New Overheating Clue
 

Just got some new information about the overheating that occasionally plagues our cars. Seems that the water pump for the 4 cylinder Subaru engines has a tendency to cavitate over 6700 rpm. It might be that our water pump has the same problem. Cavitation occurs because the water pump creates an area of lower than atmospheric pressure that allows the coolant to boil at a lower temperature, creating steam bubbles in the coolant stream. That might be why a higher pressure radiator cap would help this problem, causing the bubbles to collapse sooner, before they have an opportunity to create havoc.

I have had a lot of difficult finding an "import" size high-pressure radiator cap. I have ordered a Crucial 1.4 bar cap (approx 20 psi), from Crucial Racing Products. Will be at Pocono next Friday and will post how it works.

Any aftermarket STi cap will fit our cars. I have a 18.6psi ( I think) Sti radiator cap on my OEM radiator right now

Dan,
Good point. Now we just have to find a better water pump;)
-Bill

Lets design some better impeller blades Bill and then submit the design to Subaru and see if they pick up the design :D:D

~ Chris

Damn... and I just ordered a new water pump to go with my 60k mile service parts, too... :o

Dan,
Do the wiz kids at your firm have the tools to model the EG33's water pump? (yeah, I know we're flowing water, not air;)):rolleyes::cool:
-Bill

New Overheating Clue
 

<<New Overheating Clue>>

what kind of over heating problems do the SVX have? I have never heard of this with the SVX I had. most of the overheating problems that Subaru owners have are self inflicted (they don't know how to properly bleed the air out of the system. )

This is "track-induced" overheating, not every day, run-of-the-mill overheating;):D
-Bill

What he said. Running at high loads above 3,000 RPMs for longer-than-normal periods. :D

How do you do that:confused:

RallyBob has a good post on how they do this for the EG33RS race car. Think it's in the 285 WHP thread.
-Bill

Those of us with MT's, at the track, will continuously take the engine over 7000 rpm. I've hit 7200 at the end of straights when I didn't want to upshift before the corner.

I'm wondering if an underdrive pulley might also help. Any comments from underdrive pulley users. Any downsides? Dead batteries, overheating while idling, etc?

Bill:

Yes we have the software, and even have the expertise to do liquids. What we don't have is the expertise to do a 2 state problem. Pumping cavitation is a very very complex problem.

Just because we have a set of chisels doesn't make us Michelangelo.

Dan

Dan

Dan,
Agreed. But I thought you were hiring "budding Michelangelos" out of college?:D This would be a case to ask Subaru Engineering how they model cavitation. Shame we don't know anyone "on the inside"
-Bill

Just about every race engine I've ever built I've had to slow the water pump pulley down on. Most factory water pumps are driven a little bit faster than the crank speed. Running an engine at 1000, 2000, or even 3000 rpms higher than the factory anticipated usually results in cavitation and overheating.

For a street engine I typically slow the water pump down by the same amount it is overdriven. In other words, if the pump spins faster than the crank by 10% (example: 6600 rpm pump speed @ 6000 rpm crank speed) I will reduce that speed to 90% of the crank speed (5400 rpm pump speed @ 6000 rpm crank speed). Nothing scientific about it, just trial and error on my part. For racing I will knock the overall pump speed down by 35-40%, but this can create overheating at slower speeds so it's not recommended for road use.

Hey, Bob, not to get too off topic in this thread, but do you happen to have the part number on that PS pump?;)
-Bill

OK. So, I'm going to get an underdrive pulley. Who here has experience with them? Is there any reason I should cough up 206 bills for the Unorthodox Racing Aluminum Billet vs. the $45 job in an eBay shop?

Used the Search button. no need to reply to previous post. Seems like the Unorthodx Racing is the way to go.

Dan,
You might want to see what LAN would charge you to machine one up. He's got the set up already from the Stage 3 SC work.
-Bill

1-2-Tree..

I got a $25 pulley off ebay (OBX) and have had it over a year. I see no reason that it isn't adequate. It is metal, round, has the grooves and works. Why pay so much for something else that cost more due to advertising? 6061 is 6061.

How does the underdrive pulley work for the water pump? The pump is driven off the cam belt, so teh only thing I could see would be to machine up a large sleeve that would press fit on the water pump pulley.

