Team-BHP > Technical Stuff


Reply
 
Thread Tools Search this Thread
Old 7th July 2020, 19:21   #1
BHPian
 
MegaWhat's Avatar
 
Join Date: Apr 2020
Location: Pune
Posts: 75
Thanked: 492 Times
Default Aerodynamics, simulations and the Tesla Model S

Enzo Ferrari had once said "Aerodynamics are for people who can’t build engines".

With due respect to the man and the circumstances in which he made the above statement, a quick look at today's Ferrari 488 - or any decent production car - will show how ingrained aerodynamics is in the design of today's automobile and how far we have come from 1960 (when the above statement was made). It's one of the few factors that - by itself - affects three primary performance parameters of a car - top speed, handling and fuel consumption/range.

I work with fluids in an engineering environment and performing simulations to understand the flow is a big part of my job. It’s exciting to see how a correctly placed tiny design feature can have a massive impact on the flow around it. For example, a properly placed flat plate just 7 cm in width can make a big difference in maintaining traction at high speeds. Place it incorrectly and the fuel consumption will go up with no specific benefit at all. So I used some of my free time exploring the flow around a Tesla Model S. I also simulated some generic aero improving accessories to see the effects they may have on the car and I’d like to share some observations that I found interesting. I have broken this write-up into the following sections:
  1. Basics of aerodynamics
  2. Understanding flow separation
  3. What has been simulated and why
  4. Results - Lift, drag and power required to overcome drag
  5. Results - Effect on traction: Front wheels vs. rear wheels
  6. Results - flow behaviour

So grab your coffee and buckle up!


Name:  ModelS_Strmlinesside.png
Views: 7251
Size:  211.0 KB

Last edited by MegaWhat : 10th July 2020 at 15:34.
MegaWhat is offline   (52) Thanks
Old 7th July 2020, 19:32   #2
BHPian
 
MegaWhat's Avatar
 
Join Date: Apr 2020
Location: Pune
Posts: 75
Thanked: 492 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

But first, a quick refresher on the basics of aerodynamics
The next time you get into your car, let your brain run wild and imagine that there is a set of weightless airplane wings strapped on the roof of your car. While you are at it, also imagine that you have a parachute strapped to the boot of your car. You proceed to take your car out on the open road and pick up pace. And just like the wings on an airplane make it take-off from the ground, the wings on your car want to lift you off the road as your car picks up speed. In simple terms, what you experience there is lift. And as a parachute tends to slow the rapid descent of objects as they fall, the parachute attached behind your car tries to slow you down as well. In simple terms, that is drag.

Wing on the roof: tends to lift the car off the road --> Lift
Parachute on the cars boot: tends to slow the car down --> Drag

A more technical definition of lift and drag is as below:
Lift: Component of aerodynamic force that acts perpendicular to the oncoming flow.
Drag: Component of aerodynamic force that acts along the oncoming flow.

Name:  liftdragcomponents.png
Views: 7087
Size:  44.7 KB
Components of aerodynamic force acting on a car

To properly identify and control the lift and drag acting on the car, we define a coefficient of lift (Cl) and a coefficient of drag (Cd). These coefficients are a reflection of the shape and size of the car. Larger the Cl, more is the tendency to take off from the road. Larger the Cd, more is the fuel you need to burn to overcome this drag. Cd is only positive, while Cl can be either positive or negative. Positive Cl implies a force that tends to lift the car away from the road. Negative Cl implies a force that pushes the car towards the road, increasing traction. Ideally, we want a high negative Cl that increases traction and a small Cd to have minimal power consumption.

For the technically inclined, here's the definition of Cl and Cd for a car:
Cl = (2*Lift force)/(air density * velocity * velocity * frontal area)
Cd = (2*Drag force)/(air density * velocity * velocity * frontal area)
Power required to overcome drag = Drag force*Velocity

A generic sedan has a Cl around 0.28 and a Cd around 0.35. That means a generic sedan with a frontal area similar to a Tesla Model S will experience forces as follows:

At 30 kmph: Lift = +2.8 Kg, Drag = 3.6 Kg; Power required to overcome drag = 0.4 HP
At 60 kmph: Lift = +11.4 Kg, Drag = 14.2 Kg; Power required to overcome drag = 3.1 HP
At 120 kmph: Lift = +45.5 Kg, Drag = 56.9 Kg; Power required to overcome drag = 24.9 HP

It is important to note that lift and drag are proportional to the square of the car speed. And the power required to overcome drag is proportional to the cube of the car speed. This means that the drag & lift that the car experiences at 120 kmph is 4 times the drag & lift it experiences at 60 kmph. And it needs 8 times the power to overcome drag at 120 kmph as compared to 60 kmph.

