Another analytical method of course is to use tried and tested mathematical formulae. The Log Mean Temperature Difference (LMTD) is a popular calculation here, and relies on knowledge of overall heat transfer coefficients, the effective surface area of the radiator, a correction factor (sometimes referred to as an ‘ignorance’ factor) and a parameter called the LMTD of the two fluids in use.

A third method much along the same lines expresses the heat transfer in a different way: the difference in temperature between the air and the radiator wall, the surface area, the local air velocity and the type of radiator typified by its ‘K’ factor. From this we could calculate the heat dissipated – assuming, of course, the ideal condition where there are no leaks and the mass of air flowing is constant. The K factor of a radiator is often developed from tables of similar designs, but what if you don’t know your K factor or if your radiator is not of a traditional size or shape? Obviously you could wait for it to be installed and tested in situ in a vehicle and never know if its poor performance was down to its design or installation. Failing that, you could use some kind of radiator test rig or wind tunnel.

A radiator wind tunnel is fundamentally different from one used to test the aerodynamics of vehicles. The most obvious difference is that radiator rigs will generally have a much higher blockage factor, since all the airflow will be channelled through it. This effectively dictates the type of fan that can be used. In vehicle ‘aerodynamic’ tunnels the blockage factor of the car under test is comparatively low, and the pressure drop across the test section is consequently also very low or non-existent. The type of fan preferred will be that of an axial flow ‘air mover’ situated after the test section to avoid introducing turbulent flow to the vehicle or model. A radiator tunnel, however, filling the whole cross-section of the test item, will have a comparatively large pressure drop across it, and will therefore be more suited to the characteristics of a centrifugal blower, which will need to be placed in front of the test section.

Unfortunately the airflow of such a fan is far from uniform, and any turbulence induced by the rotating blades needs to be corrected and ‘straightened’ out to present laminar flow to the radiator. This is most efficiently achieved by slowing the air down through a diffuser and passing it through some form of ‘flow straightener’. The angle of this diffuser can be quite critical to avoid flow separation and the localised turbulent flow that results. At the same time, space is often at a premium and so a compromise of around a 15-20º taper is often used.   

After the diffuser comes the settling chamber. Designed to dampen out the last dregs of swirl from the airflow and create a uniform laminar flow to present to the radiator core, the settling chamber will include a honeycomb section placed across the path of the air. Usually hexagonal and made from thin aluminium or paper, the diameter-to-width ratio of the individual cells should be such as to straighten out the flow with the minimum of losses to the flow. For a typical radiator size of, say, 40 cm², a minimum number of these cells should be around 5000-6000 (75 x 75) but figures up to 25,000 (150 x 150) or more can be easily justified. Finally, after passing through another and much finer mesh, the air will be accelerated again through a converging section into the working section. 

And of course, once you have your radiator wind tunnel, you will be able to correlate the results from your CFD studies.

Fig. 1 - Radiator wind tunnel schematic

Written by John Coxon

These days of course we are a lot more enlightened, and the realisation that the position of the radiator and how it is installed – and not just its overall size and thickness – can make a great difference to a vehicle’s overall performance. That means carefully ducting just enough cooling air into and through the radiator core with the minimum of aerodynamic losses, something that few vehicle modifiers or special builders fully understand or manage.

For any given heat exchanger the rate of heat dissipated is proportional to the mean temperature difference between the radiating surface and the airstream, as well as (airstream velocity) ^0.6 and (air volumetric flow rate)^0.8 passing through it. This means that for a given radiator the thermal efficiency and internal drag are both reduced by slowing the velocity of air passing through the core. High-energy air passing into the ducting thereafter needs to be slowed down progressively before it gets to the core. At the core the velocity energy of the incoming air is converted to pressure energy which, having absorbed as much heat energy as it can, needs to be accelerated away back out into the airstream as quickly as possible. A divergent nozzle should be used to minimise any further aerodynamic losses.

In a number of branches of motorsport, however, many competitors choose to dispense with the radiator altogether, thinking that the reduction in aerodynamic drag will result in reduced elapsed times. In hillclimbs for instance, dispensing with the radiator can save valuable weight, but with the engine cooling jacket overheating, by the end of the climb this can cause considerable loss of power, not to say the possibility of the engine running into detonation as the engine coolant temperature climbs.

In drag racing though, competitors often to go to extremes, preferring sometimes to dispense with the water jacket altogether, or some even fill the void caused by the missing coolant with concrete or a thermosetting resin. This, it is said, is designed to stiffen the stock cylinder block and resist the prodigious torques often delivered by such engines. The Top Fuel boys also machine their cylinder heads direct from aluminium billet, omitting any hint of a coolant jacket. Desperate indeed but when you consider that the fuel they use – a mixture of nitromethane and methanol – has such a high latent heat of vaporisation that for short bursts of 6-7 s (the time taken to complete the course) the risk of seizure is minimal. And when engines are routinely rebuilt between runs, such abuse is easily justified in terms of run times.

For more practical sustainable running for the rest of us, however, the radiator installation needs to be carefully thought out.

Fig. 1 - 1965 McLaren M6A radiator exit duct

Written by John Coxon

In the automotive world the classic air-air intercooler matrix is made from sheet or tube aluminium alloy, which is pre-assembled and hot-dipped in a suitable molten aluminium-silicon brazing solution to bind and seal it. Before this though the matrix has to be cleaned by dipping it into a corrosive flux solution, all trace of which has to be removed afterwards. The corrosive nature of the chemicals involved means this is a hazardous process and subject to many rules and regulations concerning the materials involved. Having made the matrix core, and depending on the application, the end tanks will be fabricated and the whole lot welded together to produce a unit similar to the one shown in Fig. 1.

This manufacturing method therefore makes any form of heat exchanger or intercooler an expensive component which, given its proximity to the front or outer parts of the bodywork, can be easily damaged in the cut and thrust of modern racing. To reduce costs and improve product quality, OE manufacturers tend to make the end tanks these days out of injection-moulded plastics that are then mechanically clamped to the aluminium matrix core and sealed using sheet material sandwiched between the tank and the core.

