Air temperature compensation

Hello, I am confused.

If the temperature increases, we should decrease fuel delivery to maintain a stable AFR or we should increase fuel delivery to cool off the engine?

It depends on your goal -- maintain performance (keep AFR on target for maximum torque), or improve realiability -- allow the engine to run at a higher average performance by not overheating and thus loosing performance. The classic conundrum -- usually you look to the intended use -- is it short drag strip runs with time to cool off between requests for maximum performance, or extended high-speed runs across the desert?

Some ECU's have an average load channel, so that you can change the AFR target as the load continues to rise with continuous full-throttle operation such as land-speed record runs, or long road-racing straights.

Most VE and Speed Density ECU will be compensating for the Air Temperature change in the background, and any Air Temp compensation is just to add safety to prevent knock at high engine temps (ensure a richer mixture), you might also want additional fuel at cold temps (below freezing) as the fuel may not vaporize as easily.

The tricky thing too is the location of the sensor mad how quickly it responds on a hot start.

In my case, I one of my tuned cars is heatsoaking because the sensor location, I have to install it somewhere else.

Heat soak is difficult to completely compensate for so this is often a compromise. I personally find I get the best compromise with the IAT sensor fitted into the intake pre throttle as it's a little less sensitive to heat soak than being in the manifold. An ECU that uses a charge temp estimate table is also a big help here as you can bias the charge temp towards ECT or IAT as the rpm and load change.

I have a question regarding where its best to place the 0 value on an inlet air temp compensation table. In the video, Andre suggests that the best place would be at 20 degrees C as this is close to standard atmosphere. The supplied table shows the 0 point at 0 degrees C.

Wouldn't the most logical place to set the 0 value be at the local air temp of the dyno when the engine is tuned?

You could 0 it to wherever you want to as it's all arithmetic anyway to achieve a final result. Then each car ends up being a one-off custom kind of thing though. Often tuners may want to keep it more standardized so that they only make minor changes across multiple cars. You end up having to make sure that all the other tables take into account the compensation scaling. And it can get complicated if you have multiple compensation tables with all different resolution (some for fuel, some for spark, etc).

If I tune 5 cars that are all mostly similar, I may want to tune the first one, then carry over the compensation as much as possible to the other cars, adjusting only as absolutely needed. This saves time and of course time is money. Sometimes that's not possible though, as different cars will have sensors in different locations, totally different heat soak characteristics, and different ambient conditions (mountains vs sea level).

More background: The standard SAE condition is 25C and 99 kPa ambient. That's what the SAE J1349 correction factor adjusts the power to. That is, it calculates what power and torque the engine "should" be making at that condition. High end dyno facilities can actually control to that temperature, so the correction factors are minimal. So in those labs the engine can actually be consistently tuned at the baseline condition, and then the corrections can first be calculated and then spot checked & adjusted at other temperatures.