Click to Print

Technical Advices > Cooling System

Cooling System

6.1 General Considerations

In any high performance, internal combustion engine one of the principle issues is managing the heat produced by the internal combustion process. This heat can be your friend, however it can also be a major contributor to three very destructive processes: pre-ignition, detonation and alloy temper degradation.

Even very efficient engines are rarely more than 35 % efficient in converting heat to motive energy. The other 65 % is lost either out the exhaust pipe, or has to be dissipated through the cylinder head, engine block and cooling system.

The circulation of coolant through both the cylinder head and engine block cooling passages provides the mechanism for transferring this heat to the radiator system. The cylinder head accounts for the majority of the heat generation and transfer to the coolant, namely 65%, whereas the engine block makes up the remaining 35 per cent. While these are numbers are estimates on my part, they are supported by the various temperature loads that are generated in the head and block respectively and will suffice for illustration purposes.

Within the cylinder head the “hot spots” are the areas directly around the exhaust valve seats and the area around the spark plug seats. These are also the areas that are likely to become super-heated. This super-heating is likely to result in the creation of super heated steam pockets. These steam pockets have three potential negative effects.

  • If the steam remains trapped, then it will act as an insulator and this area of the cylinder head or engine block where the steam is located will prevent the transfer of heat to the coolant. This condition is “regenerative”, and the steam pocket will eventually enlarge, and an ever-increasing overheating situation will occur.

  • Once this steam reaches the water pump then it is possible for the entire cooling system to become ineffective, as modern centrifugal type pumps are not designed to pump steam, only liquid. In systems where the coolant temperature is very high, it is also possible for the water pump itself to boil the coolant on the suction side of the pump. As the pump rotates the “system” pressure is lowered, and thus if the coolant is sufficiently heated, the liquid may boil of its own accord and cause pump cavitation. This would lead to a catastrophic failure. Note: This “pressure drop” situation occurs when you open the radiator cap on a hot motor. Instant pressure drop equals instant boiling.

  • Aluminum head castings are usually heat treated to T6 hardness specifications. This means that the head casting will have been heated to 1000F initially, then water quenched for a short period of time, and finally left in a heat soak oven at 320F, for a period of approx. 5 hours, and finally allowed to cool naturally to ambient temperature. They key is to make sure that the aluminum head does not exceed this 320F temperature in normal operation, as it will lose it temper. Once this happens then it will be very difficult to maintain proper head gasket clamping forces. In addition pressed-in valve guides and seats, particularly for the exhaust valve, will be compromised.

So the idea is to keep a sense of equilibrium within the engine cooling system, noting that it would be best to keep the average cylinder head temperature well below 300F.

Before going any further, let put to bed a myth (I will admit that I fell victim to this early in my engine work).

If the coolant flows through the radiator too quickly, it cannot transfer the heat to the radiator.

At a fluid velocity of “X” the coolant will release a given amount of heat to the radiator. If we double the velocity of the flow, it will release less heat to the radiator obviously because the coolant is in contact with the radiator for half the original period of time. Makes sense so far, right??

However, since the water now makes two revolutions in the same period of time (remember we doubles the velocity of the flow), the net result is the same. So thermodynamic law says that in the same period of time the same amount of heat is released. This part of the myth is definitely busted

One reason this rumor persists is that the 2 most often cures actually work sometimes. The first "cure" is to slow the coolant pump rotation with a larger pulley. The problem that is actually solved, in that many stock pumps will cavitate at even moderate RPM. A cavitating pump is very inefficient at best and may stop working altogether, at worst. Therefore slowing the pump the pump actually becomes more efficient. The second part of the "cure", restricting the outlet of the engine (supposedly slowing the flow), actually causes a pressure buildup behind the restrictor thus pressurizes the block by a few pounds more that the overall system pressure. Therefore, the boiling point of the coolant is raised by a small amount. This works to preventing local boiling in stagnant flow areas.

Remember that steam bubbles can slow, or stop, coolant flow through small passages. The idea of a restrictor is actually a worthwhile implementation in a competition motor. The increased pressure also creates some back-pressure and this, in turn, reduces the onset of cavitations.   A more optimized solution would be to run the coolant pump at 50% of engine RPM and to place a restrictor (in a production car this would be a thermostat) wherever the coolant flow exits the block. By not overly slowing down the coolant flow we can introduce a level of “turbulence”. This will have, as a side benefit, the ability to sweep some of the stagnant areas of any steam accumulation. So I guess that this part of the myth is at least “partially” true.   Stewart Racing pumps actually built a coolant pump dynamometer and has demonstrated these effects. There is also a significant body of literature, within the Society of Automotive Engineers, addressing this very problem.     I know that some of you reading this are probably saying, “OK, I understand all of this, but how does all of this get me more performance?” Stay with me a little longer, as it will all become clearer.  

