Measuring and Improving Cooling System Performance – Part 5: Fan and Outlet

Last time, we saw that drag from the heat exchanger varies with the velocity of the flow entering it and measured its loss coefficient. Before that, we measured total pressure losses across the grill and diffuser, as well as other parameters such as static pressure that told us how well these components worked.

 
Getting air into the cooling system is just one part of the story. Just as important is how we get air out, and what that air does once it has left the cooling system or engine bay.

This Fiat 500e has a large heat exchanger package (we learned why at the end of the last post) with a single, centered fan. Many cars have multiple fans behind the heat exchangers.

Fans: State 3 to State 4
 
The purpose of fans placed in front of (for example, the Tundra cooling system in the first post) or, more commonly, behind the heat exchangers is to increase mass flow through the cooling system. You can prove this by writing out an energy balance similar to what we did in Part 4, which shows that work input by the fans must, due to the conservation laws, increase the velocity through the core and thus, for fixed core area, the mass flow rate (and, as we saw in that post, increasing mass flow rate increases cooling capacity).
 
However, we do not want the fans to run by default because:
 
1) Fans must be driven, either mechanically or electrically, and the energy to spin them has to come from somewhere. If your car doesn't have a plug, it comes from the liquid fuel. If your car has a plug, it comes from an electrical outlet. There is no free lunch.
 
2) Increasing the mass flow rate through the cooling system increases its drag (despite the small thrust from the fan, a result of its increasing the momentum of the flow). Again, you can prove this for yourself by working through the continuity and momentum equations: larger ṁ means a larger streamtube capture area and larger u_core/u_∞. As we saw, a larger velocity ratio at the core increases losses and internal drag.
 
Most cars today have electric fans so that the fans can be switched on and off; they run only when required, usually when coolant temperatures reach some value determined by the manufacturer. In the past, fans were constantly driven by a pulley connected to the engine accessory drive (and represented in the engine schematic that still serves as the Malfunction Indicator Lamp [MIL] in most cars). More recently, some mechanical fan setups used clutches to reduce fan speed at low coolant temperatures, as on my truck:

The viscous clutch, circled in green, mechanically adjusts fan speed as the temperature of the fluid inside changes. However, this adjustment is limited and the fan is never "off."

To quantify the effect of the fans on your car, you can measure dynamic pressure behind them with and without the fans running (on some cars, it is possible to force the fans to run by turning on the air conditioning). You can find the increase in mass flow rate by,

If your vehicle has a mechanically driven fan, aftermarket electric fans are usually available and can be fitted. On some popular off-road models, like the Toyota 4Runner or Jeep Wrangler, model-specific kits are common due to the benefit of being able to turn the fan off for water crossings.
 
Outlet: State 4 to State e
 
On most cars today, there are effectively two "outlets" for cooling system airflow. Just behind the core, the fan housing is combined with a short nozzle (popularly referred to as a “shroud”):

The shroud on this 2026 Toyota bZ, seen here on the left side of the image, has an abrupt area change typical of modern cars. This will incur losses but it isn't something that's easily changed by home modifiers—so my advice is to leave it as is unless you're building a cooling system from scratch.

It is easy to measure this outlet area since it's well defined and takes just two measurements (of the fan inner and outer radius) to calculate. This gives the area ratio on my car:

But behind this we have an engine bay that also has vents and outlets. Most textbooks treat cooling system outflow at these engine bay vent locations, not the fan shroud. The downside of this is, these outlets have complicated geometry and are hard to measure, so there is no clearly calculable outlet area for cooling system airflow. In this model, the fan nozzle represents a restriction in the flow path, with certain parameters like mass flow rate determined by the engine bay outlets.
 
After giving this a lot of thought (and working through a bunch of mathematical models), I think the best way to look at it is a nozzle outlet (the fan shroud) with a pressure boundary at its exit that determines the flow parameters there. The pressure boundary condition can be changed by altering the size and location of engine bay air exits, and useful quantities related to mass flow rate and cooling system efficiency like q₄ can be directly measured behind the fan nozzle outlet.
 