Last edited by MegaWhat : 10th July 2020 at 14:10.
MegaWhat is offline   (43) Thanks
Old 7th July 2020, 19:44   #3
BHPian
 
MegaWhat's Avatar
 
Join Date: Apr 2020
Location: Pune
Posts: 75
Thanked: 492 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

What is flow separation?
Flow separation is something we will be referring to later, so let's quickly see what it means.

Imagine that you're cruising in your car at 90 kmph on an empty 4 lane road. The road is straight for now so you're cruising comfortably. But out of the blue, you encounter a sharp 90 degree turn. It's tough to hold on to the road and take that turn and stay attached to the road, despite hard braking. So you end up going straight and off the road. Thankfully there's some some soft grass on the sides of the road and it slows you down and there's no damage to life or property. In aerodynamic terms, your car "separated" from the road as it stopped following the road.

Along similar lines, when air stops following the shape of a body, we say that the flow has separated. For example, when you hold a circular glass sideways under a tap, you’ll see that the water follows the circular shape upto a point, but then stops following the glass and leaves the glass surface in a rather turbulent state. That is flow separation.


Aerodynamics, simulations and the Tesla Model S-separationcyl.png
Flow separation around a cylinder


A point to note is that once your car has left the road, it does not matter if the road after the point of "separation" is straight or curved or broken, as the car is no longer "attached" to the road. Similarly, once the flow has separated from the surface of an object, the shape of the object after the point of separation does not matter as the flow is not attached to the surface to follow it.

As compared to separated flow, an attached flow usually gives lower drag and higher lift, and in most cases (especially flow over a car) air will tend to separate at one point or another. In that case, controlling the point of flow separation is better than letting it separate by itself. This is one of the reasons we have spoilers on our production hatchbacks – to ensure that the flow separates exactly where we want it to at all road speeds. Any movement in this point of separation over different speeds can lead to unexpected variation in how the car handles at high speeds.

For example, if at 80 kmph the flow separates on your car near the A-pillar, the car will not experience a lot of lift and you'll feel planted to the road. But if at 100 kmph, suddenly the flow is attached on the roof and separates only at the C-pillar, the car will suddenly experience a jolt of lift as it goes from 80 to 100 kmph and the ride will feel slightly unsettled as the car becomes "floaty". The spoiler on the hatchback ensures that the flow is attached to the roof and consistently separates only as it leaves the spoiler itself - at all speeds.

Last edited by MegaWhat : 10th July 2020 at 20:48.
MegaWhat is offline   (40) Thanks
Old 7th July 2020, 19:49   #4
BHPian
 
MegaWhat's Avatar
 
Join Date: Apr 2020
Location: Pune
Posts: 75
Thanked: 492 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

So what all has been simulated?
Apart from the stock Tesla Model S, I tried out a few accessories that work on the air around the vehicle.

But why the Tesla Model S? Why not any other car
There are a few specific reasons why I picked up this particular vehicle:
  1. The Model S is one among few production cars where a lot of focus was given to aerodynamics right from the start. It has a certain features designed specifically for aerodynamic efficiency such as:
    a. Reduction of ride height at highway speeds.
    b. Contoured chin to ensure smooth flow under the car.
    c. Carefully designed front fascia that diverts air away from the tyres.
    d. Shutters that block three grille openings unless opening them is deemed necessary for cooling.
    e. A stock rear diffuser to minimize lift and drag.
    f. Optional trunk lid spoiler to further reduce lift without increasing drag.
    All of it leads to a rather low Cd of 0.24 (based on a wind tunnel test carried out by Caranddriver magazine).
  2. Lot of documentation is available in the public domain to compare to. This helps to ensure that results of the simulations are within otherwise observed range. Here are links of 2 publications that I primarily referred to:
    https://www.tesla.com/sites/default/...n-the-road.pdf
    https://download.atlantis-press.com/...e/25884554.pdf
  3. And finally, I had access to a publicly available, fairly acceptable virtual model of the Model S that I could use for the simulations. This isn’t exactly a detailed model – as in panel gaps, contours along the underbody and all details of the grille aren’t present – but for what I wanted to do, I thought it was good enough.