Next, we need to consider both heat generation and heat release mechanisms (engine and radiator). Let me walk you through a “mental” exercise.   Consider the instance where the coolant flow-rate is just such, that the outlet temperature of the radiator is near ambient temperature. In essence, hot coolant (200F) going into the radiator, and ambient temperature (72F) coolant coming out. Not realistic but good for this exercise. This must be MAXIMUM EFFICIENCY for sure. Wrong.

Assuming that the coolant loses its heat linearly, the top part of the radiator, where the hot coolant enters the radiator, absorbs almost all of the heat, while the lower part, where the coolant is at, or close to, ambient temperature, the radiator absorbs almost none. The area in between works proportionally.  

Similarly, in the engine, where the ambient coolant enters the block, maximum heat is absorbed by the cooling liquid. Where the coolant now exits the engine (at the cylinder head), only limited additional heat can be absorbed. Thus, the cooling ability is non-uniform. Worse, areas in the head that have the highest heat load (such as around the exhaust ports and spark plug areas) may suffer localized boiling. Once a film of steam forms, almost all cooling is lost.  

Almost all current production systems utilize what I will refer to as a “bottom-up” approach to the cooling system. Coolant flow is moved by a pump, located on the cylinder block. This takes coolant fluid from the radiator and pumps it into the block first, then through the head gasket into the cylinder head and finally from the head back to the radiator system.

As discussed earlier both the absorption and release of heat is linear. This means that the “bottom” up approach severely compromises the coolant’s ability to deal with the more significant areas of heat generation in an engine, namely the cylinder head. As such, most cooling systems are much larger than actually would be required, were the system biased to where the majority of the heat is generated. We only have a given temperature delta ( ∆) to work with within a closed system. This is the difference between the input and output temperatures of the coolant as it enters and leaves the radiator.

In terms of proportional heat generation, the block cooling passages account for about 35% of the heat generation, whereas the cylinder head cooling passages account for the other 65%. In a “bottom up” arrangement, by introducing the coolant into the block first, and thereby drawing a significant portion of the block heat into the cooling fluid first, we have already narrowed the temperature ∆ available to deal with the much more substantial heat load generated by the cylinder head cooling passages. This will increase the likelihood of steam pocket formation at higher load levels encountered under racing conditions.

It may be that there is significant justification to consider adopting a “top-down”, reverse flow, approach to the cooling of the Fiat /A112 competition engine. This idea is far from new, although contemporary developments have lent some new insight. Pontiac, as early as 1956, used a coolant pump to pump water in through the heads first, and then to the block.


More recently, John W. Evans filed a patent (5,255,636)) that makes very interesting reading. I may be more than coincidental that Mr. Evans has a pending, large lawsuit against General Motors for using his cooling ideas on the LT1 Chevrolet V8 installed in the latest Corvette. This motor is capable of running at 10.5:1 on unleaded gasoline, with far greater spark advance than normal.

You might ask, “Just how much advantage can be gained by reversing the coolant liquid flow?” In terms of overall cooling performance the percentage change may only be in the high single digits, but there are other “follow-on” benefits that are much more substantial. Since the coolant is now first going to the head (the place where the most heat generation takes place) there is the full coolant temperature ∆ available to deal with the hottest part of the motor. Since detonation can take place at temperatures as low as 190F, controlling the areas around the exhaust seat and the spark plug seat provides large dividends in terms of minimizing steam pocket generation, but also allows more distributor advance and compression to be used. Head gasket surfaces are also likely to run at more even temperatures. If the overall temperatures at the cylinder head/gasket interface could be reduced by 10%, this would be a substantial additional safety factor.

A further benefit is derived with regard to the engine block. As the engineers at Chevrolet discovered, because of the relatively even temperature distribution, of the coolant entering the block from the cylinder head, the block was heated much more evenly. Even thought the temperature was slightly elevated, this turned out to be an advantage. In a “bottom-up’ cooling system, as low temperature coolant enters the front of the block and has to work its way to the rear cylinders, there is an inherent uneven heat distribution, with the rear cylinders always running hotter than the front ones. While with “top-down” cooling it is uniformly higher and therefore promotes uniform expansion. This assists in better ring sealing and lower piston/cylinder wall friction, all attributes that are important in any competition engine.

In reading Mr. Evans’ Patent, it is obvious that some careful thought went into the design of the system, particularly for a road car. Here you have to account for quick warm-up and providing coolant to the cabin heater core. In principle the system is pretty simple. Here are the items that are required.