If we measure q₄ (the difference between p₀₄ and p₄) behind the fan outlet and then measure again while we change the static pressure there by modifying engine bay venting, we can find the normalized change (i.e. a ratio or percentage) in mass flow rate as above. In this model, we will consider losses in the engine bay due to friction and separation around components like the block, oil pan, and transmission as interference drag rather than internal drag (these divisions are arbitrary, so we can classify them however we want. Just be careful, as one of my instructors put it, that we "don't count it twice, but don't count it none!"). Since most of us can't really do anything to change or streamline these components, so they represent fixed losses in the cooling system flow, I think this is the most practical way of approaching it. In just a minute, we'll look at the aspects of interference drag we can control like engine bay vent placement.

 
Theoretical Model
 
Hoerner derives a model equation for prediction of internal drag coefficient (referenced to core area not vehicle reference area! This is equation 14 in Chapter 9 if you have a copy) for a fully ducted cooling system, similar to what we have in our cars if we consider the "duct" as stretching from the grill inlet to the fan nozzle outlet:

ξ here refers not to the heat exchanger loss coefficient we measured in Part 4 but to the loss coefficient of the entire system. You can find this by the difference in freestream total pressure (given, again, by a probe on a pole at the front of the car) and total pressure at the fan nozzle exit. Then, the loss coefficient is

We'll revisit this in a couple of posts, when I modify the cooling system of my car for better performance. I made an assumption in the plot above that C_p,out = 0 i.e. static pressure at the cooling air outlet is the same as ambient static pressure. This, of course, does not have to be so—depending on how the static pressure in the engine bay varies with where you exhaust the cooling air.
 
Let's now assume ξ_system = 10. Varying C_p,out from +0.5 to -0.5 (while assuming outlet to core area ratio I found on my car) does this to the internal drag of the cooling system:

If we vent cooling air to an area of low static pressure, we can increase the momentum of the outflow (by entrainment of the flow) and potentially reduce internal drag of the cooling system (however, we have to be careful because, if outlet area is left unchanged, this will increase mass flow rate and thus drag—I'll explain more in a subsequent post). If you've read other posts on this site, especially recent ones (after I went to school for engineering and figured out all the ways I had been thinking incorrectly before), you should have an intuitive "feel" for the way momentum changes in the flow and drag are related. Venting cooling air axially (along the longitudinal axis of the car, so it's parallel to the direction of drag and thrust—roughly what we should have coming out of the fan nozzle) and increasing its momentum when you do means a small thrust component must be exerted on the car—so, overall drag is reduced. Referenced to the core area and considering only the internal surfaces of the duct, that small thrust shows up as what it is: negative drag.
 
Outlet Interference
 
But that isn't the whole story. We still have to be concerned about how the cooling air outflow from the engine bay interacts with the external flow.
 
First, if the cooling air outflow separates from the body surface or causes flow that would otherwise be attached to separate, it may increase drag. Further, if venting cooling air somewhere causes the external flow to stay attached where it otherwise would separate, this can reduce drag (and this beneficial interference can result in negative drag overall from cooling airflow, if the drag reduction from interference outweighs drag increase from internal momentum loss). So, ensure (check with tufts) that your cooling air stays attached wherever you vent it. If it separates at the trailing edge of an opening, round the edge until it stays attached.

This ZR1 hood vent blends nicely into the body surface.

Second, think about the direction of momentum in the cooling air outflow. If you vent cooling air entirely vertically, so it has no momentum in the axial (horizontal) direction, the force exerted on the airflow to reduce its axial momentum to 0 points forward and the reaction force on the car points backward: drag. On the other hand, if you vent cooling air entirely axially and at higher speed than the freestream, the force exerted on the airflow to increase its momentum must point backward and the reaction force on the car points forward: thrust. A good way to increase momentum in the flow is by venting cooling air to an area of negative C_P, as we saw above.
 
Third, be careful about the orientation of outlets and measure the change in C_P with and without the vent. If you vent air through an opening that can be defined by a plane normal (perpendicular) to the centerline axis of the car, as on the Viper below, the change in C_P could increase or decrease drag, depending on whether it gets lower (increased drag) or higher (decreased drag). In the case of the fan nozzle outlet, you can measure this change by temporarily blocking the exit flow with cardboard, coroplast, or similar—just be careful and watch your coolant temperature since this will reduce mass flow rate to 0.
 
Vents
 
Ideally, we would completely duct the cooling airflow out of the exchanger, past the fans (if applicable), and through a smooth nozzle to an exit on the body surface of the car, like the Corvette above. Adding a nozzle duct like this to your car if it doesn't have one already will likely be difficult or impossible. Neither of my cars leaves any room for such a modification.
 
