header tank cap failure - datalogging results

[Peter Carmichael]

The header tank caps are renowned for failing. I took the precaution when I installed my AIM MXL of adding a channel to log the pressure in the cooling system so I could understand this better

 

Yesterday I suffered a header tank cap failure, which I can confirm is unrelated to head gasket issues. The datalogging shows up what is really going on. This might be old news to people with the PRT thermostat. I'm just commenting on what I've measured and recorded on the datalogging.

 

The pressure sensor is mounted in the bypass pipe (no heater). This means that it isn't directly measuring the pressure in the header tank and I happily concede that different parts of the cooling system will show different pressures.

 

So... here it is:

 

From a cold start, the pressure builds and is sensitive to engine revs. It typically gets to 0.6 bar prior to the thermostat opening. The pressure drops close to 0 bar and sometimes shows negative relative pressure after the thermostat opens. After the thermostat opens there is a long period of time where the total system pressure gently builds up (this is mostly dependent on the temperature of the air space in the expansion tank). The heat soak you get on a warming down lap (less airflow) seems to cause the greatest increase in system pressure.

 

From a hot start...

... the system is already heat soaked and is at considerable pressure.

... but the thermostat is closed because the bulk temperature is below the thermostat opening temperature

... the period between the engine starting and thermostat opening is extremely sensitive to engine revs causing pressure increases

... this is when my system pressure spiked to above 1 bar and caused the header tank cap to fail (engine revs of 4000 rpm)

 

This occurred after a short period of shutdown (~5 minutes) between two runs on track with different passengers. With a longer break between sessions, (~30mins) the system had cooled sufficiently that the retained pressure in the system did not get close to broaching the header tank cap.

[shn7]

So the moral of the story would appear to be, don't get it all hot, ease off, switch off and then start off again too soon.

 

I wonder what would happen if you'd left it running whilst waiting for the next passenger and let the cooling fan run

 

Steve.

Sussex (West) AR

Not forgetting Percy the Polar Bear

[Mick Day]

Peter: are you saying the cap failed & would not hold any pressure in the system or the cap did it's job & allowed pressure release at 1 bar?

[Peter Carmichael]

The cap blew off at 1 bar and thereafter was in a failed state, allowing coolant to exit the system. After refitting a known good cap and letting it cool there was no reoccurrence of coolant loss symptoms and the pressure behaviour had returned to "normal".

[CageyH]

The pressure drops close to 0 bar and sometimes shows negative relative pressure

 

I assume you mean relative to atmospheric pressure? (I just found the above unclear).

 

So, this is with a PRRT?

Is this a downside, compared to a normal thermostat onstallation?

I am also interested to know if this is a repeatable failure, or just a one off.

 

I have been thinking of installing a PRRT, hence the questions.

 

Only dead fish go with the flow....!

[Peter Carmichael]

This is with a standard thermostat and my use of the term "relative pressure" was to mean relative to atmospheric.

 

The behaviour is repeatable. I'm still trying to make sure my explanation is feasible given where I am measuring the pressure.

 

In the thermostat closed setup, the circuit is:

 

Pump -> cylinder head -> bypass pipe (where the pressure is measured) -> thermostat housing -> pump. The water rail -> radiator -> submarine pipe - header tank circuit is all static and that should mean it is at the same or similar pressure to the reading I am seeing at the sender.

 

The build up of pressure from cold is associated with revs (flow rate). The lowest pressure in the circuit should be the inlet to the pump. The only part of the circuit between the pressure measurement and the pump inlet is the thermostat housing. This suggests that the bypass through the thermostat housing is rather restrictive.

[Jason Plato]

or just buy a tank with a conventional and reliible cap ?

 

when you say its unrelated to head gasket issues - is the initial failier of the cap unrelated or once the cap has failed the system is no longer pressurised and hence localised boiling occurs and then the secondary effect is that the firing ring gasket is comprimised ?

 

here is C7 TOP

Taffia rear gunner

 

[Rob Walker]

Peter,

 

The large diameter Freelander caps are supposed to be more reliable. I have had several stock Rover caps fail. You don`t need datalogging to spot a duff cap, the coolant will expand and when you shut down the engine put your ear near the expansion bottle and you will hear the gas escaping. Someone on Blatchat posted a neat idea, drill out the stock cap and install a tyre valve, the system can then be easily pressure tested with a foot pump and gauge.

