Diverging/Converging Radiator Ducting - Why?

[Bus_Lengths]

Okay, after 4 years of FSAE, I give up trying to understand FSAE radiator ducting.

Convention is to have a diverging inlet (slow the air down), radiator, and then converging outlet (usually with a fan at the outlet). We've typically followed this convention

Design judges support this, the Smith books support this, everyone seems to support this.
However, everyone's support for this seems to stem from WWII heat exchanger research for aircraft flying at 400 mph.

WWII aircraft Goals: Reduce Drag
FSAE Goals: Don't melt the engine, screw drag

What I mean by this is, in FSAE (at least for our team), our primary concern is making sure we don't overheat the engine when we are reaming on it at 25 mph for extended periods. Saving 4 lbs of drag at 29 mph by having a small entrance to the radiator duct means nothing.

I guess it just frustrates me because every calculation I know how to do from fluids/heat transfer/common sense, tells me to make the entrance AT LEAST the same size as the radiator if you are concerned about mass flow rate and not drag.

I just finished a CFD study of external flow over the body/ducting and it also shows highest mass flow rates at the radiator with a duct entrance area at least equal to radiator area.

What are other's thoughts on this?

[Hector]

To be honest, I'd like to hear an answer on this, too. It's been a hot point among our team members this year. I also agree that "conventional" methods are probably for vehicles traveling much faster than our cars.

[murpia]

Originally posted by Bus_Lengths:
WWII aircraft Goals: Reduce Drag
FSAE Goals: Don't melt the engine, screw drag

I guess it just frustrates me because every calculation I know how to do from fluids/heat transfer/common sense, tells me to make the entrance AT LEAST the same size as the radiator if you are concerned about mass flow rate and not drag.

If increasing mass flow was the only way to increase cooling, you may be right.

Generally race-cars use a diverging duct to the cooler core to slow the air down as it passes through the cooler core. Slower air spends more time in contact with the core and gets hotter. So, it absorbs more heat energy per kg of air. An exit duct is less critical, once the air is out the back of the cooler core any downstream ducting is probably more for drag reduction, although don't dump your air to a high pressure region if you can help it.

You ought to add 'keep it small and light with a low centre of gravity' to your FSAE goals. If you did that, then maybe a 'conventional' setup would have an advantage?

I'm surprised more FSAE cars don't take inspiration from Formula Fords. If you look at a well known auction site item 300375806610 you will see a few photos of a Swift SC93 which has a very neat cooler core installation. The cores are very angled, creating the diverging inlet duct effect without adding any width to the car.

Regards, Ian

[Wesley]

Originally posted by murpia:
Slower air spends more time in contact with the core and gets hotter. So, it absorbs more heat energy per kg of air.

Right but remember Newton's Law of Cooling. Your overall heat transfer will go down, because it takes longer to transfer the same amount of energy to hot air than it does cold air. As such, heat rejection is maximized from this perspective when mass flow approaches infinity.

q = k*A*dT

Other than modifying the heat transfer coefficients or the surface areas, increasing mass flow IS the only way to increase cooling.

[EPMAl]

I would have to agree with Wesley here, what you are seeking is not heat energy absorbed per kg of air more than heat energy absorbed, which will occur with a larger difference between the core and the air temperature. Therefore, the faster the flow in your radiator, the better should be the heat exchange, right?

[Superfast Matt McCoy]

There is a paper that Porsche did covering their development of the cooling system for the 911. They were mostly trying to reduce drag while meeting cooling requirements, but the results show a pretty much linear inverse correlation between flowrate and coolant temperature. Also a direct correlation between both flowrate & duct outlet size and flowrate & aero drag.

So if all you care about is coolant temp, get really big ducts and shove as much air through as possible. I imagine there is diminishing returns at some point, and I'm sure the judges will want some data to back up your design.

[Hector]

Originally posted by EPMAl:
I would have to agree with Wesley here, what you are seeking is not heat energy absorbed per kg of air more than heat energy absorbed, which will occur with a larger difference between the core and the air temperature.

I think this is the key.

The reason "real" racecars have diverging ducts is because each kg of air comes with a real cost - that cost being drag. The fewer kg of air that has to pass through your cooling system, the less drag you have.