So, this means the cooling system flow will incur losses due to the engine bay components as it flows around them, but it doesn't have to be the end of the world. Your car, if it was designed in the last thirty years, likely benefited from extensive CFD simulation to try and reduce losses around all these components. You may also be able to fit guide vanes or partial ducts inside the engine bay if space allows.
 
If you need ideas for vent locations, look to the plethora of options in use by OEMs:

Most current production cars that use fender side vents, like this Toyota 86 or the new Honda Prelude, exhaust air from the wheel housings—not from the engine bay.

Some older sports cars, however, have used engine bay vents on the side of the car to improve cooling system performance (and, currently, the Toyota GR Corolla). With vertical area normal to the longitudinal axis, these may have reduced or increased drag—depending on how the static pressure at the outlet changed.

Current production cars that exhaust engine bay air on the upper body do so typically using hood vents with integrated turning vanes, as on this GR Corolla (this is a good car to use for example vents because it has so many!). This can reduce cooling system drag compared to side vents that disrupt the contour of the body.

Another option is this, on the Hyundai Elantra GT racing car. Holes cut directly into the hood present almost no opening normal to the longitudinal axis of the car. However, since no provision is made to bend the flow backward, it is likely to have a lot of upward momentum here—increasing drag but decreasing lift.

A number of cars exhaust engine bay air to the front wheel housings through central vents, as on this 2026 Dodge Charger EV (top) and 2010 Toyota Tacoma (bottom)—we'll see why in just a moment.

This older Honda CR-V vents the engine bay to vertical slots along the front of the wheel arch. These may very well reduce drag by both "bleeding" cooling air through narrow slots at high speed and increasing the momentum of the flow around the bumper cover by entrainment, similar to "air curtain" ducts.

If you want to go really exotic, cooling air can be exhausted into the wake (although C_P in the wake is not as low as it is elsewhere on the vehicle). This is trivial if the heat exchangers are also located at the back of the car, as on this Lexus LFA or the new Toyota GR GT. At least one concept car, the 1982 UNICAR research vehicle, exhausted a front-mounted cooling system at the rear of the vehicle, sending cooling air all the way back through the exhaust tunnel cf. Aerodynamics of Road Vehicles p. 240.

The GR Corolla illustrates an important point about cooling airflow: it is exit area that determines mass flow rate if the inlet is sized such that it does not constrict flow (and as we saw in Part 2, we want the inlet to be bigger than the captured streamtube). Notice the fairly small outlet here (bottom), exhausting just ahead of the front wheel opening, and larger inlet (top).

Testing
 
In the past (when I didn't really know what I was doing), I measured static gauge pressure across panels (i.e. on both sides of a panel) such as the hood to try and identify good vent locations. You don't need to do this; instead, as we saw above, find the area of lowest pressure referenced to freestream/ambient, or C_P. As a bonus, this is easier to measure since it doesn't require access to the backside of panels (just make sure there's nothing important behind them if you end up cutting there!).
 
As we've done several times already, first find freestream dynamic pressure. Then, measure the pressure difference between the location of interest and a static reference at the front of the car. Find C_P by,

Do this at all the locations you think might conveniently work to site a vent; your best candidates will be the ones with the lowest C_P.

Don’t forget the underbody; here, I’ve measured at the middle center and middle side of the engine undertray/skidplate:

You can also investigate wild ideas to see if they might work. For example, looking at my car and thinking about the Honda above, I realized it would be simple to cut a slot along the outside of the front wheel housing that might serve as a good vent in conjunction with the external duct I've already situated there, along with replacing the horizontal vents I added several years ago (again, when I didn't really know what I was doing) with vertical slats to direct air toward those outside slots (as a bonus, these opposing forces on either side of the car will cancel each other out, unlike the current vents which direct air downward):

Now I know that the lowest C_P out of all the locations tested is right there, which might explain why Honda engineers chose that site as a vent location on more than one model. Alternatively, measuring C_P at the existing wheel housing vent location with the openings taped over shows C_P = -0.32, not quite as low as on the outside edge of the wheel arch but still good. Another option could be simply to leave the vent but flip it, exhausting air upward instead of down. Directing outlet air upward means the reaction force on the car will point down—negative lift.
 
Next time, we'll see what insights can be gleaned from a simple cooling system physical model you can build yourself.

Previous: Measuring and Improving Cooling System Performance – Part 4: Heat Exchanger