While minimising costs as far as the technology allows, however, even this process still renders the matrix core susceptible to corrosion and damage as a result of cyclic fatigue. Aluminium alloys have no lower stress limit below which they will operate safely forever, so at some time during use they are bound to crack, with the inevitable loss of engine boost.

Since the end tanks are currently made from plastic injection components, the next stage in intercooler development must surely be to manufacture the matrix core out of similar materials, since most plastics do not corrode and the use of extruded or welded plastics is altogether a less hazardous process. The principal objection to this though would seem to be the relatively poor thermal conductivity of plastics compared with that of aluminium alloys – but not so, according to some reports. Computer simulation has indicated that, in modelling the heat exchange process, the thermal resistance of the thin plastic tubing used compared to that of aluminium was not the major problem one might have reasonably thought. The real issue would appear to be the boundary layer between the bulk air flowing and the tube materials.

By making the heat exchanger core out of extruded polyamide tubing, which could be laser welded together under precise computer control, 100% plastic intercoolers could be manufactured quicker and cheaper without many of the expensive and environmentally hazardous processes that go with current manufacturing methods. In addition, once fully developed, intercoolers will not only be more robust but lighter too.

And I guess there can’t be many areas where engine or vehicle technology is taking its lead from the solar panel industry.

Fig. 1 - Traditionally fabricated aluminium alloy intercooler

Written by John Coxon

EPDM is the product preferred for most automotive cooling systems currently. Capable of withstanding temperatures typically in the 110-130 C range, and sometimes up to 150 C, EPDM combines good heat and ozone (weather) resistance with better than adequate protection against oil and other under-bonnet chemicals such as ethylene or propylene glycols. To enhance its performance, especially under partial vacuum (for instance at the entry to the cooling pump), it can be reinforced with aramid braided yarn to increase stiffness, give greater burst strength and yet still maintain a level of flexibility.

Properly designed, hoses made from EPDM generally last for the full lifetime of the vehicle, which is normally reckoned to be up to at least 15 years or 150,000 miles. However, a word of warning: there have been instances where corrosion inhibitors in some organic acid-based coolants– which are designed to protect modern cylinder head/block alloys containing aluminium and/or magnesium – may have leached out certain components of the EPDM, causing corrosion in the narrow passageways of the cylinder head. While we are told that this issue has now been solved, it emphasises the need to use only the approved engine coolant anti-freeze/corrosion inhibitor in any application.  

While EPDM hoses are the most likely to be found on many road transport vehicles, for applications where the requirements are more severe then silicone rubber is making headway. This is a polymerised siloxane or polysiloxane that consists of a chemical backbone of Si-O-Si-O-Si units. Unlike most other polymers, which have a carbonaceous backbone to them, polysiloxanes are inorganic and, unlike their organic counterparts, the bond angles in them are large and the bonds also vary in length. Thus, when the product is injected into the mould and then cured, the system of crosslinking between each chain of molecules gives a far less rigid product than EPDM, and it is this flexibility as well as many other attributes which is courting appeal.

Perhaps the best attribute of silicone elastomers is their incredible resistance to extremes of temperature while still maintaining their useful properties. Silicone elastomers can routinely withstand temperatures as low as -55 C and as high as +170 C, which is more than adequate for most under-bonnet applications. To counter this flexibility if used under partial vacuum – when the hose can collapse – the material can be reinforced with up to five layers of a medium-duty knitted or woven polyester fabric. A liner made of a natural rubber can also sometimes be placed inside to seal off potential leakage paths caused by any exposed fibre.

In applications where hydrocarbon fluids may be present (oil mist in engine induction systems, for example), silicone hose lined with a fluorosilicone is normally used; this prevents the migration of the fluids through the hose wall over time, causing the hose material to swell. In this way the hose will not become brittle and fail. Furthermore, the operating temperatures will be increased to up to 250 C, which makes them ideal for turbo or supercharger/intercooler hoses or on passenger vehicles that are subject to stringent emissions tests.

Above all though, silicone rubber hoses can seriously improve the appearance of your engine bay.

Fig. 1 - Seen in the paddock at the Shelsley Walsh Hillclimb, a black EPDM hose in the cooling circuit, and blue silicone for the supercharger air hose

Written by John Coxon

It’s easy to determine the heat rejection of an engine: simply measure the temperature of the coolant going in, that coming out, multiply by the flow rate taking into account the specific heat of the coolant mixture and you have your answer.

Alternately, you could always just measure the fuel flow rate and multiply by the heat of combustion of the fuel (around 43 MJ/kg). If you assume a heat rejection to the coolant of around 35% of this then you won’t be far out. Armed with these figures and a request for a maximum cooling area within the minimum of physical dimensions, and together with a few more facts (and an assumed ambient temperature), your supplier will service you with a suitable component to fit into your vehicle. Having ruled out poor heat exchanger design or inadequate heat exchanger size as the reasons for inadequate engine cooling, the only task now is to ensure a supply of sufficient cooling air at all vehicle speeds to have a fully functioning cooling system.

In the case of an existing saloon or sports vehicle, the location of the heat exchanger is generally more or less decided, and the only option now is to ensure that both entrance and (crucially) the cooling air exit is not restrictive. You may be able to guide the air more effectively in and out of the matrix core, but in general there are so many other chassis or engine parts in the way that any attempt at improving the airflow through the vehicle is almost impossible.

Where the installation allows much more freedom though, as in the case of a formula car or sports racing machine, the most efficient way is to carefully duct the air into and (perhaps more important) out of the heat exchanger back into the passing air as it exits the vehicle. The important areas to be addressed here are the leading edge of the entrance duct, which needs to be rounded to avoid separation of the airflow, and the diffuser section leading up to the heat exchanger core. This should be designed to slow the incoming air progressively without separating it from the guiding walls, thereby increasing its static pressure to aid the heat transfer process. Once through the core, the ducting needs to contract progressively to speed up the air again and return it to the same free-stream velocity of the air flowing past the vehicle at the exit.

In recent years designers have started to use turning vanes ahead of the entrance to the duct intake to ensure that the air spilling off the front of the vehicle is better guided into the duct, or other techniques to bleed off airflow boundary layers and thus improve the volume of air reaching the heat exchanger core. A science in itself, this is the area where vehicle aerodynamics and engine cooling start to merge.

