A new algorithm for uncertainty quantification for thermal conductivity measurement on polymers with the Haakvoort method using differential scanning calorimetry considering specimen height and real contact area

Abstract

A new algorithm for the quantification of uncertainty in thermal conductivity measurements on polymers according to the Haakvort method is presented. This fast and convenient method using differential scanning calorimetry has been established as DIN EN ISO Standard 11357–8 with an error margin of 5–10%, which is a rather large value when considering that this is an important material parameter for many applications and is often used in combined quantities, such as thermal diffusivity or thermal effusivity. Unfortunately, the DIN EN ISO standard does not provide useful information on the dependence of the error range on the number of specimens or important parameters, such as the height of the specimens or their real contact area. Applying a rigorous statistical approach, based on the law of large numbers (LLN) and different techniques which are also used in well-known methods, such as Monte-Carlo- or Markov chain Monte Carlo (MCMC) algorithms, we establish and investigate a method to optimize the experimental effort to a specific target, especially the number of specimens, the aspect ratio and the real contact surface of the specimen.