Tag Archives: high-density housing

Sky Rectangle

Daylighting is perhaps the biggest problem when repurposing a shopping mall as residential, and ventilation a close second. If the average width of my demonstration mall is, say, approximately 100 meters, then the closest exterior surface is as much as 60 meters away or, if the atriums are seen as sources of ventilation and daylight, 35. There are seven levels of this, five above ground and two below that, with some contrivance, could also have side lighting. Approximate dimensions are shown below and the simplified column grid is 8.4 x 8.4 metres. The black rectangles indicate elevator and fire escape stair cores.

First thought.

Treating the problem as a conventional one of giving every habitable room a window isn’t going to work as it squanders perimeter (and atrium) surface area and produces apartments 30 metres deep by 4 metres wide on either side of an artificially illuminated and ventilated central access corridor. It’s the energy-hungry Lake Shore Drive typology and would look something like this.

The central access corridors could be widened in places and, say, 8m x 8m sections of the floor removed to create atriums to bring light and create event in much the same way as shopping mall atriums do for rows of storefronts. However, the difference is that stores want to attract the attention of passers-by and to make the transition between corridor and store as invisible as possible. Residential use has different expectations for access, circulation and amenity and this makes it preferable to have individual residential zones separated from “the street” by a third zone neither circulation nor residential. With detached houses, this zone is usually a garden but it could be a porch or some other type of transitional space. I won’t reject this idea outright but it has an inherent unevenness of daylight distribution.

Second thought.

I keep thinking more lightwells are going to have to pierce the slabs in order to bring a necessary minimum of daylighting and ventilation but this would need to be done without removing any of the column and beam structure. I also thought to pinwheel accommodation around vertically shared lightwells in order to reduce the overlooking as in the example at the end of the New Squeeze post. The attempts in the middle of this next sketch made me think it wasn’t going to work. Too much space would be used to access the units and to too little effect. There was little legibility and not that much daylight either, especially on the lower levels where sideways (east-west) daylight penetration would be obstructed by accommodation evenly distributed in two directions.

The New Squeeze proposal didn’t have this problem as the first (ground) floor was dedicated to accessng the three levels above.

Third thought.

My third thought was to arrange the accommodation in “streets” primarily lit by lightwells but with the long sides of the mall (i.e. the ends of those streets) open to the east and west sides of the mall. Noontime sun would illuminate the atriums. Or so I think. As a reference, I drew upon the linear inclined mat section of Kiyonori Kikutake’s 1973 Pasadena Heights that I continue to learn from. Three apartments partially overlap with shared lightwells lighting different parts of the deep plan.

I learned from studies such as Pasadena Heights 3.0 and The Rooftop that it was possible to stack terraced apartments with very deep plans, with the deepest and most internal parts of the building used for car parking still naturally illuminated and ventilated to a degree befitting non-residential use. Whatever this new proposal turns out to be, it will be called the Mat-rix House.

I once lived in a basement flat in Kensington Garden Square near London’s Queensway. The bedroom was at the lowest level of a seven story lightwell approximately 2.5 x 2.5 m. Man, it was dim.

First look.

I thought of have a plan provided with daylight and ventilation via lightwells and for those plans to be overlapped for five or six levels both horizontally and vertically. Instead of each living unit facing open sky as at Pasadena Heights, it would face the rear of another unit across an access street. You can think of it as shikumen in a three-dimensional matrix. Apartments would be primarily lit by ambient light while direct daylight and outward views would become communal amenity at the ends of streets. Something had to give but the question is if there will be any compensating advantages.

The shaded area in the diagram below represents a living unit approximately 150 sq.m in area. This is large, and may turn out to be too large, but is a consequence of the grid dimensions. Living units are arranged on opposite sides of an 8 meter walkway, and accessed from two directions past 8-metre long lightwells separating them from the street. This arrangement will use exactly 50% of the available floor plate, not counting the lightwells. Providing two levels of accommodation per floorplate would give a per level FAR of 1.0 and a per building FAR of 10.0 for five floors, or 14.0 if the two basements are included. The staggering of access corridors and the visual links between them do more than just promote airflow. I think there would be an awareness of actually living in an inhabited matrix and this is much more than streets in the sky ever did.

The largest downside is that bedroom windows of different apartments overlook each other diagonally up and down from a (horizontal) distance of eight meters. This could be solved by stacking the corridors but then the long views through the matrix would be lost and the bedroom windows would still be eight meters away from an access corridor. I won’t forget about this but I won’t fret too much about it either. The compromises of having entire apartments overlooking narrow streets are not new and solutions and workarounds to them are not new either. Mostly, they involve some sort of curtain or blind or shutter at night.    