Has anybody got a picture of what you are talking about with respect to underdrive pulleys?

Thanks,
Matt

There is everything scientific about what you have described. If the pump speed is such that the outlet is restricting flow, cavitation will occur.

A crude method of prevention, is to drill a hole through each impeller blade so as to reduce efficiency and provide for increased RPM. This was a cure used with the old flat head Ford V8 engines. Unfortunately the SVX set up does not allow for a sufficiently oversize pump pulley, so that this could be an answer to the problem.

Dave(indirectly through SubbyRu89):

Thanks for the correction. Didn't realize the water pump was not driven off the main crank pulley. Don't expect there's any easy way to slow down the water pump. Anyboyd got any ideas?

Trevor mentioned the "old school" way of slowing it down. Cheap effective, but not exactly good practice. Reducing the efficiency of the pump may be our only route in this area

Tom

Well this is a thread of interest to me. I have been designing centrifugal pumps for about 30 years now. I had no idea this was a problem in car engines. Cavitation can drop the pressure on a centrifugal pump for sure. This usually happens at very high flow rates. Well it is just more critical at high flow rates.

The pump is variable speed as the speed goes up the pressure increases and the flow rate increases. The flow rate is determined by the point of intersection of the system curve and the pump curve.

For a fixed speed, starting at zero flow the pump normally has the highest discharge pressure. This is usually the maximum pressure at the fixed speed. As the flow thru pump increases the pressure drops for this fixed speed case. The flow rate thru the pump depends on the system resistance. The system resistance is how much pressure is required to push the volume of water thru the engine, hoses and radiator. This system curve requires more pressure as the flow rate increases. The system curve changes with the flow rate squared. The flow rate thru the pump is where the system curve and pump curve intersect.

The other part of this is that the pump requires NPSH, Net Positive Suction Head. This curve is plotted as NPSH versus the flow rate. As the flow increases the NPSH also increases. So yeah the flow can get high enough to run out of NPSH and then the pump produces no pressure.

This can be corrected by limiting the flow of the system. This involves increasing the system resistance to limit the flow. Installing a valve and throttling a valve will decrease the system flow rate and solve high flow cavitation problems.

Then there is the other problem. AIR Centrifugal pumps do not pump air and water. Air can stop a pump from pumping. The spinning impeller acts as a centrifuge. Any air in the water will get pulled to the center of the impeller at the inlet. The water goes around the outside of the air bubble. The air will stay there blocking the flow and a more air get pulled out it will add to the blockage. If there is enough air in the coolant system it will block the flow and the pump will stop pumping.

This is why it is absolutely important to get all the air out of the system after the coolant system has been refilled. My experience with the SVX has shown me that it takes at least two warm-up and cool down cycles of the SVX coolant system to get all the air out. The SVX engine has lots of places for air to get trapped. I know when the system is filled up “hard” with no air as this is the point were the coolant level stops changing when the engine cools off.

Take care of that SVX,

John

John,

Questions must be raised to ensure members receive exact information. Please do not take offence as a result of my queries. The issue probably comes down to the use of words.

For a fixed speed, starting at zero flow the pump normally has the highest discharge pressure. This is usually the maximum pressure at the fixed speed.

A rotary vane pump has the highest reliable discharge pressure at a maximum established (fixed) speed, and this is the point of reliable maximum flow. At any speed exceeding this speed, the rotor will run in advance of the flow and cause cavitation. One can think along the lines of a marine engine running at full power and turning a propeller which is too small.

The other part of this is that the pump requires NPSH, Net Positive Suction Head. This curve is plotted as NPSH versus the flow rate. As the flow increases the NPSH also increases.

You indicate that the pump requires a net positive suction head. I assume this means it must pump against a pressure at the outlet, above that which is statically present at the inlet and as a result the pump chamber must be full at start up. Primed, is the common term.

So yeah the flow can get high enough to run out of NPSH and then the pump produces no pressure.

An increase in flow requires an increase in pressure, i.e. the pump will be pumping against an ever increasing head in order to increase flow. How can there be a point of zero pressure? However there will be a level of pressure and flow which if exceeded, will result in the pump impeller inducing cavitation.

