pressure.jpg Underbody low pressure.jpg

Now you can say that you can make these pictures in photoshop or paint even, and you'd be correct. There's no way to verify this stuff so that we can be sure it is in the ballpark of reality. One thing we can do is estimate the drag coefficient from available data and see how close the drag value from the simulation is to it. Assuming 15 percent drivetrain loss and 0,02 coefficient of rolling friction we can estimate the drag coefficient to 0,371. Close enough for me.

Will be posting more stuff as soon as I complete more simulations. Cheers!
 
This is great stuff @tomislavp4 !

For visualising the flow in CFD you can use Total Pressure (or it's coefficient) - this gives an effective measure of energy in the flow and shows where it's used as it it travels over the car. And to get some measure of the flow that travels through the engine bay, you could set the radiator (grille) as a pressure outlet and then put a new (mass flow) inlet into the domain under the car at the back of the engine bay - circulation around the front of the car (what goes under, round or over) is surprisingly sensitive to how much flows through the engine bay.
 
This has to be thread of the year for me. Absolutely fantastic work mate! Really really looking forward to more. The splitter especially. It's such a shame there is only ONE decent splitter option for the 197.

A couple of questions, if I may?

1. Why the half model? Surely one half of the car being completely flat affects results, or do the calculations not go beyond the centre line?

2. Doesn't the front grille being fully filled in solid really affect airflow?

These could be fully noob questions as I've no idea what I'm talking about but I'm really interested.

Thanks for posting.
 
  • Like
Reactions: Banbury
This has to be thread of the year for me. Absolutely fantastic work mate! Really really looking forward to more. The splitter especially. It's such a shame there is only ONE decent splitter option for the 197.

A couple of questions, if I may?

1. Why the half model? Surely one half of the car being completely flat affects results, or do the calculations not go beyond the centre line?

2. Doesn't the front grille being fully filled in solid really affect airflow?

These could be fully noob questions as I've no idea what I'm talking about but I'm really interested.

Thanks for posting.

1) It cuts runtime down massively, and as the car is symmetrical across the centre line, you can make this assumption. Yes as said it doesn't take the split as an affected area, as it knows this is the centreline. OEMs do this all the time.

2) He said earlier he has done the engine as a sealed unit, again for simplicity, as this is what people do early on / do quicker runs.
Later on models mature and all these are taken into account, but it is a massive job and you wouldn't do a full or even half car, as run time would be massive, you'd do smaller areas, once you know the above data.


Great work Tomi, I have done some CFD at Uni for my degree I also work with CFD analysts and CAE analysts daily (one my best mates is a CFD Analyst working with me, my other best mate is a Crash Safety CAE engineer who just moved to another company).

Keep it up mate :smile:
 
@suj thanks for that mate. Makes sense. Although air flow without the grille must be very different to reality surely? That's a lot of air that can pass through that would make a big difference no?

LOVE this thread!
 
Fascinating thread! Clearly there's some serious brainpower on this forum :smile: thank you for taking the time to do all this, and in your spare time too! Looking forward to more.
 
So @suj already answered the questions and his answers are spot on but I'll try to give more info.

The engine bay is sealed two reasons: a) it drastically reduces the computing power needed to solve the problem and b) it doesn't involve me modeling every detail in the engine bay.

a) These simulations use 22,5 of the 24GB of RAM that I have available. If I was to do a full simulation, including engine bay flow, I would need at least twice as much RAM, probably more if I wanted decent mesh quality. Also, the time to solve the problem would get doubled as well. These simulations take about 9 hours each.

Now as I said before, these simplified simulations overestimate the lift values and underestimate the drag values. The lift (downforce in this case) gets overestimated because the air that in reality is vented under the car (and in the wheel wells) isn't being taken into account which means the pressure under the car is simulated as lower than it is in reality. The drag on the other hand is underestimated. Bascially, the air going through the radiator and engine bay creates greater air resistance and increases drag. This isn't acounted for, so the simulated drag is lower (by around 0,02 accoring to literature) than the real drag. The effect that the sealed engine bay has is greater on the lift value than it is on the drag value.

@ttrw2 mentioned setting the radiator opening as a outlet and the engine bay openings as inlets. I thought about this before running the simulations but I think it involves too much guesswork. What exacly is the pressure of the radiator opening? How is the mass flow distributed between the openings in the wheel-well and the opening in the underbody? If ttrw2 can point me to a study discussing this, it would certainly help.

Now, these simulations have their weaknesses but that doesn't make them useless. On the contrary, they are very useful in comparing different shapes early on in the development of a new car which is why pretty much every carmaker runs simulations of this kind. Here's a picture showing the difference, or lack thereof, between a Saab and the Clio. The picture of the Saab comes from a master thesis and includes engine bay airflow. Notice how the difference is only noticeable at the front of the vehicles, with the Clio having somewhat different air velocities in front of the radiator intake and around the air dam.
difference.jpg

b) Even if I had the computing resources neccessary to solve a full-car problem, I wouldn't necessarily do it because it would involve modeling pretty much every component under the bonnet. That would involve taking the car apart to see how everything looks and where it is located. Running a full-car simulation is easy for carmakers as they already have everything modeled in CAD, I don't have that luxury.
 
  • Like
Reactions: Jimmy and Sean197
Thanks for the through explanation. Makes perfect sense. It's a lot of clever mathematics and as soon as you open the front up you're going to need to give it a huge amount of much more complicated shapes to deal with.

Absolutely loving this thread mate. Really interested to see other aero elements added. Will you be adding side skirts? Any chance of seeing a sprint spoiler?
 
