There are two issues involved; one is the Overload Class of the protective device, which is designed to emulate the thermal damage curve of the motor, the other is a "cooling time constant" of the motor. Unfortunately most motor manufacturers do not provide you with the cooling time constant value. They trust that the Overload Relay curve is going to take care of that, and for the most part it does. The problem I see you having here is that you want to play with that motor thermal damage model in terms of getting the most out of the motor under a known overload condition. You are (almost literally) playing with fire here. If you want to get an idea of how complex it is, try to read this paper;
https://library.e.abb.com/public/58...Section_08p11_Motor-Protection_757291_ENa.pdf
The bottom line though is that the cooling time constant is highly variable, because it has to do with the ambient air, the method of cooling of the motor, the mass of the motor stator and Rotor, the duty cycle involved etc. etc. etc. Here in North America, we rely upon formulae made by NEMA that relate to a "Starts Per Hour" rating of a motor, allowing us to over size a motor based on typical values from known motor designs. Those would NOT be appropriate for IEC motors however, and seeing you are in Greece, you should be using IEC motor data.
For IEC motors, you have a "Duty Type" system, S1 through S10, and within that list, Types S3 through S8 are what are referred to as "intermittent duty" cycles, relating to how often they cycle, for howl long and by how much. This is a decent article on that issue, I suggest that you read it and understand what you are getting in to..
https://electrical-engineering-portal.com/10-duty-types-three-phase-asynchronous-motors