fe1rx adds Dual Oil Coolers

[fe1rx]

Not that the world needs another thread about adding dual oil coolers to the 1-series, but here goes …

Like many before me, my installation uses a pair of 25-row Setrab Series 6 heat exchangers, plumbed in series using 10 AN flex hose and the OE thermostat. Where it differs from any other installation I have seen is that my heat exchangers are mounted vertically, and I have incorporated actual inlet and outlet ducting. In all the off-the-shelf kits I have seen ducting seems to be very much an afterthought.

Mechanical Installation

The primary structural element securing oil coolers is a plate sandwiched between the bumper crash beam and the chassis. This method is very simple and easily carries the vertical and lateral loads.



The plate also serves for mounting an exit plenum that closely fits the fender liner. By mounting the cooler somewhat ahead of the fender liner, there is room for a plenum which allows for an orderly exit air flow.



The oil cooler is secured to the plate via vibration isolating grommets and aluminum channel sections. A stay connects to the bumper crash beam to react fore/aft loads. The resulting installation is very solid.



Plumbing

Plumbing is typical 10 AN flex hose with reusable AN and 22 mm Boss fittings. I have incorporated a tee in the lower interconnect line that allows for full drainage of the system (the branch being capped off at the moment but intended to permit connection of an Accusump down the line). At the oil filter housing I am using Hard Motorsports AN adapters (not impressed - I recommend spending more on any other brand).



Exit Plenum

Whereas the OE installation places the oil cooler hard up against the louver (and calls it a duct), my installation does actually provide a plenum/duct. I like the OE louver design because it protects the oil cooler from grit thrown up by the tires.



LH Exit Louver

Unfortunately BMW doesn’t make a comparable LH louver, so I made my own by trimming the flange off a RH louver.



Then I welded a mirror image aluminum frame.



The louver element was then bonded in place. The resulting LH installation is mirror symmetrical to the OE RH louver.



Design Principles

A basic design principle for ducting heat exchangers is to expand the inlet air (with a diffuser) to reduce its velocity and increase its pressure at the face of the heat exchanger, and then contract it to increase its velocity and decrease its pressure at the exit. A rule of thumb (and the one I followed) is to have an inlet area of 20% of the radiator core area, and an exit area of 25% of the radiator core area (conveniently the OE louvers provide this). For the inlet to work well, the expansion must be gradual enough not to result in flow separation in the duct.

Intake Duct

My inlet duct went through many iterations. Many models were printed on card stock, cut out, assembled and tested for flow characteristics. The final configuration was fabricated from TIG welded 0.050” 6061T6 aluminum.



I will follow up in a later post with some details of the testing process, and how the duct design ended up as it did. The ducts are identical on the LH and RH side of the car, which is somewhat convenient. I have previously described my brake cooling duct modification. As part of this installation I also built a new lower radiator/intercooler inlet duct. All three are visible in the following image.



I have had one track day since these modifications, and that was cold and damp, so not a good opportunity to see how effective the oil cooling is. That will have to wait for another day.

[Geran]

This is great looking… I believe people would be very interested in something like this.

[rackhenry]

fe1rx Nice job on the oil cooler mount. Looking at the oil cooler for my car, and you pretty much confirmed my thoughts on the AN adaptor you used. It just seems like a poor design.

[fe1rx]

Originally Posted by rackhenry

fe1rx Nice job on the oil cooler mount. Looking at the oil cooler for my car, and you pretty much confirmed my thoughts on the AN adaptor you used. It just seems like a poor design.

Thanks. I can't say the AN adapter design is bad. Quite a few manufacturers use the same basic design. As an example, ECS lists 4 different manufacturers for essentially the same part:

https://www.ecstuning.com/Search/Sit...tting_adapter/

I chose the Hard Motorsport version simply based on price.

My complaint is that the blue aluminum fittings on the Hard version are not built to standard AN/MS flared fitting tolerances and/or they are not properly deburred. I rejected the first set I got and ECS send me another under warranty and it had the same problem. When everyhing is made correctly, AN flared nuts run on to flared fittings easily with finger pressure, which makes it easy to avoid cross-threading. The threads on the Hard adapters were tight on all AN flared nuts and could not be run on with just finger pressure.

The standard for AN/MS flared tube fittings is in the public domain (MS33656) so there is no excuse for Hard disregarding established design dimensions and tolerances.

I hope the other manufacturers do a better job, but who knows, maybe the same people make all of them.

[DRLane]

Truth

[fe1rx]

I made quite a few mockup inlet ducts from card stock, installed them on the oil cooler, then used a shop vac to draw air through them and used smoke to examine the internal flow. The vacuum hookup replaced the exit louver.



Smoke was generated by using a titanium tetrachloride smoke stick, but a mineral oil boost leak smoke generator would likely work just as well. A smooth stream of smoke indicates laminar attached flow in the duct.



I found it easy to achieve this on the back (concave) surface of all my duct models but all of them exhibited serious flow separation on the front (convex) surface. This was apparent when the smoke tendril turned into a fog.



No amount of smoothing of profiles worked until I installed a splitter blade to divide the duct into two halves. Hence the final version was built with the splitter, and both channels flow well when tested as above.





The smoke test provides a good representation of how well the duct can flow air, but it is not representative of the actual inlet conditions in service. The smoke test draws from a large reservoir of still room air. In service the air is moving rapidly and is directed by the body contours. To see how that affected the flow in the duct I tufted up the final installation and installed a GoPro where it could observe the action.



Here it is apparent that the proximity of the brake cooling duct is influencing the flow entering the bottom of the oil cooler duct, as there is some separation on the two convex surfaces, particularly on the splitter blade. The high pressure above the front splitter may also be causing some upflow into the duct lower edge, with some flow separation over that lower corner resulting.

I have tufted the exit louver also. Visibility is poor but the tufts that can be seen flow laterally outboard as directed by the louver vanes. Tire rotation does not seem to particularly affect them.



In conclusion I am happy that my air flows are reasonable, and I am looking forward to a hot track day to see how well the installation actually functions.

[amg6975]

As always, amazing work. So fun to read through and see your thought process and progression.

[changster]

Looking great!

[blnk-128]

When you installed the tee fitting for an oil accumulator down the road, did you also incorporate a check valve? I've been reading people saying that you need one to flow oil in the right direction, where others say the pump does that and the check valve reduces flow.

[fe1rx]

Originally Posted by blnk-128

When you installed the tee fitting for an oil accumulator down the road, did you also incorporate a check valve? I've been reading people for them saying that you need it to flow oil in the right direction, where others say the pump does that and the check valve reduces flow.

The oil filter housing incorporates a check valve to prevent reverse flow, as is visible in the attached image. Therefore I think another check valve would be redundant, and I didn't incorporate one.