I got my new toy in the mail earlier this week and decided to try to log some data to evaluate the impact of this mod on FE.
Here's a graph showing the injected quantity in mg per stroke during the course of my test (X axis in seconds). The black line is a moving average based on 60 data points (seconds).
I just did an A-B test as it took almost an hour of driving. The cruise control was set at 90 km/h the whole time and the car was never turned off. Run A1 started at a coolant temp of 85.5 Celcius and 90 Celcius was reached at the end of the run. All remainning runs were done at 90 Celcius.
During run 1, the ECU thought the intake air temp and fuel temp were -23.4 Celcius.
During run 2 the ECU saw actual intake air temp and fuel temp which were respectively averaging 45.22 and 50.4 Celcius.
The average injected quantity in mg per stroke was as following:
A1 - 11.9172
A2 - 11.2812
Var = 5.3368%
B1 - 8.4111
B2 - 7.9727
Var = 5.2121%
Let's average the variation at 5.5% to be on the safe side and account for unknown errors.
So it looks like the injected mass was actually higher with the mod. But wait, the ECU thought the temperature of the fuel was way cooler than what it really was and ajusted the volume accordingly.
Diesel volume varies approximately 0.125% per degree Celcius. The ECU thinking the fuel was 73.8 Celcius cooler than it really was, it should have adjusted the volume by approximately 9.225%, but let's use 9% just to be safe.
So, for any given mass the actual injected volume should be 9% smaller with the mod (at least during the summer), but since the injected mass was 5.5% higher the theorical net result is an increase in fuel economy just shy of 4%.