Aerodynamics, simulations and the Tesla Model S-stock-model-s.png
Simulated stock Model S

Coming to the accessories, I did not have access to the Tesla branded spoiler of course, or details of other accessories, so I have simply created an approximate geometry of these with dimensions approximately in the same ballpark. Images for reference below.
  1. OEM type trunk lip spoiler.
  2. Roof spoiler.
  3. Diffuser. There was Tesla Model S Plaid seen at the NŁrburgring sporting a wide diffuser along with other accessories for enhancing the cars aerodynamics, so I was curious how well it could work.
  4. Front splitter. Just to see what happens. As I did not have access to the complex geometries of the usual splitters, I ended up with a rather over-sized splitter in my enthusiasm. It's a bit too big to actually be put on a Tesla, but its representative enough for a comparative study.
  5. A combination of the diffuser, trunk lip spoiler and the front splitter.
  6. Rear Wing. I would like to differentiate between a spoiler and a rear wing. A spoiler usually simply “spoils” the flow to reduce the upward lift. A wing on the other hand actively generates downward lift (or downforce as it is colloquially called) due to its own shape. A spoiler can be thought to reduce the size of the imaginary wing that we had strapped on to the roof of our car. A rear wing is more like strapping on another wing on our car, in addition to the first one, only inverted (to generate downward lift). For the simulations, I simply made a rear wing out of a contour that I know can generate a lot of lift at these speeds without significantly increasing drag.

Aerodynamics, simulations and the Tesla Model S-geometriespng.png
Simulated accessories

Last edited by MegaWhat : 10th July 2020 at 14:10.
MegaWhat is offline   (32) Thanks
Old 7th July 2020, 20:05   #5
BHPian
 
MegaWhat's Avatar
 
Join Date: Apr 2020
Location: Pune
Posts: 75
Thanked: 492 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

Results: Lift, drag and power required to overcome drag

Below are the results for the stock Model S and other add-ons. Do note that these values should be used only for getting a general idea and for comparison only among each other as they are from the same set of simulations. It is not ideal to compare these to any other number you may find elsewhere as the simulation or testing methods may vary. Cl is considered positive when the force tends to lift the car up, away from the road. Cl is considered negative when it tends to push the car into the road. Ideally, we want a high negative Cl to maintain traction while keeping Cd low.

Name:  Tablejpeg.jpg
Views: 6355
Size:  14.9 KB
Values of Cl & Cd for the stock Model S and Model S with other accessories


Letís see how much force (in Kgs) to expect at different speeds and what is its effect on the power required to overcome drag (in HP).


Aerodynamics, simulations and the Tesla Model S-lift-vs-speed-curve.png
Variation of lift with car speed

Aerodynamics, simulations and the Tesla Model S-drag-vs-speed-curve.png
Variation of drag with car speed

Aerodynamics, simulations and the Tesla Model S-hp-vs-speed-curve.png
Power required to overcome drag at different car speeds

Here are some of my observations:
On lift:
  1. The impact of the accessories is higher on the lift than on the drag. In general, itís reducing lift is easier as compared to reducing drag, and that is reflected here.
  2. The front splitter reduces lift significantly, followed by the diffuser and the trunk lip spoiler, both of which reduce lift by almost the same amount. This could possibly be due to the much larger size of the front splitter. The roof spoiler does not make much of a difference as compared to the stock car.
  3. A combination of the three is quite capable of reducing lift significantly.
  4. The rear wing effectively cancels out the positive lift generated by the car body resulting in a net negative Cl while increasing drag only marginally. The force it generates is nothing as compared to the wings we have on F1 cars, but for a production vehicle, it's not that bad, though it can get better.
  5. Letís say, at 130 kmph, the stock car tends to get lifted up by a force of ~53 Kg. The diffuser or the trunk lip spoiler, by themselves can reduce this lift to ~36 Kg. The front splitter by itself can bring the lift down to 28 Kg. A combination of these three reduces lift to 17 Kg, while a wing brings it down to -1.7 Kg.
  6. If we look at Indian highway speeds of 80 kmph, the stock Model S generates a lift of 20 Kg. This isnít much honestly and that is a good thing. However, even at this speed, a trunk lip spoiler or the diffuser can bring this lift down to 13 Kg while the front splitter can reduce it to 11 Kg. A combination of these reduces the lift to 6.5 Kg, while the wing produces close to zero lift. This means, that we should not be extremely concerned about lift at 80 kmph, but properly designed accessories can still make a difference.
  7. As lift is proportional to the square of speed, the differences are quite big at speeds of 150-190 kmph. These add-ons can be quite useful for the Model S as itís top speed is around 250 kmph.

