Cooling the NSX: Fixing a Rare OEM Oversight

[Madhatter10-6]

I recently picked up a 2005 turbocharged NSX, and I’ll be updating my build thread about it soon. When I got the car, it had several issues that prompted me to deep dive its subsystems. For context, I’ve owned a 1991 turbo NSX for about a decade, and it’s been trouble-free, so I never had to learn much about its inner workings. I've owned and worked on 1990s turbo imports—mostly 300ZXs and MKIV Supras—where I’ve done everything from tuning and engine swaps to developing custom parts. The 2005 NSX presented a new challenge: erratic and high engine temperatures caused by a cooling system that didn’t function as expected and plumbing that didn't make sense.

Understanding the OEM Cooling System
For those unfamiliar with the NSX’s stock cooling setup, here’s my grasp of it (please correct if I am wrong on this), split into two parts:

1. Fluid Flow Path: Water pump → Radiator → Thermostat → Engine
2. Electrical Control:
   • The engine coolant temperature sensor measures pre-radiator fluid temps in the block. This data feeds the Fan Control Unit (FCU) to activate the low-speed fan setting.
   • A second, binary temperature sensor in the thermostat triggers only when the fluid hits ~195°F, disabling the low speed circuit and activating the high-speed fan circuit.

This design surprised me because it feels like a rare misstep. Having built many cars over the years, I’m consistently impressed by how well thought-out the NSX platform is. My 300ZX has its share of quirks, and even the MKIV Supra has shortcomings in stock form, but this cooling setup stood out as a genuine flaw.

The Problem
The issue lies in the thermostat’s placement. It reacts to post-radiator, cooled fluid rather than the actual temperature of the fluid in the engine. This creates a disconnect: the better your radiator and fans perform, the cooler the fluid reaching the thermostat, and the less it opens. As a result, the engine stays hotter than desired, with insufficient coolant flow to stabilize temperatures. This also leads to erratic temperature swings. After reading threads about overheating NSXs needing vented hoods, it clicked—the radiator’s cooling capacity isn’t the problem; the thermostat’s location is. For forced-induction builds like mine with upgraded cooling, this flaw becomes even more pronounced.

My Solution
Fortunately, fixing this is straightforward. I tackled it in two parts:

New fluid flow path: Water pump → Thermostat → Radiator→ Engine

1. Fluid Flow Fix:
   • Hollow out the OEM thermostat to allow constant coolant flow. (Later I may CNC a thin ring to keep proper fitment of the OEM gasket)
   • Install an inline thermostat before the radiator. This lets the engine’s actual temperature dictate coolant flow—how most cars are designed.
   • I will also be adding a dual-pass radiator (in at the top, out at the bottom) for optimal efficiency compared to OEM's in at the bottom.
2. Electrical Fix:
   • For stock setups: Swap the low-speed fan output (Pin 12, GRN/BLK) with the high-speed fan output (Pin 4, BLU/RED) in the FCU. This should align fan speed with engine needs better (I haven’t tested this myself since I’m running a Haltech ECU).
   • For standalone ECU users like me: I’m triggering fan states and speeds based on the engine’s TW sensor, bypassing the stock logic entirely.

Why This Matters
Before these changes, my Haltech triggered the fan (in high mode) at 185°F based on the TW sensor. But even at that point, the thermostat housing was cold to the touch—the radiator was cooling the minimal flow so effectively that the thermostat barely opened. This left the engine hot while the system struggled to catch up. With the new setup—especially for forced-induction NSXs—I expect much more stable and effective cooling.

Closing Thoughts
I hope this helps anyone facing similar issues. I’m open to feedback or counterarguments if others have different experiences or solutions. Coming at the NSX from a fresh perspective this is just what makes the most sense to me....I may learn later this isn't optimal and have to revert back to OEM routing. But for now, this approach is working and feels like a practical fix to a surprising oversight in an otherwise brilliantly setup car.

I found these threads helpful during this process: LINK LINK

 

[JohnWayne]

Very interesting..any reason you are not using blue Honda coolant ?

 

[Madhatter10-6]

Very interesting..any reason you are not using blue Honda coolant ?

