Ava Center Shopping Mall

The building was not designed to be a shopping mall

Ava Center began as a car park.

The structure was designed, engineered and erected for parking, and in the middle of the plan there was a void — an open well running down through the building, uncovered and open to the sky. In a car park that is unremarkable. A void gives you ventilation and daylight into a space that needs very little of either, and nobody minds if it rains into it.

Then, with the frame already up, the use changed. The building was to become a shopping mall — compact, and positioned at the luxury end of the market. Adaptive reuse of a structure that is already standing, in other words, with the one complication that it was not yet finished.

That decision changed everything about the void, and it changed it in two directions at once.

The first was a problem. A hole open to the weather is acceptable over parked cars and unacceptable over a retail floor. Rain, dust, summer sun and winter cold now arrived in the middle of a space that had to hold finishes, lighting, shopfronts and people.

The second was an opportunity nobody had designed for. A car park does not care about daylight. A shopping mall cannot work without it — and a compact mall, with a plan deep relative to its footprint, cannot be daylit from its walls at any cost.

• Car park: an open void costs nothing Air and a little daylight come down the well and the rain comes with them. Nothing parked underneath minds.

• Shopping mall: the same void is a hole in the roof The weather now arrives in the middle of floors that have to hold finishes, lighting, shopfronts and people.

• Glazed: closed to the weather, open to the light A glass roof over the void keeps the weather out and still lets daylight down into the middle of a plan no wall can reach.

Section, schematic. The levels drawn are not the building's own.

That is not an opinion, it is geometry. A window usefully lights about twice the height of its head above the floor, so a 3 m head reaches roughly 6 m into the plan and no further. Beyond that the lights are on at midday. Retail without daylight reads as a basement, and people do not choose to shop in a basement.

• Side light stops; roof light lands A window lights a strip about twice its head height deep, and adding glass to the wall does not push that line further in. An opening in the roof lights whatever is under it — including the middle of a plan no wall can reach.

Rule of thumb: the side-lit zone is about twice the window head height deep. Section, not to scale.

The longer version of that argument, and the types of roof that answer it, are in our guide to skylight types.

So the void that was a leftover in a car park was the only thing that could daylight a mall — provided somebody could close it.

Why the obvious answer was the expensive one

The straightforward way to close a hole in a building is to build something across it. Masonry, concrete, a conventional floor or roof deck. We looked at those routes first, and at several others, and every one of them failed the same two tests: cost and programme.

The reason is the same in every case, and it is worth setting out because it applies to every change of use on a standing structure.

Anything heavy has to be carried, and carrying it means new load paths.

A masonry or concrete closure across the void is a significant weight landing in the middle of a plan where no structure was ever designed to receive it. That load has to reach the ground. In a multi-storey car park it reaches the ground only one way: new columns, driven down through every level below, to new foundations under an existing building.

• A heavy closure needs its own way to the ground Masonry or a concrete deck puts weight in the middle of the plan, where nothing was designed to receive it: new columns through every level below, new foundations under a standing building, and a parking bay gone wherever a column lands.

• A light roof uses the structure already there A long-span steel and glass roof lands only on the edge of the void, which is already standing. Nothing passes through the floors below, and no load path is added.

Section, schematic. The second is the route taken.

Which produces four costs, and the third and fourth are the ones that decide it:

1. The columns themselves, through each floor.
2. New foundations, excavated and poured beneath a building that is already standing — the slowest and most disruptive construction operation there is.
3. Parking bays lost, permanently. Every column lands somewhere, and in a car park somewhere is a bay. Those bays are revenue, and they are gone for the life of the building.
4. Programme. The change of use had already cost the project time. A structural intervention through every level would have cost far more of it, and until it was finished nothing else could happen.

We did not have that time, and the numbers did not work. That is the honest summary of the options study.

The answer was to not touch the structure at all

The cheapest way to close a hole in a finished building is the way that does not add a single load path.

A long-span lightweight roof spans the void and lands only on its perimeter — on the structure that is already there, already standing, and already sized to carry an edge. Nothing passes through the floors below. No new columns. No new foundations. No bays lost.

The entire solution sits above the problem.

And because the closure is glass rather than concrete, it does not merely solve the weather problem — it does the thing the mall actually needed, which is to bring daylight down to the retail floor through the middle of a deep plan. One intervention, two problems, and the second one is the one that made the building worth converting.

The roof

What spans the void is a doubly curved steel and glass surface of 400 m².

