Oil Coolers, kilowatts, and speed

[burgermeister]

Some interesting data on an oil cooler, which may be of general interest for the more techy types.

Background: I have been struggling with oil temperatures in my 3.6 (993) powered kit car. So I built a datalogger from an Arduino to log ground speed, oil & air temperature on the inlet and outlet sides of the cooler, and air speed in the cooler exit ductwork. I had an airspeed sensor for the front of the car as well, but the sensor went flaky and provided no data.

Temperatures are from thermistors, so quite accurate. Ductwork airspeed is a pitot tube and differential pressure transducer - decently accurate, but not at low airspeeds (<30MPH accuracy falls off quickly). I checked it in free air vs. GPS at 50-80MPH, and error is <10%, which I consider great.

The plan was to do this for the original cooler (7 x 22 x 2.5") and again for a custom replacement (7 x 22 x 4"). Unfortunately, a terminal oil leak forced a rebuild before I could get data on the original cooler - and while everything is drained, I put the biggger cooler in. So data is for the bigger cooler.

Grattan, low 50's temperature, no wind, and pushing the car fairly hard.

The cooler transferred an average of 21.3 KW of power to keep oil temperatures around 210degF.

For my setup, cooling power is closely correlated with GPS speed - I got about 0.3 KW/MPH. This suggests that oil cooling is greatly influenced by the average speed, and thus, the track being run. Not something I have seed discussed, but it does make inherent sense. Also this suggests running with Hoosiers improves oil cooling, as the average speed will be considerably higher than with street tires.

Also interesting was the difference in temperature across the cooler. Given the flow rates of a 993 oil pump (I think somewhere around 1 liter / minute / 100 RPM), oil cooled indeed!!

[GH85Carrera]

Wow, you win the geek of the week award.

Very cool, and I that is some very cool engineering. You knew going in the more power you make, the effect is more heat. Just a rule of physics that can't be ignored or changed. So keeping a 200 HP engine cool is easy, but stuffing 300+ HP in the same space takes lots more cooling.

I think it was Neil deGrasse Tyson that said if the power cells of Terminator were able to generate that much power it would require a radiator the size of building to cool the terminator. He would melt into a puddle of metals and plastic in short order without a lot of external cooling.

Very cool graphs and data logging.

[Mixed76]

21.3 KW x 1.34 KW/hp = 28.5 hp lost as heat.

Would be fun to see how much is being removed by the cylinder and head fins.

Sent from my Nokia 7.1 using Tapatalk

[Mr Beau]

Originally Posted by burgermeister

For my setup, cooling power is closely correlated with GPS speed - I got about 0.3 KW/MPH. This suggests that oil cooling is greatly influenced by the average speed, and thus, the track being run. Not something I have seed discussed, but it does make inherent sense. Also this suggests running with Hoosiers improves oil cooling, as the average speed will be considerably higher than with street tires.

Average time at WOT will also increase with Hoosiers so there will be more engine heat to be rejected via the oil...

[930cabman]

Does a Tesla S have these issues?

[burgermeister]

Originally Posted by Mr Beau

Average time at WOT will also increase with Hoosiers so there will be more engine heat to be rejected via the oil...

Why? I run street tires, and am mostly WOT from just before apex to braking zone. Though that clearly depends on available power - if it's Donohue's idealized car, you would just reach WOT at the end of the longest straight ..

A Tesla has to worry about cooling the battery and the motor. No personal experience, but I have heard that an S will only tolerate WOT for a rather short number of seconds before it starts self-protecting. Granted, that would be a pretty ludicrous speed for public byways. Same source says a Taycan will keep going at WOT until the battery is empty (15-ish minutes apparently). All just hearsay though - I have not seen any data.

[ClickClickBoom]

Originally Posted by 930cabman

Does a Tesla S have these issues?

No, but it looks and sounds like a Hyundai.....

[ClickClickBoom]

I would instrument the front and rear of the oil coolers to determine the pressure delta. I had a Mooney 231 come to me, the owner was having cooling issues. Every other shop said "just run with the cowl flaps open". The Mooney is close cowled, and after with the Teledyne Continental rep, I started closing every hole from the high pressure to low pressure side, about 10 hours of fab work, test flying and pulling inches of water at cruise. Turns out 1 square inch of openings between high and low pressure sides was Continentals limit. I got it to less than 1/4" square inch, took a lot of silicone baffle material and silicone sealant.
After we finished the plane would cruise at max rated power with the cowls closed at a very high ISA temp. The plane actually would climb with the cowl flaps partially closed at Vx and Vy.
I built a oil cooler shroud for the 2.7:
Oil Cooler Baffling!
Oil temperatures is a quest.

[ClickClickBoom]

Originally Posted by 930cabman

Does a Tesla S have these issues?

Funny, I was going to a memorial heading south on Hwy 5, a black Prius was "showing" me how fast they could pass me, I was doing 75-80ish. Up comes the Grapevine and I pushed the skinny pedal down slowly, wicking it up to a reasonable speed, the Prius got big in the windscreen, passed it and they thought some kind of toy car race was on. I could see them coming up fast, then the battery/motors slowed it down, and I just slowly pulled away.

[burgermeister]

The cooler is fully shrouded - no way to calculate cooling power without knowing air volume passing through the cooler!
The airspeed measurement location in the ductwork has maybe 10 or 15% more area than the front inlet - based on the cooler thickness, I think I'm doing pretty good for airflow.

The differential pressure sensor to airspeed relation is very nonlinear - while it shows a significant offset at 0 speed, this is only dither of 2 or 3 units of its resolution, and it changes based on temperature, humidity, the cheap chinese sensor's mood, who knows. The calculated airspeed at 50MPH changes by a few MPH if I rewrite the code to zero the sensor at 0 MPH - but this is a huge PITA. So even though it's way off at 0 MPH, it is pretty close at 50 MPH.

The red circle is where the flaky sensor is located - no data from that one. The duct sensor is in the cutout slot in the top of the RH duct ... the slot, sans sensor, is sort of visible in a couple of the pictures.