A pump's performance is going to be effected more by inlet restriction than by pressure on the outlet side. It's my opinion the pump should draw directly from a reservoir (not through the intercooler or heat exchanger), pump through the intercoolers, then the heat exchanger (this maximizes temperature delta between the intercooler fluid and the ambient air at the heat exchanger), and then back to the reservoir.


Off-topic to this last post, but after a few heat cycles, it is evident that the poly Ys are not going to hold up. One of them has already split again. I am going to have to weld up my own stainless ones if I can't find a reasonably priced off-the-shelf solution.

 

This will be my new setup:

 

Like Zeph said, pumps are better at pushing than pulling, so ideally, you would flow from;

tank ==> pump ==> I/C ==> H/E ==> tank.

This way you have a tank that is your reserve of cold fluid and the pump isn't trying to "pull" fluid. H/E's in a series are going to get temps down colder than in parallel.

As for H/E flow, bottom to top will make it easier to get out trapped air.

 

Thanks Zeph and Matt, that is what I initially thought, makes sense.

 

Matt is right, but there are some considerations depending on what your goals are with the system, and which side of the system is better at transferring heat.

Heat exchangers in series will result in colder water temps at the exit, but also lower total BTU dissipation capacity for that side of the system. For max cooling in terms of absolute energy dissipation, you need as much surface area as possible exposed to the hottest possible coolant. This is achieved in parallel flow configuration. Basically, putting them in series will definitely give you colder temps when your heat exchangers are outperforming your intercoolers. But at WOT, if your intercoolers are removing more heat from the charge than your heat exchangers can dissipate, then you would probably stay cooler in parallel. Another consideration is how much pressure head your pump can stand. Series heat exchangers will increase pressure head/restriction in the system whereas parallel will reduce it.

 

Very good thought process on this Zeph. I plan to try both in due course and I will advise. Cheers.

 

Further to my Pierburg CWA50 pump, can anyone advise how to bench test this pump? I sent the first one back which the vendor promptly replaced. Unlike the first one which had the same resistance across the + and - terminals as between the - and the pump housing, the new one has a very large difference as it should, the increased resistance being from the motor windings. Connecting straight to a battery with over 13.5 volts showing on the meter I still cannot get it to work. There is supposed to be a delay of a few seconds however after 10sec it will not turn on. My new Johnson 30 starts up instantly, however my OEM Bosch pump does not turn on. I was concerned with the latter when I first got the car as it suffered much more heat soak power loss than my stock E55. The dealer said all was fine with the pump. Is there internal circuitry which prevents running until a certain temperature is detected?

 

I don't know the specifics of that pump, you'll have to compare with somebody else who has one or wait for the manufacturer to respond. I'll just be sticking with dumb 2-wire pumps, lol.

 

Thanks again Zeph, I got it working. There is a delay of a few seconds for some reason. The lack of a proper plug meaning a jury rigged connection was the problem. As this pump is used on late model BMWs I will have to go to a dealer and try and buy one.

 

I just soldered directly to the pins on my extra pump. Didn't really feel like paying a ridiculous amount for a pigtail. It's straight hardwired for now but if I ever have to remove it I'll just cut the wiring and salvage a fan connector off of something.

 

Johnson, Bosch, Meziere etc are simple pumps - there are two wire connections to the motor.

Pierburg, VDO-Continental, EMP etc are smart pumps with electronic control. The motor is controlled by internal electronics, and the pump really needs an external controller to work properly.

Having said that, Pierburgs at least will run at high speed if 12V DC is applied to the control pin (as well as the power pin). Just need to wait a few seconds. The pin out is given in this thread.

Nick

 

In your opinion, is that done strictly for economy/pump life purposes? There's no reason I wouldn't want the pump running full-tilt all the time when I was intending to get maximum performance, correct?

 

To be honest, I think the ideal way to drive an IC pump is with the poly-V belt. More revs - more heat - more pumping. However, electric pumps are being used more and more these days as an economy measure, so that's definitely part of the reason.

In the case of the Bosch pump, it really doesn't last very long, due to its commutator brushes, which wear out fairly quickly (like most DC pumps). Only the clever electronic control, electronic commutator pumps like the Pierburg or EMP get round this, and are much more durable as a result. SO the Bosch is thermostatic to preserve its life.

Having said all that, I still think its worth using a proper pump controller. The Pierburg runs very fast, and in my experience it tends to froth the coolant, which is not good! I've read several other anecdotes of the same thing, so I think its a good idea to let the coolant "rest" and settle out from time to time.

Nick

 

Interesting. I definitely agree about mechanical pumps being a LOT more powerful and somewhat regulate themselves due to RPM changing at the same rate as the engine. But packaging just sucks.

 

This upgrade is STILL fighting me, lol. I got aluminum Ys made up and inatalled, and of course my pump inlet Y then went to leaking so I pulled that apart and plastic-welded it again...actually melting the plastic together as opposed to epoxy this time, topcoat with silicone and put it all back together and was only able to pull 95% or so vacuum on the system. As it turns out the x3 radiator end tank has gone to leaking again as well. So I said screw it and threw some stop leak in the system. After a short drive it appears to have stopped dripping but it also could be that it's just now warm enough that it's all evaporating before it hits the ground. In any case, I'm not about to tear the front end off again right now so I am just going to hope the stop leak takes care of it and if not, just monitor my fluid level and add as necessary. I mean I have like 4 gallons system capacity now so a slow drip isn't really a dealbreaker.

 

could some one take a water to air inter cooler,and modify it to hold a gas that would keep it colder than water?