Third thought, second look

The area per apartment was large and so, rather than having two levels, I arranged two-storey apartments based on a plan I’d already made, back to back. (This was a big decision and perhaps I made it too early because it means that there will now be two service risers per lightwell, instead of one.)

There are four options for apartment planning. All have one bedroom and bathroom at entry level, and another bedroom and the living areas upstairs. There are two positions for the bedrooms and the only variable is where they are placed. The yellow block is the open space for the apartment of the colour above.

NB: I’ve just finished reading Moshe Safdie’s recent book. These next images are not clever photorenders but photographs taken in the garden of configurations of actual (vintage) LEGO-like blocks.

Type I is a two-storey L-shape back-to-back plan I had from a previous project. The garden is replaced by the lightwell, the bedrooms are stacked and both share the lightwell by the outdoor space at entry level. As with any back-to-back plan, there’s no horizontal through ventilation.

Type II is a linear layout with both bedrooms stacked where they share the lightwell with the outdoor space and the living areas of the paired apartment. However, it has horizontal through ventilation on both levels.

Type III has the Type I layout on the lower level and the Type II layout on the upper level. It therefore has horizontal through ventilation on the upper level but the lower bedrooms share the lightwell with the open space of the paired apartment.

Type IV has the Type II layout on the lower level and the Type I layout on the upper level. There is therefore horizontal through ventilation on the lower level but the upper bedrooms share the lightwell with the open space of the paired apartment. This variation.

These four Types have minor differences for under-stair storage and such but the most important is the degree of compromise between ventilation and privacy. Externally, they all look the same. Type II has the best through ventilation but the most compromised privacy. Type I has the least compromised privacy but no horizontal through-ventilation. Type IV has horizontal through ventilation but not for the living areas where it is preferred. Type III is the one I‘m going to proceed with.

It’s not ideal on either count but is the least compromised. This is what those bedroom windows look like from the open area (left) and from the Level 4 access corridor (right) looking up to levels six, eight, and the rooftop. I’ve added utility pipes and conduit runs.

NOTE: Problems with bedroom windows could be avoided by placing both bedrooms above the open space so they face the blank wall on the other side of the lightwell. The bathroom would stay on the entry level, but the living spaces would be split into Types I–IV with the dining-kitchen and living area split across the same four positions. This makes sense because (apart from the kitchens and bathrooms no longer being stacked) all that’s happened is that the areas of the living spaces and bedrooms have been swapped.

In the final fitting into the demonstration mall, no apartment would be more than seven bays away from the outside proper or more than three from either the outside or a large atrium. I don’t generally like the artificiality of “photorenders” and how they overpromise but, given the nature of this project, even inexpert ones such as these give a better idea of the expected level of daylighting. Nevertheless, they are only an approximation of what I expect it would be like. In that sense they’re as true as any other photorender you’ll see. “Realistic” let alone photorealistic has no meaning for things yet unbuilt. I know I can push ambient light as much as I like but I simply don’t know if I’ve pushed it too far or underestimated it with respect to what the built reality would be. I see this proposal more as Walden 7 in Barcelona than Habitat in Montreal.

Having said all that, if this is a reasonable approximation of a level of illumination for apartments seven stories down and lit by lightwells, then it’s not bad. Whether it’s appropriate or not depends upon climate, latitude and – as anywhere – whether it’s day or night.

Access to direct daylight and open “space” is communal rather than private and this is how it must be if an equal distribution of both is desirable. This goes against the history of residential architecture and residential architectural aesthetics framed in terms of the abundance of space and light for some. If we want an acceptable minimum for all, then different rules will apply and a different architecture will result. 

Also worth mentioning is that it’s not unusual for a project to be fitted into some given (or desired) shape. This exercise took a given structure as the starting point and attempted to fit a project into it and, to my mind, was successful in fitting it in a way that adds value to that structure. The next thing is to apply this method with all its known advantages and shortcomings to the actual (demonstration) structure with its various atriums and cores.  

The point of this exercise was to show that some unconventional solution might result. Nobody is going to build a structure like the one I’m taking as a starting point in order to build apartments or any other kind of accommodation. If a residential repurposing of such a structure were to happen, the only financial advantage would be that the accommodation itself can be relatively flimsy. The primary structure is massive and protective and all that’s required of the secondary one is that it support itself.

Any residential proposal for the demonstration mall will be determined by the floor-to-floor height and the dimensions of the structural grid. My attempt to fit accommodation in this way, into the demonstration mall will be a separate post. Optimizing the typology itself will be another and my first thought is to use a floor-to-floor of 3.0–3.5 metres, a structural grid of 5.5 x 5.5 metres (two car parking bays, one-way road) and single-level apartments. It would have infill panels within a lightweight steel or timber frame structure. It would be a mat-rix landscape rolling over pockets of car parking.

