Intelligent multi-objective optimization of thermal comfort and ventilation performance in stratum ventilation design

Why the air around your desk matters

We spend most of our lives indoors, yet the invisible air flowing around us strongly shapes how healthy, alert, and comfortable we feel. Heating and cooling systems are usually designed piecemeal—one tool to predict conditions, another to cut energy use, a third to support decisions—leaving building operators to juggle trade‑offs by trial and error. This study shows how these pieces can be stitched together into one “intelligent” design process for a promising system called stratum ventilation, which delivers fresh air directly into the breathing zone instead of simply stirring the whole room.

Targeted fresh air instead of one‑size‑fits‑all

Traditional systems often mix all the air in a room or push it from the floor upward. Stratum ventilation takes a different route: it supplies clean, conditioned air horizontally at about head height, so occupants breathe fresher air with less effort and, in many cases, with less energy. The challenge is that comfort, air quality, and heating efficiency pull in different directions. Change the speed of the supply jet, the angle of the vent, the temperature of the air, the warmth of the wall, or even how warmly people are dressed, and these three goals can improve or worsen together in complex ways. The authors use detailed computer simulations of a typical office and turn them into data that can teach an intelligent system how these factors interact.

Teaching a computer to predict comfort and freshness

From 50 carefully validated simulations, the team trains artificial neural networks—computer models loosely inspired by the brain—to predict four key outcomes: how warm people feel on average, how long air lingers before being replaced, how different temperatures are between head and ankles, and how efficiently heating energy is used. They then let two search methods, a genetic algorithm and a “Harris hawks” strategy, automatically tune the internal dials of these networks so that predictions line up as closely as possible with the simulated data. The evolutionary-style genetic algorithm proves slightly better, reaching correlation scores above 0.995, meaning the model’s predictions almost sit on top of the original simulation results.

Searching for sweet spots, not a single perfect point

Once the computer can predict performance instantly, the authors let a multi‑objective optimizer explore thousands of possible design settings. Instead of chasing one best answer, it builds a “Pareto front” of trade‑offs: operating points where you cannot improve comfort, or air freshness, or temperature uniformity without hurting at least one of the others. The results uncover clear patterns. People feel most neutral when supply air is fairly fast but not drafty (about 1.18–1.20 m/s), mildly warm (around 22 °C), and when clothing insulation is roughly what you would wear in a light sweater. Freshness improves with small vent angles and stronger jets, which sweep old air out more quickly, while unwanted layering of warm air at the top and cooler air at the bottom is eased by wider vent angles and moderately warm wall surfaces. Remarkably, heating efficiency stays high and almost constant across all these competing solutions.

Turning a cloud of options into concrete choices

For designers and facility managers, a cloud of equally good options is still a practical puzzle. To make the results usable, the authors apply a decision‑making method called VIKOR that ranks the optimized solutions under different priorities. They build ten representative “scenarios.” One favors pure comfort—ideal for executive offices or hospital rooms. Another focuses on rapid air refresh, better suited to clinics or crowded classrooms where infection risk is a concern. Others balance comfort, freshness, and vertical temperature uniformity for large halls, gyms, or open‑plan offices. Each scenario comes with specific ranges for vent angle, air speed, air and wall temperature, and expected clothing levels, turning abstract optimization into simple knobs a building operator can set.

What this means for everyday buildings

To a non‑specialist, the message is straightforward: we no longer have to guess our way to comfortable, healthy, and efficient indoor air. By combining advanced prediction tools, automated search, and transparent ranking of options, this study offers a roadmap for tuning stratum ventilation systems to different types of spaces and priorities. In practice, that could mean offices where people feel comfortable without blasting the heat, hospital wards where fresh air reaches patients more reliably, and large venues where warm‑head, cold‑feet discomfort is kept in check. The work demonstrates that intelligent design can turn the abstract promise of better ventilation into concrete, adjustable settings that work in the real world.

Citation: Hammouda, N.G., Ahmed, Z., Omar, I. et al. Intelligent multi-objective optimization of thermal comfort and ventilation performance in stratum ventilation design. Sci Rep 16, 6272 (2026). https://doi.org/10.1038/s41598-026-36233-7

Keywords: indoor air quality, thermal comfort, stratum ventilation, energy-efficient buildings, machine learning optimization