Absolutely loving this thread mate. Really interested to see other aero elements added. Will you be adding side skirts? Any chance of seeing a sprint spoiler?
The sprint spoiler is the one K-tech sells, right? That one is coming shortly. As for side skrits and other stuff, I am open to suggestions on what to run. After the sprint spoiler I'll do the Maxton kit http://www.maxtondesign.co.uk/product/clio-iii/set-of-splitters-renault-clio-iii-rs.html and after that one I'll start running custom wing and splitter designs and other ideas.
 
Thank you man! I've seen that you've been playing with CFD as well. Maybe you can share your results so we can compare them to mine?

Be happy to share, however its really just something i've dabbled in, i'm an IT technical architect by trade, so have computation power, and computer skills, and I've been into 3d modeling for years, but aero and CFD is just an interest and my understanding of it is very basic. It is reassuring to see from your results that what I found with the cup wing was similar. I have stayed away from stating any figures as I simply do not understand enough about the subject, but I'm grasping at least some of it.

What software are you using? it is solidworks? I first started using openfoam on linux, but have recently taught myself solidworks, and have used its flow simulation a little but nothing more than a toe in the water yet. I hope to learn quite a lot from these posts!
 
@Zuban same here really. I'm a B.Sc mechanical engineering - product development. I have taken a few simulation courses at uni but they all focused on structural stuff, not airflow, so I read a bunch of studies, presentations and user manuals to gain basic knowledge on aerodynamics because it's something in interested in.

I'm using SW for modelling and Fluent for simulation. The flow simulation that comes with SW is pretty basic. What turbulence model is that thing even using?
 
This is great, I'm far from being clued up on this but well done. As for the downforce thing I think I can remember reading some official Renault thing years ago about the car producing 40kgs at 80mph. Not sure if they mean it's 40kgs of actual downforce or -40kgs of lift from what it would ordinarily have without the aero bits.
 
This is great, I'm far from being clued up on this but well done. As for the downforce thing I think I can remember reading some official Renault thing years ago about the car producing 40kgs at 80mph. Not sure if they mean it's 40kgs of actual downforce or -40kgs of lift from what it would ordinarily have without the aero bits.
I read this to somewhere but i thought it was the rear diffuser adds 40Kg at 80mph.
 
@petercronin @Dmonk197 there's conflicting information about that. I've read both 40 kg and 70 kg for the df produced by the diffuser but the speed was not mentioned. Do you have a link? As this most likely is the df that the diffuser produces it should be subtracted from the lift that the rest of the car produces to get the net down force.
 
@Zuban same here really. I'm a B.Sc mechanical engineering - product development. I have taken a few simulation courses at uni but they all focused on structural stuff, not airflow, so I read a bunch of studies, presentations and user manuals to gain basic knowledge on aerodynamics because it's something in interested in.

I'm using SW for modelling and Fluent for simulation. The flow simulation that comes with SW is pretty basic. What turbulence model is that thing even using?

from what I can find solidworks uses k-epsilon.

Openfoam which i've been using is sst k-omega, its a deeply unfreindly peice of software to use, however it comes with a reasonable motorbike tutorial which i've taken and altered for the clio, it outputs the drag and coefficients aswell as the lift/downforce for each axle. I use paraview to look at the results, so far its been working quite well, and the accuracy of the output is fine for my purposes i'm not being to scientific about it. Its quite fast aswell, I've got various i5's with 16gb of ram which dual up as 3d rendernodes and cfd comp nodes, and I can throw iterations of the model at them and get results in an hour or two depending on the complexity.
 
from what I can find solidworks uses k-epsilon.

Openfoam which i've been using is sst k-omega, its a deeply unfreindly peice of software to use, however it comes with a reasonable motorbike tutorial which i've taken and altered for the clio, it outputs the drag and coefficients aswell as the lift/downforce for each axle. I use paraview to look at the results, so far its been working quite well, and the accuracy of the output is fine for my purposes i'm not being to scientific about it. Its quite fast aswell, I've got various i5's with 16gb of ram which dual up as 3d rendernodes and cfd comp nodes, and I can throw iterations of the model at them and get results in an hour or two depending on the complexity.
Ah, ok. Realizable k-epsilon is what I use in fluent, with Non-equilibrium wall functions.

Hour or two? That's fast! What type of mesh and what size are you running? Open foam is up there with the best CFD software available from what I've heard so it should give accurate results if set up right.
 
The sprint spoiler is the one K-tech sells, right? That one is coming shortly. As for side skrits and other stuff, I am open to suggestions on what to run. After the sprint spoiler I'll do the Maxton kit http://www.maxtondesign.co.uk/product/clio-iii/set-of-splitters-renault-clio-iii-rs.html and after that one I'll start running custom wing and splitter designs and other ideas.

Hi mate, yeah exactly, the Sprint spoiler is the one K-Tec sell like this

As for a splitter, I'd be really interested to see a custom design. I have something in my head but have no modelling skills to get it out. There's nothing I've seen for a 197 splitter that I like. I'd rather not end up with the standard Renault one if possible but there's just nothing available.

Again, absolutely loving this thread, really looking forward to see where it goes.
 
Hi mate, yeah exactly, the Sprint spoiler is the one K-Tec sell like this

As for a splitter, I'd be really interested to see a custom design. I have something in my head but have no modelling skills to get it out. There's nothing I've seen for a 197 splitter that I like. I'd rather not end up with the standard Renault one if possible but there's just nothing available.

Again, absolutely loving this thread, really looking forward to see where it goes.

We're in the same boat, mate. I dislike the cup splitter (if you can call it that) as well. What I'd like to have on the Clio is something similar to the picture below. Simple and effective. Also relatively easy to fabricate with the right tools.
Porsche-911-RSR-show-floors-102-876x535.jpg