SVX Engine cooling "Again & Again"

[Dessertrunner]

This artical provides some info as well.
Tony

[Trevor]

Tony, you are trying to catch me out.

The issue I covered was your suggestion that the Subaru figures are incorrect. The figures obviously are meant to apply to a pump free of any head.

If the flow rate increases against a constant head, as a result of increased RPM, there will not be an increase in pressure, unless the head is increased, or a restriction is introduced. Therefore any restriction has a direct relationship with pressure.

You are saying that if the flow rate is increased (as a result of increased RPM), the head/restriction becomes more relevant, i.e. a bigger issue. This can be construed but it has nothing to do with the original subject, which was pump specification.

You were stating that the Subaru figures covering the pump are wrong. Their figures will not refer to the pump operating against a head, i.e. as per the Reliance Electric graph 1, which in fact proves my original point.

Graph 2 covers pressure and does in a way, relate to your suggestion regarding the importance of any restriction. However the graph shows this is may very well be less than you had reasoned. The slight curve shows that pressure increases at a rate dependent on speed, presumably with a constant head creating the pressure.

The pump curve you have shown, covers flow rate as against a variable head together with power consumption and has no real significance, unless one loosely relates power to pressure.

Back to square one, left head overheating, due to a restricted junction. The restriction is shown to be of less importance at high RPM, due to steep angle of graph 2. No problems at intermediate RPM. Why is not a small degree of restriction in the right leg, a valid remedy?

[oab_au]

Just to expand on my last post. The problem area is in the way the two outlet pipes from the engine join to flow to the radiator. As the pump speed increases, the flow out and the suction in, both increase.
With the standard system only the right bank has the suction acting on the flow so the right bank has the highest amount of the total flow. The left bank has very little suction, so it slows and overheats.



If the outlet pipe is modified to allow the left bank to flow independent of the right, it will have the pumps suction acting on its flow also. Although they still have different lengths, they should still both flow the same.



This will remove the restriction that left bank has, in trying to overcome the flow from the right bank. How the modification is done, depends on what clearance there is under the manifold.

Harvey.

[Dessertrunner]

Looks cool Harvey, I totaly agree with you. One question I am thinking over, do we acheive the same thing as you suggest if the top suction is taken from the centre of the top pipe as I positioned the ruler in the other image. Not much room under there.
You diagram is exactly as the UK rally guys do it so you are on the right track.
Tony

[Trevor]

The facts. ----

A centrifugal pump does not have the ability to develop significant negative pressure, “suction.” No negative pressure “suction,” exists within the system. The complete circuit is under the same pressure, held and limited by the radiator cap.

Pressure must be and is, virtually equalised throughout the system. Only the rate of flow can vary according to resistance within the circuit. Both left and right and left legs operate at the same positive pressure and can not become independent in this respect. Experience suggests that the greater flow resistance is in the left leg.

It would appear that the resistance occurs at the junction within the cross water pipe and the left is the most affected. It is not a matter of one flow trying to overcome the other.

Provided the revised junction overcomes the present unequal resistance, the modification could work, but most certainly not along the lines suggested. However where is the pipe work to be located, and fabrication of a convoluted structure, presents real difficulties.

In the event that resistance is unequal in some other section of the circuit, and this has not been ruled out, the suggested modification will not work under any circumstance.

All manner of hypothetical arrangements can be put forward, but the practical problem is space. If shifting the alternator and whatever is an option, whoopee do, let’s have a ball.

Balancing the circuit by restricting flow slightly in one leg, is a practical answer in every respect, and as an an experiment will cost bugger all, other than time. If time/money is not a problem, a flat but wide sheet metal fabrication, of equal left/fight section could possibly be made to fit. The sky is the limit.

[SilverSpear]

Trev and Harv, both of you are right.