So how do you maximize cooling with fewer kg? You slow down air flow, so that the air has the maximum amount of time to absorb heat: maximum cooling per kg of air.

In our cars, drag isn't a huge disadvantage like it is in most racing series. Therefore, rather than try and maximize heat transfered per kg, just maximize the amount of airflow through the radiator. The efficiency in heat transfered per kg is lower, but the overall heat transfer is higher, at the cost of drag.

[Jimmy01]

We did some simple wind tunnel testing of a radiator with different size inlet ducts last year. Mass flow DOES increase with a smaller inlet vs radiator size (in our testing anyway and obviously only until separation begins). The reason we want to slow the air down as it enters the radiator is to increase the pressure difference across the radiator. Slower air = higher pressure. The only way air moves from one place to another is by creating a pressure difference.

[exFSAE]

Originally posted by Jimmy01:
We did some simple wind tunnel testing of a radiator with different size inlet ducts last year. Mass flow DOES increase with a smaller inlet vs radiator size (in our testing anyway and obviously only until separation begins). The reason we want to slow the air down as it enters the radiator is to increase the pressure difference across the radiator. Slower air = higher pressure. The only way air moves from one place to another is by creating a pressure difference.

+1. We have a winner.

If you try to "neck down" from the inlet to the radiator, the air will just blow right back out and around. High velocity but low pressure to force through the core. You don't want flow reversal.

I'd been told an old UTA car had that issue, before they found out why in a wind tunnel. That could be just BS though..

[plohl]

I reading this wondering if someone was going the bring up the pressure difference across the radiator. +1 to Jimmy

Also, I'm pretty sure this is covered in the radiator design topic on the forum. And RCVD also has something on it.

[ZAMR]

+1 +1 +1 to the pressure diff. You just need to make sure you pick the right area ratio so the pressure doesn't overcome the "velocity pressure" as they call it in the HVAC industry and reverse flow.

I believe the recommended inlet to area ratio for our speeds is close to 1 (.8 I think). This all becomes especially important if you have limited access to clean air (eg. with wings) and have to maximize the clean air you take in.

@Bus_Lengths: Makes me wonder what kind of CFD you ran on your full body? If your radiator area is defined as a pressure outlet it might skew the results???

[EPMAl]

So if we look at the arguments, a diverging-converging radiator duct creates a pressure differential that helps to get better mass flow, hence increasing total heat transfer but also slows the air at the radiator to get a better heat transfer per kg of air. Doesn't this looks like there is no inconvenient?

I think where all saying the same thing but with different approaches. In the case of FSAE, what do you think about constant inlet area and then a converging outlet? That way you get your pressure differential (I agree lower than with a converging-diverging) but are not limited by flow separation in your inlet (did those of you who did CFD and physical testing noticed it in high converging inlets?). We still have the drag but it seems to be a consensus not to worry about it.

[Bus_Lengths]

@ZAMR: For the CFD I approximated the radiator as a porous medium to yield a similar pressure drop to what we see when driving. As we increse the diffuser angle from 0, the mass flow rate through the core just drops. Not by a lot, but it certainly doesn't go up. (granted, I know you can't always trust CFD, but I really thought it would show some advantage)

Out of curiousity, what pressure drop across the core are other people seeing? We've measured about a 200-300 Pa delta at speeds from 15 mph to 25 mph. Doesn't really seem as high as people would make you think. This was with a diffuser/nozzle setup.

[nabjab]

Originally posted by Jimmy01:
The only way air moves from one place to another is by creating a pressure difference.


Our cooling system design team has had this idea in the back of our heads for a long time. The part that we are struggling with is how to maximize the pressure difference. We had a senior design group (non FSAE members) try to test and gather data about our current core, and somehow the project transformed into duct design. Which has for the most gained us what we already knew. The duct designs they tested with CFD use all of the diverging converging combos possible. Oddly the one that gave the greatest Delta P was the converging diverging. Which counters basically everything I thought I knew about fluids. However when we number crunch by hand the D/C ducting is the winner.

This is where our teams starts to wonder about our CFD. More specifically core representation Our team doesn't have access to a wind tunnel (which we know is a problem), so we have to rely on CFD to give use a general idea of what could work. We are using Flow Simulation to do CFD and one of the properties defined for a porous material is the pressure drop per something. (I'm the draftsman not the CFDman, I just make the models look pretty http://fsae.com/groupee_common/emoticons/icon_biggrin.gif ) @Bus_lengths: Isn't the pressure drop across the core the dependent variable you are trying to find? How is it dependent if you defined it? Are there tricks to accurately define the core so the CFD might be more truthful?