Fig. 1 - Placed at the front of the vehicle, this is still the most efficient position for the vehicle heat exchanger, if a little vulnerable

Written by John Coxon

Electric fuel pumps are now common, but although direct-injection pumps have reverted back to mechanical for the time being, I wonder how long it will be before they too, in the search for better starting, will go to electric.

These days we can drive turbo-compressors and control the coolant temperature electrically, replacing the good old wax thermostat. There are even plans to introduce electric oil pumps – though possibly not just yet. These no doubt have their benefits in fuel economy in production vehicles, but the opportunity to produce the biggest benefit for the rest of us is in powering the water pump.

The problem with the mechanically driven water pump is that, being geared to the crankshaft, its installation is very much a compromise. During the engine cold-start and warm-up phase, a mechanical pump working off the crankshaft will actually delay the time to reach the ideal working temperature (around 75-85 C). This is perhaps not so important for competition vehicles, but in the world of high-performance road vehicles, and with carbon dioxide/fuel economy emission legislation becoming imminent, rapid warm of engines is seen as critical.

At the other end of the performance envelope, the pump has to transfer the maximum heat coming from the engine and dissipate it through the vehicle radiator. The pump may be selected to cater for the worst-case condition but in the case of, say, a cruise condition or when running at high speed but only part-throttle, the mechanical pump is working faster than needed for a given cooling requirement. And working faster means greater power consumption. But as well as that, all this uses fuel, producing higher fuel consumption in road vehicles and greater start-line weight in competition vehicles.

However, to me as an ardent support of club motorsport, the biggest benefit arises when the engine is not working. After completing your ten or 15 laps out on the circuit and coming straight into the pits or paddock, switching off the engine will result in a phenomenon called heat soak. Here the residual heat of combustion will steadily ‘soak’ into the surrounding water jacket which ordinarily would transfer the fluid to the radiator via the recirculating coolant to dissipate this heat in the normal way.

But since the crankshaft has stopped rotating, and along with it the water pump, under these conditions no such heat transfer can take place. The coolant temperature in the engine will therefore steadily increase until it reaches the boiling point of the fluid at the particular pressure prevailing, at which time it will start to boil. At this point, releasing this pressure by, say, removing the system header tank pressure cap will inevitably cause the fluid to boil instantly, spraying superheated coolant everywhere. Also, the excessive temperatures reached around the engine caused by this heat soak could introduce damagingly high levels of strain within the engine, leading to gasket failure or permanent distortion of block/head or possibly even both.

The introduction of the electric coolant pump allows the coolant to flow independently of the engine, such that when switching off the engine the coolant is still allowed to circulate, dissipating the heat in the normal way. When the top hose coolant temperature has fallen to a safe value, the pump can automatically switch off.

So I have to admit, not everything electrical is bad.

Fig. 1 - Afterboil

Written by John Coxon

Mounting it near the engine gives acceptable throttle response at the expense of cooling efficiency, whereas placing it on the exterior of the vehicle gives good efficiency and hence power, but results in an installation woefully lacking in engine immediacy. Some years ago the latter may have been fully excusable and tolerated as the price of high turbo performance, but these days there is no such justification. So when it comes to supplying aftermarket intercoolers to create a product that is better than that of the OE vehicle manufacturer, the task will be difficult.

For an intercooler, the critical characteristics are based around the drop in temperature against the pressure loss at a given flow rate. For a 3.0 litre engine, for instance, at peak engine performance and when cooling the compressor outlet temperature from 140 to 40 C, the pressure drop across the cooling unit may be something around say 3 psi (20 kPa). The intercooler unit itself could be 24 in (60 cm) wide by 7 in (18 cm) high and 3 in (8 cm) deep, and given the constraints in the original styling of the vehicle it will be virtually impossible to increase the frontal area of the unit.

The only option might be to increase the depth of the matrix core. Unfortunately, however, doing so leads to greater inefficiency and simply may not be possible giving the space constraint. Positioned at the front of the vehicle, this area is reserved these days for a crushable structure in the event of a head-on collision. Not particularly important in the search for improved 0-60 times, but moments before impact its significance increases out of all proportion to either cost or weight. When faced with conundrums like this though, engineers come up with some ingenious solutions.

I saw one such solution at a recent exhibition. Tasked with the problem of installing a much larger intercooler at the front of the vehicle, the ideal space was occupied by a cross-member at the front that also contributed to the crashworthiness of the vehicle as a crushable zone. Thinking laterally, and using the principle that one component can surely do more than one function – a mantra that should be familiar to everyone in motorsport – the cross-member was modified to include the intercooler, and the intercooler design was also modified to sit inside a 6 mm thick rectangular frame that doubled as the outer wall of the cooling matrix and was also welded into where the original member lay, providing structural rigidity to the vehicle and acting as a crushable structure. Furthermore, the unit was situated at the front, making best use of what air it could find and keeping intake system hose lengths to a minimum.

Indeed, it was a solution that many designers might be proud of. As an aftermarket design, and after recalibrating the fuelling system, this modification may even be better than the original installation.  

Fig - The intercooler as a structural member

Written by John Coxon

Pushing the air through has many advantages. First and foremost, operating as a simple air pump, the performance in terms of flow rate will be most likely higher. Axial flow pumps – for in effect that is all this type of cooling fans are – are good for moving large quantities of air at low pressure differentials, and any intake pressure loss as a result of the restriction of the radiator matrix will have some effect on fan performance. Placed in front of the radiator and therefore pushing the air through with the minimum of intake restriction is a good thing. Conversely therefore, placing it behind the radiator and pulling the air against the restriction of the radiator is bad.

Positioning the fan and its electric motor in front of the radiator should also have aerodynamic benefits as well. All things being equal, and keeping the radiator in the same position in relation to the engine, it will also reduce the blockage effect, assuming the engine is mounted behind it, as in most front-engine vehicle layouts. This will reduce the effective back-pressure against the flow of air, and at the limit of its performance will minimise the chance of the air stalling through the compressor blades. At this limiting condition, not only will this create unnecessary noise but stalling of the air will drastically reduce the performance of the fan, and is a not uncommon failing in a number of fan installations I have seen.