On drag:
  1. Coming to drag, thereís barely any difference below 75 kmph. At 80 kmph, the difference between the design with minimum drag and the one with maximum drag is only ~4 Kg. However, at speeds of 150 kmph, this difference increases to ~13 Kg.
  2. The car with a rear wing attached produces the most drag Ė no surprises there, while the car with a front splitter produces the least drag. The splitter works because it reduces the amount of flow going under the car and pushes it over the car instead.
On power required on overcome drag:
  1. The curves for aerodynamic power required are similar to drag curves, except that they vary as the cube of the speed. At 80 kmph, youíd only need around 6 HP to overcome drag. But at 130 kmph, youíll need around 14 horses to pull you through the air. At 180 kmph, the stock car needs ~67 HP to pull through the drag. A car with a front splitter will need around 60 HP, while one with a wing will demand 70 HP.
  2. Let's say you have only 50 HP to spend to counter the drag. With a wing you can reach a top speed of ~160 kmph. But with a front splitter or a trunk lip spoiler, you can reach ~170 kmph, while consuming the same power.

So for a road vehicle, how much lift is safe, and when do we consider it to be dangerously high?
I could not find a clear answer online with a basic search. However, I believe it depends on a few other variables such as the weight of the car and the suspension setup. For a car as heavy as the Model S, experiencing an upward lift of 80 Kg (at 160 kmph) might not be a big deal - it is a mere 4% of the kerb weight of ~2000 Kg. There is still ~1920 Kg weight acting on the tyres and helping maintain traction. However, if a car with a kerb weight of ~900 Kg experiences 80 Kg of upward lift, I believe we can expect some adverse impact on handling as almost 9% of the cars weight is taken off the tyres. Would like to hear comments on this.

Last edited by MegaWhat : 10th July 2020 at 20:21.
MegaWhat is offline   (39) Thanks
Old 10th July 2020, 15:27   #6
BHPian
 
MegaWhat's Avatar
 
Join Date: Apr 2020
Location: Pune
Posts: 75
Thanked: 492 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

Results: Effect on traction - Front wheels vs. rear wheels
Another aspect that we look into is the moment due to these aerodynamic forces. After all, a net positive lift is one thing, but where exactly on the car is it acting - the front of the car or the rear?

Imagine that your car is kept on a see-saw, in such a manner that the center of gravity is right above the pivot of the see-saw. In that case, the car should be stable and not fall either side. Next, imagine that a gigantic fan is blowing air at your car from the front. To keep the see-saw balanced, the aerodynamic lift generated by the front half and rear half of the car should be equal. Otherwise, the the arrangement will drop towards the side that has less aerodynamic lift. But achieving this balance is easier said than done.

For example, when we attach a rear wing, a lot more downward force is experienced by the rear wheels as compared to the front wheels and we may experience reduced traction from the front wheels at high speeds. This situation is reversed if we only attach a front splitter in which case we may lose traction on the rear wheels at high speeds, while maintaining traction on the front wheels. It is important to know which wheel is going to give up traction first.

Below is the variation in load on the front and rear axle with increasing speed for various accessories and the stock car. We should keep the kerb weight of ~2000 Kg in mind when looking at these numbers. Positive load values imply a tendency to lift away from the road. Negative load values imply being pushed into the road.