Harv your diagram is the better solution to the problem on hand. Here is what must be done in this respect:



And I think we can TRY to do that with MODELING CLAY. We can just fabricate this route with modeling clay below the manifold and by trial and error check if we have a clear path for a metal fabrication. If not then I suggest Trevor's method by restricting flow on the right side a little bit to solve that junction problem.

[Trevor]

=Dessertrunner

I have cut my manifold as you can see from the image the next plan if that doesn't fix the problem will be to put a new pipe inthe centre were the ruler is.
Tony

If credits are to be issued, they should be directed where they are due. Harvey did not suggest a centre take off. This much better idea was proposed earlier by Tony, as above.

[SilverSpear]

Trev, either Tony, Harv or Sov13t, I really do not care as long as we get the job done. We are all here to discuss this flaw on our engines and resolve these flaws. I do not like to argue with people on this board over stupid stuff...

I really see it childish when people put in their signatures:

First bla bla bla modification to be done
First XXX whatchamacallit used on an EG33
First BS & BS & BS....

So what? He thinks he is Einstein now? What shall I put in my signature?
First complete bodykit on an SVX? Even though people see it as ugly (and now I do myself, that was 6 years ago, I was young and stupid) it will still be the first complete bodykit... but I don't care!

There is no doubt that YT, yourself, Tony and Harv are all extremely knowledgeable people, please keep yourselves proud on a professional level and do not slip youselves on bull sh!t...

[Dessertrunner]

With the graze of god and a lot of luck we may have one of these moded pipes on a motor that is in the racing buggy tomorrow. So with any luck we should know if it is a total waste of time or an major improvment.
By the way in case I forgot to mention if your AC is off it is impossiable to run fans flat out. They will only run slow and medium.

[shotgunslade]

I think we are on the cusp of solving this problem and ought to stop throwing darts at one another about certain details and about specific authorship. We are sort of a team here, even if somewhat prickly. Here is what we know about the problem.

1. It isn't about slow steady increasing coolant temperature registered on the gauge which would suggest a simple inadequacy of the radiator to reject the heat from the engine. Instead it is characterized by a a sudden spike in temperature that suggests a steam bubble passing the sensor. This symptom suggests localized boiling somewhere in the coolant system. Note that if the problem were simply that overall water flow was inadequate or heat transfer surface of the radiator was inadequate we would get a steady slow rise in coolant temperature, rather than a quick spike that would just as quickly subside. Note also that this phenomenon is often accompanied by burping water into the overflow tank, further suggesting steam generation in the system.
2. Localized boiling could arise from two sources, a. cavitation at the entrance to the water pump, or b. some portion of the block that locally has inadequate flow so that it overheats causing local boiling.
4. I was an early proponent of cavitation, but I hadn't thought it through. The coolant system on an automobile is not a truly open system. It is a semi-closed system. The radiator cap causes it to operate at a pressure of approximately 2 bar rather than 1 bar as on a completely open system. Therefore, if the coolant temperature at the inlet to the water pump is approximately 100C, then the negative pressure at the pump inlet would have to be 1 bar relative to system pressure in order for localized boiling to occur. That is very unlikely.
5. Jack Lavety's use of an electric water pump doesn't solve the problem. Given that it would have a different inlet configuration than the stock water pump suggests that the problem isn't cavitation.
6. The experience from OZ suggests that the left side of the block is the overheating culprit. YT's observation of the unexpectedly small size of the opening in the crossover pipe exposes the likely cause of that overheating because of reduced flow across the left side of the block.
7. Trevor is correct that the easiest way to balance current flow in the 2 legs of a parallel circuit (it's easier to think about this using the electrical analogy) is to add resistance to the leg with the smaller resistance, however that will add to the overall resistance of the circuit, requiring greater voltage (pressure from the water pump) to get the same amount of current (coolant flow). A better course is to reduce the resistance on the leg with the larger resistance to the extent easily achievable and then afterwards, increase resistance on the other leg to balance current flow.
8. Remember that the current (water flow) balance doesn't have to be perfect, only that the flow through the high resistance leg (left cylinder bank) is adequate to avoid local wire meltdown (coolant boiling).

