Cooling System Questions

[Chris Duncan]

Ordering the radiator for the GT40 and Tim, the resident expert at Ron Davis Radiators, is saying that 1-1/2" coolant lines will be fine. He has a mid engine chevy V8 car that has a 1" line to the radiator and a 1-1/4" return, that he says cools fine. (Apparently Ron Davis makes radiators for ERA so the GT40 design is already on file, you don't want to know the price though.)

Previous research shows that the original cars and the Mark V's had 1-3/4" lines. That said though 1-1/2" would be easier in the crowded area where the lines go from tunnel to engine and reduce overall weight with smaller tubing and less coolant volume. It's also noted that the Ford water pump outlet I.D. measures 1-3/8", and I've also seen where larger lines can adversely affect flow. Bathroom sink drains go slow if the drain is too large.

Another question is what affect does a double bypass radiator have on flow. Some good info in the Stewart article below but it says that it takes 16 times the pressure to pump coolant through a double bypass radiator. Tim says that's wrong and he's argued with the Stewart people online and supposedly they are going to retract that statement.

http://www.stewartcomponents.com/html/tech_support/techtip1.asp

The double bypass is more desirable on a GT40 because both inlet and outlet can be placed at the bottom of the radiator making the plumbing in the front much more sanitary.

And here's another good article on water and oil cooling problems and solutions in a Pantera. This guy has an inline thermostat that has better flow than the typical manifold housing type.

http://www.spacecitypanteras.com/Technical/Pantera_Cooling-JTaphorn.pdf

Maybe we can get Adam Christian involved, it would be interesting.

 

[Adam C.]

Since I've been specifically called out on this one I'll chime in.

The main culprit we are talking about here is FRICTION.

It occurs betweent the pipe wall and the moving fluid.

The formula for calculating the pressure drop due to wall friction is:

deltap=(4*f*L/D)*(1/2)*rho*u^2

where:

dletap = pressure drop between two points in a pipe
f = friction factor
L = pipe length between the two points
D = pipe diameter
rho = fluid density
u = fluid mean velocity

Please forgive the crappy naming convention, but we normally work in greek letters.

So if the length of pipe from the radiator to the engine is fixed, what is the difference in pressure drop due to the change in pipe diameter?

f, the friction factor is basically constant over the range of velocities we're looking at. L is constant, as is rho. So all that is changing is the pipe diameter and the velocity. To find the velocity we must find the change in cross-sectional area:

A = pi*(D^2)/4

where:

A = cross-sectional area
pi = 3.14....

so the change in crossectional area is

A2/A1 = (D2^2)/(D1^2)

For us that is (1.5^2)/(1.75^2) = .735. So what does that do to velocity?

u = Q/A

where:

Q = volumetric flow rate

So if we say that we want the same volumetric flow rate.

u2/u1 = A1/A2 = 1.361

so D goes down, u goes up and everything else is fixed. What does this do to our pressure drop through the tube?

deltap2/deltap1 = (u2^2*D1)/(u1^2*D2) = (1.361^2)*(1.75/1.5) = 2.161.

So if you can maintian the flow rate, the pressure drop, and therefore power necessary to create it, is approximately double for the 1.5" tube.

On the other hand, if you can only maintain a certain pressure drop, the flow velocity is going to decrease to:

u2/u1 = 0.92

and so the volumetric flow rate will decrease to:

(u2/u1)*(A2/A1) = 0.92*0.735 = .680

So your flow rate would be approximately 68% that of the 1 3/4 pipe.

The truth is something between these two cases. As you increase the restriction to the pump, the pressure will increase some, and the flow rate will go down some, but at least this gives you some idea.

Also remember that we are talking about the the pipes alone. Anyone with an electrical background can think of the radiator and pipes as resistors in series. If the resistance of the radiator is >> than that of the pipes, then the pipe diameter does not matter.

So with the double pass radiator you have effectively reduced your cross-sectional area by half, and increased your length by double. The same calculations above can be used to find the diference in pressure and flow.

Maybe Kalun can run through those for us!

If we found out how many, what size, and length the radiator tubes are, then we could find the relative resistance of the radiator to the coolant pipes.