Another though: Like EPMAI said, the goal we are all striving for is the same, just with approaches. If we want to raise the pressure in front of the radiator, we could in essence drop the pressure behind it as well. Would it be better to develop both sides at the same time, or have teams found out that focus on one side over the other is better?

[Jimmy01]

I hear a fan does a pretty good job of reducing pressure on the outlet http://fsae.com/groupee_common/emoticons/icon_wink.gif Because we have always run a fan, I don't see much benefit in spending time on designing the outlet. We just make sure there is plenty of area that the air can escape to. Drag is also down my list of Priorities, so trying to get the exiting air at the same speed as the flow seems pointless.

My advice to radiator duct design:

-Think carefully (and probably do testing/simulation) about your duct placement.
-If possible, use a small inlet with a long duct (this will reduce chance of separation)
-Make sure the duct is well sealed (especially to the fan)


So if we look at the arguments, a diverging-converging radiator duct creates a pressure differential that helps to get better mass flow, hence increasing total heat transfer but also slows the air at the radiator to get a better heat transfer per kg of air. Doesn't this looks like there is no inconvenient?

I think where all saying the same thing but with different approaches. In the case of FSAE, what do you think about constant inlet area and then a converging outlet? That way you get your pressure differential (I agree lower than with a converging-diverging) but are not limited by flow separation in your inlet (did those of you who did CFD and physical testing noticed it in high converging inlets?). We still have the drag but it seems to be a consensus not to worry about it.

Remember that when we increase the pressure differential, mass flow will increase and thus the air speed through the radiator core will actually be higher than the less efficient design. Don't worry about your deltaQ/kg, as was said earlier, it is always better to have more airflow and thus a higher temperature diffference (higher rate of heat transfer) between the airflow and radiator.

[Vittorio]

Wow! Nice topic!

I thought a couple of hours about all your arguments and tried to use my aerodynamics knowledges.

My conclusion: cooling depends mainly on radiator area and then on mass flow.

Once fixed the area having a slower flow is detrimental, so using a divergent to slow down the flow is simply negatively affecting mass flow and then undesired.

In fact, using a divergent-convergent is only useful for fast cars, as previously said.

Let's move to how to maximize mass flow.

Remember that we want to improve delta p in order to improve mass flow.
That delta is across the radiator, if you want to relate it to mass flow.
Reasoning in terms of mass flow is easier, trust me.

Let's start with the easiest case; the real case is different but the conclusion remains valid.
Let's compare a conv-div with a restricted area A equal to radiator area and a div-conv with the same area A after the div.
Without losses air would see the same area at inlet and outlet and it's as if we had a straight tube, in which it would flow a mass flow equal to the velocity of the air (equal to vehicle speed) times the area.
Within the duct it's clear that mass flow is constant.
In the first case the inlet area is bigger and so will be the mass flow.

Are you persuaded?

I hope I didn't make any mistakes.

Regards

Vittorio

Turin Polytechnic, Italy

[Wesley]

Originally posted by Vittorio:
My conclusion: cooling depends mainly on radiator area and then on mass flow.

Once fixed the area having a slower flow is detrimental, so using a divergent to slow down the flow is simply negatively affecting mass flow and then undesired.

...

Without losses air would see the same area at inlet and outlet and it's as if we had a straight tube, in which it would flow a mass flow equal to the velocity of the air (equal to vehicle speed) times the area.

...

In the first case the inlet area is bigger and so will be the mass flow.


Well, first, core area is directly related to mass flow through the radiator, so your comment about it depending secondly on mass flow is really the same thing. At a given airspeed, a bigger core area will have a higher mass flow. While it's true that mass is conserved within the duct, remember, only air that ENTERS the duct will have to exit. Thus the problem here isn't conserving mass through the duct, but rather adjusting pressure ratios to allow more air into the duct from the free stream. Thus, the higher the duct pressure, the lower free stream air is likely to enter the duct at a given speed.

