GerAPlanO - A new building design tool: design generation, thermal assessment and performance optimization

Building practitioners (architects, engineers, energy managers) are showing a growing interest in the design of more energy efficient and livable buildings. The best way to predict how a building will behave regarding energy consumption and thermal comfort is to use a dynamic simulation tool. However, the use of this kind of tools is difficult on a daily basis practice due to the heuristic and exploratory nature of the architectural design process. To deal with this difficulty, the University of Coimbra and three companies have been working on the development of a prototype design aiding tool, specifically devoted to the space planning phase of building design, under the project GerAPlanO (Automatic Generation of Architecture Floor plans with Energy Optimization). This project aims to combine the capabilities of design generation techniques, thermal assessment programs, and design optimization methods to provide assistance to decision makers. This paper presents the overall concept, as well as the current status of development of this tool.

💡 Research Summary

The paper presents GerAPlanO, a prototype decision‑support tool that integrates automatic floor‑plan generation, dynamic thermal performance assessment, and multi‑objective optimization to aid architects, engineers, and energy managers during the early, space‑planning phase of building design. Recognizing that dynamic simulation tools (e.g., EnergyPlus) provide the most reliable predictions of energy consumption and indoor comfort but are rarely used in day‑to‑day practice because of the exploratory nature of architectural design, the authors set out to bridge this gap.

GerAPlanO’s architecture consists of three tightly coupled modules. The first module translates user‑defined functional requirements (room types, required areas, adjacency, orientation, circulation) into a set of mathematical constraints. A meta‑heuristic engine—implemented with genetic algorithms and particle‑swarm techniques—searches the constrained design space and produces thousands of candidate floor plans. Each candidate is automatically screened for geometric violations such as overlapping spaces, unrealistic aspect ratios, and accessibility issues.

The second module links each viable plan to a dynamic thermal simulation engine (EnergyPlus). Geometry is converted into building envelope data, window placement, internal loads, and HVAC system specifications. The simulation runs a full‑year hourly analysis, delivering key performance indicators: annual heating and cooling energy demand, peak loads, and thermal comfort metrics (PMV/PPD). Results are fed back to the design variables, establishing a closed‑loop evaluation.

The third module performs multi‑objective optimization. Two primary objectives—minimization of total energy consumption and satisfaction of thermal comfort criteria (e.g., PMV within 0 ± 0.5)—are simultaneously pursued. A Pareto front is constructed, allowing designers to visualize trade‑offs between energy savings and comfort. Users can assign weights to prioritize one objective over the other, and the system will suggest the most appropriate design alternatives.

A user‑friendly interface supports drag‑and‑drop manipulation of spaces and real‑time visualization of simulation outcomes, enabling rapid iteration. The tool is built with an open API, facilitating integration with existing BIM or CAD platforms, which is essential for adoption in professional workflows.

The prototype, co‑developed by the University of Coimbra and three industry partners, has been tested on several case studies (residential, office, educational). Results indicate that the optimized plans generated by GerAPlanO achieve up to a 15 % reduction in annual energy use compared with conventional manually produced designs while maintaining comfort within the target range.

The authors acknowledge current limitations: (1) the constraint formulation struggles with highly irregular or curvilinear building geometries; (2) full dynamic simulations are computationally intensive, limiting the number of design alternatives that can be evaluated in real time; (3) the collaborative workflow between architects and engineers is not yet fully defined, requiring further user‑experience research and training. Future work will focus on extending the constraint library for non‑rectilinear forms, incorporating surrogate models or machine‑learning predictors to accelerate thermal assessments, refining the UI for better stakeholder communication, and linking the tool with broader design‑management platforms.

In summary, GerAPlanO represents a significant step toward embedding energy‑performance intelligence directly into the conceptual design stage, offering a scientifically grounded, yet practically accessible, means to generate, evaluate, and optimize building layouts for both sustainability and occupant comfort.