PDA

View Full Version : Understanding Intercooler Efficiency and Pressure Loss


Mr Paddle.Shift
11-19-2005, 09:59 PM
Often we know why a FI application requires an intercooler but not many talk about computing the efficiency of an intercooler. We see that the intercooler is a standard part of a turbo kit. Lately, due to high spinning supercharger compressors, we have seen a rising number of supercharger kits with intercoolers as well.

Many of us have used the term intercooler loosely to describe an air-to-air heat exchanger and an aftercooler as an liquid-to-air heat exchanger. The terms however have a deeper history behind them. Technically, if a heat exchanger is placed after the compressor, it's called a aftercooler. The word "intercooler" came about when in the old days of fighter planes where more than one compressors were used, a heat exchanger was placed BETWEEN the compressors. Thereby, coining the term "inter-cooler'.

That aside, we move on to understand what kind of efficiency are we looking for when testing an intercooler. The common myth is a bigger intercooler is always better. Not always true. Reason is while the thermal efficiency has improved, the pressure efficiency may not. Sure you increase the boost of the system to compensate the loss, but our objective really is about maximizing thermal efficiency and minimizing pressure loss.

So just how do you go about doing that?

Thermal Efficiency.

For intercooler thermal efficiency, the ratio is defined as:

http://www.bimmerfest.com/forums/attachment.php?attachmentid=61809&stc=1

T_in : intercooler inlet temperature (or post compressor temperature).
T_out : intercooler outlet temperature.
T_ambient : ambient temperature.

Intuitively, what this ratio is telling us is that ideally you want an intercooler that can cool the charged air back to an ambient temperature. A simple arithmetic calculation will show that is a ratio of 1, or 100% efficiency. But of course in the real world it's quite impossible to attain 100%.

I have a supercharged system setup with pressure/temperature sensors mounted before the intercooler and after the intercooler. My intercooler temperature sensors are yet to be hooked up to the datalogging system, hence no graphs to show for now. I have however in the past notice my T_in to be as high as 115C and at the same time T_out to be 60C. T_ambient was 40C or so. This happened a few times during track sessions in summer here in California. A simple computation arrives at 73.3%. Not bad given the narrow bumper mouth I have. Note T_out also depends on vehicle speed and area of exposed intercooler core.

Pressure Loss.

Again, the two words we have heard so much in the FI community, but not a lot of folks have hard data to show what exactly is pressure loss.

Corky Bell has suggested that the car be placed on a dyno and with pressure gauges mounted on the inlet and outlet of the intercooler. The pressures will be read at engine redline.

The other common laboratory test is to have air blown through the inlet of the intercooler and measure the inlet and outlet pressures. But the potential flaw in this experiment is that the flow is in steady state. When we step on the gas, we hardly keep the engine RPM steady, there by keeping the belt-driven compressor spinning at a constant rate for that RPM. Note, I use the word steady. In a laboratory environment, one can create a transient state to obtain data from. But in real life, this is hardly the case.

In my setup, I have two analog pressure sensors, one mounted before and the other after the intercooler. Both send signals to a A/D convertor and into a micro-processor. While a pressure gauge can give a more continous needle reading, a A/D converter captures data at a preset sampling rate. This means I have analyze pressure gradients rather than instantaneous pressure points.

Here's a sample of a data I gathered from one track session at Buttonwillow. Graph shows Pre-IC and Post-IC pressure data. Please note: 0.1kgf/cm^2 is about 1.42psia.

http://home.comcast.net/~vsengineering/typical_pressure_graph.jpg

If you look at the graph, you can make out a general pattern that the blue line is always above the orange line. This is because the outlet pressure is always less than or equals to the inlet pressure. You also notice that on a couple of occasions the inlet pressure shot up to above 1kgf/cm^2. I am running a 6.5psi setup. Why is this so? When the throttle body plate closes during a gear shift, this creates a back pressure through the piping system, while the compressor is spinning. This datalogging device has captured the moment after the plate closes but before the bypass valves open to release the pressure.

Now, we delve into a sample collection of pressure points to compare the differential (or gradient) of the data. The RPM line has been normalized between 0 and 1, eg 0.4 means 4000RPM.

http://home.comcast.net/~vsengineering/sample_pressure_graph.jpg

If you look carefully, the gradients of the inlet and outlet pressure are similar. We compute a gradient difference of less than 0.07kgf/cm^2 (1 psi). Thereby reaffirming that the intercooler I have in my supercharged setup is a differential pressure loss of less than 1psi.

