This paper presents a thermal simulation validation study of the typical precision that a trained thermal simulation engineer can expect to obtain for the simulation of a room connected to a naturally ventilated double skin facade. The open source building thermal simulation tool EnergyPlus is used to predict air and surface temperatures in a free running weather exposed test cell. The validation approach used does not allow for fine tuning the simulation model by trial and error and includes a sensitivity analysis on the impact of different simulation options such as: number of thermal zones, ground heat transfer and accuracy of solar radiation data. The analysis of the results is consistent with the engineering application goal by focusing on standard indicators: average error, bias and difference between predicted and measured daily maximums. The results show good agreement between simulation and experiment, with an average simulation error in air and radiant temperature of 1.4 °C and an average daily maximum error of 2.5 °C.