This paper introduces a thermodynamically grounded framework that replaces conventional energy-centric building simulation with a heat-based reconstruction method, enabling the use of standard-year weather data without bespoke weather files. High-accuracy predictive models are developed to estimate seasonal heating and cooling loads and to derive convective heat transfer coefficients for transitional seasons. Models are calibrated against extended AMeDAS records for Kurume, Fukuoka Prefecture, demonstrating improved capture of local climatic variability compared with standard practice. Methodologically, the study establishes time-series regression relationships linking outdoor air temperature, incident solar radiation, and wind speed to instantaneous thermal demand and surface convective rates. These empirical relations feed a reconstruction algorithm that adjusts simulation outputs to standardized climatic conditions by rescaling heat flows and convective parameters rather than altering meteorological inputs. Results indicate that reframing simulation inputs in terms of heat enhances reliability and transferability of performance predictions and reduces uncertainty associated with variable weather. The paper examines essential assumptions affecting applicability, including the assumed linearity of convective correlations, representativeness of the reference building envelope, and the use of virtual walls for parameter identification. It further proposes a novel procedure to compute a physical adjustment coefficient for the convective heat transfer coefficient, addressing a parameter gap in macro-scale models while acknowledging challenges for conventional comparative validation. The authors recommend future work on automated calibration routines and on extending the method to tropical and arid climates to broaden applicability and support foundational reform of building energy simulation.
Tatsuhiro Yamamoto (2025) studied this question.