Now remember I mentioned that I am running a 6.5psi kit. Usually when SC or turbo manufacturers claim a certain boost level for the kit, they are referring to the manifold boost level. By intuition, one would think that the compressor has to produce more than 6.5psi to compensate the loss through the intercooler. Again, this can be clearly seen from the plot above where the PreIC pressure peaked at 0.70kgf/cm^2 (9.94psi) instantanteously.

Well, I hope you have enjoyed reading this article. Feel free to question/comment/criticize. :thumbup:

iateyourcheese
11-20-2005, 09:09 PM
I hope you get the data logger for the temp sensors working soon. I'd be interested in efficiency vs. rpm, but it would be a tough one to get. It would also be interesting to see efficiency vs. speed, since the heat transfer is obviously linked to the car's speed.

Then, does efficiency scale linearly with speed, or do you get crazier effects dominating (boundary layer limiting), etc.?

How much of the pressure drop is from the air being cooled, and how much is flow related? I guess you could figure this out through experiments where deltaT is small (like on a dyno with no fan).

Mr Paddle.Shift
11-21-2005, 06:53 PM
Thanks for your input.

Yes, note taken about eff vs RPM, eff. vs speed. I have a RPM signal, so no problem there. I am planning to use static pressure on bumper to estimate vehicle speed though. Let me know what are your thoughts on that.

Ideally, I would think that relationship bet eff and speed is linear in a transient state. Then again, nothing in the world is hardly linear!

You brought up a good pt about pressure drop and temperature drop. What I am lacking now is air flow rate (mass or volume). I *could* use the MAF readings to get the flow rate, given that the this is a controlled volume system. Then we can use a PV=nRT for inlet and outlet to compare theoretical results with actual ones.

I hope you get the data logger for the temp sensors working soon. I'd be interested in efficiency vs. rpm, but it would be a tough one to get. It would also be interesting to see efficiency vs. speed, since the heat transfer is obviously linked to the car's speed.

Then, does efficiency scale linearly with speed, or do you get crazier effects dominating (boundary layer limiting), etc.?

How much of the pressure drop is from the air being cooled, and how much is flow related? I guess you could figure this out through experiments where deltaT is small (like on a dyno with no fan).

LmtdSlip
11-22-2005, 07:42 AM
This is way over my head at this level but I do have a VERY basic understanding of the subject.

My question would be: Arent there some examples or precident for what you are seeking in the racing world? It would seem to me that a Champ Car team or series where FI was permitted would have some solid data on the optimum size of an intercooler for a given engine displacement/FI air volume?

Or would the rules constraints artificially limit the data you seek?

Just wondering.

iateyourcheese
11-22-2005, 02:20 PM
Thanks for your input.

Yes, note taken about eff vs RPM, eff. vs speed. I have a RPM signal, so no problem there. I am planning to use static pressure on bumper to estimate vehicle speed though. Let me know what are your thoughts on that.

Can't you estimate speed from rpm, tire size, and gearing?


Ideally, I would think that relationship bet eff and speed is linear in a transient state. Then again, nothing in the world is hardly linear!


Not sure that it's linear... if we look just at wind resistance, the power needed to maintain a given speed goes as velocity^3. Logically, then, the mass of air moving through the IC won't be linear. I'd have to look also at my references for cross flow heat transfer to tell the effects of speed on the cold side.

Mr Paddle.Shift
11-23-2005, 06:09 PM
Can't you estimate speed from rpm, tire size, and gearing?


Or that too. Just that the recent work on measuring static pressure gets me so hyped up. I will get more data this weekend

Not sure that it's linear... if we look just at wind resistance, the power needed to maintain a given speed goes as velocity^3. Logically, then, the mass of air moving through the IC won't be linear. I'd have to look also at my references for cross flow heat transfer to tell the effects of speed on the cold side.

You're probably right about that. I also need to dig my references too. Thanks for the note.

Mr Paddle.Shift
11-23-2005, 06:16 PM
I have searched and asked around for some basic data to compare. Maybe just a data trend view or something. But nothing thus far, especially pertaining to size of IC.

In the world of aftermarket FI, I am even quite surprised no one really did a study on pressure loss. Or that these data are strictly confidential. Usually what folks do to overcome pressure loss due to an oversized IC is to crank up the boost.


