Due to negative environmental impacts caused by the building industry, sustainable buildings have recently become one of the most investigated fields in research. As the design technique itself is mainly responsible for building performance, building energy design optimization is of particular interest. Several studies concentrate on systems, operation, and control optimization, complemented by passive strategies, specifically related to the envelope. In building physics, different architectural considerations, in particular, the building’s shape, are essential variables, as they greatly influence the performance of a building. Most scientific work that takes into consideration building geometry explores spaces without any energy optimization or calculates optimization processes of a few basic variables of simplified space geometries. Review studies mainly discuss the historic development of optimization algorithms, building domains, and the algorithm-system and software framework performance with coupling issues. By providing a systemized clustering of different levels of shape integration intensities, space creation principals, and algorithms, this review explores the current status of sustainability related shape optimization. The review proves that geometry design variable modifications and, specifically, shape generation techniques offer promising optimization potential; however, the findings also indicate that building shape optimization is still in its infancy.

The building industry is responsible for a significant degree of energy consumption in the world, causing negative climate changes and energy supply uncertainties due to low energy efficiency as well as the high resource demand of construction. Consequently, energy design optimization has become an important research field. Passive design strategies are one of the most definitive factors concerning energy-related building development. The given architectural problem calls for a method that can create all potentially feasible building geometries, thus guaranteeing the optimal solution which is addressed in the current paper. To reach this requirement, the necessity of a modular space arrangement system and architectural selection rules were determined, focusing on the relationship between the rules and the generation of geometries with mathematical rigor. Next, the architecture-based congruency analysis performed, further reduced the number of simulation cases. With the simulations, it is illustrated how the building shape versions affect the heating energy demands: the performance of the configurations themselves. Results clearly illustrate the importance of the synthesis step of the architectural design.

Optimal building design in terms of comfort and energy performance means designing and constructing a building that requires the minimum energy demand under the given conditions while also providing a good level of human comfort. This paper focuses on replacing the complex energy and comfort simulation procedure with fast regression model-based processes that encounter the building shape as input. Numerous building shape descriptors were applied as inputs to several regression models. After evaluating the results, it can be stated that, with careful selection of building geometry describing design input variables, complex energy and comfort simulations can be approximated. Six different models with five different building shape descriptors were tested. The worst results were around R² = 0.75, and the generic results were around R² = 0.92. The most accurate prediction models, with the highest level of accuracy (R² > 0.97), were linear regressions using 3rd power and dense neural networks using 1st power of inputs; furthermore, averages of mean absolute percentage errors are 1% in the case of dense neural networks. For the best performance, the building configuration was described by a discrete functional point cloud. The proposed method can effectively aid future building energy and comfort optimization processes.

The vision of the research project is to find an energy optimal building configuration, suitable for specified requirements and restrictions. The first step on this way is to create a measure to compare building configurations, faster than explicit energetic simulations. The current study examines the applicability of multivariate linear regression to support the solution of building optimization problems. During the study, multivariate linear regression models were created to estimate the expected annual heating energy demand of building configurations and examined their accuracy Between examinations, the models were modified so that the complexity was increased only to such an extent that the approximation was still sufficiently accurate. The result was a multivariate linear model that estimated the expected output for unknown descriptive variables with a 0% relative error and a 1.6% standard deviation. The R2 point of the estimates was 0.9884. Based on these, the model was considered applicable in the search space defined by the training patterns.

The emergence of large language models (LLMs) has opened new avenues for integrating artificial intelligence into architectural design workflows. This paper explores the feasibility of applying generative AI to solve a classic combinatorial problem: generating valid building geometries of a modular family house structure. The problem involves identifying all valid placements of six spatial blocks under strict architectural constraints. The study contrasts the conventional algorithmic solution with generative approaches using ChatGPT-3.5, ChatGPT-4o, and a hybrid expert model. While early GPT models struggled with accuracy and solution completeness, the hybrid expert-guided approach demonstrated a successful synergy between LLM-driven code generation and domain-specific corrections. The findings suggest that, while LLMs alone are insufficient for precise combinatorial tasks, hybrid systems combining classical and AI techniques hold great promise for supporting architectural problem solving including building geometry generation.

Visual identification of objects is an important challenge today. Main target of frequently applied methods is to identify or classify complex objects. These methods are far less effective when objects are small and less complex, and thus less descriptor features are on hand. The main reason for this is that these features can significantly change on object occlusion or appearance of noise. The presented solution performs identification of simple, small (size is 17 × 13 pixels) objects with elliptical shape. High pass filtered normalized cross correlation is used for region of interest detection and a simple deep neural network is used for classification of selected regions. The proposed method detected objects on a noisy image with accuracy of 96.2%.