This can be corrected by limiting the flow of the system. This involves increasing the system resistance to limit the flow. Installing a valve and throttling a valve will decrease the system flow rate and solve high flow cavitation problems.

Throttling a closed loop system at any point, will increase input pressure as the affective means of reducing the upper flow rate. If the flow rate at a point of increased speed, falls below the capacity of the pump, the impeller will run in advance of the flow, resulting in serious cavitation. Any form of throttling will exacerbate the problem of cavitation within the pump at increased RPM.

The cure I have suggested, i.e. to reduce the efficiency of the pump, could allow for operation at higher speeds without cavitation.

Trevor
I won't attempt to answer all your questions but when thinking about the water pump on a car you need to remember that the pressure on the out let is also the pressure on the inlet less the friction loss. That is why on start up it needs to most power.
Tony

Trevor:

A few clarifications of terminology. As you mention, a centrifugal pump creates positive pressure on the outlet side and negative pressure on the inlet side. It is also adding kinetic energy to the fluid by accelerating it. As the fluid circulates through the system, friction dissipates the positive pressure induced by the pump.

The problem occurs when the fluid reaches the inlet of the pump. Theoretically, all of the positive pressure induced by the pump will have ben dissipated by the flow at that point. If the negative pressure induced by the pump inlet causes the absolute pressure of the fluid to fall below the vapor pressure of the fluid at that point, it boils. This pressure drop at the inlet is high, because the fluid is being accelerated from a low velocity before the pump inlet to a high velocity at the pump discharge. The higher the pump speed, the greater this velocity difference, and the greater the pressure drop through the pump inlet. That is the net positive suction head problem. The warmer the fluid, the higher the vapor pressure, the bigger the problem. It is similar to propellor cavitation, except that a propellor is not in a housing (a volute). Because of this housing the cavitation is mostly limited to the throat of the pump.

Increasing the pressure drop near the pump outlet is effective, because it increases the pressure drop through the system, thereby decreasing the fluid flow rate at any specific impellor speed, thus, decreasing the velocity at the pump inlet. The pressure drop at the pump inlet is thus decreased, increasing the absolute pressure of the fluid at that point and avoiding boiling.

Actually, the solution to our problem might be a Griswold flow control valve. This is a needle valve that is held open against the flow by a spring. Greater pressure on the upstream side pushes the needle into the orifice against the spring pressure, reducing the open area. They are designed to maintain nearly constant flow over a wide range of upstream pressure. We use them all the time in the HVAC industry to facilitate hydronic system balancing. I will look into this.

I agree
 

I agree with shotgunslade. A flow control valve is the most simple solution. Using a swirl pot and/or electric water pump would also help. I haven't seen any real good swirl pots out there. You're probably better off making one custom anyway to get the best fit and performance. A poorly designed one can make things worse. I'll try to post pics/diagrams later.
I prefer the Davies Craig electric water pump to others. It has a good impeller design and its temperature controller only runs the pump as fast as it needs to to keep the temperature you set. It can be used as a total replacement or as a booster.
You could get a nice aftermarket radiator to keep temps down. You could also go all out with stainless steel braided piping and AN fittings.

Thanks for a well written decisive explanation. Thankfully you are no doubt correct in your interpretation of the text.

I had the affect of kinetic energy as a result of the circulating medium in mind, but had discounted this in view of the wording of what was described. (I am immediately aware that there are those who will interpret my words as a means of bragging/excusing, that I knew it all anyway, but who cares?)

Your practical experience brings another possible means of correction into the picture, which certainly warrants investigation. Your further thoughts will prove valuable and will be awaited with sincere interest.

More good good stuff. :)

I was told many years ago, that a major problem at high RPM can be the pumping of the water at a pressure, whereby the speed of flow through the radiator becomes such that there is too short a time for the fluid to be cooled via the radiator. This makes sense, but I have never read anything along these lines.

Comments and or experienced advice regarding this, could be of value.