Improving delta P involves changing airspeeds, which means a nozzle or diffuser. Thus, at any speed, a duct is useful. It's just a tradeoff between weight of components and cooling capacity gained.

Like ZAMR said, (though it's less likely to cause a problem in our case unless you build a giant nozzle into the radiator) a problem is designing your ducts so the pressure at the radiator surface is greater than the pressure generated by forward velocity: in other words, when your duct pressure reaches flow stagnation pressure. That's not really a concern for our cars, since our speeds are so low.

[Vittorio]

Originally posted by Wesley:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Vittorio:
My conclusion: cooling depends mainly on radiator area and then on mass flow.

Once fixed the area having a slower flow is detrimental, so using a divergent to slow down the flow is simply negatively affecting mass flow and then undesired.

...

Without losses air would see the same area at inlet and outlet and it's as if we had a straight tube, in which it would flow a mass flow equal to the velocity of the air (equal to vehicle speed) times the area.

...

In the first case the inlet area is bigger and so will be the mass flow.


Well, first, core area is directly related to mass flow through the radiator, so your comment about it depending secondly on mass flow is really the same thing. At a given airspeed, a bigger core area will have a higher mass flow. While it's true that mass is conserved within the duct, remember, only air that ENTERS the duct will have to exit. Thus the problem here isn't conserving mass through the duct, but rather adjusting pressure ratios to allow more air into the duct from the free stream. Thus, the higher the duct pressure, the lower free stream air is likely to enter the duct at a given speed.

Improving delta P involves changing airspeeds, which means a nozzle or diffuser. Thus, at any speed, a duct is useful. It's just a tradeoff between weight of components and cooling capacity gained.

Like ZAMR said, (though it's less likely to cause a problem in our case unless you build a giant nozzle into the radiator) a problem is designing your ducts so the pressure at the radiator surface is greater than the pressure generated by forward velocity: in other words, when your duct pressure reaches flow stagnation pressure. That's not really a concern for our cars, since our speeds are so low. </div></BLOCKQUOTE>


I meant cooling depends firstly on radiator area and secondly on air velocity; once fixed the area there is a fixed relationship between mass flow and air velocity.

Eventually all the arguments lead to a direct correlation between inlet and/or outlet area and mass flow, isn't it?

Vittorio

Turin Polytechnic, Italy

[dazz]

Everybody seems to agree that the pressure differential accross the radiator is the key thing here, and more delta P equals more flow giving a greater average delta T through the core, hence transfering more heat allowing a smaller and lighter radiator to be used. Right?

Also most of us agree that aero efficiency is not of great concern at the relative speeds we deal with. (I'm not saying that it shouldn't be considered though) With this in mind, I can't see how a diverging inlet duct helps us? We want high pressure in front of the readiator, not lower. Opposite story on the outlet, a converging duct will only create a bottle neck (literally!) for the hot air, therefore expanded, that we need evacuated from behind the radiator.

So I would:

1 - Make sure I have enough ducts, vents to effectivly let the hot air out. Better still suck it out using air flow over the car.

2 - Emply ducting to channel plenty of air into the front of the radiator. With out going soo far as to create unnecessary aero drag.

3 - As Jimmy suggested, throw a fan on the back of the radiator to provide a greater delta P when required.

If I have some massive hole in my logic here, please correct me. Preferably with the use of diagrams to highlight how little I understand on this topic... http://fsae.com/groupee_common/emoticons/icon_biggrin.gif

[VFR750R]

I think what hasn't been layed out here, but may be a driving factor is the cross section of the radiator is not simply the height times the width. The tubes take up considerable space so an opening that represents the available area of the radiator will be much smaller then the actual radiator. I know nothing of radiator ducts but I would start with an opening that would have the same representative area. This would create the same velocity at the opening as through the core, or if you know what velocity you want then you could create some sort of ratio between the openings).

A diffuser as long as possible is also ideal assuming packaging is not an issue.

Rhetorical question - with a proper diffuser added, which side of the diffuser should the fan be?

[Jimmy01]

Originally posted by dazz:
Everybody seems to agree that the pressure differential accross the radiator is the key thing here, and more delta P equals more flow giving a greater average delta T through the core, hence transfering more heat allowing a smaller and lighter radiator to be used. Right?