Aside from aerodynamic considerations, there are however also the practical ones. Placing the fan along with its electrical motor in front of the radiator ensures that the motor remains as cool as possible. Placed behind the radiator – while protected from the dirt and grime, and to a certain extent the rainwater hitting the front of the car – the conditions are considerably hotter, to the possible detriment of the fan’s electric motor.

So while the decision to put the fan in front of the radiator to ‘push’ the air may appear to be the correct one, taking into account aerodynamics, the practical considerations of protecting the electric motor often rule the day, and suggest that pulling it through may be best.

Fig. 1 - Radiator fan assembly

Written by John Coxon

As the years passed, however, we learned that vehicle bonnet lines needed to be low, and with the universal inclusion of the water pump to forcibly circulate the coolant, radiators with the same frontal surface area could be low and wide. Instead of having an increased number of short but vertical tubes leading from top to bottom of the radiator, it would make sense for the coolant to flow from side to side, with a small collector tank at each side. The downside was that with the radiator mounted so much lower in the chassis, a totally separate header tank would be needed elsewhere higher up on the vehicle. So when cooling systems would more or less automatically vent air when being filled or when in use, the new cross-flow systems would not be quite so accommodating. And when we come to high-performance engines of course, the problems become worse.

In essence, the problem is all down to air and/or steam becoming trapped inside the engine or radiator during use, displacing the engine coolant. In a carefully designed cooling system, since the coefficient of thermal expansion of the coolant ( at 207 x 10^-6/K) is greater than the volumetric equivalent of aluminium (69 x 10^-6/K) or iron (33 x 10^-6/K), the coolant expands greater than the water jacket containing it. As the engine warms up, the system is therefore pressurised and the coolant will be ejected past the relief valve/cap and into the catch tank. If perhaps at some point, say when making a long pit stop and the cooling effect of the radiator passing through the air is lost, then as a result of heat soak the coolant would expand even more – perhaps even boil – and push out even more fluid into the catch tank. Returning to the track afterwards and with less coolant in the system, the inevitable is just around the corner – literally!

The obvious remedy is to make sure the header tank is large enough and placed high enough in the chassis and linked into the bottom hose of the radiator. In addition, if the total volume of the cooling system void is greater than that of the coolant at its maximum running temperature, provided that little air is circulated with the coolant then no further coolant should be expelled. If air or steam is carried around then some kind of swirl pot may be needed. This will centrifuge the coolant at some point in the system, separating air from the coolant and bleeding the air into the low-velocity zone at the top of the header tank.

Fig. 1 - Header tank mounted on top of the engine with the top hose inlet tangential, thus creating a centrifuge effect and separating any air

Written by John Coxon

Efficient combustion is obviously the main aim in most engine designs, and with a third of the heat energy going to the coolant jacket and roughly another third rejected to the exhaust gas there is still quite some way to go. Little wonder therefore that people are increasingly interested in extracting the heat from the exhaust gas or engine coolant in other ways, rather than dissipating the heat into the atmosphere, as we do now.

The most popular way of ‘harvesting’ this heat is nothing new. A traditional turbocharger expands the exhaust through the turbine and uses the work produced to compress the intake charge. Nevertheless, as the exhaust gas leaves the turbine wheel there is still more of this ‘waste’ heat to be recovered. Sending this gas through yet another turbine and expanding it still further, and then feeding the power generated back into the driveline, should in theory at least produce performance or fuel economy benefits – and, if you like, is in effect another form of ‘heat’ exchange. But while the idea is simple, the issues associated with turbocompounding are not without difficulty.

The greatest issue perhaps is to ensure that the pressure drop in the exhaust system does not adversely affect overall engine performance by increasing engine pumping losses. In creating an additional obstruction in the exhaust system, the pressure upstream at the exhaust port is likely to be increased. This will affect the expansion of the gases through the primary turbine, which in turn can affect the intake pressure out of the compressor, reducing engine performance to the detriment of efficiency. Carefully matching both the primary and power turbines to the application is therefore critical.

While clearly there is much work involved in specifying the turbines, there are many other aspects of the installation to be covered.

Protecting the turbine blades of a turbine wheel travelling at 150,000-plus rpm from the ravaging effect of the torsional vibrations of the driveshaft in a reciprocating engine that’s doing only 5% of the wheel’s speed is an absolute necessity. This can be taken care of by using either long quill shafts or, more commonly these days, some kind of fluid coupling. Excited by the inherent dynamics of a four-stroke engine, the power turbine loading is best introduced at a ‘quiet’ place in the driveline, more usually at the flywheel end of the crankshaft.

An increasingly attractive approach, however, is to power an electrical generator feeding the power back into a battery or supercapacitor device. This has the added advantage of being able to control the turbine speed to run at its best efficiency by varying the torque generated through that generator. Secondary turbines can optionally be used to power engines auxiliaries – water pumps and alternators, for instance – and save valuable engine power this way

It may not be everyone’s definition of a heat exchanger, but as a way of increasing raw engine performance – or, more importantly these days, engine fuel efficiency – it no longer can be ignored.

Fig. 1 - Turbocompound general arrangement

Written by John Coxon

But even with adiabatic efficiencies of up to 80% or more on some of the better compressor wheels, the charge temperature can increase significantly to the point where compressor outlet temperatures of the order of 200 C are not uncommon. While part-load fuel efficiency might be impressive at these temperatures, engine performance in terms of raw power is certainly not. Consequently, to compensate for these high charge temperatures, the most obvious solution is to install an aftercooler (commonly known as an intercooler) which, by cooling the intake charge, has more advantages than we might think.

First and perhaps most obvious, in keeping charge air temperatures low the air density will be greater, allowing more air to enter the cylinder. The greater the density of the incoming air, the greater amount of air trapped in the cylinder. Important at all engine speeds and loads, at high engine loads a temperature drop of even 30 C can increase the engine charge air density by around 10% with, all things being equal, a similar increase in power output. In gasoline engines, not only is the power increased because of the increased charge density but the engine’s sensitivity to combustion ‘knock’ or detonation may also be reduced, enabling possibly greater ignition advance and therefore even more engine power. Likewise, in diesel units when operating near the smoke limit at high load and low speed, the black smoke produced through incomplete combustion will be significantly reduced, enabling slightly more fuel to be ingested and greater low-speed performance.