 

Rob

[elie boone]

Or use a cap from a Volvo or a Saab

[Mr Locust]

Are you saying that the cap has a poor hysteresis characteristic in terms of relief and reseat behaviour or that it is a one shot 'pressure fuse'? ie. once the cap has blown is it goosed?

 

What you have also illustrated is that maybe the cooling fan should be wired to allow fan operation after the engine has been shut down to help with hot shutdown/afterboil issues causing a high system pressure before the engine is restarted.

 

What you have demonstrated is that in this particular case, there is merit in the old fashioned practice of warming the engine gently until the thermostat is open and you have reasonable oil temperature.

 

How about this for an alternative approach: If the system needs protection an alternative to a PRT might be a temperature based rev limit to minimise pressure spikes. Whether your ECU can do this easily is another matter.............

 

Ian

 

Green and Silver Roadsport 😬 and Mrs Locust's Blue Roadsport Academy 😬

[Neal Worth]

I know that it's adding weight and possibly unnecessary complications, but if the problem only arises after a hot start, would fitting an auxiliary electric pump be any benefit?

Reasoning being, once the engine is stationary the coolant no longer flows, if the fan runs after switch off it will only be of limited benefit as a cooling aid. By fitting a small electric pump, coolant flow is maintained and there would, hopefully, be a decrease in the heat soak and less of a chance that there will be a pressure spike.

The auxiliary pump could be run by a controller very similar to a turbo timer. That way the length of time it would run for could be adjusted for the best results.

 

Would this work?

 

Is this a problem just with the K series? Or are other engines susceptible also?

 

 

[Ivaan]

I drilled a small hole at the top of my thermostat, to prevent air locks when refilling the system ( this does slow down getting the engine up to temperature ).

Would this help to prevent any pressure surges, when the stat is closed, protecting the cap?

 

Clive

[Peter Carmichael]

Lots of questions here. I'll try and keep up.

 

You don`t need datalogging to spot a duff cap, the coolant will expand and when you shut down the engine put your ear near the expansion bottle and you will hear the gas escaping.

 

I've used this method to diagnose failed head gaskets. A failed cap with a good head gasket results in no pressurisation and no hiss at shutdown (unless the engine is significantly heat soaked). I was able to open the cap (having verified zero pressure reading on the gauge) and there was no rise in coolant level in the tank and no escaping gas.

 

Someone on Blatchat posted a neat idea, drill out the stock cap and install a tyre valve, the system can then be easily pressure tested with a foot pump and gauge.

 

I'm contemplating something like this to set a pressure "preload" in the system. This then gets around the issue of putting a new cap on a hot system and not having any pressurisation.

 

or just buy a tank with a conventional and reliible cap ?

 

The pressure characteristic will still be there. Now I've got a way of monitoring the pressure I'm interested in understanding it. I won't just be plonking a new cap/tank into the system and turning my back.

 

when you say its unrelated to head gasket issues - is the initial failier of the cap unrelated or once the cap has failed the system is no longer pressurised and hence localised boiling occurs and then the secondary effect is that the firing ring gasket is comprimised ?

 

I think the initial failure is unrelated, but engine damage follows from running with an unpressurized cooling system.

 

Are you saying that the cap has a poor hysteresis characteristic in terms of relief and reseat behaviour or that it is a one shot 'pressure fuse'? ie. once the cap has blown is it goosed?

 

I've seen quite a few caps that have "gone bad" and replacing the cap cures the immediate symptoms. Unfortunately many of these engines go on to suffer head gasket failure.

 

What you have also illustrated is that maybe the cooling fan should be wired to allow fan operation after the engine has been shut down to help with hot shutdown/afterboil issues causing a high system pressure before the engine is restarted.

 

I'm still at the stage of trying to understand the cause. There are lots of things that can be done, but I am not interested in panic measures that just try and cool hot things down. The system relies on an airspace in a header tank providing pressurisation. I'm still looking at the data, but it looks like road speed is a significant factor in loss of pressurisation - this could be that the siting of the header tank results in it being actively cooled by airflow under the bonnet. The answer may be to re-site the header tank. I'm going to keep looking for ways to characterise and understand what the system is doing.