Green is what was in it when I opened the system up so that’s what I put back in until I am done on the Dyno. After that I will probably flush it all and make the switch to blue.

 

[blaketjones]

Paging @Kaz-kzukNA1, very curious to hear his thoughts on this.

 

[neuronbob]

I posted a link to this thread in the FB NSX Owners Group to get more comments.

 

[sca037]

Interesting thread, thanks for posting your findings here!

 

[SWFL_NSX]

For context, I’ve owned a 1991 turbo NSX for about a decade, and it’s been trouble-free, so I never had to learn much about its inner workings. [snip] The 2005 NSX presented a new challenge: erratic and high engine temperatures caused by a cooling system that didn’t function as expected and plumbing that didn't make sense.

I love enthusiast modifications of these cars, but am not sure the cooling system, particularly the location of the thermostat, is the issue with your '05.

Before re-engineering the cooling system, it might make sense to post what issues you're having so others can check your troubleshooting.

========================
Late edit-I'm not an expert, but are you sure your understanding of how the existing system works is correct? I always assumed the thermostat was between the engine and the radiator, but honestly never bothered tracing the pipes.

 

[nis350]

Green is what was in it when I opened the system up so that’s what I put back in until I am done on the Dyno. After that I will probably flush it all and make the switch to blue.

why do you think Honda design the cooling system differently? I assumed there's a good reason?

 

[MotorMouth93]

I do not think the analysis of this problem is quite right. I have a NSX water manifold here on my desk and the passages in it do not align with what you describe.

This is the flow through the manifold. Red is the hot water coming from the cylinder heads, from there it is either recirculated (yellow) or goes to the radiator (red). At first glance, it does seem like the thermostat sits in the flow of cold water.



But if we look more closely, we can see that recirculated hot water is pushed directly over the thermostats wax element, to the point that very little cool water will ever come into contact with it. That hole there in the middle will always have hot water come through it in some quantity which is what is largely controlling the thermostat. The hotter the water coming out of the heads, the more the thermostat will open. (and the less hot water will be recirculated)

There is going to be some intentional mixing of hot/cold water here and that is necessary for the thermostat to properly regulate.



Your solution also has a few issues. The first of which is that you are using the ECUs temp sensor (which monitors the hot coolant at the very end of the cooling loop before water leaves the head) to control the radiator fan. This makes no sense, because the actual operating temp for the C30 is about 85C which is 185F, which is where your fan comes on. The radiator fan needs to be triggered based on the radiator outlet temp, which is why the factory fan temp sensor was located on the thermostat housing directly in the return coolant path.

The 2nd issue is when you deleted your thermostat you also removed the metering device that controls how much coolant is recirculated. Coolant will take the past of least resistance, and without the thermostat there to control the amount of hot coolant that is immediately recirculated you just introduced a massive and potentially dangerous unknown to your cooling system as well as likely reduced its maximum cooling capacity.

My recommendation would be to put the thermostat back where you found it (well, a new one, since there's a chance the old one was bad and it's now destroyed), and fix your cooling fan activation issue, since that seems to be what started all of this. As far as the radiator goes, if you can get your hands on a MASIV setup that is the best option, otherwise double or triple pass converted Koyo radiators seem to be good.

The fluid flow path you mention is also backwardsish, the water pump pushes coolant into the block, where it then flows around the cylinders and up to the heads, and then out into the 2 legs of the water manifold. So the correct flow is:

Water pump -> engine -> thermostat/radiator -> water pump again

As far as blue vs green coolant go, both are fine as long as you change them at proper service intervals. IIRC the Honda blue stuff is good for 4 years while the parts store green stuff is 2 years.

 

[Honcho]

I really hate to post this, especially considering how well thought out your post is....but you've got the cooling loop wrong.

"Cold Loop"
On our NSX, the cooling loop starts at the water pump, which pumps cool water directly into the engine block. This water picks up heat from the cylinders, goes through the heads and exits via the two curved aluminum passages on the water distribution housing. When the thermostat is closed, most of this hot water is re-directed to the water pump via the coolant return pipe that sits on top of the engine block. This process ensures the engine will warm up quickly. You can see the path in the diagram below.



