A multimodel approach to building thermal simulation for design and research purposes

The designers pre-occupation to reduce energy consumption and to achieve better thermal ambience levels, has favoured the setting up of numerous building thermal dynamic simulation programs. The progress in the modelling of phenomenas and its transfer into the professional field has resulted in various numerical approaches ranging from softwares dedicated to architects for design use to tools for laboratory use by the expert thermal researcher. This analysis shows that each approach tends to fulfil the specific needs of a certain kind of manipulator only, in the building conception process. Our objective is notably different as it is a tool which can be used from the very initial stage of a construction project, to the energy audit for the existing building. In each of these cases, the objective results, the precision advocated and the time delay of the results are different parameters which call for a multiple model approach of the building system

💡 Research Summary

The paper presents a comprehensive multi‑model framework for building thermal simulation, embodied in the software tool CODYRUN, which is intended to be useful from the earliest conceptual design phase through to the energy audit of existing structures. The authors begin by categorising the principal user groups—building designers/operators, government policy makers, and building‑physics researchers—and highlighting that each group has distinct requirements regarding model fidelity, computational speed, and the type of output needed. Designers need rapid, coarse‑grained feedback on orientation, shading, and gross heat loads; policy makers require aggregated, building‑stock level estimates for national energy targets; researchers demand high‑resolution, physically accurate models to investigate underlying heat‑transfer mechanisms.

The paper reviews the evolution of building simulation tools, noting that early codes such as ESP, BLAST, and SERI‑RES were large‑scale, main‑frame applications with limited flexibility. Subsequent generations (e.g., QUICK‑TEMP, CODYBA, COMFIE, COMIS) introduced multi‑zone capabilities and natural ventilation, yet they still relied on a fixed set of models, forcing users to compromise between speed and accuracy. This limitation motivates the authors to propose a truly modular, hierarchical approach.

CODYRUN’s architecture decomposes a building into three hierarchical entities: zones (thermal volumes), inter‑ambiances (interfaces between zones, including doors, windows, and walls), and components (the physical elements themselves such as wall layers, glazing, HVAC equipment). For each entity a dedicated data structure stores geometric, material, and operational attributes. Crucially, for every physical phenomenon—heat conduction, convection (both indoor and outdoor), long‑wave and short‑wave radiation, airflow, moisture transfer, and HVAC operation—a library of alternative sub‑models is provided, ranging from simple empirical correlations to detailed physics‑based formulations.

Model selection is driven by two user‑defined parameters: desired accuracy and allowable computation time. The software can automatically recommend an optimal combination of sub‑models, but the user retains full control to override the recommendation. For example, in early design a wall may be represented by a single‑resistance conduction model and outdoor convection by a wind‑speed‑based linear correlation, yielding results within seconds. In a detailed design or retrofit audit, the same wall could be modelled with a multilayer conduction‑radiation network, while airflow through large openings would be handled by the Walton model that couples pressure differences with thermal buoyancy, at the cost of longer simulation runs.

The authors describe each model family in depth. Airflow modelling is highlighted as particularly demanding due to its non‑linear pressure system; CODYRUN integrates both network‑type solvers and the Walton approach to capture mechanically ventilated and naturally ventilated scenarios. Sky temperature, often approximated by outdoor dry‑bulb temperature, is instead offered through several empirical formulations that incorporate cloud cover, humidity, and atmospheric emissivity, improving night‑time long‑wave radiation estimates. Convection coefficients for both interior and exterior surfaces are drawn from ASHRAE, ISO, and recent wind‑tunnel studies, allowing users to select the most appropriate correlation for their climate and façade type. Heat conduction models span from simple R‑value calculations to multilayer resistance‑capacitance networks and even coupled conduction‑radiation solutions for high‑performance envelopes.

The simulation workflow proceeds from data entry (via a Windows‑based GUI) through model selection, time‑stepping calculations, and finally result export in graphical, tabular, or CSV formats. Outputs include annual energy consumption, hourly temperature and humidity profiles, and sensitivity analyses that help designers evaluate the impact of design alternatives. Because all models reside within a single platform, designers can quickly iterate with low‑fidelity models and later refine the same project with high‑fidelity models without data migration.

In the discussion, the authors argue that the multi‑model approach delivers three key benefits: (1) it aligns simulation fidelity with the specific stage of the design process, improving decision‑making efficiency; (2) it provides extensibility, allowing new research findings or regulatory changes to be incorporated by adding or swapping sub‑models rather than redeveloping the entire code; and (3) it facilitates communication between practitioners and researchers by offering a common environment where both coarse‑grain and fine‑grain results can be compared and validated. The paper concludes that CODYRUN, through its hierarchical data structures and flexible model library, bridges the gap between rapid design‑stage tools and detailed research‑grade simulators, thereby supporting more sustainable and energy‑efficient building practices.