Radiator Design

[Sam Zimmerman]

I am interested in the different ways teams go about designing their cooling systems. How detailed is your math model? How much testing do you do? Do you just turn it over to somebody else? SWAG method?

My current work is with two phase fluid flow, heat transfer, and thermodynamics. I didn't have anything to do with the cooling system on our FSAE cars, so I am just curious if many of you are doing the same basic heat exchanger analysis I am doing - minus the two phase part - and how many are forging forward without developing their own models.

[Adam Vaughan]

I sized ours using the e-NTU method with our Q approximated to ~30% of the total fuel energy under an assumed 1/3rd duty cycle (that had been discussed in the radiator thread). I also looked at historical temperature data to get an idea of the worst case scenario would be. Additionally, our CFD car aero and radiator runs were pretty useful in characterizing the flow and convective coefficient that the radiator would experience.

We put together a simple thermometer and flow meter setup for our engine dyno, but the essentially steady state data were still sort of an approximation of what the actual system parameters would be. I'm hoping that the awesome hardware in the loop, CarSim based engine test setup that a team mate is working on will be able to better quantify the racing duty cycle and associated "real world" Q.

All in all, for a moving car, I think it's really difficult to optimize without real-world test data... something that's hard for us to come by without a place to drive in NYC.

Ultimately, I conjured up a radiator that was bigger than what our radiator supplier (PWR) recommended, so I undersized it slightly, but not down to what they suggested. Just looking at the radiator though, a voice in the back of my head says it may be a bit too big for the car, but more testing remains.

Basically, without real data our method is close to, but not exactly, SWAG... Incidentally, if you do go the e-NTU route, try to find ways to say Cmin as often as you can... ;-)

[Sam Zimmerman]

The heat load is always the big question. http://fsae.com/groupee_common/emoticons/icon_biggrin.gif

It would be helpful if you had access to one of these (http://www.keithley.com/products/dmm/integrasystems/?mn=2700), one of these, (http://www.keithley.com/products/dmm/integramodules/?mn=7700) one of these, (http://www.jlcinternational.com/liquid_turbine_flowmeters_pvdf.htm) a couple of these, (http://www.omega.com/ppt/pptsc.asp?ref=TC-NPT) and of course one of these. (http://www.dell.com/)

Then you are designing to real numbers gathered from running an old car on a track. The Kays and London part becomes easy after that. I know we wouldn't have flailed on our HXs if we would have had real data rather than using the 30% rule and some steady-state data from the dyno.

[Maverik]

I did three years of work into a fun little tool my team should benefit from now on. It's all derived from real world data, Kays and London methods, and a little bit of in-depth creativity. At one point I had a pretty cool model to validate analytic heat load numbers coming out of the cylinders and going into oil/water. Then using two phase theory to model the transfer into the water from the head. That hardest part by far was figuring out that pesky air side coefficient at the radiator...

[Mike Hart]

I have to say, from my experience last year, you can calculate stuff to your hearts content, but it will rarely do what you predict.

Best thing to do considering the amount of time you have is to do some rough sizing calcs to justify your choices, get an electric water pump or modify the mechanical one to shift more volume at lower revs and get a big properly sealed fan. Test, test and test some more and adjust ducts, fan speed etc to get a good balance and aim for temps around 100deg C with a pressure cap.

Don't what ever you do try and save weight unless you already have a perfectly working system. Just make something that works reliably first and then worry about making it smaller.

[Chadwick Huntington]

So, what would you guys recommend for a first year team? We don't have a previous car to test. So far, we've been using the e-ntu method and then heavily compensating for environmental anomalies. What do you guys think?

[Mikey Antonakakis]

This may be incredibly stupid of me, but the stock radiator should be more than enough, right?

[D Collins Jr]

This should work pretty well for you:
1. Acquire a stock radiator. One off a 600 cc bike should work pretty well.
2. Stick a high cfm fan on it, with a shroud for extra results.
3. Make sure that there is nothing blocking the ducting to the radiator (should only be a problem if you have wings or no opening in the sidepod/duct)
That should make your first or tenth year car not explode, aka overheat. First year, have no concern for making it too big and heavy. If I remember the entries right, you guys at Houston are going to Cali? It gets hot there. Fires can happen. make sure you're engine doesn't overheat. That's more important than saving three pounds.

[Wesley]

If you want a good starting point, look at a stock radiator, and find one with comparable fin density, tube size, material, and core area.

All the theory in the world won't replace real numbers. I can go out with two thermocouples and a flow meter (or calibrated RPM vs. flow curve,) and in 15 minutes know the actual motor heat rejection at any speed.

I would say do testing for basic heat rejection before you even bother with the heat exchanger theory. Not to mention, even if you can determine your "ideal" numbers with whatever model you devise, the chances of a radiator company building you whatever you design are slim. Stick to what you can buy.

[Chadwick Huntington]

yeh, thats the same response i got from our rad source as well. ima 1-up the size from the math numbers i got (add 2 inches in height).

good call on the entries! Funny thing is, that part of Cali only gets around 80F high in june. Which, i dont think is too bad. I wouldnt say its that much worse than texas =)

[Scott Borg]

Fontana is quite a bit further from the sea than Los Angeles or Orange County... I can guarantee you will routinely see above 90F during June, nearly 100F if it is a hot day. Google "weather history," the data is out there.

[Wesley]

Yeah, in 2007 we hit 100ºF I think, and that was just ambient! Track temps were ridiculous. I think our air temp sensor logged 115.

[damufunman]

I've been working on getting numbers for 2 different radiators, and there are huge discrepancies radiator UAs. Part of that might be from the engine only barely sputtering in the fall. But just got the new dyno built and running well, and the numbers I'm getting are pretty ridiculous. The effectiveness is different, depending on whether I use the hot or cold side numbers.