Adam

 

[Bill Musarra]

Damn,
I haven't seen that much physics since I left grad school!!!!!
On a better note, My question is what type/kind of radiator are people using in their 40's. Single pass? Dual pass? number and size of cores or rows? How thick is it, are you sticking with 2" or are you going wider? Is a "Be Cool" necesary or will a "Griffin" do the job. Is anyone having problems with their electric water pumps?
I'm trying to decide what I will need.
Thanks guys,
Bill

 

[Jim Rosenthal]

We haven't got as far as the radiator yet but I believe Safir use a radiator made by BeCool, packaged with their fans (2) and it is three rows of pipes, I think. And, while we're talking about cooling system issues, is there such a thing as too much coolant flow- does there need to be a dwell time of coolant inside the engine for heat transfer? Just curious.

 

[Chris Duncan]

Yes Bill I was looking for 2 answers really. What Adam thought about flow as it relates to tube size and what everyone was running on their 40's.

If it works on paper I'm going to use 1.5" pipes with a double pass 2" aluminum rad. With a short Ford Motorsport water pump.

 

[Lynn Larsen]

Well guys, I have 1 1/2" pipes and a Griffin radiator. And while my car hasn't hit the roads and been in summer stop and go traffic with the radiator 6-18" above 160+ degree tarmac, it has set in my garage on a warm Carolina day and idled for as long as 2 hours without ever getting above the temp where my fans kick on. For what ever that is worth. My radiator is something like 12x24 two core. I don't think it is double pass as my inlet is high right and the outlet is lower left. One thing I have noticed is that there can't be to much differential pressure across it because I was not getting much flow through my heater core which if fed by Ts in the lines on either side of the radiator. Either that and/or it was still air locked.

I plan to put a Stewart inline pump just after the heater T on the return side with a thermostatic activator. I thought I had bought one the other day on eBay, but some nut went and bid it up to within a few dollars of new. They are expensive at ~$400 and change, but they flow 50gal/min and look just too cool coming in 4-5 anodized colors or polished. On drag cars, they have the impeller removed from the engine driven water pump (still carries belts if needed.) But on our cars, I don't see any harm in having both. If I am wrong on this, someone educate me, please! (addendum: Stewart's website now describes its use for cars that experience over heating at low RPM with their engine driven pump.) I had placed a pic of this pump in the parts area of the old forum, but it is probably gone by now. I think they are so much more compact and easier to fit than the Australian one everyone raves about and they cost about the same. Frankly, I really don't understand why no one has given them a try as I know several of the Aussie ones have been installed. They also make an integrated radiator/water pump unit:

Find out more here:EMP Stewart Components

Lynn

 

[Chris Duncan]

Jim,

You should be ok with one fan, the MK V I know only has one, unless your stuck in Phoenix rush hour or something.


"Is there such a thing as too much coolant flow- does there need to be a dwell time of coolant inside the engine for heat transfer?"

from the Stewart tech article, pg 3

"A common misconception is that if coolant flows too quickly through the system, that it will not have time to cool properly. However the cooling system is a closed loop, so if you are keeping the coolant in the radiator longer to allow it to cool, you are also allowing it to stay in the engine longer, which increases coolant temperatures. Coolant in the engine will actually boil away from critical heat areas within the cooling system if not forced through the cooling system at a sufficiently high velocity. This situation is a common cause of so-called "hot spots", which can lead to failures."

Adam,

I measured the rad tubes and stuff, going to have to wait till I have time to plug in the numbers to your formula, and my calculator got wet under my shade tree engineer's shade tree, which isn't really a shade tree because it's always raining in Seattle.

One thing I noticed is the ends of the tubes in the rad tank have sharp edges, not your nice bell shape like induction intake air horns.

After the number crunching you didn't tell us what size coolant pipes your GT40 is going to have, wait a minute you said it is going to be a MK II, nevermind. Tell us what size tubes you would run if you were building a MK I.

 

[Adam C.]

I haven't given it much thought. Coolant tubes are so far away for me. In general, we have shown bigger is better. The frictional losses are lower, there will be more capacity in the system, the increased diameter gives more surface area to radiate heat from, etc.

The down side is that the heat is being radiated to the passenger compartment, and that you need to fit the big tubes through the central tunnel, along with all kinds of other things.

As for the bellmouths, I wouldn't worry about it too much. For induction systems the inlets are a significant portion of the flow losses. Because of the long lengths of tubing, the frictional losses dominate in the cooling system. It is that resistors in series thing again.

Let's crunch some more numbers and see what dominates, the radiator or tubes.

