Killer B Motorsport 2022+ FA24 BRZ/GR86 Oil Retention Baffle Development

[dsc_pat]

An overfill, and heavier weight oil and no oil cooler could be tested before a baffle, as if you prevent transient starvation with filling, and also increase baseline oil pressure with a proper viscosity and no loss from a cooler, we could very well be all good without having to rely on an aftermarket, warranty-voiding part.

[gcmak]

Originally Posted by dsc_pat

An overfill, and heavier weight oil and no oil cooler could be tested before a baffle, as if you prevent transient starvation with filling, and also increase baseline oil pressure with a proper viscosity and no loss from a cooler, we could very well be all good without having to rely on an aftermarket, warranty-voiding part.

i'm done 17 mixed events on my car w/no cooler using 5w-30/40, but do not have the live data to prove anything...aside from clean bills of health via oil analyses...so I've mostly been doing what you're thinking.

however, a light footnote (discussed before & worth reiterating) - the dealer/corporate often gets opts for an oil analysis for engine failures, which can reveal a heavier weight oil was used, which can point to improper maintenance.

...and yes, we've read about different world markets w/heavier weights written in owners manuals and the 'tricks' of keeping some used 0w-20 around to swap back into the motor...and other shenanigans.

[Arthur-A]

Originally Posted by gcmak

i'm done 17 mixed events on my car w/no cooler using 5w-30/40, but do not have the live data to prove anything...aside from clean bills of health via oil analyses...so I've mostly been doing what you're thinking.

Are you adding extra oil? If yes, how much? Also, do you monitor oil temperature? I'm curious what's the peak temp on the race track.

[900BRZ]

Wanted to share the results in detail on this thread before summarizing elsewhere, because this is where the bulk of this history about this product lives. We have already established baseline data that shows that the pressure drops are not unique to my car. For this post, I’m only going to share data from my car. I will show data before and after the baffle install.

Testing setup

On May 22, 2023, we tested my car at Thunderhill 5-Mile Double Bypass, with myself and Joe McGuigan driving. (Joe is a well-known local instructor and private coach and you can find more videos of his driving here: https://www.youtube.com/@JoeMcGuiganRacing)

On July 1, 2023, we tested my car again at Thunderhill 5-Mile Cyclone/Bypass (AKA West Bypass) with the Killer B prototype baffle.

I tried to control as many variables as possible, so I ran the same 5W30 motor oil filled to the same level. At or a few mm above the max fill line on the dipstick. I looked at both sides of the dipstick as well.

Unfortunately, we could not control for weather, and we experienced higher temperatures on July 1, up to 107° F ambient. 5W30 was, in my opinion, too light and I had lower nominal pressures (Note: I’m using the word nominal to describe the consistent oil pressure level above 4000 RPM, when there are no pressure drops).

My car saw nominal pressures down to the mid-40s, which is lower than I’ve seen before and, honestly, lower than I’m comfortable with. I’ve included oil temp in the channels present in the telemetry screenshots. Unfortunately this difference points to a difference in observed viscosity, and that could affect the dynamic behavior of the oil in the pan. I chose to use 5W30 to reduce variables, but it introduced a separate variable. We are not professionals and are funding these tests out of pocket, so it is difficult to construct a perfect test.

The other cars we had instrumented were using 5W40 and maintained higher nominal oil pressures, in the mid-50s. I know for lots of track rats, this will seem obvious (and it’s why the other owners swapped to 5W40), but I didn’t appreciate how big a difference it would make until I saw it for myself. The oil temps themselves were pretty well managed despite the lower pressures.

Overview of results

When you’re looking at this data, please note that the difference in configuration between Double Bypass and West Bypass is 1 turn: Turn 5 East. The faster corner is called Bypass and the slower corner is called Cyclone. Joe estimates that Cyclone is roughly 2 seconds slower than Bypass.

You can see they take different paths in the track map in the bottom left of the screenshot. Please do not confuse the difference in this corner between the 2 track days as a difference in performance of the control and experiment. You can see that Race Studio 3 auto-generated segments 8-10 are T5E. Again, please ignore these segments when analyzing this data.

