Waste heat recovery through cycle humidification is considered as an effective tool to increase the operational flexibility of micro Gas Turbines (mGTs) in cogeneration in a Decentralized Energy System (DES) context. Indeed, during periods with low heat demand, the excess thermal power can be reintroduced in the cycle under the form of heated water/steam, leading to improved electrical performance. The micro Humid Air Turbine (mHAT) has been proven to be the most effective route for cycle humidification; however, so far, all research efforts focused on optimizing the mHAT performance at nominal electrical load, and no thermal load. Nevertheless, in a DES context, the thermal and electrical load of the mGT needs to be changed depending on the demand, requiring both optimal nominal and part load performances. To address this need, in this paper, we present the first step towards the development of a control strategy for a Turbec T100 mGT-mHAT test rig. First, using experimental data, the global performance, depending on the operating point as well as the humidity level, has been assessed. Second, the performance of the saturation tower, i.e.The degree of saturation (relative humidity) of the working fluid leaving this saturator, is analyzed to assess the optimal water injection system control parameter settings. Results show that optimal mHAT performance can only be obtained when the working fluid leaving the saturation tower is fully saturated, but does not contain a remaining liquid fraction. Under these conditions, a maximal amount of waste heat is transferred from the water to the mGT working fluid in the saturation tower. From these data, some general observations can be made to optimize the performance; being maximizing injection pressure and aiming for a water flow rate of 5m3=h. However, having a specific control matrix, that allows setting the saturation tower control parameters for any set of operational setpoint and the inlet conditions would be of more interest. Therefore, future work involves the development of a control matrix, using advanced data post-processing for noise reduction and accuracy improvement, as well as an experimental validation of this methodology on the actual test rig.