Simulation of thermal building state in summer - study of results uncertainties for thermal ambient prediction
Citation
International journal on architectural science, v.1, no.3, 2000, pp. 126-148
Abstract
In this paper, the authors have developed a double study on the building thermal comfort models. The first step concerns the comparison between three models well known in specialized literature: Fanger's thermal comfort model taken as the reference model, elaborated by Denmark University of Lingby; Deval and Berger's model, elaborated by Centre National de la Recherche Scientifique of France; and Sherman's model, elaborated by University of Berkeley, USA. After a compete description of the three models, they only compared their results, the predicted mean vote, with a protocol defined in the text. This leads to predict the user's thermal sensation PMV with each model, on two climatic sets, warm and hot in summer, taking an example of a building. Comparison shows quasi constant analogy or differences between the reference model and others. With the warm climatic set, Fanger's and Sherman's model provide almost the same results, and the results obtained with Deval and Berger's model have a constant 0.5 PMV unity difference. With the hot climatic set, a 0.25 PMV unity constant difference is observed for the Sherman's model, whereas Deval and Berger's model results are almost similar.
The second step of the study concerns the propagation of data uncertainties on the PMV output of the three models. Two methods are used in order to define the uncertainty result interval: the probabilist Quasi Monte Carlo method and the finite differences differential analysis. Authors present those two methods and define hypothesis for each data uncertainty field to that of finite differences differential analysis method. They conclude on the quality of finite differences differential analysis method, which has proven its reliability and its performance with its very low computation time compared with the Quasi Monte Carlo method. Then, they obtain uncertainties with 1 predicted mean vote unity wide domain. Finally, they notice that the accumulation of uncertainties due to the modelization and the data uncertainties can provide important variations of results, which must be known by building engineers and designers.
(1) Introduction
(a) Analysis of results variations between models (fig. 1)
(b) Analysis of variations due to data uncertainty in the models (fig. 2)
(2) Comparison of the models
(a) Preliminary observation
(b) Description of the models
(i) Heat exchanges by convention Ecv(X)
(ii) Heat exchanges by radiation Ery(X)
(iii) Heat exchange by skin diffusion Eps(X) (perspiration humidity transfer)
(iv) Heat exchange by latent respiration Ehr(X)
(v) Heat exchange by dry respiration Ecr(X) (convection respiratory)
(vi) Heat exchange by evaporation of sweat secretion Esd(X) (sudation)
(vii) Thermal sensation
(c) Exploitation hypotheses - definition of the data vector E
(i) Definition of vector A(t)
(ii) Definition of vector U
(iii) Preliminary observations on the models
(d) Comparison of results on thermal sensation Y(t)
(3) Propagation of data uncertainties on the results Y
(a) Method of study
(i) The probabilist Quasi Monte Carlo method (QMC)
(ii) The finite differences differential analysis method (FDDA)
(b) Exploitation hypothesis of the methods: data perturbation
(c) Analysis of results: uncertainty domains of the PMV (=Y)
(i) Warm climatic set (figs. 9 and 10)
(ii) Hot climatic set (figs. 11 and 12)
(4) Conclusion
(a) On the result obtained
(b) On the methods used to define the uncertainty fields
(5) Nomenclature
(6) References
Description
Notes: Heating
Subject
Type
Article
Format
Date
2000
Language
en