 

It can certainly be done, but it becomes an increasingly impractical setup the colder you go. A killer chiller type setup can get the fluid down to about 0 degrees F if you run propane as the refrigerant and give the system long enough to cool down, but since there is no phase-change cooling taking place in that type of system then you need to have a very large system capacity (in terms of coolant mass), otherwise the temperature rise even just during a 1/4 mile run is going to be very large without the benefit of active cooling (from a front mount heat exchanger) during the pass. And obviously if your fluid is colder than ambient, you can't use an ambient heat exchanger because it will just help warm it up faster.

The better solution would probably be to use a colder type of ice than just water ice in a reservoir. If you had a mix of water and antifreeze frozen into ice cubes, or water/methanol, etc you can get a lot colder melting point and take advantage of the latent heat of melting to offset about 144x more heat before getting any temp rise than liquid water would hold. This allows for a much smaller system mass while maintaining colder temps, but again, it's practicality is limited to pretty much a drag-race only situation since you've gotta bring a cooler full of ice and drain liquid off each run in order to add more. And if you're not using just water ice, you definitely have to catch what you drain and recycle it.

Other options would be like I think you're suggesting, eliminating the liquid coolant altogether and spraying CO2 or some other sacrificial refrigerant through the cores. That would probably work, and pretty well, but would be a drag race only setup and would take something like 4 lbs of CO2 per 1/4 mile run, so realistically you'd only get 2 passes out of a 10-lb nitrous bottle full of CO2.

One other problem is that when you get something that cold, atmospheric humidity is going to condense out of the charge air and instantly freeze on the core, eventually blocking the airpath and choking the engine. So you'd only be able to run it in very low humidity in terms of the atmospheric conditions, or for very short time periods.

Another fact to consider, as I've hypothesized before, is that on a turbo car it most likely won't even offer you any significant benefits to cool the air any further than necessary to ensure the computer isn't reducing boost or timing. On a SC car you'd see some gains from the reduced drive cost from the lower boost pressure it would take to make the same power. But on a turbo car that's already running the turbos to the limit of the compressor map the limiting factor as it relates to max power is going to primarily be air density available at the compressor inlet...pressure differential across the compressor is going to have little to no effect on the turbo's maximum mass flow capacity since most turbos have their peak mass flow at a pressure higher than what we are dropping to by redline. You could most likely get more peak power out of these turbos on a somewhat smaller displacement engine at higher pressure. There are no peak flow gains to be had that I can see by reducing boost pressure, in fact there will either be a reduction in mass flow output (if you stay on the compressor map), or a significant reduction in efficiency if you manage to run off the edge of the map.

 

Phew!

Isn't the maximum mass flow rate a function of the intake density?

ie: where you variables like pressure and flow that affect mass flow, aren't these usually normalized to a particular intake pressure/temp/density?

I always thought the lower the temperature the better (as long as you don't freeze).

Nick

 

Yes. A function of the density at the turbo inlet. Increasing density in the intake manifold only makes the engine breathe better, not the turbo. CO2 chillers in place of the air filters however would affect the whole system.

 

The challenge then becomes implementing something that increases density from cooling more than it decreases it from pressure drop.

 

To clarify what I mean a little, here's a flow map for a borg warner EFR 6258, which is a bit bigger but probably the closest relative to the S600 twins that I can actually find a map for. You can see that it hits a hard edge at 44 lb/min and the flow rate does not increase at lower pressure. It does decrease a bit at the top of the pressure capability of the turbo, but that's not where we're operating so it's irrelevant for our setup. Many turbo charts don't actually extend the line past the body of the map, in which case they usually appear to actually be capable of flowing more at higher pressures (although I don't think that's the case, I just think the efficiency is far worse at the lower pressure but it will still flow up to the max rate). Since our turbos are obviously operating along that hard right edge (pressure dropping as RPMs and therefore mass flow rise) then operating with less head pressure by intercooling doesn't help us get any additional mass flow out of the turbo.


The result from intercooling is that the engine is capable of ingesting more air mass at a given pressure due to the greater density/lower temperature, but if the turbo can't deliver that greater air mass then the pressure will just drop further than it would have otherwise. The net mass flow will remain constant since the increased density is acting only upon the engine, not the turbo. To get the turbos to flow more, we need to improve their ability to breathe on the suction side through less restrictive filtration and/or chilling the air prior to it entering the turbo (with as little pressure drop and/or volumetric displacement of the air charge from vaporization cooling as possible).

 

Meaning .... the loss of pressure across the cooler?

 

OK, I think I see where you're coming from. But doesn't that chart only apply to a specific inlet air temperature / pressure condition (presumably 298K / 1013mbar or something similar).

Still thinking..... I was about to ask what difference intercooling would make to that chart, but it wouldn't make any difference, would it?

Intercooling affects the engine inlet temperature, not the turbo inlet temperature.

Nick

 

Yeah, I think now you're getting it. A flow map is standardized, yes. But what I'm saying is that it will be affected by inlet pressure/density but not so much by outlet pressure/density. So an intercooler between the turbo and engine is largely just for the benefit of the engine. If the turbo is at max mass flow for a given inlet preasure then you aren't magically going to get it to flow more by cooling that air/reducing the pressure between the turbo and engine. And yes, I meant in order increase the turbos mass flow capability, a pre-cooler would have to chill the incoming air enough to generate a density improvement large enough to more than overcome any density loss from pressure drop across the core. When you've only got atmospheric pressure to work with, the density penalty is much higher for a given amount of pressure drop. So it would have to be a really free-flowing heat exchanger that also happens to be extremely efficient in terms of cooling.

 

Just a heads up guys, there are some brand new 12V Stewart EMP water pumps on ebay right now, the guy had like 7 of them and I bought 2. I made an offer of $250 each for 2 of them and it was accepted. That's practically half price. This is the king of all intercooler pumps. 3 of them left, jump on it!

http://www.ebay.com/itm/282195356531