Also most of us agree that aero efficiency is not of great concern at the relative speeds we deal with. (I'm not saying that it shouldn't be considered though) With this in mind, I can't see how a diverging inlet duct helps us? We want high pressure in front of the readiator, not lower. Opposite story on the outlet, a converging duct will only create a bottle neck (literally!) for the hot air, therefore expanded, that we need evacuated from behind the radiator.

So I would:

1 - Make sure I have enough ducts, vents to effectivly let the hot air out. Better still suck it out using air flow over the car.

2 - Emply ducting to channel plenty of air into the front of the radiator. With out going soo far as to create unnecessary aero drag.

3 - As Jimmy suggested, throw a fan on the back of the radiator to provide a greater delta P when required.

If I have some massive hole in my logic here, please correct me. Preferably with the use of diagrams to highlight how little I understand on this topic... http://fsae.com/groupee_common/emoticons/icon_biggrin.gif

I would say that's all pretty sound logic! A diverging outlet duct helps reduce drag. I'm not sure how a diverging duct helps with our deltaP, my gut feeling would be that it wouldn't (but gut feeling and aero generally don't go together). So I agree with you dazz, just give the hot air as easy a path as possible!

[Wesley]

@dazz

Faster air=lower pressure.

Diverging duct slows down air, increases pressure.

[Hector]

Originally posted by Wesley:
@dazz

Faster air=lower pressure.

Diverging duct slows down air, increases pressure.

It's a self-regulating system.

If I try and push a lot of fast-flowing air through the radiator (low pressure according to your post) and the delta P isn't high enough to push the air, what happens?

It slows down and pressurizes in front of the radiator until the delta P is high enough to flow the air.

I don't understand how the two are different.

[Kirk Feldkamp]

*insert monkey wrench here*

Has anyone done much investigation into core thickness? With the speeds that FSAE runs, I would think there would actually be an advantage to using a thinner core than would be run on a higher speed vehicle. A high delta P across the core doesn't seem like a good thing if it is simply because of high drag losses through the core. There has to be a balance somewhere between air mass flow and delta P for a higher total heat rejection. After all, the goal isn't to change the temperature of the air as much as you can... it's to take as much heat out of the water as you can. I'm interested to hear the outcome of any testing that teams have done in this area.

-Kirk

[Wesley]

Originally posted by Hector:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Wesley:
@dazz

Faster air=lower pressure.

Diverging duct slows down air, increases pressure.

It's a self-regulating system.

If I try and push a lot of fast-flowing air through the radiator (low pressure according to your post) and the delta P isn't high enough to push the air, what happens?

It slows down and pressurizes in front of the radiator until the delta P is high enough to flow the air.

I don't understand how the two are different. </div></BLOCKQUOTE>

What happens when your duct entrance reaches this high pressure due to flow stagnation?

The idea of a diverging duct allows inlet pressure to remain low while increasing pressure at the radiator surface.

[Jimmy01]

That's a good explanation Wesley. As you say, a non-diverging duct will increase the pressure at the inlet and this is certainly not beneficial for flow. This isn't the best example but might help some people understand: If a windscreen is removed from a car, at high speeds it will not feel that windy as a high pressure zone is created in front of the 'hole', forcing air around the vehicle. This same concept can apply with poorly designed ducts, remember that the radiator is a significant flow restriction and will increase your pressure at the inlet duct

[nabjab]

Sheesh there is a lot of information from this thread. It is all great insight.

The General understanding I am getting from this thread is:

- diverging/converging is the basic shape. Because the diverging duct increases pressure by slowing the air down, rather then slamming it against the radiator, which are basic pressure velocity fluid equations.

- By diverging the inlet ducting a "bubble" of air will not form on inlet causing air to go around the radiator.

- A thinner radiator might be more effective because could decrease drag across the radiator, which in turns allows could allow more air flow and cool better.

- A large negative pressure on the back side helps pull air through (a fan). This could also be done by getting the speed of the air back up to vehicle velocity. Which could be done with vents?

This is I do have a few concerns. I understand that a fan on the aft ducting is probably the best way to suck air out of the duct. This is also the best way to get the turbulent air from the wheel in the ducting as well (instead of relying on laminar flow), but we wont even go there.