"Hot Loop"
If you examine the thermostat both in the diagram and inside the housing, you'll see that the thermowax pellet sits on the "hot" side of the housing and is bathed in the hot coolant that is coming off the heads/block. Once this coolant reaches 78C, the wax will start to (1) move the bypass plate away from the housing and (2) open the plunger on the cold side. By 90C the thermostat is fully open (10mm) and the radiator coolant loop is functioning. Instead of re-circulating into the engine block, the hot coolant is sent to the radiator. Returning cool water from the radiator enters the water distribution manifold via the thermo housing and is returned to the water pump via the return line on top of the engine block. As this cool water flows over the thermostat, the wax pellet will start to close until the temp reaches 78C again and then the whole cycle starts over. The water temperature in a normally-functioning NSX engine will fluctuate between 78C and 90C regardless of external conditions. See below.



If the water temperature goes over 90C, the thermostat will remain fully open and now you are relying on the Kcal/hour heat exchange capacity of the radiator to manage the water temperature of the system. If the heat in the coolant exceeds the cooling capacity of the radiator, the water temperature will continue to rise unless you either increase the cooling capacity of the radiator or reduce engine heat.

By hollowing out the thermostat you are deliberately creating a condition similar to a stuck thermostat, which on the NSX is always stuck open. This exposes the cold water to the radiator cooling loop immediately and you will find that the car warms up much more slowly and appears to have "better" cooling. However, eventually you will reach the thermal exchange capacity of the radiator and the water temp will continue to rise as along as engine heat exceeds capacity. It may also be that by increasing the opening size of the thermostat, you're increasing the flow rate of the coolant, which in theory might slightly improve the cooling capacity of the system. With the standard thermostat, the plunger opening is a relatively small "donut" shape and the return water is slowed when passing through the opening. Now with a big hole there, there is little restriction and more water can pass through the cooling loop in the same amount of time. However, you removed the thermostat and now the engine is not able to self-regulate its temperature. Instead, it will always go the equilibrium state of the radiator capacity. You may encounter issues with ring seal, emissions and AFR due to a cold engine block. Same with oil temp- you'll have cold oil in the system for a long time. Thus, I recommend returning the thermostat function to the system.

Virtually all NSX turbo systems use a significantly larger dual or triple pass radiator to drastically increase the Kcal/hour heat exchange capacity of the cooling system. The benefit of running this solution is that when the car is driven normally, the thermostat regulates the engine temp. But when you get into long periods of boost (like at the track), the big radiator takes over and keeps water temps down.

One potential improvement could be to keep your hollowed OEM thermostat for the flow benefit and fit a large-opening aftermarket thermostat into the "hot" feed side line of the radiator. This way, you maximize the coolant flow rate of the water pump, but keep the vital thermostat function for engine health.

 

[Madhatter10-6]

I really hate to post this, especially considering how well thought out your post is....but you've got the cooling loop wrong.

That’s exactly how I understood the loop. Sorry if my explanation wasn’t clear. Your diagram represented it much better than I could have described. But we’re 100% on the same page about how the coolant flows. I should have started with a diagram.

One potential improvement could be to keep your hollowed OEM thermostat for the flow benefit and fit a large-opening aftermarket thermostat into the "hot" feed side line of the radiator. This way, you maximize the coolant flow rate of the water pump, but keep the vital thermostat function for engine health.

^^^^^

I really hate to post this, especially considering how well thought out your response to my post was...but this is 100% what I said I was doing

My Solution
Fortunately, fixing this is straightforward. I tackled it in two parts:

New fluid flow path: Water pump → Thermostat → Radiator→ Engine

1. Fluid Flow Fix:
   • Hollow out the OEM thermostat to allow constant coolant flow. (Later I may CNC a thin ring to keep proper fitment of the OEM gasket)
   • Install an inline thermostat before the radiator. This lets the engine’s actual temperature dictate coolant flow—how most cars are designed.
   • I will also be adding a dual-pass radiator (in at the top, out at the bottom) for optimal efficiency compared to OEM's in at the bottom.