 

[Lynn Larsen]

The time factor in heat flow has been a raging argument for years. Those who rely on experience will say that if you don't put some sort of restrictor in place of the thermostat if it is removed, the engine will run hot. From a heat migration point of view, the rate of flow is determined by the delta in temperature between the two sides and the material between them. On a molecular level it doesn't make any difference if the same molecule (of water) is there or not, it is the difference in the temperature of the molecules on one side and the molecules on the other since the material between them remains the same. So the way time comes into it is as the flow rate slows the delta between the molecules on either side will drop, thereby reducing the rate of heat flow. As the flow goes up, the delta T should also go up increasing the rate of flow. When the flows are such that the amount of heat being lost to the atmosphere through the radiator equals the amount of heat being generated by combustion, the system should reach equilibrium and have a constant temperature. Decrease the flow rate of the coolant and the systems temp should go up; conversely, increase the coolant flow and the temperature should go down (this assumes the ambient temperature remains the same and the power being produced is constant. (The flow rate of the air across the radiator would have the same effect.)

I believe this to be correct, but I know it won't put an end to the argument. I also believe that when Adam models the system with an equasion, the result will support this argument.

Lynn

 

[Devin]

Adam,

I was thinking the same thing...that the heat from the pipes running through the cockpit would/could be more of an issue.

I was thinking that the pipes could be wrapped in the same foam insulation material that your water pipes in a house or building is wrapped in would help to keep cockpit temperatures down a bit (more of a concern for those of us that deal with 100+ degrees F on a daily basis for several months).

Anyone insulate their pipes during their build up? Is the heat from these a real concern?

Devin

 

[Chris Duncan]

Wolfman,

"that the heat from the pipes running through the cockpit would/could be more of an issue."

That brings up another question. Do you use aluminum pipes which save weight but transfer/radiate a lot more heat than most metals, or do you use stainless pipes which transfer a lot less heat than most metals?

I've seen both aluminum and stainless pipes on the original cars.

So this adds to the question everyone, what diameter and what material coolant transfer tubes are you using in your 40's?

 

[Paul W]

Has anybody considered using copper pipes in their 40 ????

PW.

 

[Lynn Larsen]

I insulated mine with foam that theortically should be able to stand up to temp of the pipes. We'll see in a couple of years. I also wrapped the foam with a spiral of insulation tape. The heat was the major concern, but also noise was considered. Both the noise of possible vibration (mostly of stuff touchning the pipes, but also of air/steam bubbles. The first time I drove my car around the block, there were some real hidious noises coming from the pipes. Now that I have purged all of the air from the system, a process whose importance should not be underated, this won't happen again, hopefully. I had heard stories of warm to hot tunnels too. I considered filling the tunnel with foam, which has been done believe it or not, but I made an effort to allow and encourage air to flow from front to back to help exhaust any heat escaping from the pipe's insulation.

Lynn

 

[Chris Duncan]

This has gotten really difficult really quick. (for me anyways, but that's what makes it fun)

>>The main culprit we are talking about here is FRICTION.

agree

>>The formula for calculating the pressure drop due to wall friction is:

>>deltap=(4*f*L/D)*(1/2)*rho*u^2

>>where:

>>dletap = pressure drop between two points in a pipe
>>f = friction factor
>>L = pipe length between the two points
>>D = pipe diameter
>>rho = fluid density
>>u = fluid mean velocity

>>Please forgive the crappy naming convention, but we normally work in greek letters.

Nonsense, your doing us a favor by translating.

>>So if the length of pipe from the radiator to the engine is fixed, what is the difference <<in pressure drop due to the change in pipe diameter?

So you haven't really used your full deltap formula to find the pressure drop, your just comparing the 2 diameters to find what the difference in the drop would be?

>>f, the friction factor is basically constant over the range of velocities we're looking at. >>L is constant, as is rho. So all that is changing is the pipe diameter and the >>velocity. To find the velocity we must find the change in cross-sectional area:

>>A = pi*(D^2)/4

so * means multiply? and ^2 means squared? (sorry I've never seen math on a basic keyboard)

>>where:

>>A = cross-sectional area
>>pi = 3.14....

>>so the change in crossectional area is

>>A2/A1 = (D2^2)/(D1^2)

>>For us that is (1.5^2)/(1.75^2) = .735.

so the cross section of the 1.5 D pipe is .735 or about 3/4 the area of the 1.75 D pipe.

I don't understand how you went from the circle area formula to only comparing the diameter squared? Why are you comparing the diameter squared and not the area?

>>So what does that do to velocity?

