The Water Pump

There are water pumps and there are water pumps – “yer pays yer money and yer takes yer choice!”

To quote the following from Barrie Jones, TD/TF Technical Specialist for the ‘T’ Register of the MGCC:

“The original water pumps had a shaft with a shoulder in the middle, so the bearings were positively located each side of this built-in spacer. The impeller position was set during manufacture and it could not be altered. The castle nut on the end only held the pulley on, so you could tighten it as much as you liked to hold the pulley secure.

Some replacement water pumps have a straight shaft onto which the bearings are pressed. There is a nyloc nut on the end which must be adjusted to ensure there is a .020” gap between the impeller and the pump body. This must be done off the car. If you over-tighten the nut, the impeller will be pulled into contact with the pump body, causing the pump to seize solid, but if you don’t tighten it enough, the pulley will be able to wobble on the shaft. Obviously, the pump must then be fitted to the vehicle with the pulley already attached to it. It gets worse – if you do this more than two or three times, the bearings become a sliding fit on the shaft, and then it is impossible to get the adjustment right, so basically the pump is scrap.”

The following article has been provided by Steve Turner of Racemettle. Steve responded very promptly to my request for details of his product and I hope that TTT 2 readers find the information useful.

XPAG/XPEG Water Pump Re-Design


In the late summer of 2006 the company that manufactured most of the classic car water pumps in the UK decided to close, probably due to the influx of cheap imports. At this time Racemettle had already re-designed and was manufacturing aluminium water pumps for the Triumph TR range of cars and had come to an agreement to buy up the drawings and patterns from the company. As part of this transaction, we obtained drawings, patterns and a box of castings for the MG ‘T’ type water pumps and this box of bits languished in a corner of our workshop almost forgotten.

During 2008 Mr Alan Atkins walked into our offices asking about one of our products he was interested in purchasing and as we talked, putting the world to rights, the poor quality of his now leaking water pump came up in conversation. Alan bemoaned the fact that whilst the pumps were cheap to buy, the effort involved in changing them every couple of years was disproportionate and he would rather pay more for a reliable pump. Alan also told us that when motoring in Europe, in the hot weather, his MG had a tendency to overheat when standing in traffic. We specialise in re-design of automotive components so suggested we could probably do a better job and produced one of the forgotten castings to talk over the issues; the seed was sown! Subsequently an old sample pump was provided by Alan and the work started.


First consideration in the pump re-design was availability of components. The bearing manufacturers produce special pump bearings that comprise a shaft and two bearings all in a sealed for life shell. This is very durable and easy to assemble but unlike a few years ago when they were available in almost any size you wanted, now the manufacturers will only make a limited range and then only if you purchase 500 or 1000 off.

The best scenario would have been to use the same bearing and seal as the TR pumps we were already manufacturing but unfortunately they were just too short to do the job. When we located a bearing with the correct length shaft it was too small in diameter to take the existing seal we were using. Nothing is ever simple!

We managed to persuade both the bearing and seal manufacturers to make us a sample so that we could re-design and test the pump. This was a major step forward and allowed us to design the casting internals and the impeller as a reasonably straight forward job.

It is important that the seal is under the correct amount of compression so careful consideration must be given to tolerances. In the past we had used monobloc seals that can be arranged to grip the shaft so the compression of the seal is obtained during fitting. However we were told the new two part seals supplied for this type of application required the correct compression from the impellor. This arrangement was less prone to failure due to its ability to tolerate some eccentricity in the shaft rotation to absorb vibration and bearing wear. We decided to use this arrangement. We had already designed a centrifugal impellor blade for the TR pumps which was effective and so this basic design was used again in modified form to suit the MG pump.

The only problem left was the pulley. It would not fit and the bearings were not available for a keyway pulley as they were designed to be used with a push fit pulley. Due to the eccentric design of the fan attachment it was considered that manufacturing new pulleys from scratch would be too expensive. When we looked around to purchase pulleys we could only find 5 in the UK available for sale so it was obvious we were going to have to re-use old pulleys whenever possible.

Fortunately existing pulleys could be re-bored to fit the new pump as a push fit unit.

Finally we had a design that improved the original in every area. A sealed for life modern purpose built bearing that pressed into the casting and was mechanically contained; a modern high speed two part seal placed under the correct compression with tolerance to bearing wear and vibration; a redesigned impellor that would increase the volume and pressure developed by the pump; a push fit pulley that was not prone to wear on the keyway or over or under tightening of the attachment.


First consideration was that we only had one seal and one bearing so we could not afford to make a mistake. Whilst we had lots of castings, the bore of them was too large to take the new bearing and the pattern would need modification. The cost of this modification and casting new bodies on an untested pump was a risk, so we decided for the development pump to sleeve one of the existing castings. Drawings were produced and arrangements made with our machine shop to modify a casting and manufacture a rig so that they could be pressure tested.

The first unit was very carefully assembled checking and double checking all dimensions. When we were satisfied that everything was assembled correctly it came to the moment of truth as the pump was attached to the test rig and 15psi was applied; no loss of pressure was apparent so it was pushed up to 20psi and left for 30 mins. resulting in a satisfactory pressure test.