Sometimes, tone can get lost in posts like this. I’m not trying to be argumentative or take anything away from other people’s setups—I just see this as a poorly designed system.

Over the years, I’ve noticed a strange tendency in car communities to have an almost religious attachment to OEM. When I first started running the IACV differently than OEM on 300ZXs with a Haltech, there was pushback. When we first started using VVTi heads on the 2J, the biggest complaint was wiring complexity (which was just one wire) and tuning difficulty. Now, it’s common in well-optimized street setups. People tend to get attached to one way of thinking. Same thing we when redesigned the shifting system for the V160.

That said, as I clearly stated above, I’m not sure I’m correct on this and genuinely welcome thoughtful feedback.

Also, why has no one addressed the OEM feeding hot side of the radiator into the bottom?

 

[MotorMouth93]

That’s exactly how I understood the loop. Sorry if my explanation wasn’t clear. Your diagram represented it much better than I could have described. But we’re 100% on the same page about how the coolant flows. I should have started with a diagram.

If that's what you understood then why did you write this objectively incorrect statement?

Fluid Flow Path: Water pump → Radiator → Thermostat → Engine

It doesn't seem like you actually want feedback or discussion if all you can do is double down and say "well actually I was right even though what I wrote was wrong."

Also, why has no one addressed the OEM feeding hot side of the radiator into the bottom?

Because it really doesn't matter which side you pump the coolant into. Some radiators have the tanks on the side for example. If this bugs you it would be pretty easy to make custom hoses to "fix" it though, and I'd be interested to see before/after testing showing it makes a difference or not.

The bottom line is you really need to get your system working properly and understand how it works before you reach the conclusion that it has a design flaw, and your system was not working properly and you did not have a proper understanding of how it worked. I'm about the last person to be religiously attached to OEM, I will modify anything and everything if I think it could make it 2% better, but so far I haven't seen that in this thread.

 

[Honcho]

Understanding the OEM Cooling System
For those unfamiliar with the NSX’s stock cooling setup, here’s my grasp of it (please correct if I am wrong on this), split into two parts:

1. Fluid Flow Path: Water pump → Radiator → Thermostat → Engine

The Problem
The issue lies in the thermostat’s placement. It reacts to post-radiator, cooled fluid rather than the actual temperature of the fluid in the engine. This creates a disconnect: the better your radiator and fans perform, the cooler the fluid reaching the thermostat, and the less it opens. As a result, the engine stays hotter than desired, with insufficient coolant flow to stabilize temperatures. This also leads to erratic temperature swings. After reading threads about overheating NSXs needing vented hoods, it clicked—the radiator’s cooling capacity isn’t the problem; the thermostat’s location is. For forced-induction builds like mine with upgraded cooling, this flaw becomes even more pronounced.

No offense taken at all, but I guess my concern is the above. First, the coolant path as stated is wrong. The OEM path is: Water pump --> thermostat --> radiator --> engine.

Next, your description of the "problem" is not correct. The thermowax pellet on the thermostat is on the "hot" side, so it is reacting to the hottest possible coolant coming directly off the engine pre-radiator. I tried to explain with those diagrams, but maybe this makes more sense. There is nothing wrong with the location of the thermostat. I hope that is more clear!

 

[nis350]

MotorMouth93 & Honcho, you guys really know the NSX

 

[fastfwd]

why has no one addressed the OEM feeding hot side of the radiator into the bottom?

That arrangement -- water flowing UP through the radiator -- ensures that all the water will flow through all the radiator all the time, no matter what might happen to the passages within the radiator.

With water flowing DOWN through the radiator, internal obstructions or damage -- clogs or air/vapor bubbles, or holes where they shouldn't be -- can allow water to bypass most of the radiator and just flow directly through the short-circuit path of least resistance between inlet and outlet.

 

[Valhalla]

Running a turbo system, you are adding in more heat than what the car was designed for. I am running two (2) turbos and track the car in Arizona. I still run the OEM cooling system but had to make some changes as the cooling system couldn't keep when ambient temperatures were over 110F. Here are four (4) things you can do.

