Walk into any large shopping mall and you rarely think about the air itself. It just feels right — until, on one level, it doesn't. A corner near a glass façade feels sticky. A seating area under a skylight feels colder than it should. An indoor fountain leaves the nearby food court oddly humid. None of this happens by accident, and none of it gets fixed by accident either. It gets fixed — or avoided entirely — before construction, using computational fluid dynamics.

This article walks through a real category of project we see often: a large-format retail atrium where the mechanical design looks correct on paper, but the client wants proof that it will actually deliver comfortable air at every level, in every season, before a single duct is installed.

01 Why "Comfortable Air" Is Harder Than It Sounds

Thermal comfort is not just a temperature setpoint. It's a combination of air movement, temperature stratification, humidity, and the heat sources scattered through a space — people, sunlight through glazing, roof gain, kitchen exhausts, and in this case, decorative water features. A design can hit the right numbers on a spreadsheet and still fail in the real room, because air doesn't move in straight lines. It rises where it's warm, pools where it's still, and gets pulled sideways by anything with a fan attached to it.

That gap between "designed on paper" and "behaves in the room" is exactly what CFD is built to close. Instead of guessing, engineers build a full 3D digital twin of the space and simulate how air, heat, and moisture actually move through it.

02 The Brief: A Vertical Garden Atrium in a Middle Eastern Mall

The project centered on the central garden atrium of a large mall — a soaring, multi-level void roughly 7,800 square meters in floor area and 37 meters tall, finished with landscaped planting, seating, decorative interiors, and several artificial fountains cascading down through the levels.

The mechanical design already existed: supply air grilles positioned at every level, an underfloor cooling system supporting the ground plane, and return air grilles pulling air back out. The question wasn't whether the system existed — it was whether it would actually work, level by level, season by season, with fountains running.

There was one added variable most HVAC designs don't have to account for: standing and moving water at every level. Fountains add continuous evaporation, and evaporation adds moisture to the air. The client's real question, beyond comfort in general, was specific — would the fountains push relative humidity past a comfortable range?

03 Building the Digital Twin

A CFD model is only as trustworthy as the geometry behind it. For this study, the model reproduced the atrium in real detail — not a simplified box, but the actual obstructions that change how air behaves: planting beds, decorative columns and interiors, seating and tables, and the fountains themselves, each represented at their real size, position, and water surface area.

Every supply and return grille was placed exactly where it sits in the mechanical drawings, at its designed flow rate. Heat loads were layered in one by one: occupant load scaled to expected footfall, roof transmission gain, and solar sensible gain through the skylight and window glazing, applied where the sun actually strikes the space rather than as a flat average. Evaporative moisture release from each fountain was added as its own boundary condition, since a static water body and a splashing fountain don't add moisture to the air at anywhere near the same rate.

Two Seasons, Two Very Different Problems

• Summer condition — high external heat load, glazing gain at its peak, cooling system working hardest to hold 24°C uniformly across a 37 m vertical span.
• Winter condition — lower external load but a wider acceptable band (20–24°C), where the risk shifts from overheating to uneven stratification between the ground floor and upper levels.

Both were run as steady-state simulations, solving simultaneously for velocity, temperature, and relative humidity across the full volume — not just at floor level, but through the entire height of the atrium where people actually sit, walk, and linger near the fountains.

04 What the Simulation Found

This is the part that makes CFD worth doing: the model didn't just confirm the design — it found where the design fell short.

In the summer run, temperature and relative humidity were not uniformly maintained across every level. Some floors tracked close to the 24°C target; others drifted warmer, particularly where solar gain through the skylight combined with people load in a way the original grille layout hadn't fully accounted for. The winter run told a different story — overall loads were lower, but temperature stratified unevenly between levels, a common outcome in tall atrium spaces where warm air rises and pools near the upper floors while lower levels run cooler than intended.

"A design that looks balanced in a spreadsheet can still be unbalanced in the room. Height, glazing, and water all change the story — and only a full 3D simulation shows you where." — Simulation methodology note, SuperDesignTech Building Physics Practice

Turning Findings Into Fixes

Identifying a problem in a model is only useful if it leads to a buildable fix. The recommendations that followed were deliberately practical:

• Grille relocation at levels where supply air wasn't reaching the occupied zone effectively, rather than adding new equipment.
• Adjusted supply air temperatures, raised or lowered per level and per season, instead of applying one blanket setpoint to a 37 m tall space.
• A second verification pass — the revised design was modeled again to confirm the fixes actually closed the gap before anything was finalized for construction.

That second pass matters. A recommendation is a hypothesis until it's re-simulated. Running the CFD study again against the revised design turned "we think this will work" into "we confirmed this works."

05 The Bigger Lesson for Large-Space HVAC Design

Projects like this point to a pattern that shows up across malls, atria, airports, and large places of worship alike: the bigger and taller the volume, the less a single design assumption can be trusted to hold true everywhere at once. Water features, glazing orientation, footfall patterns, and vertical height all interact in ways that are genuinely difficult to predict by hand.

CFD doesn't replace the mechanical designer's judgment — it pressure-tests it. It's the difference between hoping a large space will feel right and knowing, level by level and season by season, where it will and where it still needs work.