Tuned header = Lumpy VE?

My engine is a 3L Porsche 944 Turbo 4 cyl/2V, the cam is a mild performance cam with a small overlap, ECU is a Haltech Elite 1500, Car is a dedicated track car, but but registered & "street legal". i did my tuning on an engine dyno last winter with an exhaust system very similar to what is on the car...the primary difference being that the wastegate had it's own exhaust pipe and did not dump back into the 4 inch primary.

Overall pretty please with results above 0Kpa torque is pretty flat fron 3000 all the way to 6500 at about 375 Ft lb. However, when i was doing this tuning i noticed that the VE had some very prominent narrow peaks and valleys (lumps) that become pronounced below 0Kpa. see attached file. I did add some RPM rows to accomodate them, but I was focused mostly on 0 Kpa and above so i didn't spend a lot of time at the lower MP. I've been driving on both the street and a couple of short track sessions with LTT trim and WBO2 control on below 0 Kpa. Logs show WOT throttle is fine, no significant change from Dyno results. Partial throttle is still a work in progress, but the lumps are still there and not getting any flatter.

1) I am wondering if this is possibly the result of the 4 into 1 tuned header that is on the engine....if that's not likely then any other thoughts as to a cause of the lumps?

2) should i intentionally smooth these lumps? or just let the LTT do it's thing over a long period of time? I try not to spend a lot of time in these cells but it does happen on ocassion and i need to be able to drive thru it without bogging down.

Peter

That is very possible, as the entire intake and exhaust systems are interacting and affecting the cylinder filling at different rpm and/or loads. If working together you will get better filling that will require more fuel and less timing, whereas where they are conflicting the cylinders will fill less efficiently and require less fuel and more timing. If you have a power/torque curve you should see a dip that matched that of the graph you provided.

There is an old expression "the engine will tell you what it wants", which in context means to map to what it needs, not what you think it 'should' get.

[edit for spelling]

I agree with Gord's mention of "the engine will tell you what it wants" in this case, but only after exhausting potential issues that could be causing it.

Some examples include verifying fuel pressure is stable, no intake or exhaust leaks, if VVT equipped check positions etc., verify whether injectors are operating in a non linear region and whether the tune if fully accounting for that non linear behavior, etc.

If you're certain there's nothing "wrong" then I'd say yes that's how the engine is behaving, and I would re-do axis values for extra resolution in the non linear VE region so you can best characterize it.

Thanks Gord and Mike.

The torque curve @WOT from the dyno follows the VE shape ...t's relatively flat from 3K to 6500.

Mike i had prevoiusly added rows around the anomaly so i have rows at 2000,2250,2500,2750 3000.Seems like adding rows every 100 RPM between 2000-30000 would be overkill....correct?

ECU should compensate for fuel pressure on the fly ( but i will double check the logs) NO VVT. The lumps are much more prounounced at low MP, 2-4K RPM range - this is low-mid range operation so injectors are unlikely to be non-linear here and not sure that would cause such pronounced lumps at specific RPMs

I tend to start with break points 500 RPM apart, then remove ones that aren't needed because behavior is linear to save me time and CPU resources.

If an engine is rather non linear in an area or few, which mostly happens on highly tuned NA engines, then I add more break points where the VE peaks and troughs are based on observed lambda deviations, until I get a smooth lambda trace without corrections through the region.

When choosing where to put extra break points, I wouldn't set your mind on a fixed interval like every 100 or 250 RPM. Maybe you have points at 2000, 3000 RPM, and 2340-2557 RPM has unique behavior, so perhaps you add a break point at 2300 which is the end of linear behavior, then one at a VE peak at 2340, another at a VE trough at 2489, another at 2557 where the oddity ends. Maybe you find the point at 2000 RPM no longer serves a purpose and can remove it now that the 2300 RPM break point is present.

I wouldn't be afraid to add a few extra points, I'd just remember to place them where they do you the most good, and avoid retaining break points which really have no impact because the flow of VE is linear through that area anyway.

Thanks Mike, those are helpful pointers regarding non-linearities in VE map

Awesome, glad to hear it.