1) Get a hood with a duct in it. They are made by several manufacturers. It will make a huge difference by itself.
2) Seal all openings on the high pressure (front side) of the radiator. All air entering your radiator shroud will be forced to enter the radiator, maximizing that air.
3) Get a MASIV radiator, if you can find one. They work extremely well but be sure to add, or make, some screening to protect it from road debris. It's core is brass and copper (that is why it works so well) but is prone to leaks. But they can be repaired by any radiator shop.
4) I added twin SPAL fans which were able to increase airflow thru the radiator by 3 times. They still use the OEM circuit for activation. Problem solved.

Good luck.

 

[Honcho]

Running a turbo system, you are adding in more heat than what the car was designed for. I am running two (2) turbos and track the car in Arizona. I still run the OEM cooling system but had to make some changes as the cooling system couldn't keep when ambient temperatures were over 110F. Here are four (4) things you can do.

1) Get a hood with a duct in it. They are made by several manufacturers. It will make a huge difference by itself.
2) Seal all openings on the high pressure (front side) of the radiator. All air entering your radiator shroud will be forced to enter the radiator, maximizing that air.
3) Get a MASIV radiator, if you can find one. They work extremely well but be sure to add, or make, some screening to protect it from road debris. It's core is brass and copper (that is why it works so well) but is prone to leaks. But they can be repaired by any radiator shop.
4) I added twin SPAL fans which were able to increase airflow thru the radiator by 3 times. They still use the OEM circuit for activation. Problem solved.

Good luck.

This 100% is the solution for the turbo NSX. Not that I expected any less from @Valhalla! It's not an accident that this setup is essentially what all NSX race cars do and have done for 30 years. If you can't find a MASIV, Ron Davis, PWR and Koyo all offer some beefy dual or triple pass radiators that are more than up to the task. I think Shad still offers them too.

That arrangement -- water flowing UP through the radiator -- ensures that all the water will flow through all the radiator all the time, no matter what might happen to the passages within the radiator.

With water flowing DOWN through the radiator, internal obstructions or damage -- clogs or air/vapor bubbles, or holes where they shouldn't be -- can allow water to bypass most of the radiator and just flow directly through the short-circuit path of least resistance between inlet and outlet.

You beat me to it and I agree 100%. I was also going to mention that natural convection may also play a role here. Hot water rises and cold water falls, so by putting the hot coolant in the bottom of the rad, the pressurized upflow will be assisted by the natural convection of the water, similar to how natural convection coolant loops work in many Gen III and III+ pressurized water reactors.

 

[Madhatter10-6]

View attachment Dyno Run.MOV



I spent the day on my dyno dialing in the NSX. No catastrophic failure, I didn’t get the cancer, no dangerous conditions ruining the car—just 600hp on a Dynojet. The car wanted a lot more, but I got nervous about head lift and wanted to see how this power level performed on the street. 600hp in an NSX is a handful. Sidenote: those of you pushing higher on street cars are wild, lol.

And this is still with the radiator plumbed backward. Speaking of which—if you're arguing against top-to-bottom radiator flow, take a look at almost every production car, especially race cars. This design isn’t just a convention; it’s based on physics and practical engineering. Challenging this concept is a bold stance and requires some serious mental gymnastics to justify as a superior method.

When hot coolant enters the bottom side of a radiator instead of the top, it disrupts the natural convection and efficiency of the cooling system. Radiators are designed to use gravity and thermal dynamics—hot fluid rises, and cooler fluid sinks. If hot coolant enters at the bottom, it forces the cooling process to work against this principle, potentially leading to uneven heat dissipation. The hottest coolant may not get enough time to transfer heat properly before being recirculated, reducing overall cooling efficiency. This can also cause air pockets to form in the system, leading to hot spots and potential overheating. Additionally, since most thermostats and temperature sensors are positioned to read coolant leaving the engine, improper flow direction can cause inaccurate temperature readings, leading to poor thermostat regulation. In extreme cases, this incorrect flow can create localized boiling within the engine, leading to cavitation, coolant breakdown, and long-term engine damage.