>>u = Q/A

>>where:

>>Q = volumetric flow rate

>>So if we say that we want the same volumetric flow rate.

>>u2/u1 = A1/A2 = 1.361

>>so D goes down, u goes up and everything else is fixed.

so pipe diameter goes down, velocity goes up IF volume/flow remains the same

>>What does this do to our pressure drop through the tube?

>>deltap2/deltap1 = (u2^2*D1)/(u1^2*D2) = (1.361^2)*(1.75/1.5) = 2.161.

>>So if you can maintian the flow rate, the pressure drop, and therefore power >>necessary to create it, is approximately double for the 1.5" tube.

The math is lost on me at this point but I can understand your end result.
To achieve the same volume/flow going from a 1.75 tube to a 1.5 tube takes twice as much HP at the pump because friction increases with the increased velocity. That's assuming that the pump is capable. This also brings into question how much HP does a water pump use.

from the Stewart site


"Stewart high-flow water pumps deliver up to 180 GPM (gallons per minute) of coolant flow (at 8,000 RPM), yet consume just 2.26 horsepower (at 4,000 RPM)!"

so a decrease in tube size would bring waterpump HP drain to 4.5HP. Maybe that's why the original 40's used 1.75 because they were chasing a couple HP, like all racers do, and the person saying 1.5 works fine is correct as far as he's concerned because the system cools just as well and the HP loss is not noticeable. OR what if the rad already has more resistance than the 1.5 tube, would it matter then?

>>On the other hand, if you can only maintain a certain pressure drop, the flow velocity >> is going to decrease to:

>>u2/u1 = 0.92

>>and so the volumetric flow rate will decrease to:

>>(u2/u1)*(A2/A1) = 0.92*0.735 = .680

>>So your flow rate would be approximately 68% that of the 1 3/4 pipe.

Is this figure dependent on a waterpump not providing ANY increase in flow/volume? This figure compares closely with the direct size comparison of .735.

>>The truth is something between these two cases. As you increase the restriction to the >> pump, the pressure will increase some, and the flow rate will go down some, but at >>least this gives you some idea.

>>Also remember that we are talking about the the pipes alone. Anyone with an >>electrical background can think of the radiator and pipes as resistors in series. If the >>resistance of the radiator is >> than that of the pipes, then the pipe diameter does not >>matter.

Until that pipe gets so small that it has more resistance than the radiator? Now this is how we're going to find out how small of a pipe is possible.

So your saying that if the rad resistance is greater than the 1.5" pipe then going to a 1.75" pipe isn't going to decrease the waterpump HP drain? I'm kind of lost on this one too, it doesn't make sense to me. Are you saying that if the loss at point B is greater than the loss at point A the the loss at point A does not exist?

>>So with the double pass radiator you have effectively reduced your cross-sectional >>area by half, and increased your length by double. The same calculations above can >>be used to find the difference in pressure and flow.

>>Maybe Kalun can run through those for us!

ha ha ha, (don't laugh at me, laugh with me) (they didn't teach this level of math at the school of hard knocks that I went to)

>>If we found out how many, what size, and length the radiator tubes are, then we >>could find the relative resistance of the radiator to the coolant pipes.

>>Adam

OK here are the sizes, carefully measured with wire gauges made up of bailing wire, wire feed wire, electric fence wire (reminding me of a shocked cat), solder wire, and a couple other wire sizes that were laying around. Inserted down in one of the rad tubes via the inlet neck of the radiator. The bailing wire came the closest at .048.

The outer dims of the rectangular tubes are 1.00 x .080. So with the .048 wire fit, that means a wall thickness of .016 which sounds about right. So the inner dim is .968 x .048.

The ERA spec rad has 2 rows of tubes with 64 tubes per row so total 128 tubes @ .968 x .048 x 12" long. Or 64 tubes 24" long with a double bypass. At first I thought why not just add up the cross sectional area, which totals 5.95 sq. in. which is way more than the 2.4 sq. in. of the 1.75 tube. But I'm guessing you can't do that because of that FRICTION culprit, which exists at the walls of the tubing. So somehow you have to bring the wall area into the equation.

I think you need to compare the ratio of wall surface to cross section area of the total rad tubes to the same ratio of the transfer tubes.

I'll compare total cross section and total wall area. I'm assuming it's correct to only compare the transfer tubes going one way to the radiator because of the series/resistance thing. (Maybe that's why on some production cars the return line is smaller than the supply line to the radiator, hey why can't we save some size/room/weight here too?)