Skylight area

400 m²

Steel structure assembled on site

14 days

New columns or foundations

0

Best Special Steel Structure in Iran

2020

Free-form, because the geometry suited the opening and because a curved surface carries load more efficiently than a flat one — curvature does structural work that a flat roof has to do with depth, and depth over an atrium is depth you look at from underneath for the life of the building.

• Flat: bending Most of the section does very little. To span further the member goes deeper, and the steel grows faster than the span.

• Curved: membrane action Load resolves into tension and compression carried within the surface, so every part of the section works. Curved in two directions, the surface gains this both ways.

Sections. Deflection and curvature exaggerated.

A free-form skylight is not a larger rooflight. Once the surface stops being flat or spherical, every panel is different, every node resolves at a different angle, and the structure has to be found rather than selected from a range.

For Ava Center the surface was modelled parametrically, then rationalised into a panel set that could be cut, glazed and installed without improvisation on site. The steel was engineered around that panel set, not the other way round.

That order matters more than it sounds. A geometry drawn first and rationalised afterwards produces a panel schedule where every unit is unique, which is expensive to make and slow to install. A geometry rationalised as it is designed produces families of repeating panels and nodes, which is what makes a surface like this both affordable and fast.

• Unrationalised: 24 positions, 24 parts Every panel has its own drawing, its own cutting file, its own check and its own place on the scaffold.

• Rationalised: three panel families Panels grouped to share a radius, a mould or a cutting pattern within an agreed tolerance — made, checked and fixed in batches.

Illustrative. The counts are the drawing's, not this roof's.

Three things were resolved in the model before anything was cut:

Panelisation. How the surface divides into panels that can actually be made and glazed, and how many of them repeat.

Node geometry. Every intersection on a doubly curved grid meets at its own angles. The node set has to be rationalised into families, or every connection becomes a bespoke component.

Daylight. Illuminance and glare modelled rather than assumed, so the roof performs as well as it reads — because a skylight over a retail floor that produces glare on the shopfronts has solved the weather problem and created a trading one.

The glazing was specified against span, climate and load. On the systems we build, overhead glass runs 40 – 70% light transmission, a solar factor of 25 – 50% with a solar-control coating, and 1.0 – 1.6 W/m²K double-glazed. The trade that cannot be avoided is that transmission and solar gain move together — more daylight means more heat — and on a Tehran roof, where cooling load dominates, that balance is set deliberately rather than by default.

• Clear Daylight and solar heat arrive together and pass together. More light means more heat, one for one.

• Solar-control coating A coating reflects more of the invisible infrared than of the visible light, so it separates the two — partly. Some light is always lost with the heat.

• The trade you cannot avoid Asking for the most daylight and the least solar gain is asking for two different glasses. Choose which one the climate says matters more.

Arrow weights illustrative, not measured.

Fourteen days

The steel structure of the roof was assembled on site in fourteen days.

That figure is the whole argument for resolving geometry in a model rather than on a building. The steel grid of a 400 m² doubly curved roof, every node at its own angle, going up in two weeks — which is only possible if every member arrived knowing where it went, and if the node it arrived at was already in the right place.

It mattered commercially as much as technically. The change of use had already taken time out of the programme. The roof was the element that had to give time back rather than take more of it. A masonry closure with new columns and new foundations beneath a standing building would have been measured in months, and the mall could not open until it was finished.

What CORTEX did, and what we did not

Worth being clear about, because the building has a strong architectural identity that is not ours.

The concrete diagrid that wraps the elevations and carries the glazed upper floors is Fluid Motion Architects' work — Catherine Spiridonoff and Reza Daneshmir. It is what the building is recognised by from the street.

CORTEX's scope was the free-form skylight over the central atrium: the geometry rationalisation, the structural engineering, the panel and node set, the fabrication and the installation. The main contractor was Paya Takht-e Jamshid Holding, the group CORTEX belongs to.

The project was recognised with Best Special Steel Structure in Iran, 2020.

What this project demonstrates

Four things, and the first one is the one worth remembering:

• Adaptive reuse on a standing frame is a load-path problem before it is an architectural one. The question that decides the budget is not what the closure is made of — it is whether it needs a new way to the ground. That holds for every change of use, not only this one.
• A lightweight long-span roof can be the cheapest intervention available, even when its rate per square metre is higher than the alternatives, because it lands on structure that already exists.
• Geometry resolved in the model goes up in days. The steel structure of a 400 m² doubly curved roof, every node at its own angle, assembled on site in fourteen days.
• A void is worth what the building below it needs. In a car park it was dead space. In a compact mall it was the only route daylight had to the retail floor — and covering it correctly is what turned a liability into the feature the plan is organised around.