A gap in standardization of quantitative infrared thermography (IRT) directly leads to a lack of measurement pattern for determining in-situ U-values of heavy multi-leaf walls. Three groups of causal factors might influence the estimation of this build quality indicator: operating conditions, thermophysical properties and technical conditions. Focusing on the last one, previous studies underlined the difficulties of measuring below 3¿h. In contrast to active IRT, no algorithms have been found to process images, despite playing an important role in the effectiveness and robustness of IRT. The traditional approach involves analysing from 120 to 7200 thermograms with a data acquisition interval of 1¿min up to 1¿s respectively. The aim of this paper was to critically assess the test duration that is traditionally used. Six real heavy multi-leaf walls were tested under a stationary regime as a stochastic process of underlying data. For the first time, a research based on two U-value time series analyses (statistical tests and a signal modelling technique by MATLAB) demonstrated the feasibility of short-lasting IRT tests. Moreover, this research posed an innovative data management tool to automate this non-destructive testing (NDT) in mid-term, stopping IRT tests in real time once the right level of accuracy was achieved.