My concern is how to get the ducting more efficient when the fan isn't on. I've seen many teams run fanless, so it can be done. I however don't want to run our car fan less. The extra weight is worth keeping our engine from detonating. In the data we collected for the capstone it was apparent that our fan turned on and off every 15 seconds or so. My thoughts are that is just a little too often. That basically tells me that duct design is SOO ineffective that it only really cools when the fan is on. The amount of time the fan is on increases as our speeds increase. I know one solution is to say "let the fan run," but from a more engineering stand point, what else could be done to get the pressure drop greater? If everything depends on the speed of the air at the inlet and outlet, shouldn't the focus of the design be on how to keep the air speed at those two locations the close to car velocity as possible?

-Nabil
UofL Powerdrain design team

[dazz]

Originally posted by Jimmy01:
That's a good explanation Wesley. As you say, a non-diverging duct will increase the pressure at the inlet and this is certainly not beneficial for flow. This isn't the best example but might help some people understand: If a windscreen is removed from a car, at high speeds it will not feel that windy as a high pressure zone is created in front of the 'hole', forcing air around the vehicle. This same concept can apply with poorly designed ducts, remember that the radiator is a significant flow restriction and will increase your pressure at the inlet duct

So there is a high pressure zone in front of the big gaping hole left by the missing windscreen, therefore everything inside the car is at this increased pressure, no? Thus creating a high delta P across any small holes opening to the rear of the car, which is what we are after in the case of our radiator.

[Hector]

Originally posted by Wesley:
What happens when your duct entrance reaches this high pressure due to flow stagnation?

The idea of a diverging duct allows inlet pressure to remain low while increasing pressure at the radiator surface.
Still not sure about this. How is a bubble of carbon fiber with a small hole different than a large hole covered by a bubble of air? I still (intuitively) think the system reaches an equilibrium - pressure builds until max flow is reached - but I'll be the first to say my intuition has been wrong in the past!
[EDIT: Still assuming relatively low speeds here, not much faster than 100kph]

Without access to a wind tunnel (or time this year to test!) I won't know for sure. Someone mentioned wind tunnel testing farther up, but without the actual test setup and method used I'm skeptical. Not saying that it was wrong, and by all means was probably done right, but some pictures would go a long way...

Same with the CFD that predicted C/D to be the best. I'd take a crappy real-world test over a nice CFD any day.

[murpia]

Can I take a little step back here?

What are people's priorities for their design (duct(s) plus cooler core plus optional fan)?

Low drag, low mass, low centre of gravity, low cost, high reliability, speed range of effective cooling, something else? I'm sure there are a few different opinions out there?

Some of the ideas above, such as minimising duct spillage or using a thin core, might conflict with other priorities.

Regards, Ian

[poe21]

we just do a simple duct that converges to a puller fan behind the radiator, and we have no problems with overheating. I see a million teams that just slap tiny fans all over the radiator and they are only pulling air through certain areas with little effectiveness.(I'm assuming that you are not one of those teams) Since the average speed over the endurance course is somewhere around 30mph, I would say that while you may be able to make a fantastic carbon fiber D/C shroud that has CFD and wind tunnel numbers to back it up, you also might under design and melt the block if you see a slower course with tight turning.

But really with a 10x7x2 inch core, and a light weight shrouded 6.5 inch puller fan, which drops the temps nicely, we seem to have little to worry about. We just determined that putting mass efforts into this would not net large gains, so maybe your motives are different than yours.

Another thing is radiator placement. I see many mounting the radiator behind the motor. Most of these I have seen have been largely over-sized with huge fans on them.(and rightly so) Our motor gets so hot during just autocrossing that you can't touch it. These rear mounted radiators are just pulling latent heat off of the engine which decreases delta T and gives you less effective heat transfer out of the system. So what that tells me is if you don't mount the radiator behind the motor, you can probably cut your radiator size down by a decent amount without worrying about overheating. Weight savings? I think so. We have a 12 inch fan sitting down stairs and it weighs 3 times as much as our 6.5 inch. Once again, our motives may be different, but I just figured I'd chime in on what works for us.

PS, it's easy to package a 10x7 radiator in a sidepod; Not so much with a 10x14 with 2 bulky shrouds.

