Work

AI-Driven Optimization of Sanitary Facilities in Office Buildings: A Machine Learning Approach Using LSTM Neural Networks

Year: 2025

Type: article

Abstract: This study introduces an AI-based approach to planning sanitary facilities in buildings, with a focus on Long Short-Term Memory (LSTM) neural networks. By examining factors like occupancy patterns, daily fluctuations, and building-specific characteristics, the model delivers datadriven guidance for fixture allocation. Validation was carried out using empirical data from a Singaporean office building, resulting in strong predictive performance and improved operational efficiency. The LSTM model surpassed traditional methods, with a 39.26% boost in Mean Absolute Percentage Error (MAPE) over queueing theory. In a real-world scenario, its predictions were within 2.63%–4.76% of actual needs, whereas prescriptive codes deviated by 12.07%–19.05%. Sensitivity analysis showed that total occupancy, time of day, and gender ratio exerted the greatest influence, enabling gender-specific recommendations. The findings also indicate a potential 15% cut in overall fixtures with no compromise in service quality, providing considerable cost and space savings. By uniting advanced computational modeling with real data, this research demonstrates how AI can elevate sanitary facility planning toward evidence-based decision-making, efficient resource use, and higher user satisfaction. The results may influence policy, regulatory standards, and performance oriented design in contemporary buildings.

Source: Applied Sciences

DOI: 10.3390/app15052499

Authors: Samson Tan, Teik Toe Teoh

Institutions: Victoria University, Nanyang Technological University

Language: English

Topic: Building Energy and Comfort Optimization

Subfield: Building and Construction

Field: Engineering

Domain: Physical Sciences

SDGs: Sustainable cities and communities, Clean water and sanitation, Industry, innovation and infrastructure

Open Access status: gold

Estimated APC paid: $2,896