Mr Alan Atkins had agreed to performance test the unit for us so after a phone call the unit was on its way to him. We expected Alan to just test the pump on his car but unbeknown to us he built a small, unsophisticated but effective test rig. By pumping water over a fixed amount of time Alan determined that the pump would produce about twice as much volume as the old pump. By fitting a pipe and holding it up in the air to see how high the pump could raise the water he also determined that the pump could produce over twice the head of the old design.

As with all centrifugal pumps, they produce only sufficient head to overcome the resistance imposed on them. The volume produced varies and is determined by the head developed. A centrifugal pump will normally be provided with a graph showing a curve that determines what volume they will pump against the head developed. This curve is affected by a number of other factors including temperature and NPSH or net positive suction head (amount of pressure at the inlet).

Probably available by now are complex computer simulations but for most of us the only practical way to determine this curve is to use empirical data. As the equipment and effort to produce this data is out of proportion to the application it was decided it was unnecessary and as Alan had shown, the new pump was capable of out performing the standard pump in all areas.

Only one slight problem occurred fitting the pump to the car. The car had been modified with a 5 speed gearbox and a steady bar had been fitted under one of the bolts that secured the pump. As the pulley was already fitted on to the shaft it was impossible to get the longer bolt that had been used under the pulley because of the decreased space. A quick call to the gearbox supplier confirmed that the steady could be reversed with the thinner end at the pump and consequently the bolt could now be inserted. So with the pump fitted Alan set off for the south of France. On his return, a tanned and more relaxed looking Alan called to see us and reported how well the pump had performed throughout his trip and how he had none of the problems previously experienced with overheating when standing in traffic.


The pump had proved itself to out perform the original design and probably just as importantly, it had been designed and manufactured with modern bearings and seals that should last for many more miles than the original pump with little or no maintenance.

We have only been manufacturing the MG pump for 4 years (and Rolls Royce pumps, that use the same bearings and seals for 3 years) but we have been manufacturing the TR pumps that are essentially the same for 15 to 20 years and we have not had one in service seal or bearing failure reported to us yet!

Every pump has a pulley pressed on and before despatch is pressure tested to ensure it is performing up to design specification. Purchasing one of these newly designed and better performing pumps should be the last time the customer ever has to replace or repair such an item.


We have just obtained a supply of standard pulleys and very shortly will be able to supply every pump with a newly manufactured pulley. Recently we had been asked by a customer for a pulley for use with an electric fan. Always eager to oblige we have designed and can now also supply the pumps fitted with a new aluminium pulley without the fan boss.

Steve Turner, Racemettle www.racemettleltd.co.uk April 2012

Ed’s note: Racemettle is a trading name of Racetorations. Some comments and photos follow.

I started this article with the saying “yer pays yer money………………..” Yes, you can buy a new water pump for a lot less than the cost of the pump offered by Racemettle and it may well give you excellent service if you are careful with the fitting, albeit this was not Alan Atkins’ experience, hence the moves to get a new pump designed and manufactured.

To put the quote from Barrie Jones into context it was in response to a number of complaints that had been referred to him as the MGCC ‘T’ Register Technical Specialist for the TD/TF models about reproduction XPAG water pumps and loose pulleys. As a Technical Specialist he was giving his opinion on what might be the cause of owners’ dissatisfaction with their purchase.

Another option to replace your water pump if it is giving you trouble is to send it to EP Services in Wolverhampton for overhaul. A friend of mine with an L2 sent his leaky pump to them and is pleased with the service offered. I intend to contact them to see if they are prepared to let me have an article for publication in a future issue.

Whilst on the subject of cooling, Barrie Jones, who has read the article before publication, has come up with a couple of suggestions for other cooling issues as follows:

“May I suggest that there is also an opportunity to re-design the thermostat (as fitted to the TB TC and TD) so that it used a modern wax-stat?

Also, the water outlet pipe on the TF is made of an alloy that seems to have a life of less than 2 years. Re-manufacturing these from a better metal and with a thicker flange would probably be popular”.

Photo 1 – the re-designed XPAG/XPEG water pump manufactured by Racemettle.
Photo 2 – Showing bearing arrangement

Photos 3 & 4 – two more views of the XPAG/XPEG water pump manufactured by Racemettle

2 thoughts on “T-TYPE OVERHEATING – PART 3

  1. Richard Tinkler says:

    I would recommend using Evans waterless engine coolant with any type of pump used on a TA. The advantages are very well worth the extra cost, which is really a saving in the long term. With a boiling point of 182 C, detonation (pinking) is completely eliminated, corrosion is eliminated and with zero pressure in the system, there is little possibility of leaks or core plugs blowing out.

  2. Tim Willis says:

    I have a 1953 MG TD, I was having high temperature problems, after checking everything out and finding no basic problems with the cooling system. The operating temperature was around 200F, I read the above article and contacted Steve Turner. As I am in Canada he did not say that the pump would solve my problem, but I thought that increasing the flow of coolant through the system would certainly help. After installing the pump the engine operates at about 175-180F, it runs very much better and at Idle after highway driving it only goes to 190-195F, lots of margin between the temperature and the boiling point as it is a non-pressure system.

Comments are closed.