But the introduction of the aftercooler doesn’t just lower the inlet charge temperature, the combustion temperature and the exhaust gas temperature will also be lower. A reduction of, say, 50 C at the inlet valve will most likely see a similar reduction in the exhaust gas temperature, with the possible benefits to the quality of the exhaust valve material required and its durability. The opportunity to use a lower quality material could therefore be examined. If you like, including an aftercooler could therefore save you an awful lot of money on exotic valve materials 

But including an aftercooler means extra weight and greater induction system volume, leading to transient control issues and the obvious extra complexity of installing it into the vehicle. Is it all really worth it? That particular conundrum is, in addition to the increased performance delivered, also down to a parameter known as the effectiveness of the device. Defined as the ratio of the actual heat transfer as a fraction of the maximum possible heat transfer, the recovery of the charge density through the aftercooler is quite an important feature, and one that not only encompasses its design but also its installation.

To be highly effective, the aftercooler will need to be positioned close to the extremity of the bodywork, introducing large manifold volumes and accentuating the infamous ‘turbo lag’. Alternately, a larger unit in a less ideal environment could have the same thermal effectiveness but being nearer to the engine, the manifold volume would be reduced. The size and weight of the cooler might be increased, but that might be offset by the reduced weight of the hoses and so on. In essence you pay your money and, well, you fit it in the most optimum place you can find!

Installing an aftercooler might at first seem a simple undertaking, but to get it right takes a lot of effort and understanding.

Fig. 1 - An aluminium aftercooler 

Written by John Coxon

These days though there are no excuses for oil leaks. There is an entire system of aircraft-quality hoses and hose ends available that can easily be assembled in the workshop and, when assembled correctly, not only do not leak, burst or come undone but are visually attractive and a credit to any vehicle engine bay. However, as with most things that are apparently simple, there is a lot more to this issue than meets the eye.

In any flexible hose connection there are generally two places where leakage can occur. The first is where the metal union connects to the component – for example pump, cooler or oil tank – while the second (and more difficult) place is that of the join between hose end and the hose itself.

The former is a simple matter of mechanical engineering design to ensure that the conical seat of the male part contacts the marginally differently angled conical cone of its mating part for all its 360º. So long as the clamp force is maintained, the joint should not leak. The connection between the hose and hose end is however a different matter, and depending on the diameter of the hose and the fluid pressure flowing through it, may require either a single nipple design or one using a nipple and cutter.

The single nipple design is, as you might expect, the simplest. With the female socket part of the hose end slipped over the braided hose, the male, threaded part of the hose end is inserted into the hose. When tightening the female socket, the hose gripped on its outside diameter by flutes in the inside of the female socket is compressed firmly into the space between female and male parts. Hence the wedge thus produced forms the one and only seal.

When pressures are greater or the degree of safety required is higher (as in the case of modern hydraulics, for instance – the nipple and cutter design is probably more appropriate, because as well as the single male nipple that fits tightly into the inside of the hose, a second nipple or cutter concentric with the first cuts into the hose elastomer material. During the assembly process the hose is grabbed by the flutes of the female socket, forcing the elastomer of the hose onto the nipple to create not one but two annular chambers [Fig. 1]. Should, for instance, the inner, primary seal leak, a secondary seal will prevent the fluid from leaking outwards – a sort of ‘belt and braces’ approach but one that is simple and, above all, reliable.

Fig. 1 - Nipple and cutter design

Fig. 2 - Selection of hose ends

Written by John Coxon

Of course though, the heat transfer characteristics of air are not as efficient as those of the primary constituent of a traditional engine coolant – water. For instance, the specific heat of air at normal atmospheric conditions is about 1.0 kJ/kg deg K, while that of water is about 4.2. However, the density of water is something like 840 times that of air, so even if we ignore the effects of the heat transfer coefficients using air it will take at least 840 times the volumetric flow rate of traditional engine coolant to absorb the equivalent amount of heat energy. To ensure that the air traverses all the parts of the engine, a fan of some description generally has to be used in, around and between the cylinders.  

Perhaps the most famous manufacturer using air-cooled engines for many years, including its high-performance and racecars, is Porsche. Famed for its reliability if not perhaps for its ultimate out-and-out performance, such was the technology surrounding the head gasket design of liquid-cooled engines that designer Ferdinand Porsche’s would say that an engine of his had never lost all its air. Even so, the compromises in engine design would be considered too great today since the space between bores needs to be considerably larger, and cylinder configurations are limited. Porsche in particular was famous at one time for its six-cylinder flat or ‘boxer’ engine for roadcars, and flat eight, 12 and even 16 cylinders (on one occasion) for its race machines.

For the roadcars, fans were mounted horizontally and belt-driven off the nose of the crankshaft, whereas for racing, drive would come from the crankshaft via a bevel gear. The fan would therefore invariably be positioned centrally and directly above the engine. The 1.5 litre eight-cylinder Formula One engine of 1962 had a 250 mm diameter axial flow fan using 17 blades, giving an airflow of around 890 m³ per minute using only about 6.5 kW of crankshaft power. Axial flow fans are very good at shifting large amounts of air at low pressure differentials, and by ensuring that there were no restrictions to flow made them ideal for the application. By the end of the 1970s the 4.5 litre flat 12 engine, producing around 440 kW at 8400 rpm, used a slightly smaller version, this time with only six blades.

Critical to the whole design is the internal ducting and cowling necessary to ensure that the engine is cooled uniformly along its length, and that the least internal stress is built up due to any temperature variations. Likewise, the decision to ‘blow’ down into the engine rather than ‘suck’ up through it was essential to the cooling of critical areas – in particular the cylinder barrels and the high-temperature zones between.

In cooling the engine using only air, more stress is inevitably placed on the oil circuit and the additional necessary cooling. Air-cooled engines therefore run slightly hotter and can be more marginal on lubrication. To ensure better reliability, air-cooled engines tend to be fuelled more on the ‘rich’ side to keep piston crown temperatures down. Air-cooled engines therefore tend to be less fuel efficient, which is partly why we don’t see them in modern vehicles.