OMG, I just had a flashback to my junior year fluid mechanics course!:lol::lol:
-Bill

I would support your point Trevor about the "Water going to fast". From my experiance with refrigation and blast tunnels these things don't happen. In its simplest form a radiator is a means to relocate heat from the water to the air passing past or throught the cores. The factors that effect the performances of the unit are:-
Difference in Tempreture of the water and air, put a higher pressure radiator cap on you raise the tempreture of the water going throught the core that in turn means 1 cubic metre of air can take away more heat. Higher the difference the greater the amount of engery relocated to the air.
Volume of air passing the radiator at any given time and the amount of fin surface area it contacts.
Volume of water in the cooling loop. The point being that if the water passing through the engine is to slow or to hot it will not be able to relocate the heat as fast as the engine generates it.

In my opinion the SVX engine under extreme loand needs the return line to the pump redirected and the thermostate to drill a small hole. You need to keep the biggest possiable difference between the water into the engine and out of the engine possiable. The same as you need the biggest possiable into and out of the radiator.
Tony

I love fluid mechanics. It makes my day.

Trevor:
The issue of the water moving so fast thru the radiator that it has a lower temperature drop is correct. However, since the heat transferred from the water to the air is a function, on the water side, of both the temperature drop in the water and the mass flow of water through the radiator, the greater mass flow of water more than makes up for the lowered temperature drop. In general, with a fluid to fluid heat transfer device, the more fluid you have moving through the two sides, the more heat transfer occurs, even though greater mass flow results in a lower temperature change on each side of the heat exchanger. Higher flow rates also mean higher velocity through the radiator passages, which increase the heat transfer coefficient between the radiator tube walls and the coolant.

The only place higher flows will really hurt us is on the inlet of the pump, because it is both a constriction and because it is the farthest away from the pump outlet, thus has the the lowest static pressure in the system. It only hurts us there if the static pressure falls below the vapor pressure of the fluid.

I've been running Water Wetter and a 33% anti-freeze concentration. While the lower anti-freeze concentration may reduce the boiling point of the coolant, it also increases the specific heat of the coolant, so a given temperature change at a given mass flow rate results in a greater heat transfer. Water wettter reduces surface tension and increases the surface film heat transfer coefficent of the coolant.

Ahhhh, physics.:D:D

OK I'll ask again. How do you get all the air out?:confused:

Like he said. Fill the system. Run it through a heat cycle. I also recommend running the heater to help purge the air. After it heats up, allow it to cool. Make sure your reservoir is full. Do the same thing a second time. Then drive it keeping an eye on your reservoir.

Why reinvent the water pump if running higher pressure prevents the cavitation? If you are building a race car, do what rallybob does to lower the pump speed. Should be very easy to machine off some of the diameter on the water pump.

By the way, STi caps are about half the cost of the crucial caps and are very close in their pressure settings.

STi's have a different water pump impeller. They also have heat issues though, so I'm not sure that this cures the problem we have been discussing.

Question is, how is rallybob's car running now that he added a new cap? Still no problems? That would be the bar. If he isn't having problems, none of us will.

Bob had a great idea for filling the system. A large(tall) funnel in the cap of the radiator and a lot of patience to make sure all bubbles are out. Once the bubbles stop, leave the funnel in and start the car and let it get to running temp so the thermostat has a chance to open. Once no more coolant is allowed to get in, remove the funnel (a drip pan and quick hands are a must here) and replace the cap. top off the resevoir and allow the car to run with the cap on and take it for a 20 min drive. Bring it back in, let it cool completely. Remove the cap and add fluid as needed.

Tom

A caution here is that it may take a while (20 min?) for the engine to come up to temperature (thermostat opening) without the rad cap on.
-Bill

When I need to fill the empty/drained cooling system, I fill the radiator with coolant (just to the top of the fins), then I disconnect the top hose from the radiator and use the hose to fill the engine with water. Reconnect the hose, top off the radiator.

This fills most of system and it works out to be close to 50/50.

Then I run it with a funnel in the cap hole and top it off as a few bubbles come out.

We should be talking on this over a beer. :D

I understand exactly what you are saying and agree with you. Therefore the early theory put forward must have been wrongly attributed and the results must have come about through eliminating cavitation.

“Not Ahhhh, physics.” Quite the reverse, an interesting and worthwhile sorting of the issues involved. Thank you.

Cheers, I am in fact having that beer. Trevor. :lol: