I have been working through the process of improving the volumetric efficiency of my engine recently, by improving flow. Headwork is an area often advocated to improve the engines performance, with undoubted results, but what about the other parts of the system the charge must travel on its way through the engine? As power can often be gained by chasing restrictions in an engine, and turbos pressure at the compressor can be much higher than pressure at the intake manifold, I decided to try and limit these restrictions and have the turbo as efficient as possible, limiting heating of the charge.

An intercooler has the job of the optimum mix of cooling the charge (heat exchange), increasing charge density, while not restricting the flow to an extent that pressure at the compressor outlet is significantly increased due to restriction. Increasing pressure means increased heat (assuming compressor efficiency remains constant - moreover compressors often get less efficient as pressure PR rises).

To illustrate a turbo running at 70% efficiency and 1 bar at the compressor cover (2 absolute) of boost with an ambient temp of 25c discharges the charge at 117celsius, if the pressure must be increased to 1.2bar at the compressor cover due to restrictions to "see" 1 bar at the plenum, the turbo discharge temp is 131c - so we are actually adding heat into the charge, to overcome a restrictive intercooler that said intercooler must then try and remove. Even when an efficient intercooler runs at 75% (some icebox types can run over 100%) we are on the losing end. Higher temps lower charge density, and lower charge density is not good.

Intercoolers for sale often quote pressure drop across the cores at certain boost values, quality core should reject the maximum amount of heat while causing the minimum amount of restriction. One reason cheap ebay core are often dismissed is that they are too inefficient at rejecting heat, perhaps through a lack of turbulators, or cause too much restriction trying to reject the heat (poor turbulator design).

Reading, always dangerous, such as Graham Bells Forced Induction and Corky Bells Maximum Boost, I believe I need 282cm2 (square centimetres) of intercooler end tank based on their calculations for 400bhp on a tube and fin, at 5in/sq per 100bhp multiplied by a factor of 2.2 as less than half tank area is tubes (bar and plate suggests 8in/sq per hundred). The best way to increase end tank area was to go to taller/thicker or for the greatest gain a vertical flow intercooler. While they are more difficult to source I managed to find a custom made unit, which consists of two of Mazda RX7 cores side by side, in a vertical setup, giving a tank area of 590mm long by 65mm wide for a total end tank area of 383cm2, so enough for 530bhp based on the formulas.

I decided I would attempt to test these intercoolers for flow, but not heat rejection as I didn't have appropriate equipment, so this was only one aspect of the intercooler but an important one as detailed above.

The 4 test candidates are in increasing size, and will be added to in the future:

Stock OE Fiat coupe 16vt
Stock OE Mitsi Evo 6
HDI GT2 competition core Evo 6-9 (rated 780bhp)
Custom twin RX7 core vertical

Their sizes were:

Coupe OE 16vt
Core end tank size (cm2) 123cm (190mm x 65mm)
Area of tubes 13*65 at 6mm = 5070mm2

Evo 6 OE (rated 400bhp)
Core end tank size (cm2) 156cm (240mm x 65mm)
Area of tubes 24*65 at 3mm = 4680mm2

Evo 6-9 Race (rated 780bhp)
Core end tank size (cm2) 182cm (260mm x 70mm)
Area of tubes 16*70 at 7mm = 7840mm2

Vertical flow (2 x RX7 cores)
Core end tank size (cm2) 383cm (590mm x 65mm)
Area of tubes 42*65 at 3mm = 8190mm2







While it was going to be difficult to test the heat transfer abilities in the garage I hoped I could test and confirm restriction. I had seen it done a few times on the web using a home built nanometer and a leaf blower but to be honest I was pretty skeptical if meaningful data could be made of it. I thought perhaps it would at least confirm the restriction of the OE coupe IC being high, the Mitsi evo as a good upgrade and that I should be trying my best to use the twin RX7 cored IC. This was even though with an inlet top and bottom, it would be by far the most difficult to fit.

My trusty echo leaf-blower is rated at 380cfm at no pressure, which while a long way short of a large turbo at full noise and 20psi I hoped would be enough to show variations in flow restriction, An hour or so in the garage later and I had made the test rig, it is made of couple of Samco hoses, a tyre valve with the valve removed. There may look like duct taped joints but they are air tight and in all cases press fits, the duct tape is only to ensure no leaks.



It is fair to say I had prejudged the result as the RX7 being the best with the biggest core and greatest tube area.