Popular among many classic car owners, you could say these are the fans of the fan cars.

Fig. 1 - Air-cooled Porsche racer engine

Fig. 2 - Air-cooled Porsche 911 engine bay

Written by John Coxon

In the world of pumping there are many differing methods. Positive displacement pumps, for instance, can be used for low flow rates and pressures. For the highest flow rates and low-ish pressures though, axial flow pumps are often the preferred option. However, for medium flow rates having to work against high pressure ‘heads’ – as for example that created by the presence of a vehicle radiator – the only serious option is that of a centrifugal flow pump.

Reminiscent of a turbocharger’s compressor, the principle behind a centrifugal pump is much the same – flow is introduced into the centre of a spinning wheel (the impeller) and centrifuged out, thus increasing its velocity as it flows outwards. Guided by the vanes in the impeller, the additional velocity energy imparted to the fluid can subsequently be converted into pressure energy by the action of the diffuser in the outlet of the pump, and so while these pumps may not be the most efficient around (roughly 80% could be the maximum of a well-designed pump) simplicity rather than efficiency is usually the name of the game here.

In converting this velocity energy into pressure energy, care has be to taken to ensure that the diverging, conical section of the diffuser does not cause the formation of flow eddies at the wall and introduce undesirable losses in this portion of the pump. For some impeller designs, simple radial vanes might be the approach, whereas for other, more efficient types, backward-swept vanes may impart a much higher efficiency, though produced perhaps within a much restricted performance ‘window.’

In general, for forward-angled blades at the periphery of the impeller, the pressure head of the pump will increase with increasing flow rate, whereas if the impeller blades are swept backwards, more efficient pumping may be attained. In designs like this when the blades are swept backwards, the pressure ‘head’ achieved by the pump will most likely fall as the flow rate rises. This is quite the opposite to those facing forwards to the direction of travel which, as a rule, will increase the pressure head as the flow increases. 

Using computation fluid dynamic studies, the complex behaviour of the engine coolant into and through the pump during operation is now being more fully understood, but somehow, although we may have more powerful tools, the exacting nature of modern engine requirements doesn’t seem to make the challenge of pump design any easier.

Fig. 1 - Classic radial impeller water pump design

Written by John Coxon

The first point to remember in any design is that the density of hot water is less than that of cold water. At 90 C, a temperature typical of the top of a vehicle radiator, the density of pure water is 965.163 kg/m³. At 85 C, a typical temperature at the base of the radiator, this increases to 968.508 kg/m³, an increase of 3.3 kg/m³. While this difference is relatively small, the natural tendency of the cooling medium will still be to sink towards the base of the radiator, allowing hotter fluid from the engine to take its place. 

On the other side of the engine cooling system the cold water circulating through the engine will pick up heat and rise up, passing through the cylinder head and into the top hose. If this hose is linked directly to the top of the radiator then the combined effect is to generate a circular motion of cooling fluid sometimes referred to as a thermo-syphon. To speed up the process we also normally add some form of water pump to work along with and in the same direction as this thermo-syphonic action, when any attempts – however small –  to work against this natural order of things will compromise any cooling system to a certain degree. Ensuring that the engine outlet hose is linked to the top of the radiator(s) without air locks or sudden changes in overall flow area, and that the radiator outlet hose feeds back into the base of the engine (via the water pump), is therefore a golden rule.

Another point to remember is that the most economical radiator design is one with the narrowest of matrix depths and therefore, for a given level of heat dissipation, the greater the frontal area of the radiator then the more efficient it will be. As radiators fall in cross-sectional area and become thicker, their efficiency also falls, because as the air passes through it picks up heat and the temperature difference between it and the medium to be cooled – so essential to the process – will be reduced. In many racers this is evidenced by very large but thin radiators being laid down at an angle to the direction of flow. This reduces the cross-sectional area exposed to the direct flow of air but also has the effect of lowering the centre of gravity – so essential for other aspects of vehicle dynamics. 

When requiring the best radiator design to be used it also makes sense to ensure that the radiator itself is cooled in the most efficient way. Making sure that the cooling air is properly ducted to and from the radiator, and that all cooling air is encouraged to pass through it rather than around it, are all also important. Smooth internal ducting converging slightly to accelerate the flow of cooling air is therefore essential for best results.

So while the design of the radiator itself is important, to dissipate the heat efficiently, so are many other aspects of the installation.

Fig. 1 - Formula One Lotus Renault turbo radiator installation (Photo: John Coxon)

Written by John Coxon

In the end, however, no matter how little heat is left, what remains has to be dissipated into the surrounding environment. For the exhaust system, this is dumped straight back into the atmosphere, but for the heat rejected into the engine cooling system the task becomes a little more difficult. For the racecar designer therefore, where heat rejection to the atmosphere costs, this means smaller, lighter, more efficient radiators with fewer in the way of aerodynamic losses.

In the past, racecar designers have taken much of their inspiration from aerospace, adopting advanced composites or titanium and other more exotic materials. So when it comes to advanced heat exchanger technology it seems only sensible to look in that direction as well.

Thus in preference to the traditional fin-and-tube heat exchanger design, it seems that microchannel technology may in future find its way into automotive applications. Consisting of a multitude of very small rectangular or circular coolant channels [see Fig. 1], the advantages stem from the greater heat transfer rate compared to traditional fin-and-plate designs and produce a considerable increase in the surface area of the liquid side compared to that of the air on the other face.

These small channels can be either circular holes or rectangular channels flowing through the radiator ’slabs’ at right angles to the flow of air. When circular, the holes are typically somewhere around 200-700 microns in diameter, but since there are so many of them, the surface area in contact with the liquid to be cooled is extensive. On the other surface, however, where the heat is dissipated into the air, the air has a very low thermal conductivity and in general will act as a barrier to heat flow. At this surface therefore the thermal resistance will often dominate the overall performance of the unit. By increasing the area in contact with the liquid side compared with that of the air side, the overall efficiency of the unit can be increased.

However, as is normal in life, things are never always so simple. To provide a volume and weight benefit, microchannel heat exchangers are generally much thinner than their more traditional plate-and-fin alternatives. Air-to-liquid microchannel heat exchangers usually have short airflow lengths and long liquid flow lengths. This results in low airside pressure drop compared to the much higher pressure drops on the liquid side. Reconfiguring them to produced a much fatter, cubic shape more suitable for automotive applications will inevitably trade this shape against performance.

