Engine suddenly running rich, but lowering fuel table VE to unrealistic values somehow "fixes" it

Helped a friend tuned his JDM 2007 STI EJ20 on the dyno over a year ago.

However, after a recent "oil change" the car suddenly started running extremely rich, had reduced power and very poor gas mileage. We tried diagnosing the issue on the road but could not find identify cause. Finally we had some to put it on the dyno and here are the initial result that we have found after a few short 3000 to 5500/6000 rpm pulls.

Here is the initial pull without changing anything on the tune:

Here is the log from the initial tune a year ago:

As you can see comparing between the logs, even with a lightly lower injector duty/PW, the car is running MUCH MUCH richer compared to the old logs.

Here is after "adjusting" the fuel map until the AFR is back to "normal":

The injector duty/PW needed to be dropped significantly compared to before in order to achieve a reasonable AFR.

Here is the original vs the "modified" fuel map (I only dialled in the part that is relevant to the WOT pull):

Cutting VE all the way down to the mid 80s just not look right to me. Unless there is something seriously wrong with the engine, the VE should not be that low. Another strange thing we noted was, after "correcting" the AFR, the power measured by the dyno did not show a significant drop in power compared to the previous tune (even with a slight drop in boost level). If the VE indeed had dropped so drastically, power should drop too.

Before, adjusting the fuel, we also tried advancing/retarding the cam via VVTi by 10 degrees to rule out whether the cam belt have skipped but that did not improve the AFR and also showed less power on the dyno vs the original tune.

Is it possible somehow, the ECU/engine is injecting more fuel than it actually is showing?? If you look at the logs, the injector duty was indeed similar to the original log even though AFR was running extremely rich and when the AFR was "fixed" the injectors duty was significantly less.

While trying to diagnose the problem before putting the car back on the dyno, we did hook up a oscilloscope to the injectors and the PW at idle does match what is shown on the ECU.

Other points to note:

For the initial tune, the car had an aftermarket hood without proper ducting for the top mounted intercooler so the IAT was pretty high. The OEM hood had since been reinstalled and it ran without any problems.

TGV and EGR deletes were installed after the initial tune but they were installed AFTER the issue have arise so it shouldn't be the cause. We initially suspected the TGV might be causing the issue that's why we installed the delete.

Have you replaced (or confirmed with another wideband) the O2 sensors to make sure reading reflect reality.

I notice you don't have a fuel pressure sensor data in your graph. A failed fuel pressure regulator providing excess fuel pressure would explain the behavior.

Hello David, I will connect another O2 to the tailpipe to double check tomorrow but the dyno results does seem to suggest the lambda reading is accurate as the engine was indeed significantly down on power when it was running extremely rich and power was back to the previous level once the AFR is adjusted back to normal.

The car does have a fuel pressure sensor gauge but it is not connected to the ECU. We did watch the pressure on the gauge as we do the pull and noted nothing out of the ordinary. The differential fuel pressure does droop a little on WOT (down to 2.7 bar from 3.0) but it also did that on the original tune. The droop would also result in a high VE to be set in the fuel table rather than a low VE.

Just though I should also post the dyno plots here if that would help:

Dotted green is the original tune from 1 year ago.

The solid lines are the runs made today with the issue on:

Green = original map

Red = Cams advanced by 10 degrees

Orange = Cams retarded by 8 degrees

Magenta/Blue = Gradually pulling fuel in the fuel map

Light purple (looks almost white in the photo) = Fuel map adjusted to "normal" AFR

Any chance the ECU received a firmware update or tune may have been accidentally altered to change fuel calculation modes?

Fuel pressure dropping less than before during significant fuel demand is a good guess.

The turbo dying or a significant boost leak are my next guesses.

Have you confirmed the baro reading and all sensor readings are sane and not causing an unwanted and inappropriate fuel compensation?

Have you verified all cams are on target?