Are there times that hot in bottom is done? Sure. Does it work. Yes. But for most cars and race applications, top-in is simpler, more reliable, and thermodynamically sound.

The new NSX (NC1) is plumbed in this top down way…is it wrong?

What I’m suggesting isn’t some wild experiment—it’s just how most cars already handle cooling. The NSX’s unique setup is what made me notice it in the first place that it was plumbed oddly. The diagram Honcho posted is what I should have included in my original post to clear up the confusion, so I’m using it here to clarify:

• Point A is my proposed new thermostat location.
• Point B should remain open with my modified gutted housing. As Honcho pointed out, making this wider than the OEM thermostat could even have additional flow benefits.
• Fan trigger should be based on fluid exiting the block…not post cooled in the thermostat housing. The low fan speed temp sensor is what should be the fan reference for the entire system.

Why This Is Better:

1. OEM design creates the highest temp gradient in the system, forcing the coldest fluid to mix with the hottest at the thermostat, which acts as the gate at this turbulent point. This results in inefficient thermostat control, with slow response times and poor temperature regulation.
2. OEM places the high fan speed trigger at the coldest point in the system, which doesn’t make sense. Triggering the fan based on radiator-cooled coolant leads to over-cooling during engine warm-up or under load, reducing overall system efficiency.
3. OEM design results in the least uniform temperature distribution across the system. My new approach ensures the thermostat sees consistent engine-out temperatures, resulting in smoother temperature transitions that help prevent thermal stress, improving engine longevity.
4. OEM has more restricted flow due to less efficient routing. By using a high-flow thermostat in my new setup, coolant flow is improved, which increases the system's cooling capacity—especially crucial when pushing more power or under high-load conditions.
5. My setup makes servicing the thermostat easier. While not a huge deal, having direct access without disassembling the cooling system or dealing with complex routing saves time and reduces the potential for mistakes during maintenance.
6. This is simpler—two temperature reference sensors instead of three. Fewer sensors means fewer failure points, less complexity, and more reliable readings, which contributes to overall system reliability.
7. The OEM radiator is plumbed backward! The reverse flow doesn’t optimize the cooling process and causes inefficient temperature dissipation. My setup follows the natural top-to-bottom coolant flow, improving heat transfer and cooling efficiency.

At the end of the day, this isn’t some radical idea, I am not suggesting to fill up your radiator with bananas and only drive backwards—it’s just applying fundamental cooling principles that are standard on almost every vehicle.

I’m not proposing that everyone should rush to make this modification. The OEM cooling system is perfectly capable of maintaining proper coolant temperatures, and this isn’t a necessary upgrade. However, for the reasons mentioned above, I believe this setup offers improvements that are even more pronounced on FI standalone cars . What’s interesting is that I may be the only one who has run an NSX both ways—OEM and modified—and can confirm that, at least on my car, this works better. Yet, some may still argue against it from behind a keyboard.

 

[Miner]

Take it easy Francis.

Without being able to sit down and really read this entire thread, it sounds like you spent a good bit of time researching the cooling loop, but most of the others have 10-20 (or more) years working on the NSX and many of us are technical enough to have good thoughts on the subject. A long time ago, “Q” would get my head excited, today keeping my battery charged is my speed.

Perhaps it is ok to agree to disagree, and move on.

 

[Madhatter10-6]

On a somewhat related note, I think this radiator might be a good upgrade for the project. It's very close to the spec of the NSX radiator but is thicker, dual-pass, and comes with fans and a shroud, all at a lower cost. I'll cut off the fill neck, weld it shut, add pegs to use the OEM attachment points, and likely weld on AN fittings as well.

55mm Universal Performance Radiator Module (PWR Open 28" x 16") W/ Dual 12" SPAL Fans, 1-1/2" Inlet and Outlet

Please allow 4 weeks to manufacture if the part is available on backorder.

www.pwrna.com

Anyone tried this?

 

[Honcho]

On a somewhat related note, I think this radiator might be a good upgrade for the project. It's very close to the spec of the NSX radiator but is thicker, dual-pass, and comes with fans and a shroud, all at a lower cost. I'll cut off the fill neck, weld it shut, add pegs to use the OEM attachment points, and likely weld on AN fittings as well.