Microchannels are however highly sensitive to particulate contamination of the coolant circuit so when or if aerospace finally adopt the technology it may be a long time before we see it in mainstream automotive applications.

Fig. 1 - Microchannel cooling

Written by John Coxon

By their very nature, centrifugal compressors - indeed all air compressors - are inefficient to some extent, and this inefficiency is expressed as an increase in the air temperature during compression. Unfortunately, however, increasing the charge temperature has the double whammy of not only reducing the charge density and therefore the amount of air going into the cylinder (and hence power) but also, by increasing the charge temperature, once inside the combustion chamber it is that much closer to 'knock' or the uncontrollable detonation in a turbocharged engine. For any competition engine therefore, the design of the intercooler is critical.

Efficient charge cooling is a matter of creating a large, efficient thermal transfer surface and the necessary airflow geometry to create efficient heat transfer. In the case of an air-to-air cooler, this means the engine air flowing through it and into the engine as well as that flowing through and around the heat exchanger matrix, dissipating the heat to the surrounding air.

At first one could be forgiven for thinking that the larger the matrix, the more effective the heat transfer, since a large surface area and minimum pressure drop through the engine intake side will be most beneficial to engine performance. However, large intercooler volumes create transient engine response issues as well as weight and high aerodynamic drag once installed in the vehicle. Furthermore, unless it is carefully marshalled, the tendency will be for the air to flow through one particular part of the matrix, meaning that the matrix will not be as efficient as expected and will fail to generate the heat transfers desired. Unlike most naturally aspirated power units, turbocharged and intercooled engines are very much more an integral part of the vehicle, and what may be best for engine power may not always be beneficial to the aerodynamics/weight of the vehicle once the intercooler is fully installed in the chassis.

Inside the intercooler the airflow has to be carefully managed to ensure that the flow passing through or around the many metal tubes and fins is as turbulent as possible without generating too much of a pressure drop. Laminar flow - when the lower layers of air remain adjacent to the heat transfer surface - is the killer of efficient heat transfer, since in this instance the air here is close to being totally static and acts more as insulation rather than an efficient heat transport medium. Let's not forget that the air gap in double-glazed windows is designed to insulate against, not promote, heat transfer, so unless the flow is fully turbulent where it matters, the effect inside the cooler could well be that of insulation.

High heat transfer through large flow areas (and hence low intake pressure losses) may be good for the engine, and small, lightweight intercoolers with minimum cross-section (and hence aerodynamic drag) may be good for the chassis. As a designer balancing all this and producing the best possible transient performance for the engine when installed in the chassis will no doubt keep you awake at night.

Fig. 1 - Intercooler installation

Written by John Coxon

A common part might be, say, a top or bottom radiator hose. If money was not a problem, a motor accessory shop or tame vehicle parts store could be the first port of call, for in those days they would often have bins full of suitable hoses and one could spend hours looking for that exact shape and diameter. When money was tight or even non-existent, the vehicle breaker was the place where, for a few coins, a radiator hose could be acquired which when hacked about would do the job, albeit not altogether perfectly.

In those far-off days little thought was given to the material or its suitability, but we were young then and oblivious to the potential dangers. And, let's face it, it was a wonder that more of us weren't hurt more - I for one remember scalding-hot water spurting out all over me at one time from an inadequate hose.

These days of course we live in enlightened times, and the practices of the past are quite rightly no longer tolerated. As well as that, we now have a choice of a number of companies supplying a full range of pre-formed hoses for various applications and of all shapes and diameters at competitive prices. No longer is there an excuse therefore to make do and mend.

But not only can these hoses be made to fit more or less any application, the superiority of their materials are an added bonus. The most significant of these is silicone rubber, which when reinforced with woven polyester fabric can cater for just about any automotive application you can think of. Silicone rubbers, you see, have a natural resistance to extremes of temperature - as much as 300-plus C under certain conditions and down to -100 C at other times. Furthermore, the elastomer retains a far higher tensile strength and elongation at rupture, as well as better resistance to tearing than any of the other materials frequently found.

But in vehicle engine compartments it is not normally about strength or compression 'set', or indeed even the material's exceptional electrical insulating properties. In the real world the main concern - on top of things like resistance to the latest lubricant formulations or organic acid coolant additive technology, or indeed the effect of various solvents - is its general resistance to catching fire.

In the wider world, silicon elastomers are often used in applications where the risk of fire cannot be ignored. In some industries, if hoses have to used at all then silicon elastomers may be one of those technologies allowed. The reasons for this are very simple - silica-based inorganic compounds are less easy to ignite and the non-flammable silicon forms a protective barrier on exposure to flame. When that flame is removed the substance extinguishes itself and enables the fire to be put out quickly and safely, minimising the risk of a major conflagration.

So they may be readily available in all shapes and sizes and for all automotive applications but most of all, a shiny gloss cellulose finish of blue, red or even yellow will make any engine installation an attractive prospect.

Fig. 1 - A silicon hose makes an attractive engine bay finish

Written by John Coxon

To understand many of the issues, it's as well to look at what we are trying to achieve in drawing or pushing air through a vehicle's radiator. First and foremost, a cooling fan is an air mover, designed to move relatively large amounts of air against little or no back-pressure. For this type of application an axial-flow air mover is ideal.

Having selected the type of fan most suited to the application, it is now important to design it so that it imparts to the air stream a uniform velocity and pressure over its entire area. And to ensure that the radiator is working as efficiently as possible it is imperative to select the largest fan diameter that can be accommodated. Sometimes, because of the size or shape of the radiator or its location, there is insufficient room for a single large unit and therefore two smaller ones have to be substituted. Be aware, however, that even if you ignore the blockage effect of the central hub motor, it takes for instance two 8 in diameter fans to give the same cooling performance as one only 3 in bigger.