  1. A pump capable of reasonable coolant velocity so as to keep the interior surfaces of the head cooling area and block cooling area well scrubbed of steam bubbles, yet not rotating fast enough to cause serious cavitations.

  2. A method of routing any steam bubbles in the top of the head, away from the cylinder head and condensing them back to liquid, to then join the normal fluid flow again. The Patent information is informative in this regard.

  3. A restriction on the output of the block, either in the form a thermostat, bi-directional thermostat, or a restrictor plate.

  4. A condensing unit.

Items #2 and #4 are THE SECRET to making this work. Anyone familiar with refrigeration, or air conditioning, design will quickly realize what is required. A small line, or a larger line with an internal restriction orifice, must be attached to the top of the A112 head and then routed to the remote header tank in the engine compartment. Is that all there is to it?

Well, yes. You see, as the coolant pump is now pumping into the top of the head, and there is a restrictor on the side of the block (where the current water pump sits) there is a pressure differential between the coolant in the block/head and the coolant in the remainder of the system, including the header tank. By forcing the steam/coolant through a small orifice, it will be subject to a pressure drop and the surface area of the line, as well as the overall surface area of the header tank to which the other end of the line is attached, is sufficient to cool the steam and cause it to condense to a liquid state. If there was any doubt as to this you could always make a header tank that incorporated a finned heat sink. This is the same principle that takes place in a modern air conditioning or refrigeration system where gas is converted back to liquid, after carrying away heat.

Another option-

Another way to deal with the same problem would be to use a “dry deck” configuration. In this implementation a head gasket would be used that had NO water passages. Water would be channeled into and out of the head first and then routed to the block. This may also be a viable solution as no steam venting/condensing port for the head is required, although one could be implemented as a safety measure. Overall the amount of heat load to the radiator system would be the same as a conventional cooling system. This type of system would likely need an electric pump, remotely located from the engine.

I will not go into how I would implement these systems in detail, but I believe they hold much promise in being able to run higher dynamic compression ratios, and greater ignition advance, without the onset of detonation.


John W. Evans – US Patents 5.255,636 and 4.550,694

John De Armond – Hot-Rod Magazine

6.2 Radiators

In modern vehicles radiators are most oftem made of either all aluminium or a combination of an aluminium cores with plastic end tanks. For competition purposes I do not recommend the second type, as the stress placed on them by racing may cause a premature failure.

For competition purposes radiators can be made from either aluminium or brass. The heat dissipation qualities of the two metals is similar, with of course the aluminium radiator weighing less.

Careful attention should be paid to the number of fins on the core tubes, and the overall thickness of the radiator. If the fin spacing is too close, or the core is too thick (or perhaps both), then insufficient air will travel through the radiator core. Radiator efficiency depends on air traveling through the core.

If you are going to run a front and rear radiator, then the flow should be as follows:

Engine thermostat housing to top tank of rear radiator
Bottom tank of rear radiator to top left of front radiator
Bottom right of front radiator to input to water pump
Output of water pump to input of engine.

If you are only using a front radiator on a rear engined car, then the following would apply:

Engine thermostat to top of rear mounted expansion tank
Bottom of expansion tank to top left of front radiator
Bottom right of front radiator to input to water pump
Output of water pump to input to engine

In both cases there should be a small, manual valve installed on the top left tank of the front radiator to allow for venting of any trapped air during system filling. As an alternative, a small tube could be run from the top left tank of the front radiator to either the top tank of the rear radiator or the top half of the rear mounted expansion tank. In this way and air or steam bubbles trapped in the front radiator will automatically be dealt with.

6.2.1 Air Flow – Radiators require air flow. The more flow the greater the efficiency of the radiator. This also means that all the air “caught” by the radiator should be made to go through the radiator, rather than around it. Air will always take the course of least resistance, so if there is a gap around the radiator, it will always flow around it rather than through it.

The radiator should be shrouded, to direct the air through the core. This hold true for systems where the movement of the car forces air through the radiator, as well as systems where the air flow is wholly dependent on an air conveyor as in the Fiat 600 and 850 type cars. Air conveyor systems do depend on engine RPM, and therefore may not be able to handle extreme competition temperatures as well.

6.2.2 Radiator Sealing – Most modern radiators have a radiator cap, and this includes Abarths and Fiats. What is generally not known is that the radiator neck on Fiat cars is different to most other cars. It is slightly deeper. Therefore the use of an aftermarket cap, even if it says 22 lbs on it, will generally only result in a 3-4 lb cap. So, if you are using a standard Fiat radiator, use a standard Fiat cap. It will at least give you 12-14 lbs of system pressure.