 

What you have demonstrated is that in this particular case, there is merit in the old fashioned practice of warming the engine gently until the thermostat is open and you have reasonable oil temperature.

 

I think this is valid advice.

 

How about this for an alternative approach: If the system needs protection an alternative to a PRT might be a temperature based rev limit to minimise pressure spikes. Whether your ECU can do this easily is another matter.............

 

I think it should be possible to design a cooling system that doesn't suffer from this effect.

 

Is this a problem just with the K series? Or are other engines susceptible also?

 

I think the problem is specific to the K-series. Certainly the K-series seems to be less able to cope with problems in the cooling system than many other engines.

 

I know that it's adding weight and possibly unnecessary complications, but if the problem only arises after a hot start, would fitting an auxiliary electric pump be any benefit?

 

I've previously considered that a turbo-timer type drive of an auxiliary pump might help avoid hotspots in the engine on shutdown, but I think this is a different characteristic. The auxiliary pump wouldn't remove very much heat from the engine bay and it seems likely that the header tank would heat soak just the same.

 

I drilled a small hole at the top of my thermostat, to prevent air locks when refilling the system ( this does slow down getting the engine up to temperature ).

Would this help to prevent any pressure surges, when the stat is closed, protecting the cap?

---

 

I don't think this will help. The only part of the coolant circuit between the point I was measuring pressure and the low pressure side of the pump is the thermostat housing. That is a bigger hole than any reasonable hole that might be drilled in the thermostat itself.

 

I'm still checking the data. There is a chance that vehicle speed played more of a part than I was giving credit to. A simple test will be for me to apply temperature lagging to the expansion tank. I'll be trying that soon enough and I'll report back.

 

 

 

 

 

 

 

 

[Bricol]

I'm feeling a little left out here. Four, sorry, five, Rover engined motorcars, and not one of them has suffered from a header tank cap failure . . . daren't say the other, but none of them has suffered that either . . . Debs Polo did require a head gasket change tho'.

 

😬

 

Bri

[Colin Mill]

This is quite interesting. I have for a while been contemplating changing the header tank arrangements as I think the existing header is rather hit and miss. I don't like the fact that the pressurisation of the system is delayed until the header tank contents comes up to temperature - the air space normally left in the top of the header means that very little pressurisation comes from the expansion of the water and significant pressurisation comes only from the vapour pressure over the hot liquid in the header tank. I wonder, for example how much the airflow through the engine compartment cools the header especially as there is little or no through flow of coolant in the header as the ball-valve at the head end of the pipe inhibits this unless there is gas in the head. The header seems to be heated by convection from the submarine section which is one of the coldest bits of the system *confused*

 

What I plan to do is get a swirl pot like the one on this page but with the top fitted with a Stant cap like this.. I will then need to arrange an unpressurised overflow bottle (possibly the existing header tank with the top hose fitting taken to a vent pipe to the bottom of the engine compartment).

 

I think this should allow the system to come up to pressure much quicker - with no air space under the cap it will tend to pressurise by water expansion.

 

Edited by - Colin Mill on 27 May 2008 22:40:01

[Johnty Lyons]

I don't like the fact that the pressurisation of the system is delayed until the header tank contents comes up to temperature - the air space normally left in the top of the header means that very little pressurisation comes from the expansion of the water and significant pressurisation comes only from the vapour pressure over the hot liquid in the header tank. I wonder, for example how much the airflow through the engine compartment cools the header especially as there is little or no through flow of coolant in the header as the ball-valve at the head end of the pipe inhibits this unless there is gas in the head.

Ah but a DVA mod which i did years ago is to remove the ball valve completely and allows a constant bleed of near boiling water into the Header tank I'm running a 240 BHP K and in 11 yrs have never suffered a pressure cap failure despite lots of heat soak hot restarts on Track days

 

---

jj

N.I. L7C AR 🙆🏻

Membership No.3927.

240BHP 1900cc K Series 40th Anniversary

 

[Colin Mill]

Yes, I'm sure that is a good modification and one I might do anyway as I think an airlock could get trapped in the head if that valve sticks shut. I'm not sure why Rover considered the valve necessary - perhaps it makes sense in the installation in Rover cars.

 

At the moment I top the header tank up so the air space is so small that the expansion of the water in warming up puts this under significant compression.