Aerodynamics, simulations and the Tesla Model S-front-axle-load.png
Variation of load on front axle at different speeds

Aerodynamics, simulations and the Tesla Model S-rear-axle-load.png
Variation of load on rear axle at different speeds

These plots show that for the stock Model S and all accessories except rear wing, with increasing speed, traction increases on the front wheels while traction is lost on the rear wheels. This means that for all accessories except rear wing, most of the positive lift is generated by the rear of the car. This tendency is strongest with the front splitter and least with the diffuser.

The numbers are interesting.
At 150 kmph, the front axle of the stock Model S experiences a downward force of 29 Kg, while the rear axle experiences an upward lift of 107 Kg. So as the speed picks up, it is the rear of the Model S that will give up traction first. That makes sense as at least there will be control over the steering through the front wheels.
For the front splitter, at 150 kmph, the front axle experiences a downward force of 76 Kg, while the rear axle experiences an upward lift of 125 Kg.

Addition of a rear wing reverses this trend and we see that with with increasing speed, the car will experience increasing traction on the rear wheels while reducing traction on the front wheels. However, the numbers are more balanced in this case. For example, at 150 kmph, the car with the rear wing imposes an upward force of 35 Kg on the front axle and a downward force of 42 Kg on the rear axle.

The diffuser gives the most balanced axles, along with a decent reduction in lift as the difference between the loads experienced by the front and rear axles is the least with the diffuser.

Last edited by MegaWhat : 10th July 2020 at 20:52.
MegaWhat is offline   (26) Thanks
Old 10th July 2020, 15:29   #7
BHPian
 
MegaWhat's Avatar
 
Join Date: Apr 2020
Location: Pune
Posts: 75
Thanked: 492 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

Results: Flow behaviour
The next aspect is understanding why do these add-ons behave the way they do? So we look at the streamlines and the drag generating regions around the car. Streamlines essentially show the path followed by the air as it flows across the domain. As the car goes through air, there will be regions where the energy of the air has been reduced due to the car's presence. These can be referred to as drag generating regions and are visualised using what is known as total pressure coefficient. It is a reflection of the total energy contained within a fluid element in the domain. A region where this total energy has been reduced to zero is a reflection of the drag generating areas.

A generic sedan also produces regions of swirling flow – called vortices - in certain areas around the car such as the A-pillar (where the flow interacts with the cabin) and near the C-pillar as well (where the flow leaves the cabin and interacts with the trunk). Vortices are interesting features and can be thought as like a local mini cyclone. These create drag and also some noise but can also help in mixing of the flow around it's vicinity - similar to how stirring a glass of lemon juice needs energy but ensures that all the sugar is well mixed up.

Aerodynamics, simulations and the Tesla Model S-apillarflow765x1024.jpg
Vortex structures around generic automobiles (Ref:https://mechanixillustrated.technica...-aerodynamics/)


Aerodynamics, simulations and the Tesla Model S-rear-streamlines.png
Streamlines seen from rear of the vehicle


Aerodynamics, simulations and the Tesla Model S-cpt-iso.png
Drag creating regions around the Model S

I’m not an expert in automobile aerodynamics and I can be way off in my understanding of the flow features as I have listed below. If I’m interpreting something in a wrong manner or missing out on something, I’d really appreciate if someone more experienced points it out.

Remember the swirling vortices that I mentioned earlier? You can see their effects well on the streamlines. The swirling of these vortices can potentially bring in high energy air (that is otherwise away from the car) and mix it with the low energy air closer to the car - thereby increasing the energy of the flow in these regions, leading to a slight reduction in drag.

The stock Model S seems to be designed to use the vortices well. The streamlines show some high energy flow from the sides of the car flowing into the rear of the car which is among the primary drag creating regions. The trunk lip spoiler further enhances such mixing and the vortices further cut into the drag creating region of the car, reducing drag a little bit more. The roof spoiler creates an extra region of drag generation near itself, but it seems to slightly alter the position of the larger vortices, leading to a marginal reduction in drag. The diffuser pumps in a lot of high energy air directly into the drag creating region, reducing drag. The front splitter essentially amplifies the aerodynamic effect of the cars exterior by moving a lot more air on top of the car than under it, so the mixing due to the vortices is enhanced as well. The rear wing adds its own drag creating regions and reduces the strength of the vortices around the car, thereby increasing net drag.

Let’s look at streamlines in the cross-section below.