So let's cheer for those who are pursuing decreasing the resistance on the left cylinder bank (Dessertrunner and YT) and wait to see their results. In the end they may need add some flow resistance to the right bank or maybe not.

Finally, lets all congratulate ourselves for getting to the root of the problem. Remember at the end of the day, no one will remember exactly who contributed exactly what detail to the final solution. Furthermore, no idea from a single author will make it thru the process unchanged. The solution will be a group effort and the basis of congratulation should for supporting the overall effort of the group.

By the way, Trevor, we need to have a discussion about the differences between positive displacement and centrifugal pumps. This is what I do for a living and you somewhat mischaracterized the suction characteristics of centrifugals.

[RallyBob]

Originally Posted by shotgunslade

5. Jack Lavety's use of an electric water pump doesn't solve the problem. Given that it would have a different inlet configuration than the stock water pump suggests that the problem isn't cavitation.

Agreed. But it did help a bit. He went from overheating 4-5 laps into a race to eventually (after changing the radiator, higher psi cap, electric pump, and surge tank) overheating 17-18 laps into a 22 lap race. So progress was made, although the problem is far from solved. It probably doesn't help one bit that during the progression of cooling system changes he has continuously made power advances and rpm increases, which certainly makes isolating cause and effect difficult.

He has not run the latest configuration changes at the track, which includes the machining work at the front of the block (water pump inlet).

The best part seems to be that now the heat rises far more slowly, rather than instantaneously. And dropping the rpm shift points from 8200 to 6500 or so lets the engine cool off now, compared to before when a 6500 rpm peak was enough to boil over within 5 laps!

Anyway, I think that with all the information being thrown about here between Tony, Trevor, Harvey, Tom, etc....the cause and perhaps the solution is not too far away. I'm just sitting back and observing for now.

[oab_au]

Originally Posted by Dessertrunner

For the recod the big book says 100lpm @ 3,000rpm and 200lpm @ 6,000rpm.

I have cut my manifold as you can see from the image the next plan if that doesn't fix the problem will be to put a new pipe inthe centre were the ruler is. That pipe will be 40mm as opposed to 30mm it is now. Next change the water pump inlet pipe from 30mm to 45mm if possiable. Also both inlet and outlet on the radiator changed to suit

I feel sure the inlet and out lets of the motor are both 30mm will cause a problem. Centrifugal pumps always have a larger inlet then outlet to prevent caviation.
Great the way everyone is jumping on board I think if a few of us run tests we will beat this problem finally. Have a great day or night.
Tony

Yes Tony the suggested center tap will do the same thing as the Y join, if it will fit in.

We have to keep in mind of what we are doing here. We are trying to cure a problem that is caused by another problem.
If you are stuck with using the original pump, then any of the rerouting, or restricting the right bank will help to overcome the problem, it really depends on how much work you want to do.

If in the case of a race engine, then the root cause has to be fixed. That is still too large a flow, flowing too fast. An electric pump will remove the problem of too fast a pump speed, but not if you use a BIG pump that pumps too much too fast, you are back to the same problem.

This is a case of " the bigger number, is not the best". Water is a bad conductor of heat, it takes time to heat it, and time to cool it. If the water flows too quick it will not have time to remove its maximum amount of heat from the engine block, and will not have time to loose it to the air through the radiator metal.
The boundary layer will heat and cool, but the main flow in the middle of the stream will charge straight through, unchanged.

The engine was designed to operate at 3000 to 4000 rpm, all day long. At this speed the pumps flow is about 120 lpm, this is probable the flow rate that provides the best transfer of heat. We really should be trying to replicate this flow, to restore the system to the ideal flow regardless of the sustained engine speed.

I think the best we can do is to do the Y or center tap change, and use an electric pump of about 120 lpm.

Harvey.