The Cyclone also slightly offsets the rest of the lap’s data, since I’m using a distance base in this chart. The Cyclone is a slightly longer distance and RS3 does not correct for this. It’s relatively easy to see this offset in the data but it’s an exercise left to the viewer to correct for it in their mind when analyzing the data.

Here are the telemetry screenshots I have attached:

1 & 2) Joe McGuigan driving control and experiment laps at a similar pace. Note that of the July 1 laps, I selected the earlier lap even though it’s slightly slower, because it has lower oil temps and is more consistent with May 22. For May 22, I have included the both laps, one per screenshot.

3) 900BRZ driving control and experiment laps at a similar, slower pace. You can see that the Double Bypass lap gains time at the East Bypass, but then loses time for the rest of the lap. This shows that, outside of the Cyclone, the pace West Bypass lap was slightly faster.

4) 900BRZ driving, max-line fill and 500 ml overfill laps. Unfortunately, this came at the end of a very hot day and I had, by this point, lost the ability to go fast. Unlike the previous comparisons, which compared May 22 (without baffle) to July 1 (with baffle), in this comparison, both laps are from July 1. Thus, you can look at data from Cyclone as well. The lap without the overfill (orange) is faster throughout the middle of the lap, but loses time at the end to the overfill lap. I only downshift once, into 3rd, on T14-15, so the pressure drops that we typically see at the end of the lap are not present.

Analysis of results

Based on this data, we can say that the Killer B prototype baffle does not significantly reduce the pressure drops we are seeing at Thunderhill 5-Mile. We were not able to test one of the turns included in the original video, T8W (as that is bypassed by these track layouts).

Joe’s data is the most representative because he is a fast, consistent driver. Most of the pressure drops are consistent between the two track days, but there are more drops through the connector from East to West (segments 17-19) with the baffle, which is unexpected. However, in T5W, the magnitude of the pressure drop without the baffle does seem to be slightly larger, even if the lowest pressure recorded in that corner is similar.

My laps show pressure drops and the performance seems to be slightly worse on the July 1 test, but my pace is also a little faster if you account for the slower Cyclone corner.

In both of these tests, it’s not possible to precisely attribute the slight increase in pressure drops (with baffle installed) to: 1) the baffle, 2) differences in pace, or 3) the temperatures affecting nominal oil pressure. We can say though that throughout the laps, there are not significant improvements with the baffle. There may be subtle improvements in the relative depths of the pressure drops in Joe’s laps in T5W and T14-15E, if you consider that the pressure drop was happening from a higher nominal temp in May 22.

Regarding the 500 ml overfill: These results look roughly as expected oiling performance on slower laps. My opinion is there is truth behind the anecdotal higher failure rates among advanced drivers who are able to consistently achieve higher cornering forces. The pressure with the overfill looks more consistent in T3E, but the pace is slower through that section on the lap with the overfill.

Overall, I don’t think we can draw many conclusions from this overfill test, unfortunately. The pace is just not fast enough. Luckily, this possible mitigation is a cheap and easy to implement. I recommend further testing in this area, although I don’t believe we will see a “fix” here.

Edit: Full resolution screenshots available here: https://drive.google.com/drive/folde...yB0quUr8qOSfmC

Edit: Updated images to obscure T5E when comparing 5-Mile Double Bypass to 5-Mile West Bypass.

[OldBiker]

So baffle didn't do it... probably didn't hurt, but didn't solve the problem either. What is next, accumulator, dry sump, or LS swap?

[gcmak]

Originally Posted by Arthur-A

Are you adding extra oil? If yes, how much? Also, do you monitor oil temperature? I'm curious what's the peak temp on the race track.

Roughly 0.25L (so 5.25L total per oil change).
My 20 minute sessions keep the oil temps between 240-260F. Generally glancing at the dash I see 253F (which isn't necessarily the 'hottest' measurement point).

[900BRZ]

Originally Posted by OldBiker

So baffle didn't do it... probably didn't hurt, but didn't solve the problem either. What is next, accumulator, dry sump, or LS swap?

For now, I'm going to take some time and think things through.

There are other baffles that can compliment or replace this one. We have data on some of them already.