The other alternative is to have the radiator neck replaces with a standard aftermarket one. Now you can run any high performance radiator cap you want.

Here is an illustration of several stant competition caps that range in pressure from 19-22 lbs. While on the subject of radiators, all hose connections should have a raised lip on the end of the spout, so that once the hose is secured with a clamp, if cannot slide off..

The importance of being able to run higher system pressures will become obvious later in this section.

6.2.3 Radiator Hoses and Clamps – At a minimum I would recommend replacing the radiator hoses at least once each racing season. These items are often neglected and forgotten (at least until one fails and a motor is ruined). These hoses should be installed with NEW screw type hose clamps, and then the hose clamp wrapped in 3-M self-bonding electrical tape.

Other hose connections have been developed specifically for high performance purposes. These in include both fabric and stainless steel jacketed material with AN type screw on fittings, and Wiggins type connectors for joining and terminating solid cooling tubes.

6.3 Water Pumps

Mechanical – All of the mechanical water pumps currently available for the various type of Fiat blocks (817.843,903, 965, 982, 1050) are quite sufficient in terms of transport of fluid, even if using a remote front mounted radiator, provided they are properly driven.

By this I am not referring to whether you use a v-belt or some type of toothed belt, but rather to the size of the respective crankshaft and waterpump pulleys, and the ratio between them. The crankshaft pulley should always be 50% the diameter of the waterpump pulley. This will insure that the waterpump runs at 50% of the engine RPM. This will not only mean that it uses less horsepower, but also that it be less likely to cavitate.

Electric – There are now several type of electrical water pumps available that can be adapted to the Fiat/Abarth type motors. Some are meant to run at constant speed, whereas others incorporate a “controller”, to regulate the speed of the pump. With a controller the water is only enough water is moved to maintain a given, preset water system temperature. I know of one competitor who has quite successfully used two “water guppy” type water pumps in his race car for over 10 years, so the idea is not at all new.

Oil Coolers

Lubricating oil has two distinct purposes. Obviously, one purpose is lubrication, however a second purpose is to carry away heat from vital components within the engine. Components such as cam bearings, main and rod bearings can only stand so much heat before they fail. Likewise, boundary layer interfaces such as rocker arms and rocker shafts, also rely on sufficient oil flow though the interface to carry away heat.

So oil is an important “vital fluid” and its temperature must be maintained at an operating level that keeps it from breaking down. The “terminal temperature” for most organic oils used in automobiles is around 300 degrees Fahrenheit (148 deg. Celcius). For synthetic oils this number is slightly higher, however anything over 350 degrees will get dangerously close to the “flash point” temperature of most oils.

Fluid –to-Air Coolers

The most common oil coolers are similar to a conventional water radiator, only smaller. The idea, for any oil coller, is to provide enough temperature delta to maintain an oil temperature between 220-270 deg. Fahrenheit (104-132 deg. Celcius). In the fluid-to-air oil cooler oil is fed through a number of tubes, over which air flows to carry away the heat. To assist in heat dissipation, the tubes generally have small, thin fins attached.

Generally, oil from the oil pump should be fed to a filter first, and then to the oil cooler, before returning to the main oil gallery of the motor. In this way the cooler is protected from debris, and cooled oil is presented directly to the camshaft, crankshaft and connecting rod bearings.

Fluid-to-fluid Cooler

This type of cooler is better referred to as a “heat exchanger”, and has been used on the maritime industry for many years. Here water, (fresh or salt water) is routed through the large fittings, and oil is routed through the AN fittings. This is then 2-way system.

In a racing application several different versions have been utilized. In one implementation the oil cooler element is placed in the low delta side of a standard aluminium radiator. Thus, using a dual pass radiator, the first pass of the cooling fluid through the radiator sheds much of the engine generated heat, then the water passes around the exchanger element in the left header tank and dissipates the oil system heat in the second pass through the radiator before returning to the motor.

The unit above is a self-contained exchanger and a portion of the water flow, at the low delta temperature point returning to the motor, is fed through the exchanger.

The beauty of this type of oil cooling method is that the oil cooler can be placed almost anywhere with in the vehicle, obviously keeping it as low as possible, and it does not need to be in the air stream in order to be effective.

In the original Abarth TC and TCR implementations there was a dual radiator, placed in the front-mounted radiator shroud. This required both water and oil fluid lines to be routed from the back of the car to the front, and back again. Using a self-contained heat exchanger could save some weight in eliminating one set of the long lines to/from the front of the car.

Both types raise the overall water temperature slightly, so it is important to make sure that the water cooling system is up to the additional task with sufficient temperature delta to handle the extra load.

Specialist Services
Calendar and Race Reports