Demand flexibility is the capacity of demand-side loads to change their consumption at any instant. It makes electricity more affordable by helping customers to use less power when prices are high. On the other hand, demand flexibility can also help to increase the reliability of the power grid when is highly stressed, by reducing demand for power or for integrating renewable generation. Thermostatic controlled loads (TCLs) are one of the most promising options among demand response (DR) solutions however, conventional methods for controlling single TCLs are not easily extensible to aggregated TCLs since it may cause them to synchronize. In addition, the ambient temperature may significantly influence the power flexibility offered by the TCLs. In this paper, a metric of flexibility is applied along with a modified control algorithm to de-synchronize the TCLs with the aim of fairly comparing the different control approaches applied to aggregated TCLs. Furthermore, a sensitivity analysis, considering variations of temperature in several periods of time, is performed over the TCLs power flexibility. The results were validated in Simulink/MATLAB using real demand and generation data from UK national system and simultaneously, temperature data for the same region and time frame.