Aerodynamics, simulations and the Tesla Model S-csstreamlines.png
Streamlines along the length of the Model S

The trunk lip spoiler effectively separates the air at the trunk lip and moves the air upwards, away from the road, reducing lift. The diffuser does the same and the wing does so more effectively. The front splitter reduces the amount of air going under the car and improves the effectiveness of the stock diffuser to bring about an upward movement of air as it exits the rear of the car. A combination the front splitter, diffuser and trunk lip spoiler effectively pumps the air upwards and reduces lift.

A spoiler is meant to cleanly separate the flow and the roof spoiler does this job well. It leads to separation of flow and alters the shape of the car (as seen by the air) drastically. Note that once the roof spoiler has separated the flow, the air does not follow the shape of the rear glass or the trunk. This means, a trunk lip spoiler placed behind a roof spoiler is more or less useless. However, I do see some photos of cars sporting both of them together. More often than not, that’s a money spending activity, nothing more.

Overall, the Tesla has been very carefully engineered when it comes to aerodynamics which is neat for a production car. It will be a good benchmark for others to learn as well. We also see that well designed and well integrated accessories can be helpful at speeds above ~100-120 kmph as well, at least in terms of improving handling if not reducing drag.

I’ll keep on adding other simulations to this thread as and when I’m able to do them. I’d like to run some simulations on the cars sold in India or any add-ons that we have available here. But it’s tough to find good virtual models that can be used. So if someone here has a good 3D scanner, some free time and can provide a closed 3D stl model of the car, I’m game .

A word on simulations & accuracy
Using computers to simulate fluid flow is known as Computational Fluid Dynamics or CFD. It involves numerical methods that break down complex equations into a manageable form and solve them in an iterative manner. CFD methods and processes are usually benchmarked against tests and are used for designing new components only after they have been validated using existing test data. A variation of 5-10% as compared to lab tests is common for a basic CFD simulation. This difference can be reduced to ~3-5% for a well calibrated approach and reduced further by better simulation tools & processes. Due to such variations, CFD methods are frequently used as a comparative tool where we only look at the change in results due to design changes instead of relying on absolute values from the simulations as an output. The method is refined whenever an absolute value is needed as an output.

In my opinion, the results from the current simulations come quite close to publicly available information. For example, Caranddriver magazine tested the full scale Tesla Model S in a wind tunnel with the Tesla OEM trunk lip spoiler and a low ride height of 117mm. They arrived at a Cd of 0.24 and a Cl of ~0.13. The current simulations were done with a ride height of 160mm and a roughly created approximate OEM type trunk lid spoiler. The Cd for the simulation came to 0.25 with a Cl of 0.19. I don’t think it’s supposed to be very close, because ride height makes a big difference in Cd and more in Cl. But I'm satisfied that the results are at least around the same ballpark number. In any case, these results will only be used for a comparative analysis among each other.

Assumptions for the simulations
The below aspects have been assumed for all the runs within this simulation:
  1. Speed = 90 kmph. Cl and Cd are calculated based on forces experienced at this speed. Once calculated, Cl and Cd are more or less independent of speed. If we simulate the car at 120 kmph and recalculate Cl and Cd at that speed, these will come out to be same because the forces also scale up with speed accordingly.
  2. The ground under the car is assumed to move back at 90 kmph as the car moves ahead at that speed.
  3. Wheels do not rotate. This is a big assumption. Rotating wheels usually reduce the drag by a small amount.
  4. Car height is 160 mm.
  5. No flow through the grille. This is an ideal case as it leads to slightly lower drag.
  6. The underbody region is completely flat. No details at all.
  7. Panel gaps not simulated.

To all those who have spent the better half of an hour reading till the end, I thank you and I burden you with a few more links you might find interesting.

https://mechanixillustrated.technica...-aerodynamics/ - A good resource for basics of automobile aerodynamics.
http://www.saea.com.au/resources/Doc...-A-Seminar.pdf - A presentation by Ford on it's aerodynamic practices and some competitive assessment of Honda.
https://www.ara.bme.hu/oktatas/letol...cleaerodyn.pdf - A very basic introduction to automobile aero and how it has evolved over time.
https://www.scientificamerican.com/a...ay-in-the-air/ - An article on how we don't have a concrete theory on lift yet. Does not mean that we can't design aircraft or cars though.
http://www3.eng.cam.ac.uk/outreach/P...wwingswork.pdf - My personal favourite hypothesis on how lift is generated. Some aspects such as viscosity are still missing from the math. Work is in progress by the teams AFAIK.