One last thing, during enduro, I have seen teams that have to sit there and wait for traffic to pass until they can get back on the course, and teams have to poke along behind a slow moving car for half a lap. All of this happens without the car moving (or moving very slowly) which means ducting is really worth next to nothing at this point. Maybe they are minor possibilites, but with the fan, I have no worries.

[Wesley]

Originally posted by Hector:
Still not sure about this. How is a bubble of carbon fiber with a small hole different than a large hole covered by a bubble of air? I still (intuitively) think the system reaches an equilibrium - pressure builds until max flow is reached - but I'll be the first to say my intuition has been wrong in the past!
[EDIT: Still assuming relatively low speeds here, not much faster than 100kph]



Well, consider an alternate condition - a vehicle with the vent blocked from the free stream, or placed in a zone of recirculation (I.E our 2007 car with the ducts right behind our snowplow wings)

The pressure is low there due to turbulence and you don't have any flow, and can in fact generate a negative pressure gradient on the radiator. This can also occur if the inlet pressure reaches stagnation pressure, and you form a zone of recirculation around your duct. While the pressure is high, there is no flow. Since there's no flow through the exit of the radiator, there can't be a pressure drop across it.

So if you create such a high inlet pressure that you stall the flow, you also stall exit flow and no diffuser will allow you to increase your pressure differential at zero flow. Thus, no flow. If inlet pressure = outlet pressure, no flow.

[ZAMR]

A huge low pressure zone behind the wing and in front of the rad matches the low pressure behind it and cancels flow.

But then why wouldn't you just increase the bejesus out of your pressure in front of the radiator? I think you would, but not by using a huge converging scoop.

You want to increase your pressure in the area RIGHT in front of the radiator, not at the entrance of the duct. Say you have a CONVERGING duct. The pressure at the entrance will fall into equilibrium with the velocity pressure, like everyone has been saying. But then your pressure will decrease as the air moves towards the radiator. You will still have enough pressure and velocity to maintain flow through the radiator, and there will still be an even lower pressure behind the radiator. If you look at the pressure "continuum" throughout the process it will go from inlet=high, front of rad=mid, behind rad=low.

Ok, now use a diverging duct. The static pressure at the inlet will be atmo, the pressure right in front of the rad will be high, and the pressure behind the radiator will be low. So the pressure continuum goes: mid -> high -> low. The velocity pressure we have all been talking about will keep the flow moving forward through the mid -> high pressure region, aka the duct inlet to radiator. The optimal duct design will make the best use of this velocity pressure and flow the most air, more than a converging duct would. The highest pressure differential DIRECTLY across the radiator will produce the most flow (since the radiator is the restricting element). You need to find a balance however, between this pressure differential and the air velocity, to obtain the maximum mass flow rate. I believe the area ratio, A,rad/A,inlet, will be VERY close to 1, but less than 1.

If you go fast enough, like in a WWII fighter, the diverging-converging duct will actually produce thrust, like a mini pulsejet. However, in FSAE we cannot hope to go this fast, but the same flow-maximization concept applies.

[Wesley]

Well yeah. The best duct design for FSAE is really one that just has access to clean air. Other than that the gains are going to be minimal, whatever your ducting angles.

[ZAMR]

This is straying off topic, but has anyone used an airfoil directly above the radiator duct exit, maybe in combo with louvers, to help drive the radiator flow?

Creating a lower pressure zone below the foil would help suck air through the duct, minimize drag by smoothing the flow transition, and increase downforce while you're at it...

[Hi-oct]

I am a newbie working on the design of a first year car. The discussion is very informative. However have little idea on how to design crucial parts like the sidepod. I feel there are a lot of parameters to be tested and analyzed, and in addition they seem to have high correlations w.r.t each other.Would do me a great favor, if someone could show me a reference in the form of text book or research paper on how to design the experiment for optimizing sidepod parameters for the highest aerodynamic and cooling efficiency.
If i am in the wrong discussion, please guide me to the right one. Thank you.

[Homemade WRX]

Originally posted by Jimmy01:
We did some simple wind tunnel testing of a radiator with different size inlet ducts last year. Mass flow DOES increase with a smaller inlet vs radiator size (in our testing anyway and obviously only until separation begins). The reason we want to slow the air down as it enters the radiator is to increase the pressure difference across the radiator. Slower air = higher pressure. The only way air moves from one place to another is by creating a pressure difference.