Also, while we have test results with this baffle on a long course with lots of different styles of corners, I wouldn't be opposed to running it some more and seeing if there are corners that it helps with. Maybe that's wishful thinking, but I was specific to note the conclusion that it doesn't help significantly at Thunderhill 5-Mile, not necessarily universally.

That said, we have results in long right-handers preceded by a significant braking event, we have quick right handers (T5W), and we have multiple right-handers with elevation change (T3E, T9C).

Other things we've learned is that it's possible to drive around this problem, to some extent. If we short shift in T5W, T8W, and T14-15E, it significantly reduces drops. The goal was to find a solution that didn't require changing driving style, but at least there is a way to track the car and reduce the pressure drops, even if at the cost of some lap time.

[OldBiker]

Many thanks to 900BRZ et al. You guys are doing a fantastic job characterizing this problem. And of course missing in all this is Subaru. I would hope some VP in Subby of America would eventually reach out to you. Then again Japanese style management and the need for face saving does not help in these situations. To wit, you can follow the travails of Honda in MotoGP (or F1 for that matter). This whole thing reminds me of the 996 era IMS debacle... if memory serves, it took a class action to get Porsche to respond... and the problem persisted until they finally came up with a new engine many years later. (Subaru, lacking deep resources, would probably just cancel the BRZ if pushed.)


As for our current dilemma, I would love to see the oil pan instrumented. It would be good to characterize exactly when the pickup gets starved for oil and when supply returns, all synched with the pressure graph and G-meter. There are a lot of theories, but a real solution requires more info.


Thanks again guys.

[gnarjunkie]

Thanks again for all the testing. Bummer to see the easy button didn't solve it.

I'm curious... 900brz have you had any oil analysis done? Could be a good way to see if bearing wear is occurring and to what degree.

I had a friend a couple years ago that had a CSF oil cooler / radiator combo on his STi that caused a massive pressure drop. Enough that his oil pressure was down into the 30s pretty much constantly at WOT -- and that did show up on a UOA.

As a side note, he actually still got the rest of the track season out of that engine, put a bigger turbo on, made 400whp, then ran most of the next season on it too before it lost a bearing. That's after tracking it for a good 4-5 events with consistent max oil pressure in the mid 30 psi range, plus a lot of street miles.

[zeroomega]

After look through the data together with @900BRZ. We both conclude that we didn't see any measurable improvement from KillerB's baffle at Thunderhill 5 Miles. In other words, KillerB's original hypothesis that the large opening at upper oil pan to the timing chain cover is the culprit of oil pressure drops in cornering, might be wrong. Which is disappointing and surprising. If this is wrong, what makes FA20 to have slightly better oiling than the FA24 as we observed in our tests?

FA20 and FA24 has a almost identical lower oil pan and if we looking at the pictures in https://www.ft86club.com/forums/show...7&postcount=51, we can conclude that the upper pan is also very similar, except for that opening near the timing chain cover. It is understandable KillerB made the assumption that this large opening could be the culprit of the oil pressure drops since oil can surge into timing chain cover easily through here during braking. With the test results we have, at least we can conclude this big opening is not the main issue. Then what else could make FA20 has less oil pressure drops than FA24 considering the pan structures are very similar?

I have a hypothesis (Disclaimer, I am not an engine expert, this is just a hypothesis from what I observed in our testing): "Maybe the FA24 internal design makes it have worse oil return flow rate than FA20 under large lateral G, and both FA20 and FA24 has the same level of oil sloshing issue under lateral g". If this is true, then it can explain a lot of behaviors we have observed. For example, 1) FA20 and FA24 has oil drops at almost the same corners at tracks we have tested. This may due to the fact that both engines have very similar upper and lower oil pan design. 2) we observed both FA20 and FA24 have oil pressure drop at T8W at Thunderhill West, but the duration of the drop is around 2 sec on FA20 but 4 sec on FA24. This can be explained as: FA20 has better oil return flow under hard right turn, then it took a bit more time for the oil sloshing to the point that oil is partially moved away from the pickup tube. And after the car leveled, the higher oil return flow makes the oil move back to the pickup tube faster, so the oil pressure recovered sooner. If this hypothesis is correct, then this is a bad news. Only Subaru can change/revise the engine design to improve the oil return flow in an economical way. For users like us, we probably need an external oil source like accusump or dry sump system to fully solve the oil starvation issues, which would be very expensive. Unfortunately I don't think there is any good way to design an experiment to test this theory.