55mm Universal Performance Radiator Module (PWR Open 28" x 16") W/ Dual 12" SPAL Fans, 1-1/2" Inlet and Outlet

Please allow 4 weeks to manufacture if the part is available on backorder.

www.pwrna.com

Anyone tried this?

PWR used to offer a NSX-specific fitment dual pass racing rad. They may still be able to fab one up for you. It was the hotness for the turbo/SC track crowd in the early 2000's.

 

[Madhatter10-6]

PWR used to offer a NSX-specific fitment dual pass racing rad. They may still be able to fab one up for you. It was the hotness for the turbo/SC track crowd in the early 2000's.

I looked into some used ones, but the prices seemed too high. I called PWR, and they were very helpful and pleasant to talk with. They are willing to either make one of the older models for me or modify it however I want. However, it’s not cheap. Their company policy is that any custom work requires full blueprints and designs, with minimum fees. For example, adding two $15 AN fittings and omitting the filler neck would cost almost $1,200 more than the one I listed. Which is why I will prob just buy that one and mod it myself. My buddy is a dealer for them and with his hookup I think I can be all in around 1K for the one in the link above.

 

[Zoom]

When hot coolant enters the bottom side of a radiator instead of the top, it disrupts the natural convection and efficiency of the cooling system. Radiators are designed to use gravity and thermal dynamics—hot fluid rises, and cooler fluid sinks.

Engineer here. If there's no air in your system (as designed) the flow direction doesnt matter here, as the pump will overcome the natural convection by so much it's not even measurable. It would matter in a fully passive system but it's not (flow is forced by the pump). Heat transfer through the radiator is purely a factor of flow and surface area, flow direction doesnt matter at all.
As for racing cars other considerations like how quick and easy you can change parts are just as important (sometimes more, e.g. for endurance racers), and there placement of things matter.

 

[Old Guy]

Yeah, I am tending to agree with Zoom that you are over selling the natural convection bit. There are / were some engine designs that relied on natural convection; but, those designs are pretty far south of the 21st century. I would also observe that there are a number of modern cars that use cross flow radiators where convection inside the heat transfer tubes has absolutely no effect. Heat transfer in modern engines relies on nucleate boiling in the engine and a strong flow of coolant to scrub those bubbles off the cooling passage surfaces. I expect that strong forced flow pretty much negates any convection effect in the radiator. If the NSX radiator is internally symmetrical on the inlet and outlet sides you can have at it if you feel the need to express yourself. However, you might want to make sure that there is nothing in the radiator internals that would impair the operation of the radiator if you reverse the direction of flow. The only down side is that it may look odd and you have complicated getting hoses in the future.

I expect that the radiator inlet and outlet in a forced flow system is more a function of packaging considerations. There may have been some other design consideration that dictated the radiator connections in the NSX.

As an observation, if you are going to use it for more than a few seconds at a time 600 hp kind of puts you way outside the design window for the existing NSX heat transfer surfaces. As noted, ducting to increase air flow through the radiator would have been a more obvious first step for me. If you are going to experiment with radiators with more cooling capacity I might be inclined to think about electrically operated coolant pump(s) because turbocharged engines can produce more power and heat at lower RPM than an NA engine. You might also want to consider supplementary oil cooling with all the potential nasty negatives (oops, I meant complications) that brings with it.

 

[MotorMouth93]

To echo what @Zoom and @Old Guy said, thermodynamics actually have very little to do with radiator orientation, but if you don't actually have any real knowledge of thermodynamics it might sound good to put the hot hose on top simply because "heat rises, duh, silly engineers!", but in the real world that effect has a negligible effect on cooling system efficiency where flow is determined by the pump. In most cases, it's just way more convenient to have the hot water go in the top because that's closer to where the cylinder heads are with a standard front engine layout, and its more convenient to pull from lower down as it helps ensure the water pump won't try to suck air.

Reversing the flow might make you feel good, but at the end of the day you aren't really accomplishing anything other than a smug sense of having "outsmarted" someone.