Cheers!

Last edited by MegaWhat : 10th July 2020 at 20:59.
MegaWhat is offline   (59) Thanks
Old 11th July 2020, 05:42   #8
Team-BHP Support
 
Join Date: Feb 2004
Location: Mumbai
Posts: 9,915
Thanked: 39,871 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

Thread moved out from the Assembly Line. Thanks for sharing!
Aditya is online now   (4) Thanks
Old 11th July 2020, 07:10   #9
BHPian
 
Join Date: Mar 2019
Location: Pune, Bangalore
Posts: 47
Thanked: 432 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

Let me be the first to thank you for this incredibly detailed analysis into aerodynamic aspects of vehicle design. I have had some exposure to CFD myself. Your article has been helpful in shedding light on many post processing queries.

Some good weekend inspiration this! What with lockdowns and restrictions all around, I might just use your simulation and results to get myself up to speed on some aspects of CFD!

Curious to know what tools have been used. I'm assuming this to be a freeware/cloud based solver? Is it SimScale?
AlQuazi is offline   (2) Thanks
Old 11th July 2020, 07:40   #10
BHPian
 
Join Date: Nov 2013
Location: HR 51/HR 29
Posts: 952
Thanked: 3,514 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

An excellent thread. Aerodynamics may not be too relevant for the Indian car scene, but boy did this soothe the engineer in me. Please enlighten us on the software used for the simulation. Will be great if you can do a similar study with an SUV. The differences will be very interesting.
Shreyans_Jain is online now   (2) Thanks
Old 11th July 2020, 09:09   #11
BHPian
 
MegaWhat's Avatar
 
Join Date: Apr 2020
Location: Pune
Posts: 75
Thanked: 492 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

Quote:
Originally Posted by AlQuazi View Post
Curious to know what tools have been used. I'm assuming this to be a freeware/cloud based solver? Is it SimScale?
Thanks! And yeah I did use Simscale.
Running a 2 million mesh on my laptop would require it to toil continuously at full power for 2 days and 2 nights. It wouldn't survive and I would run out of patience.

Quote:
Originally Posted by Shreyans_Jain View Post
An excellent thread. Aerodynamics may not be too relevant for the Indian car scene, but boy did this soothe the engineer in me. Please enlighten us on the software used for the simulation. Will be great if you can do a similar study with an SUV. The differences will be very interesting.
Thanks!
Indeed it's more relevant to the US highways and more so for the German autobahn. But I think we're at a point where some Indian highways let you cruise at 100-120 kmph and that's just at the gateway of noticeable aerodynamic impact. Unfortunately, we do have many road users who go beyond and easily reach ~150-160 kmph simply because their car allows it. At those speeds - though not too high - a decent impact of aerodynamics can be seen. The effect of drag may not be of importance to them, but a loss in traction at the wrong time due to a light car + too high a Cl could lead to an unsafe situation. That's why I'm curious to see how the Indian cars fare in these aspects. I'll check if I can get my hands on a good virtual model of an SUV.

Normally I use a paid software for my professional work, but that can get expensive for so many runs. So for the current simulations, I used SimScales community plan, which provides cloud based simulation tools free of charge as long as the projects are kept open for public access. As I got the Tesla model from a public project, I decided to keep these runs public as well. Being free, I could try out a lot of variations for no cost as such. However, there are restrictions and I had to optimise the process a fair bit to ensure that I get results that are reasonably accurate.

For example, the maximum runtime is limited. This limits the mesh size that I can work with, because with a larger mesh, the solution never converges in the limited runtime, leading to an inaccurate result. So I have to deploy detailed mesh only in regions of absolute importance. In all, the current simulations had a mesh count of ~2 million and it took around 2.5 hours for each case to run on 16 cores. Ideally, I would like to have around 10 million cells for a good simulation, but good things come at a price
MegaWhat is offline   (7) Thanks
Old 11th July 2020, 10:07   #12
Distinguished - BHPian
 
Join Date: Oct 2012
Location: Delhi
Posts: 5,345
Thanked: 19,268 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

Thanks very much for this wonderful thread, fascinating.