This is way over my head at this level but I do have a VERY basic understanding of the subject.

My question would be: Arent there some examples or precident for what you are seeking in the racing world? It would seem to me that a Champ Car team or series where FI was permitted would have some solid data on the optimum size of an intercooler for a given engine displacement/FI air volume?

Or would the rules constraints artificially limit the data you seek?

Just wondering.

mwette
11-23-2005, 10:56 PM
MPS,

The classic way to measure velocity (relative to wind velocity) is pitot tube.
My guess is that using wheel speed may be more accurate.

Maybe one way to get insight would be to model the IC as a (radially)
well-mixed gas traveling through a pipe. For the thermal part might
look something like d^2 T / dLdt = 2pi*r*k*(Tamb - T). For the pressusre
use the Bernoulli equation. I think the mass flow may have to be thrown in
too, maybe proportional to engine speed. [edit: This is looking wrong. I
need to dig a little.]

This is making me feel old. I haven't had any mass transport stuff
for > 20 years. Let me see if I can find my old transport books.
[edit: I did find it. Need to see if my brain can still do this stuff.]

Matt

E2R41L
11-24-2005, 12:58 AM
wow, must have taken you a while to write all that :) .

mwette
11-24-2005, 01:04 PM
http://bimmerfest.com/forums/attachment.php?attachmentid=62225

I am relearning from Bird, Stewart and Lightfoot: "Transport Phenomina".
I am still missing something ...
[edit: and making corrections as I find them.
The Bernoulli eqn is an energy balance. I still need the momentum balance eqn.
OK, I have added what I think is the momentum equation (d rho v^2 + dP = 0)
]

Mr Paddle.Shift
11-25-2005, 01:23 PM
mwette,

Thanks for the contributions. When I started this thread, I was hoping to attract your attention. ;) Did you just type that up in Latex?

My specialization is not in transport phenomena but have done some basic work/research in flow mechanics and such. I am glad that iateyourcheese brought up the suggestions or else they would have slipped my mind.

Regarding using a Pitot tube, my understanding is that Pitot tube measures the total pressure. And along with static pressure, we can derive dynamic pressure. This dynamic pressure will give us the vehicle's velocity.

Now, can't we also use static pressure to get vehicle velocity? Suppose I have a pressure sensor amounted on the front of the bumper (B). The sensor is mounted such that it's measuring oncoming flow pressure. Using Bernoulli's Equation,

http://www.bimmerfest.com/forums/attachment.php?attachmentid=62232&stc=1

and setting V_B to zero (since this is static), can't we derive V_\infty as the vehicle's speed? I gathered some data from a recent track event (http://home.comcast.net/~vsengineering/track_reports/WS_Nov_2005.htm) at Willow Springs. There is a 1/2 mile front straight. Using the above reasoning, I arrived at 100.2mph, which I believe is quite close to the true speed.

I have mounted a couple more sensors on various locations on the bumper to gather more data this weekend as well.

mwette
11-25-2005, 02:18 PM
Thanks for the contributions. When I started this thread, I was hoping to attract your attention. ;) Did you just type that up in Latex?
I did use LaTeX +dvips +ghostscript (to gif). I assume you are doing similar.

My specialization is not in transport phenomena but have done some basic work/research in flow mechanics and such. I am glad that iateyourcheese brought up the suggestions or else they would have slipped my mind.

Regarding using a Pitot tube, my understanding is that Pitot tube measures the total pressure. And along with static pressure, we can derive dynamic pressure. This dynamic pressure will give us the vehicle's velocity.
I haven't done this stuff for 25 years. I am struggling to relearn. Anyway, a pitot tube measures the difference between static and total pressure. See this description (http://www.grc.nasa.gov/WWW/K-12/airplane/pitot.html). However, I take back my statement that it is not going to be accurate.

Now, can't we also use static pressure to get vehicle velocity? Suppose I have a pressure sensor amounted on the front of the bumper (B). The sensor is mounted such that it's measuring oncoming flow pressure. Using Bernoulli's Equation,

http://www.bimmerfest.com/forums/attachment.php?attachmentid=62232&stc=1

and setting V_B to zero (since this is static), can't we derive V_\infty as the vehicle's speed? I gathered some data from a recent track event (http://home.comcast.net/~vsengineering/track_reports/WS_Nov_2005.htm) at Willow Springs. There is a 1/2 mile front straight. Using the above reasoning, I arrived at 100.2mph, which I believe is quite close to the true speed.

I have mounted a couple more sensors on various locations on the bumper to gather more data this weekend as well.
What you have is correct, I think. I wonder if you can set up something to measure dynamic minus static pressure directly. You will avoid the error hit from multiple sensor noise and quantization.

Mr Paddle.Shift
11-30-2005, 11:12 AM
mwette,

Thanks for the link to the Pitot tube. I really enjoy these online discussions. Oh, before I forget, I met up a F-18 pilot at the Willow Springs event. After hearing about my project, he gave some feedback how the engineers measure pressure pts on their figther aircrafts. That was the inspiration behind using a Pitot tube on my car now.

However I am a little disappointed with what happened last night. Believe it or not, I accidentally erased some of the data in the removable CF. :( Dumbest thing I did. I have very little data left from the pressure points as of now.

The other issue I am facing with road testing is that foreign objects tend to block the pressure sensors intake ports. When this happens, the pressure reading is way off. I am assuming the same problem might arise with a Pitot tube.

I think I am doing too many things at one track event. :tsk:

PS: Did we meet at a SCTS tech session a while back? Let me know whether you will be at the next one. We can meet up and share some thoughts. :)

mwette
11-30-2005, 04:46 PM
About the dirt in the pressure sensor, tought one. Maybe some sort of L-type tube with
a catch at the elbow for the dirt.

I think we may have met at one of the B&M sessions or at the breakfast for one of
the Death Vally rides. I have not gone for a while but would go again if one comes up.
(I missed the ones in Arcadia, though I live close.)

Sorry about the loss if data. I have had to deal with that before.

I am still working on modeling the intercooler. Most of the text deals with constant
density, which we don't have. I think I will try to solve in rectangular coordinates
rather than cylindrical coordinates. The approach is to use mass, momentum and
energy balances.

Mr Paddle.Shift
12-02-2005, 09:07 PM
About the dirt in the pressure sensor, tought one. Maybe some sort of L-type tube with a catch at the elbow for the dirt.

Actually, it was a loose contact. So everything is good now. I change the pressure sensors to ones with a narrower range. I figure, the most pressure you can get out of running on the straightaway, with my snail mobile, is most likely less than 0.5psi. This is based on the equation I posted above with a theoretical speed of 135mph.


I think we may have met at one of the B&M sessions or at the breakfast for one of the Death Vally rides. I have not gone for a while but would go again if one comes up. (I missed the ones in Arcadia, though I live close.)

Yes, then I do remember meeting you at B&M. That was a long time ago. I am sure the gang will have another session in Arcadia.


Sorry about the loss if data. I have had to deal with that before.

Yeah I know. These days, gathering data is just as exciting as driving on the track. Especially so for this event since I am trying to see a trend in frequency response of vertical g-load between my old PSS9 setup and my new coilover setup. I think I posted that in the motorsports forum a long while ago. Did you read that thread?


I am still working on modeling the intercooler. Most of the text deals with constant
density, which we don't have. I think I will try to solve in rectangular coordinates
rather than cylindrical coordinates. The approach is to use mass, momentum and
energy balances.

Isn't it quite complicated to model the cooling surface of an intercooler? Or do we simplify it as a block of multiple rectangular channels of air flow? Sometimes I wonder do manufacturers just fab an intercooler and gather empirical data from lab testing as opposed to a CAD + FEA + heat transfer analytical approach.

Yeah, we are dealing with compressible flow unfortunately.

Hey, thanks for the insight nonetheless. No need to lose a hair or two over this.;)

Mr Paddle.Shift
01-17-2006, 10:08 AM
mwette and all,

Not even sure if this has anything to do with FI anymore. LOL! For those still interested in the development, I do have some Pitot data from last event (Jan 15, 2006). Working on it. Will post something hopefully before end of the week.

Thanks for reading.

jimmyz2
01-28-2006, 01:04 PM
"Lately, due to high spinning supercharger compressors, we have seen a rising number of supercharger kits with intercoolers as well."

I thought it was more due to higher boost levels(more than 6psi) that are being used(and thus more heat).And if I am not mistaken,as a general rule:intercoolers for turbos and aftercoolers for superchargers especially in a BMW where engine space is lacking.IMHO.

Mr Paddle.Shift
01-28-2006, 05:49 PM
"Lately, due to high spinning supercharger compressors, we have seen a rising number of supercharger kits with intercoolers as well."

I thought it was more due to higher boost levels(more than 6psi) that are being used(and thus more heat).And if I am not mistaken,as a general rule:intercoolers for turbos and aftercoolers for superchargers especially in a BMW where engine space is lacking.IMHO.

It is implied.

If we use the technical definition of "intercooler" and "aftercooler", then we are referring to the same air-to-air heat exchanger. In that case, it's the same for a SC or Turbo setup. There is actually a history behind the "inter" of an "intercooler". Do a search on google and you will find something very interesting.

However, of course, the forum or street definition of "intercooler" and "aftercooler" is very different. Often the former refers to air-to-air and the latter liquid-to-air. Believe it or not, you will see more SC setups with intercooler by this definition than aftercooler. AA, Technik Engineering, ESS etc. There are aftercooler retrofit kits for ESS Vortech but I believe it's not so much of a space constraint as it is because of the compressor setup. Last I heard CarTech has a solution for retrofitting an intercooler for Vortech on E36s. Not sure about E46s or other models.

indyzhp
02-11-2006, 03:54 PM
A few answers to some unanswered questions:

Yes you can measure the speed of the vehicle by using the wheel diameter, RPM and gear ratio. The best way to measure the wheel diameter is to just use a flexible tape measure and measure the circumference of the tire, or alternately, you could mark a spot on your tire and the ground, roll the car until the spot hits the ground again (loaded rollout measurement) Usually the static diameter measurement is sufficient. The ideal speed measurement is from the front wheels, since the rear tires get wheelspin. Perhaps you can tap into the car's wheelspeed or mount your own.

You CANNOT measure a vehicles speed with static pressure only, you need the dynamic pressure (Total - Static). All you need is a differential pressure sensor (much more accurate for pitot tubes than using two absolute sensors. Keep in mind however, that you won't be able to use the Bernoulli equations straight from the textbook. There will be some installation "fudge factor", due to flow separations on the vehicle (ie you will not be able to measure free stream velocity unless you have a very tall boom) It would kind of be pointless, since you would need to determine the installation factor from the wheelspeed. To avoid track debris clogging the tubes, just mount it higher.

I wouldn't bother modeling the intercooler itself, just figure out what requirements you have and then contact the manufacturer to see what they have. If they can't provide any specifications on the performance of their products, I would not even bother messing with them.

Mr Paddle.Shift
02-13-2006, 01:36 PM
Hey, thanks for reviving this thread.

I did end up writing a report regarding the use of a Pitot tube. Read here: vsengineeringtrack.com, look for the Jan 16, 2006 event. Feel free to comment.

Not modeling the IC though. At least not for now. Some of my focuses are mentioned in the blog.

indyzhp
02-14-2006, 08:21 AM
Hey, I read your track report, looks like you came to the same conclusions.

Just a hunch, but I think a good location to mount the pitot tube would be to pop your hood emblem out and drill a hole in the hood :yikes: under the emblem. Unless there already is a hole. Hey, it's for the sake of science right? :) Then mount the pitot tube upright. Just make sure you mount the tube solidly, they can vibrate quite a bit at speed. I can't take a picture, but I've used a tube similar to yours if not the same exact one (Dwyer?). I've just used a bent up piece of steel to essentially clamp around the static tube out the side. Then just use some dual-lock velcro (3M makes some) to attach it to the underside of the hood. You might even slide a grommet over the tube to seal with the hood. Excuse the crudeness of the attached picture, didn't want to waste time with a cad drawing. You can kind of see what I'm talking about. Black = hood, red=mount, blue = pitot tube.

Now this still won't be freestream velocity. You would be amazed at how far away from the car's body it is untill freestream velocity. This is where the aformentioned "fudge factor" comes into play. The flow should however be laminar (hopefully!). The front of the car is kind of blunt.

Anway, good luck with your experiments.

Mr Paddle.Shift
02-14-2006, 07:23 PM
Hey Eric,

Gosh..where have you been for the past couple of months since I posted this? :D

Anyway, it's not TOO late really. But yes, the idea of drilling the hood has been contemplated before. Friend of mine, who's a tech, suggested that a while back. I am not ruling that out because if you pry the emblem out, there are two holes and IIRC, they go through the hood. The Pitot tube I use have a standard diameter of about 0.125". So it's entirely feasible.

But thanks for your insight. I can visualize the setup and mounting already. Yes, Dwyer. Who else really?

Btw, I found something on Bimmerforums that you might be interested in:

http://www.lmsr.net/images/dmuller1.bmp

Yup..it's THAT high and it's interesting to see that they use the emblem location wisely. :rofl:

indyzhp
02-15-2006, 09:09 AM
I think it would be difficult to drive around with the Eifel Tower on your hood.:)

One other note, the pitot tube doesn't have to mount horizontal to the ground, you really want it pointed straight into the airflow, as pitot tubes are EXTREMELY yaw sensitive.

Hey Eric,

Gosh..where have you been for the past couple of months since I posted this? :D

Anyway, it's not TOO late really. But yes, the idea of drilling the hood has been contemplated before. Friend of mine, who's a tech, suggested that a while back. I am not ruling that out because if you pry the emblem out, there are two holes and IIRC, they go through the hood. The Pitot tube I use have a standard diameter of about 0.125". So it's entirely feasible.

But thanks for your insight. I can visualize the setup and mounting already. Yes, Dwyer. Who else really?

Btw, I found something on Bimmerforums that you might be interested in:

http://www.lmsr.net/images/dmuller1.bmp

Yup..it's THAT high and it's interesting to see that they use the emblem location wisely. :rofl:

Vam
03-02-2006, 10:15 PM
To start with I don't drive a BMW, HOWEVER while browsing through the Internet for a solution, I found you article very relevant and factual.

I fixed a CT12 turbo to my vehicle a couple of months ago, and was enjoying my ride with all the torque, the added bhp. Last week I suddenly got smart and decided to fix a Intercooler (Air to Air) to my vehicle (ofcourse Toyota does not have intercoolers for my turbo, or atleast for my vehicle my country, so I decided to use a Mitsubishi, Montero (SUV) intercooler unit. which is also the same engine capacity as mine (2.8 Lt). All above part are used part.

Now suddenly I find my turbo only produces 8psi at maximum spooling, where as it used to hit 10 psi before the intercoolers installation. :((

Now I'm totally confused, is the intercooler helping me and giving that additional power, or am I loosing that power with the loss of the 2 psi. Is there anyway to fix this? Any help is much appreciated.

Sorry again for posting this in the BMW forum, but this was the closest I got to on the Internet, as far as my problem was concerned.


Thanks
Vam

MattMc
12-16-2008, 09:03 AM
I was reading through this thread as we are about to intercool a friends 335i with an intercooler I designed and developed for my Toyota 4x4.

It breaks a few rules.

Air to water.

Minimal pressure drop (can't have cooling with out pressure drop) well now you can.
Very short flow lengths so minimal lag. I mean Very.
Huge cooling ability and no heat spikes.

Pre and post MAP pressure taken from the inlet and outlet of the intercooler the only pressure drop is due to the air getting colder and denser. Shown in the graph below, measuring intercooler air temps at the same points.

http://i81.photobucket.com/albums/j212/FJ40_LOCKED_N_LOADED/IMG_0001-12.jpg

http://i81.photobucket.com/albums/j212/FJ40_LOCKED_N_LOADED/IMG-22.jpg

William Smith
12-21-2008, 06:56 PM
Thats very interesting. I am building a turbo E36 325i and am currently doing a few investigations into the viability of water-air intercooling (aftercooling, charge cooling, whatever!), as opposed to the more traditional oversize front mount air-air method. The main reasons I considered this is that firstly the cetification authorities here in NZ are not usually happy with you messing round with the crumple zones in the front bumper, usually a neccesity to fit a decent sized cooler. Secondly I want to eliminate as much plumbing and bends as possible. I thought of maybe directing the turbo flow straight into a heat exchanger mounted atop the rocker cover, out the other side, through the throttle body, into a plenum that curves round on itself 180 degrees and then into runners.

What are some thoughts about water to air intercooling as opposed to air to air. Fabrication is not really a worry and I am dropping and planning to dry sump the engine a couple of inches anyway, so there will be a reasonable amount of space avalible.

Some books I have read say air-water is no better than a good air-air set up, while others give coolings effiencies of 30-40% better.

I'm new to this forum and any help would be much appreciated.