Thank you!!! I'm glad someone carried the pressure factor into the equation!

It's a simple venturi. Keep in mind the FLOW area of the radiator, not front area, is what matters for your equation. Then all the details like laminar flow, etc, play into it if you really wanted to dork it up.

Our body guy completely failed this on our 05 car and we had heating issues. He actually made the ducts converging and we had no pressure to push air through the cores.


Edit: VFR750; I see you hit on the flow area too.
Kirk; I had looked into core thickness with some CFD help from the school's windtunnel and numerically decide the thinner core would cool better for the same given pressure but at the trade off of size (or weight with two cores) as surface height x width would have to increase to meet the same 'time' of transfer.
Similar issue on a NASA TT-R car that I sponsor. We chose the smaller opening but thicker core IC, siting drag as our concern...since we are running nearly 190 mph at VIR.

[Owen Thomas]

Hi-oct,

I'm sure you mean the radiator ducting or "shroud" and not the sidepod itself, as that is what was discussed in this thread. The actual sidepod serves to house the radiator (if you decide to put it in a sidepod), and maybe add some curves to your bodywork.

It is unlikely that there is a specific text or paper with the information you are looking for, however you can check out the SAE publications (http://www.sae.org/pubs/), and you may get lucky to find something close. A regular old Heat Transfer and/or Fluid Mechanics text will get you well on your way, and I can guarantee they have both at your school library. Also talk to the people on your team who have taken one or more fluid mechanics courses.

In the spirit of this forum, I should tell you that as a first year team, you should not be "optimizing" very many things. Cooling efficiency is definitely way down low on the list of picking things to optimize for points. Do some basic energy rejection calculations, pick a radiator that you can afford, and put a big fan on it. Make a sidepod that fits around it out of whatever material you think is best, and worry about ducting it properly when your team has the time and resources to look into a smaller, more efficient cooling system.

[AndrewTC]

I'm gonna have to agree with Owen on this one, Hi-oct. For a first year car it is more important that you get it sufficiently cooled than efficiently cooled. Generally speaking pick up a radiator that packages well and is cheap! For a 4 cylinder with moderate power anything with surface area of more than 150^2in is sufficient. Put the diverging inlet in clean air, and the converging outlet in an area of low pressure. Slap on a big fan with a shroud and call it good.

[Hi-oct]

It may not be for using it on our car this time, but to build up my knowledge on how this is being done. Say for future reference or some in-depth understanding. I would like to satisfy my curious mind on how a duct has evolved into such complicated shapes. I have seen research papers (journals) about analyzing various factors of a bent S shaped duct. Though I may not do all those analysis, I would like to know how they got their form or at least the science behind it.
Thank you for your consideration and support (esp. Andrew and Owen).

[L B0MB]

Has no-one mentioned the air will heat up as it passes through the radiator? Air expands as it heats up, therefore velocity will increase for a constant area and pressure will decrease. So even if you had straight inlet and outlet ducting you would achieve a delta pressure across the radiator to cause flow.

Also note the radiator core takes up space in the air stream so there may be a ~30% reduction in the actual area for air to flow through the core. Therefore a diverging-converging setup will help to keep airflow through the radiator at the same speed as the free stream air around it. This all seems to be a drag reduction exercise IMHO.

As a minimum I'd recommend a rear mounted fan and a shroud sealed to the perimeter of the radiator core, so the fan draws air from the entire core.

[AndrewTC]

I don't think the expansion of the air is too significant in this scenario. The air only absorbs something like 15-20degF if I recall, resulting in a very small expansion.

However, you have brought up a point that no one seems to emphasize. The radiator is a massive restriction! Much of the area is blocked by material with only small gaps for the air to pass through. This is the reason pressure drop is so essential; to achieve flow through this restriction. That's why Converging-Diverging systems don't work, you're shoving a lot of air into the radiator but it has no ability to travel through it.

[Homemade WRX]

Originally posted by AndrewTC:
The radiator is a massive restriction! Much of the area is blocked by material with only small gaps for the air to pass through.

I guess I simply didn't explain enough when I touched on it..?


Originally posted by Homemade WRX:
Keep in mind the FLOW area of the radiator, not frontal area, is what matters for your equation.

It's like when ricers put screen in-front of their turbos and don't think of the flow restriction now infront of their inducer.