[Dessertrunner]

I think there are a number of "DON"T KNOWS" which is also a concerns and a number of "DON'T UNDERSTANDS". For example:-
:- What is the real water flow rate at the different revs, friction in water is the square of which in simple terms means double the flow from 100lpm to 200lpm and the amount of friction goes up by 4 times. This means the opening up of the coolent flow pasages becomes more important such as Rally Bob suggested. We need to run some trial to figure this out.
:- If there is restrictions were are they and how do we find and fix?
:- Why does Subaru on most of there higher performance engines use oil coolers, correct me if I am wrong Rally Bob but I thought you guys installed one on the engine you built.
:- How we could orderly develop a test plan to conduct orderly trials to find the soluation.

Maybe the best way forward is to develop a list of things we need to know to fix the problem so as we are not clutching at straws.

Here is my first suggestion.
1 What are the coolent flow rates at different revs.
2 What is the tempreture the oil is running at and what should it be.

Tony

[Trevor]

Tony, good thinking.

Restrictions:- These can be identified by eye and using common sense.

Oil Coolers:- An advantage on any engine often running or expected to run, at or close to maximum power. The lubricating oil constitutes a direct method of reducing temperature in the bearings and the bottom end in general. Normally the sump/pan serves this purpose.

1&2. The actual figures are not important, but could br useful guide lines. The objective will be to simply make improvements, whatever.

Cheers, (Home brew in hand at this moment.) Trevor.

[Trevor]

Originally Posted by oab_au

Yes Tony the suggested center tap will do the same thing as the Y join, if it will fit in.

We have to keep in mind of what we are doing here. We are trying to cure a problem that is caused by another problem.
If you are stuck with using the original pump, then any of the rerouting, or restricting the right bank will help to overcome the problem, it really depends on how much work you want to do.

If in the case of a race engine, then the root cause has to be fixed. That is still too large a flow, flowing too fast. An electric pump will remove the problem of too fast a pump speed, but not if you use a BIG pump that pumps too much too fast, you are back to the same problem.

This is a case of " the bigger number, is not the best". Water is a bad conductor of heat, it takes time to heat it, and time to cool it. If the water flows too quick it will not have time to remove its maximum amount of heat from the engine block, and will not have time to loose it to the air through the radiator metal.
The boundary layer will heat and cool, but the main flow in the middle of the stream will charge straight through, unchanged.

The engine was designed to operate at 3000 to 4000 rpm, all day long. At this speed the pumps flow is about 120 lpm, this is probable the flow rate that provides the best transfer of heat. We really should be trying to replicate this flow, to restore the system to the ideal flow regardless of the sustained engine speed.

I think the best we can do is to do the Y or center tap change, and use an electric pump of about 120 lpm.

Harvey.

I have been pushing the speed of the coolant flow for quite a long time, over years in fact, with negative response. It is good that Harvey and I agree on this, and more so emphasis this as a being valid for early consideration.

There is no necessity for changing the pump with all the subsequent alterations. Reducing the efficiency will do the trick and is a well proven technique. It will be pointed out that this will also effect flow at all levels of engine speed, but at lower RPM the loading and heat delivery is less intense, and has not in other applications proven a problem.

Speed of flow is now identified as possible problem, therefore some restriction in the right leg will assist rather than detract. Direct cost is nil.

Drilling or reducing pump the rotor is easy and the cost can not exceed the value of a used pump.

P.S. Flow can also be reduced by fitting a restrictor within the thermostat housing. This is worth a try as a first experiment, as so little is involved. However the pump mod. provides a reduced performance somewhat in line with RPM.

As an experiment the result of an increased flow can be ascertained by doing without the thermostat. Some have done this and wondered why a negative result was the outcome, and this incidentally proves the point regarding water possibly flowing too fast.

The combined labour cost, is nowhere near any of the other suggested methods. This combination should be top of the list, even if considered as only an experiment. However it is very important that each alteration should be done and tested separately for individual evaluation, in line with correct experimental practice.