Anyway, the tests:

To get a benchmark I tested the OE coupe IC, at wide open throttle on a fully warmed blower. The result showed 2cm at idle, and 15cm deflection at wide open throttle. The OE IC actually uses surprisingly large internal tubes at 6mm, which is big for an OE intercooler, but the overall unit was the smallest by a fair margin with total end tank area of 123cm.





Coupe OE 16vt at full load showed 15cm of deflection.

Next off the rack was the Mitsi Evo OE. The surprising thing is that at these pressures the evo flows near identical to the stock coupe unit, at 2cm at idle and 14cm at WOT. There was 156cm (20% more) of tank area but smaller 3mm tubes rather than the 6mm of the coupe. I am not sure what higher flow rates would show. There are however obviously benefits from non-plastic end tanks (coupe tanks known to crack/break). There are other benefits, such as a FMIC mounting point to give better airflow and hence heat rejection, a higher number of smaller tubes which likely aid in cooling the charge more efficiently, and bigger frontal area again cooling the charge. I suspect could I test heat rejection it would be far superior but purely in terms of restriction at this flow level was a little better (6%). It should be noted 400bhp @ 25-30psi is where the mitsi guys consider this OE unit to be at maximum.





Evo 6 OE at full load 14cm

The HDI GT2 competition core was next and flowed far better. This is an "ebay" IC but one of the more expensive ones - it has less than half the shown restriction of either the OE or stock mitsi evo cores at 6cm deflection to 14cm on the OE Evo and 15cm on the OE coupe; it has as large a frontal area, 3inch inlet and outlet, alloy end tanks, and the same frontal area as the OE evo with a slightly deeper core, but unlike the stock evo's 24 tubes at 3mm, it has 16 tubes at 6mm. The HDI literature would suggest the cooling is excellent which I cannot test but the restriction is certainly lower.





HDI GT2 evo6-9 competition flow load 7cm

Last to test was the one I assumed would have the lowest restriction based purely in end tank area - with an end tank of 383cm2 it had triple of the OE, twice the others. While the restriction was lower than the evo and coupe at only 10cm of movement, it was still significantly higher than the HDI evo core which was surprising, it has more tank area, more tube area (though only slightly more than the HDI core). There are a few other factors to consider - on all the others the airflow blows straight at the core tubes; on the RX7 twins the air must turn 90 degrees to enter the tubes from the end tanks. It also must turn 90 degrees twice to exit though a port not right at the end of the tank as can be seen in the pic below. It must hit that end wall and turn itself out the hole facing the camera in the picture below.





Twin RX7 core vertical flow full load 10cm

So the table below shows the summary data for the testing.

Coupe OE 16vt
Area Of Tubes 5070mm
Displacement at Idle 2cm
Displacement at full throttle 15cm

Evo 6 OE (rated 400bhp)
Area Of Tubes 4680mm
Displacement at Idle 2cm
Displacement at full throttle 14cm

Vertical flow (2 x RX7 cores)
Area Of Tubes 7840mm
Displacement at Idle 1.2cm
Displacement at full throttle 10cm

Evo 8 competition* (rated 780bhp)
Area Of Tubes 8190mm
Displacement at Idle 1cm
Displacement at full throttle 7cm

So a surprising result, and one that didn't sit well in my mind or in theory.

I had read a test of end tank designs on the web saying a poor design, even square boxes rather than properly flowed and shaped end tanks, can add 7% additional flow restriction to an intercooler, so in the case of the RX7 end tank where the air effectively "hits a wall" before swirling and taking a 90 degree out of the IC must be far worse than 7% was my educated guess. To test this further I decided to cut the end off the end tank and test again as I really believed this added a lot of restriction.

So the RX7 twin core required modification. I would not have used it at that level of restriction - with the end cut off the charge could exit as if straight into a large tube with no turn required - it can be seen in the pics,



and again hooked it up to the deflection rig



The result was a more impressive 2.5cm of deflection at WOT. This value was 10cm with the air exiting as above. This value of 2.5cm when compared to the other results at WOT (15cm/14cm/ and 7cm) really speaks for itself and for the benefits of high end tank areas. Also the preliminary finding shows the charge air does like sharp edges and this result now puts it far above the evo competition core for flow - though adding the exit tube will need to be done carefully as not to add restriction back into the system.



So take from it what you will. My belief is that unnecessary restriction on the way to the head is a negative with no benefit. I will therefore be running a vertical flow IC on my coupe with the new build. I should stress though a vertical flow IC on a standard coupe would be a very difficult fit. My coupe has many mods that will aid in its fitment, the AC removed, a custom radiator, the bumper is modified as is the bonnet slam panel. Also it will be or has been said that the IC tubes are too short on a vertical flow for full cooling to take place. I have also heard it said that the majority of the charge cooling takes place in the first 4-6inches of a IC tube, so additional length has little benefit - I guess I will see. I also think though it shows the importance of end tank design in the main, and I suspect the OE evo vs OE coupe may show the sweeping coupe plastic tank to be superior with no sharp edges when compared to the evo. I plan to test one of those GTR type IC's with the very well shaped tanks.

I am currently looking to source a vertical flow, with a 4" core, and well designed end tanks for testing - I suspect this will be the solution I will use, assuming I can address the IC routing issues such as the Treadstone TRV25-S, if you read this Treadstone please send a unit for testing ;-).

Cheers

Very useful topic!

I guess next is to try and find a way to test cooling efficiency... Which is I believe the most important.

Can you tell us the length of the intercooler core tubes for each of the four intercoolers? Useful to calculate the frontal area and the total area exposed to the cooling airflow (although the area added by turbulators is obviously not included)

sure, tube lenghts as below

Standard coupe 295mm (13 of)
Evo competition 560mm (16 of)
Vertical flow RX7 165mm (42 of)

OE evo 6 I don't have here any longer but I believe it is circa 450mm (24 of)

Thanks for that.

I did a quick check to convert 15cm of water to psi - and the result was 0.213 psi. If that is correct, even on the worst performing intercooler the pressure drop seems pretty minimal in the scale of things. Even at double the flow (380CFM approximates to 260 BHP so say, 520BHP) that would surely only double to 0.426 psi - so is it at all significant?

I ask in genuine ignorance of the answer.....

good one Paul!

It's not a measurement of pressure drop but rather a measure of flow resistance, you'd need two measurement points (one either end) to establish pressure loss. I'd prefer to look at the results as a %, they are all within reasonable use (you hope so wouldn't you) but some are better than others. Presuming that the frontal area is subjected to ambient air flow (it must pass through) it'll cool just fine (within the first few inchs as Mr Bell tells us and I'm pretty sure he'll know).

I appreciate your comment re pressure drop V resistance.

However, in this case measuring the pressure at the "other" end of the intercooler is surely not necessary as it will be at ambient atmospheric pressure, there being no further restriction after it, so any pressure measured on the inlet side is relative to that.

Also, I know of no other measure of "resistance" to airflow other than the pressure dropped across a resistance. Is there some other measure?

Is 165cm, as in the case of the RX7 cores, "long enough"?

Differential?

Is that to me?

If so, the differential is surely 15cm H2O V 0cm H2O in the case of the standard intercooler above....

i.e. the pressure measured one side versus the pressure measured (or reasonably assumed in this case) on the other side shows the loss.

Of course, it must be remembered that WITHOUT a pressure differential there would be ABSOLUTELY NO flow.....

Yes, sorry, I just hit the quick reply.

I see it as a % of restriction rather than a pressure drop, can't explain it any different to that I'm afraid. I see intercooler pressure drop as being something thats measured at either end and different to this test. I think I'm tangling myself up trying to understand why you see there being little difference between the best and worst, I don't understand why psi comes into it. Maybe just not looking at it the same way as you, maybe you could explain to me more simply.

Yes, I see 165 mm as being enough.

Still fail to see the actual relevance if not compared to heat transfer. For all I care I can make a straight tube without any surface increasing trickery and it'd be crap at cooling anything but offer very good flow. The standard 20vt has the same chicken wirey stuff inside the tubes as it has outside the tubes, the better flowing ones would have cores that offer cooling along the length, so not getting in the way of any airflow.

What good cooling without flow ability? All intercoolers cool air.

From BELLINTERCOOLERS site:

What is flow loss thru an intercooler?

Flow loss is what is measured in the pressure loss and is the restriction presented to the smooth, easy air flow through the system. Essentially, the drag. It is measured by a pressure difference between the air charge entering the intercooler to that exiting the intercooler. This flow loss is due to the aerodynamic drag offered by the shape, the net area of the tubes, the length of the tubes, and the density and style of the turbulators.

But that's not what Paul is showing in this test, it's a different thing altogether, the quote from Bell is what I was describing earlier. This is about flow restriction not loss.

Again, from the BELLINTERCOOLER site, pressure drop is a function of FLOW (rather than boost pressure) so I will correct the above. The pressure drop increases with the square of the increase in flow, so in this case (760cfm/380cfm)^2 * 0.213 which equals 0.825 psi.