Some questions:

I was looking at the formulae you quoted:

Quote:
Cd = (2*Drag force)/(air density * velocity * velocity * frontal area)
What is it that actually causes drag force? Is it similar to an aircraft wing, simply put high pressure at the front / low pressure at the back?

These days, a lot of the noise we hear in the cabin is likely to be related to air movement across the carís surface, protrusion (e.g. mirror) etc. Is air noise simulated during car design, does some of the simulation you are doing here helpful for that purpose or would it require a completely different approach?

Thanks
Jeroen
Jeroen is offline   (3) Thanks
Old 11th July 2020, 10:50   #13
Senior - BHPian
 
srishiva's Avatar
 
Join Date: Nov 2006
Location: Bengaluru
Posts: 4,141
Thanked: 1,661 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

Thank you very much for the technical explanation. Really good read. There was a time when cars drag coefficient was more than 0.3. Now a days most have it within that. With fuel efficiency becoming more and more regulated, I guess there is more effort into these aspects.

I guess for better cornering you need more drag ?
srishiva is offline  
Old 11th July 2020, 12:48   #14
BHPian
 
MegaWhat's Avatar
 
Join Date: Apr 2020
Location: Pune
Posts: 75
Thanked: 492 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

Quote:
Originally Posted by Jeroen View Post
What is it that actually causes drag force? Is it similar to an aircraft wing, simply put high pressure at the front / low pressure at the back?

These days, a lot of the noise we hear in the cabin is likely to be related to air....
Thanks for the kind words Jeroen, I'm glad you found it useful!

You are spot on in terms of what causes drag. For the car, it is primarily due to high pressure at the front and low pressure at the back. The high pressure at the front is caused as the car essentially rams into the air, while the lower pressure at the rear is caused as the air leaves the car in a separated state. It is hence important to contour the hood to minimise this "ram" effect and minimise separation (drag generating regions) at the rear. Apart from this, there is also friction drag generated as the air rubs over the surface of the car. It depends on surface finish, contamination and overall surface area that the air rubs across. But it is the pressure drag that is the primary contributor - similar to rowing a wide oar through water.

The noise we hear is indeed related to air disturbances - more so at high speeds. The simulations I did cannot be directly used to quantify noise, but they can provide an input for a separate noise prediction simulation. AFAIK, the noise models that we have today are not very accurate and predicting it is slightly challenging. A lot of computational effort can be needed to accurately predict noise. There is significant work going on in development of these models.


Quote:
Originally Posted by srishiva View Post
I guess for better cornering you need more drag ?
Drag essentially slows you down. Think of it as the resistance you face as you swim through water; but with a car, it is air that resists the car's motion. You want to minimise drag so that the car can move through the air without much resistance.

For cornering, you need traction, and it is the lift coefficient that affects it. To improve traction, you want the air to push the car towards the ground so that the tyres can grip the road better. This essentially means a high downward (negative) lift.

Last edited by MegaWhat : 11th July 2020 at 12:57.
MegaWhat is offline   (6) Thanks
Old 11th July 2020, 13:39   #15
BHPian
 
MegaWhat's Avatar
 
Join Date: Apr 2020
Location: Pune
Posts: 75
Thanked: 492 Times
Default Re: Aerodynamics, simulations and the Tesla Model S

Quote:
Originally Posted by Jeroen View Post

These days, a lot of the noise we hear in the cabin is likely to be related to air movement across the carís surface, protrusion (e.g. mirror) etc. Is air noise simulated during car design, does some of the simulation you are doing here helpful for that purpose or would it require a completely different approach?
Okay so there is a LOT of research going on for noise simulation. There are software available in the market for simulating NVH (Noise Vibration and Harshness) which takes into account all the sources of noise - engine, road, structure and air. I need to read up a bit more to find out how exactly these work. But at the same time, a lot of work is still ongoing to take these simulation models further.

You might find some of these papers worth a read. One is by VW.

http://www.j-mst.org/On_line/admin/f...2980-2989_.pdf
https://www.researchgate.net/publica...nergy_Analysis
MegaWhat is offline   (3) Thanks
Reply

Most Viewed
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Copyright ©2000 - 2020, Team-BHP.com
Proudly powered by E2E Networks