OK, a 1.5D tube has a cross section area (A = pi*(D^2)/4) or (pi x radius^2) of 1.77 sq. in.

the wall surface of that same tube with a circumference (pi x D) of 4.71" x 108" = 508 sq. in.

So the ratio of cross section to wall surface of the 9' x 1.5" tube is 1.77/508 or about .0035.

The radiator has 128 tubes with a total cross section area, wait a minute, let's go with a double bypass. So the radiator has 64 tubes .968 x .048 so total cross section of 64 x .968 x .048 = 2.97 sq. in.

The total wall surface of the rad tubes .968 x 2 + .048 x 2 x 24" x 64 = 3,121 sq. in.

So the ratio of cross section to wall surface in the dbl bypass rad is 2.97/3,121 or .00095.

so comparing the two ratios .0035/.00095 = 3.68. So this radiator has 3.68 times the resistance/FRICTION of the 1.5"x 108" tube in question.

This is shade tree engineering math only, it may be TOTALLY incorrect, please grind the numbers Adam.

Ron Davis ERA spec dbl. pass aluminum rad.

64 12" tubes going up and 64 12" tubes going down so in effect

64 tubes 24" x .968 x .048 inside dimension compared to 1.5" and 1.75" x 108" tubing.

If you have time compare this to a standard rad with

128 tubes 12" x .968 x .048 so we can debunk the Stewart claim that a dbl bypass has 16 times the resistance.

and analyze/compare the possibility of having an even smaller return transfer tube. (time dependent also)

++++++++++++++++++

I think this information could be valuable to manufactures and scratch builders.

This car is very crowded/ compact to build. One of the most crowded areas is in the firewall to engine area. The rad transfer tubes both go through this area, if they could be reduced in size it would only help matters.

Another thing I don't think has been considered is the advances in radiator technology. The move to aluminum and the change in core tubing shape means a more efficient radiator than the brass unit in the original 40's. Radiators have changed but the transfer tubing size has stayed the same. If the radiator takes less flow to achieve the same cooling then why can't the transfer tubes flow less and thereby be smaller?

This gains advantages in weight and cost by reduction in tube size and coolant volume and the afore mentioned space considerations.

And don't think that this amount of weight reduction doesn't matter. A race cars weight is reduced in this very fashion. After the change from iron to aluminum heads, or cost prohibitive changes like a carbon fibre body, other weight reductions are on a much smaller scale and a large reduction is achieved by many smaller ones.

 

[Rick Muck- Mark IV]

Pi are square, except for the one my wife baked /ubbthreads/images/graemlins/shocked.gif

All's I know is that P1116 had serious cooling issues...I took the rad out and had it checked (although it only had 500 miles on it) and it was OK. I then turned to the water pump. I ordered a Ford Motorsport (was a Stewart pump) hi-flow and did a real good bleed out of the system. The pump that was origianally installed was a rebuilt unit John Vermersch got from the parts store next door... the new pump worked great, could leave the car idling on an 85 degree day with the fan on and NO overheating. Previously, a lap of the Glen at low speeds had the coolant puking outta the tank!

I think the whole issue is flow and whatever you can do to promote good flow and proper contact with the hot surfaces and a maximized temperature differential will do the job.

Rick /ubbthreads/images/graemlins/shocked.gif

 

[Steve Toner]

One of the articles in the new Ford GT book from SAE talks about the design of the cooling system, and there's realy quite a bit of engineering involved. For what it's worth, the new GT radiator has different-sized pipes for the inlet & the outlet. Inlet is 1.25 inches, outlet is 2 inches.

From the paper (and this should come as no surprise):
[ QUOTE ]
Coolant passage flow paths were analyzed using CFD (Computational Fluid Dynamics) and modified to minimize coolant pressure drop and increase total coolant flow.

[/ QUOTE ]

 

[Adam C.]

Kalun D,

Good try, you are on the right track. No laughing here, I just like to see others give it a shot so I don't feel like I'm just lecturing into cyberspace and the whole class is asleep.

Using your dimensions for the radiator, I worked the problem using an assumed flow rate of 90 GPH. This is half what stewart warner quotes at 8,000 RPM. The calculations are somewhat long, so I am just going to post them as attachments.

Buckle your seatbelts, here we go!

 

[Adam C.]

page 2 of 5

 

[Adam C.]

pg 3 of 5

 

[Adam C.]

pg 4 of 5