Yes, it's by how much that sets them apart, WHILE keeping a good flow. Cooling = surface area.

Who cares if there is a flow restriction if there is no flow loss involved? However, if there IS a flow loss it can be read by the pressure difference:

Flow loss is what is measured in the pressure loss - it can't be said any clearer than that.

I give up, like Paul said "So take from it what you will."

Hey - its how a flowbench works. Just ask GC and he will explain it to you I'm sure.

Nice post, Saint. Good research.

Whatever the size and shape of the intercooler, there are just two things that concern the likes of us (ie. not designing and manufacturing IC): the pressure drop and the efficiency. Fortunately they are both easy to test.

To measure pressure drop simply fit a boost gauge just before the IC inlet and then just after the IC inlet. The difference between the two readings is the pressure drop. The maximum acceptable pressure drop mentioned in various books/sources is 1.7 psi.

Efficiency of an intercooler is its heat-shedding ability - again nice and easy to test using a DVM with a K-type thermocouple wire attached:

First just stand outside and note down the ambient temperature.

Next loosen a hose close to the IC inlet and pass the wire into the pipe enough for it to be exposed to turbo airflow.
Next WOT it to redline and note the maximum temperature reached (better still look where you're driving and get your passenger to note it!).
Next do the same for the IC outlet.

Then some maths: (IC in temp- IC out temp) / (IC out temp - ambient temp)

The answer is a percentage

I did this on my car a few years bag and posted it up in my Project thread. On the setup I had at the time the IC inlet temp was something like 95 deg and the outlet was 30-odd (IIRC, can't be arsed to click thru and read it!) The intercooler efficiency was 83%. Again, the min. efficienty quoted by various sources is something like 75%.

It's good to know a bit about the design issues that go into making an intercooler. But fortunately as buyers of intercoolers we don't need to worry ourselves about them. As long is the pressure drop and efficiency is acceptable, it's all good!

It seems like simple math

As I titled the post this was about flow resistance rather than heat transfer, I have to say though 3 of the 4 are OE cores (Fiat/Mazda/Mitsi), all tube and fin, and are of the design in the third of your pics, the 4th aftermarket is similar also (I will photograph)

Being OE type cores they are similar internals so it would suggest to my eye the heat transfer capabilities are similar per square centimeter- there are 3mm and 6mm cores with internal turbulators - The few 6mm tubes, like the coupe OE, have less surface area directly contacting ambient air but effect of this I haven't tested

I don't think any of these cores will be particularly awful at heat transfer - they are hardly straight tubes, but heat transfer wasn't what I was testing

I'm comfortable that any of these cores would cool the charge - I did question "would 165mm tubes be long enough" and from my reading by people far more experienced - ie graham bell the answer would appear to be yes

Thanks guys for the good feedback

Was interesting stuff to test

Cheers

Paul, I have battled the same question with myself (and still am) From what I can gather, when you take a vertical core and crossflow type of same frontal area, the only change is to the charge air flow window.

The advantage of less system restriction in the vertical core is obvious but will the shorther tubes of a V.C. cool the air charge as effectivly (or to a level acceptable) as a HC ? I honestly don't know, but one respected tuners take on it is that; Because of the enlarged charge air window, consider the velocity of air through the VC is much slower than crossflow (HC), therefore it spends much more time in the intercooler core. This gives the charge time to be cooled to a level simlar to the Horizontal version, while neting a much lower pressure drop / flow loss.

One major problem is distributing and collecting the charge smoothly because of the massive area change from core to pipework as you have found.

Got that water heater working yet ?

Are you sure

To answer to your question it is all in the design of the intercooler tubes.

Time is irrelevent in a heat exchanger. The major varibles for heat transfer are velocity and surface area. It's easy to make an intercooler flow better, just make the tubes larger, less restrictive. The velocity will decrease however and this will reduce the efficiency of the cooler (thicker boundry layer etc) so you need a larger surface area (i.e. larger intercooler) to make up for the reduction in efficiency.

Most automotive intercoolers are over engineered, i.e. they are more efficient than they need to be. This is to insure they work adiquately under all conditions. i.e. high ambient temperature, low RPM/velocity etc etc. They usually become very restrictive under good conditions and high engine outputs.

I would argue that the major variables for heat transfer are surface area and the delta T parameter.

Velocity does play a role (both sides of the core), but simplified isn't velccity a fancy name for speed ? ie; distance over time ??