This hypothesis doesn't mean there is no way to mitigate this issue to a certain extent. In our testing we saw baffles like SYMS and TOMEI reduces oil pressure drops at T5W and T8W at Thunderhill West, but the drops are not fully eliminated. And both baffle seems to make oil pressure drop at T2 Sonoma a bit worse. This behavior still matches our original theory that, the reason the oil pressure drops only happens in right turns is due to the fact that oil pickup tube is mounted at passenger side. If there is an oil pan baffle that put a wall at the left (driver side) of the oil pickup tube and put a one way trap door at bottom, potentially it should improve the oil pressure drops and work better than horizontal only baffles.

Thoughts?

[dsc_pat]

Originally Posted by 900BRZ

4) 900BRZ driving, max-line fill and 500 ml overfill laps. Unfortunately, this came at the end of a very hot day and I had, by this point, lost the ability to go fast. Unlike the previous comparisons, which compared May 22 (without baffle) to July 1 (with baffle), in this comparison, both laps are from July 1. Thus, you can look at data from Cyclone as well. The lap without the overfill (orange) is faster throughout the middle of the lap, but loses time at the end to the overfill lap. I only downshift once, into 3rd, on T14-15, so the pressure drops that we typically see at the end of the lap are not present.



Regarding the 500 ml overfill: These results look roughly as expected oiling performance on slower laps. My opinion is there is truth behind the anecdotal higher failure rates among advanced drivers who are able to consistently achieve higher cornering forces. The pressure with the overfill looks more consistent in T3E, but the pace is slower through that section on the lap with the overfill.

Overall, I don’t think we can draw many conclusions from this overfill test, unfortunately. The pace is just not fast enough. Luckily, this possible mitigation is a cheap and easy to implement. I recommend further testing in this area, although I don’t believe we will see a “fix” here.

Thanks for doing all the heavy lifting here. We all appreciate you.

Low IQ me absolutely see a potential mitigation measure, much more so than any complicated fluid dynamic hypothesis and baffle development. We want to mitigate risk here, not resolve a design flaw.

Edit : I was looking at the wrong capture. Sorry. I support your conclusion, especially the need for more testing.

[sns.brz]

Originally Posted by zeroomega

This hypothesis doesn't mean there is no way to mitigate this issue to a certain extent. In our testing we saw baffles like SYMS and TOMEI reduces oil pressure drops at T5W and T8W at Thunderhill West, but the drops are not fully eliminated. And both baffle seems to make oil pressure drop at T2 Sonoma a bit worse. This behavior still matches our original theory that, the reason the oil pressure drops only happens in right turns is due to the fact that oil pickup tube is mounted at passenger side. If there is an oil pan baffle that put a wall at the left (driver side) of the oil pickup tube and put a one way trap door at bottom, potentially it should improve the oil pressure drops and work better than horizontal only baffles.

Thoughts?

Like this?
https://www.apexi-usa.com/products/k...s-86-brz-13-21

[nomtimesthree]

Originally Posted by 900BRZ

Other things we've learned is that it's possible to drive around this problem, to some extent. If we short shift in T5W, T8W, and T14-15E, it significantly reduces drops. The goal was to find a solution that didn't require changing driving style, but at least there is a way to track the car and reduce the pressure drops, even if at the cost of some lap time.

Thanks so much. Would love to hear more about short shifting as a mitigation if possible. Like, how much of a pressure drop delta at what rpm did you end up seeing when short shifting one of these corners? Is there a rpm number that would be ideal to stay under when in a condition that triggers low oil pressure?

[900BRZ]

Originally Posted by sns.brz

Like this?
https://www.apexi-usa.com/products/k...s-86-brz-13-21

I can’t tell if that’s a 1-way door, but hopefully it is. That might be an interesting product to try.