 

My reason for going for a Stant cap is that it is a known quantity with a known operating point and a good reliability record. It can also easily be visually checked for defects (perished seal rubber or broken spring). The Rover part may be OK but how can you tell if it's gone duff?

 

Edited by - Colin Mill on 27 May 2008 22:38:40

[Peter Carmichael]

Colin, I'm interested in your thinking for modifications as I've been planning something similar.

 

I'd like the air gap to be above a point that sees the total flow i.e. radiator circuit and bypass flow. That swirl pot you link to could achieve this. If you're planning on a Stant type cap, your thinking is very similar to my own.

 

 

[Colin Mill]

A good point. After Johnty suggested the valve removal modification I was thinking that perhaps this would bring the temperature of the swirl pot up quickly if the pipe from the top of the head was plumbed into the pot. However, we could blank off the existing bypass outlet and have a rearward pointing 16mm outlet on the swirl pot from which the bypass could be taken so that the full flow went via the pot.

 

I would like to keep the top of the pot as high as possible so I'm going to eye up what the nose-cone will allow.

 

As far as I can see the tapping on the submarine section can be blanked off once we fit a swirl pot as the existing header gets relegated to an unpressurised expansion bottle. I was wondering if the expansion bottle should be a heavy PVC bladder (as per 1960's Fiat windscreen washer reservoir) as this would be light and have no air inside to dissolve in the reserve coolant.

 

Edited by - Colin Mill on 28 May 2008 08:37:06

[Johnty Lyons]

Colin is there not a danger with the proposed new bypass routing that with a dead end at the rear of the head the ECU temp sensor might need relocated as its position at the end of the club foot is no longer in the flow and there will be quite a lag in temp rise that the ECU sees 🤔 🤔

 

---

jj

N.I. L7C AR 🙆🏻

Membership No.3927.

240BHP 1900cc K Series 40th Anniversary

 

[Colin Mill]

Hi Johnty

 

On mine I think the temperature sensor is in the main outflow from the head rather than in the bypass take-off tube but I'll have to double check.

[Johnty Lyons]

Yep thereare two the one nearest the head feeds the Guage [singlewire] the one in the output stub [two wire] goes to the ECU But then again the club foot and the temp sensors has changed over the years. Perhaps I just ned to upgrade to the latest K spec CF

Peter any comments??

 

---

jj

N.I. L7C AR 🙆🏻

Membership No.3927.

240BHP 1900cc K Series 40th Anniversary

 

[Peter Carmichael]

I run my temp sensors out of a submarine pipe in the bypass hose. In fact, I run one temp sensor (ECU) and one pressure sensor (AIM MXL dash). The ECU then sends its temperature information to the dash over the CAN link.

[Colin Mill]

Shall I do a bit of a sketch and pop it on my website so we can kick the idea about till we have something that looks right? Perhaps then we could get a few made for those interested in trying it.

[Bricol]

I'd be a little worried about reducing too much, or removing the air gap in the expansion bottle.

 

Heat water up and it expands - water isn't particularly compressible, so the expanded water has to go somewhere - and on the old fashioned systems, that was up past the rad cap and down the tube into the collection bottle somewhere near it. When it cools back down, it takes up a reduced volume again, and atmospheric pressure drives the displaced water back up the tube and into the system - assuming there was enough water in the bottle to do so. The only thing assisting raising the boiling point of the coolant was the spring in the cap.

 

A modern sealed system has the air gap in the expansion bottle to allow the expanded water to go somewhere, as air is nicely compressible. Water in the expansion bottle can be cold, it's the heated, expanded water in the rest of the system that pressurises the system, not heated air vapour. The air acts as the spring, allowing the system pressure to remain within design limits, assuming the bottle cap seals. Of course, as the air heats up, there will be a slight increase in pressure.

 

No air in the system, and the pressure will go a lot higher, releasing more coolant from the system as the cap vents/fails. Then there is the addition strains imposed on hose joints, radiators (expecially crimped on header tanks), water pump seals and hoses themselves.

 

Within the machinary I used to design, you would see pressure rises in small, heated systems. Open a valve and bring in more, cooler solution, and the pressure would drop - two main, simplified reasons. The heated, expanded, liquid now had more room in the additional piping/tanks to expand into. And the new, cooler solution, chilled the original circuit, causing it to contract and reduce pressure. Bit like a stat opening on an engine cooling circuit.

 

Bri