Although apparently simple, the design of a fan is a lot more complicated than many might think. First, each individual blade will be narrower towards the tip where the velocity is highest and progressively widen towards the central hub. Also, the blade profile, an aerofoil in shape, will be selected for its optimum efficiency of lift versus drag over the anticipated speed range, and the blade will twist from a minimum angle of incidence with the airflow at the tip to a maximum at the other end as it adjoins the central hub. In this way, the reducing velocity of the blade as we move in from the tip is compensated for by the increasing angle of incidence to the airflow and a uniform velocity and pressure across the outlet results.

But of course, once you have created your uniform distributed airflow in the most efficient way, the last thing you want is to let it escape or come from a place where it doesn't pass through the radiator matrix. Shrouding the zone between the radiator and the rotating tip of the fan is therefore imperative, while at the same time making sure that the pressure at the outlet is not restricted to any degree. Such restrictions, for instance the close proximity of an engine, may cause the airflow to stall or circulate around only the blade and not through the radiator core.

The electric fan may be ubiquitous in modern vehicles but it still takes careful design and installation to get the best out of it.

Fig. 1 - The fan blade must generate a uniform velocity and pressure of the airstream over the entire area

Written by John Coxon

Let's start by looking at the basics. Since power is roughly proportional to speed, an engine operating on wide-open throttle at 6000 rpm will deliver roughly twice as much power than at 3000 rpm. That means twice as much heat rejected to the cooling system, requiring more or less twice the coolant flow rate. As coolant pumps are designed to cope with the extremes of heat, the mechanical pump geared off the engine crankshaft would seem to be a most efficient way of doing things. However, engines don't always run at wide-open throttle for long periods, so an engine running at part-load will deliver a lot less heat and therefore may not require the pump to run at anywhere near the performance level required for full-load running.

In the past, pumps with adjustable shrouds, which can alter the flow rate without altering the speed of the unit, have been tried but the extra complication and durability concerns have always outweighed the advantages. At a time when all things related to vehicle onboard control results in more electrical/electronic components, the practical solution would seem to be the electric coolant pump, digitally controlled.

When powering the pump from the vehicle's battery, which is not being recharged by an engine-driven generator, the power gained in not driving the coolant pump at 2-3 kW must be worth having. It follows that so long as the total installation weight is no greater than that of the more traditional mechanical approach, the use of an electric pump is a 'no-brainer'. Thus, disciplines such as hillclimbing and sprinting, where the duration of each run does not require the use of an alternator, are obvious applications for the electric coolant pump.

When you introduce an alternator to recharge the battery it is necessary to consider the expected duty cycle and the battery-charging strategy. If the battery is to be recharged continuously, at engine full load the conversion efficiencies between generating the electrical power and then powering the pump have to be taken into account. However, if the battery is to be recharged only when at part- or off-throttle then the case for an electric pump is much more compelling. In either case, removing the direct relationship between pump speed and engine speed allows for potentially more efficient cooling, particularly where pump speed is under the control of the engine management system.

This needs to be considered in the context of the type of racing. The flexibility of the electric coolant pump can clearly pay dividends in rallying and endurance racing, more so than in wide-open throttle oval track sprint racing. At Le Mans the engine has to be shut down during refuelling, and the electric pump permits coolant to continue to circulate, reducing heat soak.

The use of an electric pump removes the need for a thermostat, and might also help in terms of overall car packaging, since it can be mounted off the engine. Thus the use of an electric coolant pump needs to be considered as part of the whole cooling strategy, and in many applications it clearly has a lot to offer.

Fig. 1 - Mechanical coolant pump

Written by John Coxon

Split evenly between heat in the exhaust gas and that delivered up to the cooling water jacket, that means for a 450 kW race engine somewhere in the region of 500 kW of heat is rejected to the coolant - that's roughly 160 three-bar electric fires, as my old thermodynamics lecturer used to say! And all this has to be dissipated into the surrounding air in as efficient a way as possible.

But in designing a cooling system there are certain limitations. To start off with, the temperature difference across an engine when running at full load from the inlet to the coolant pump to the outlet of the cylinder head should be no greater than 3-5 C. Anything greater than that risks unacceptable thermal gradients across and through the unit, leading to extra strains as well as problems with combustion. The coolant flow rate through the engine should be such as to maintain this. And if the temperature rise across the engine is 3-5 C then by the conservation of energy and the fact that the coolant flow rate is the same for both engine and radiator, the temperature fall across the cooling system will also be around 3-5 C. The role of the radiator therefore is to dissipate all this heat in the most efficient way with the minimum mount of weight.

As an efficient conductor of heat, copper has few equals, but its specific density of 9 g/cm3 make it too heavy for most automotive uses. For modern vehicle radiators, especially those used for motorsport, aluminium (density 2.7 g/cm3) is the only option, while anything culled from an OE manufacturer's parts bin will almost certainly have plastic end caps clamped at either side.

Within the core of the radiator there are two types of construction. The first and most common for automotive use is the so-called tube and fin arrangement. Consisting of thin (2 mm) oval-shaped tubes running between inlet and outlet end caps, these tubes are separated by thin corrugated metal sheeting. Providing hugely increased surface area while at the same time producing a radiator core that is mechanically stiff, these are produced in standard core sizes of 22, 32, 42, 55, 66 and even 86 mm widths with fin densities of between 14 and 24 fins per inch.

Spaced at anything between 7 and 10 mm apart, the size (cross-sectional area) of the radiator matrix depends on the vehicle aerodynamics and the allowable drag induced, while the thickness of the core is a function of the heat to be dissipated. The greater the cross-sectional area of the matrix, the better the heat dissipation but at the expense of aerodynamic efficiency. And the thicker the core, the less efficient the heat transfer and the heavier the solution. As in everything engineering today, specifying a radiator design is all down to the compromise between cross-sectional area and core thickness.

The other core matrix design, and the one most used in high-level motorsport because of its design flexibility, is the plate and bar design. Presenting many advantages over the tube and fin approach, this radiator design can not only be externally baffled but internally baffled as well to increase the wetted area and improve the heat transfer coefficient between coolant and the radiator material. Typically two or three times more expensive than the fin and tube design, for performance at minimum weight they are however the racer's first choice.

Fin and tube, bar and plate: in the process of designing your cooling system, choosing the core of your radiator is only the first of many decisions to be made.

Fig. 1 - Typical fin and tube core

Written by John Coxon