The I constant is defined as the time, which brings the I-part of the PI controller to the same value as the P-part. (Unfortunately, this is not the case of my sketch below sorry for that. The P-part and I-part should be of the same height.)
See the behaviour in time for constant controller deviation (setpoint – actual value): in the first moment there is only the P-part (with the P constant set in the UC300 as control deviation at which the valve goes to 100%), then in the course of time the I-part is increasing. This should compensate the inaccurate P controller. In reality, the controller deviation is decreasing in time (because of the increasing “y” value (opening of the valve)), which results in decreasing the P-part and of course the I-part, too, the output curve goes more flat, may hunt several times, and then goes stable to a certain value of combination of P-part and I-part. Something like the red curve below. This is the final and required state of output, until next change of the actual value, etc.
Numerically, the I part is in seconds. For room applications, it is recommended to set it say to 15 to 30 minutes (e.g. 900 to 1800) or even more, depending on the size of the room and the output power of the heating body. Of course if you set it higher, the controller will need more time to reach the final stable position, if you set it too short, it will most probably hunt. The default value is 3600 s, which is high enough for large rooms, with sufficient response at the same time. Please do not set it too low, values like 120 seconds are totally wrong here. (This is suitable for fast processes, such as supply air control at an air handling unit, hot water control in a water heat exchanger, etc.)
If the I constant is set to 0, the I-part is disabled, and the controller acts as a P-controller only. Note that some versions of domat.exe do not allow to set 0 here.