Conference Year
July 2018
Keywords
Climate, Weather Data, Building Energy Simulation, Köppen-Geiger classification
Abstract
Several studies couple simulations of building systems and statistical techniques in order to draw findings which can be generalized under given constraints. To this extent, one of the most important inputs to deal with is climatic conditions: indeed, the weather data and the localities chosen for the analysis can seriously affect the representativeness of the simulation outcome with respect to other regions. Nevertheless, the first question we should answer regards the domain to which one or few reference climates should be representative. As a common practice, national guidelines, heating and cooling degree-days scales and worldwide recognized climatic classifications are adopted. However, in some cases, these kinds of categorization are suitable only for specific applications. For example, the well-known and used Köppen-Geiger classification with the later Trewartha’s modifications is based on annual or seasonal air temperatures and cumulative precipitation and highlights mainly the relationship between climate and vegetation. Consequently, while this classification can be very effective to distinguish ecological systems, it can be insufficient for building energy analysis. Similar considerations apply to ANSI/ASHRAE 90.1 and 90.2 classification. In this work, we propose a critical discussion of the main climate classification system adopted in Europe and present a clustering and classification analysis on 66 European locations. The aim is to identify a limited number of climatic zones and, for each one, a reference climate to be used for energy simulations. The hourly weather data of dry bulb temperature, relative humidity and global horizontal irradiation reported in typical and reference years have been used as input. The clustering analysis has been performed with two approaches with different levels of complexity: (1) a simplified approach based on the calculated monthly averages of the weather variables and (2) a more detailed one based on the hourly profiles, through non-parametric techniques such as the Kolmogorov-Smirnov test. The obtained climatic classes have been compared to Köppen-Geiger traditional ones, underlining the main changes and the impact for building energy simulation analyses.