• • • 

The Japanese Machiya

The machiya (町家) is a Japanese urban housing type. Their name translates as “town-house”. They’re basically rooms around a two-storey lightwell. We’re familiar with Tadao Ando’s 1976 Sumiyoshi House that’s said to be a modern interpretation.


Waro Kishi’s 1987 Kim House is less familiar. Having the bridge on the side makes its courtyard brighter and less oppressive than the courtyard in the Ando house. The machiya is simply one of those housing typologies that works irrespective of time, materials and construction. You’d probably have to be Japanese to appreciate the “closeness to the seasons” aspects of them that we, from a warm distance, see such charm in. Despite the Japanese tendency to over-asetheticize the beauty of seasonal change rather than add some insulation, these little houses have much embodied intelligence. The annual average humidity in the Kansai region and Kyoto in particular is around 70% but when combined with temperatures above 30°C, is extremely oppressive. Moisture doesn’t need much prompting to condense. Evaporative cooling is your friend and thus so is cross ventilation. Luckily, wind speed picks up in summer. These tight-packed houses work better with a central courtyard that, like courtyards around the world, contains cool night air longer into the day and generates heat-shedding convention airflow at night. This recent detached house uses those same principles on its 55 sq.m plot with 50cm setbacks on all sides to ensure ventilation around and across. Somewhat humbly, it reminds me of one of mine from years back except, in Gravesend, Kent, I was more concerned with overlooking than cross ventilation. These houses were capable of being either row houses or cluster houses but any window on an external wall was a bonus for daylight, ventilation and view.


Even Ando’s Sumiyoshi/Azuma house has paired small windows low down on the walls of each room to facilitate airflow, if not cross ventilation. You can see two of them in this next image. Please notice that rear corner while we’re around the back. Waro Kishi’s 1986 Kim House was rebuilt in 2011 to have a central internal void rather than a courtyard.


This is another recent incarnation of the machiya as two joined houses. It’s close to perfect. I particularly like the distinction between the habitable rooms and the non-habitable spaces on the ground floor. Notice how it’s possible to go in one house and out the other without passing through a habitable room? You can see it’s an updated version of this next plan. It hasn’t changed much. In fact, it’s the same. The room on the left is a shop facing the street and with an inner room that faces another inner room across a courtyard. Many machiya have shops on the street side – it’s the reason why they’re in cities.  In this next example, the front shop and the non-habitable spaces all have earthen floors. (You only have to change shoes once.) This next is a house above a restaurant for maybe ten people. Family meals are cooked in the shop. The central courtyard has become a void above the kitchen, topped by a skylight opening into a third-floor courtyard/lightwell/airshaft.

There’s the shop at the front accessed from the lobby that leads into the house proper via the kitchen. Above the kitchen is a double-height space onto which all rooms face ensuring cross ventilation. Heat from the kitchen would generate a stack effect in that double height space to induce cross ventilation even when there was no wind. Much like Arabian wind catchers, it’s not that effective, but it’s better than nothing. The principles work and for some people they’re enough – but obviously not these people, as evidenced by the disguised A/C compressors.


We need more typologies like this. The closest we have to such perfection in Western/European architecture is probably the corridor-accessed one-bedroom apartment. This next example is perfect.


The circulation space isn’t an extra – it’s what makes it work. You can’t have more than six doors opening into 2 sq.m. The extra 1 sq.m comes from having to walk past the bedroom wardrobe and the kitchen units to get to two of the doors. It’s beautiful. There’s other things interesting about this project but “like clouds floating across the landscape” isn’t one of them.

Some of the plans in this next project for social housing in São Paulo by Vigliecca & Associates are extremely evolved.

Their type C and type D are good

but I particularly like their 2-bedroom type A not least of all because its rigour reminds me of the Type A from the 1928 Types Study by the team led by Moisei Ginzburg and whose story we shall resume shortly.

A four- and five-person Type A


How Other Folks Live

The standard “Western-style” apartment block typology is fairly resistant to change. LMvdR hit it bang on at Lake Shore Drive. Central access core, a long corridor, party walls linked to structure, central bathrooms and kitchens mechanically-ventilated, peripheral habitable rooms with views. 

from a 1957 marketing brochure
from a 1957 marketing brochure
openable windows

And nothing much has changed since. This post is about how the apartment building typology is being improved upon in other countries. It’s not some condescending “critical regionalism” based on “others” trying to be modern in their own way – usually by adding lots of lovely hardwood shutters and shading devices to your standard white building. Instead, on the very basic level of architectural response to climate, these other apartment buildings are, on the whole, better – apart from in Australia. 


Bangalore, India. This is Mantri Tranquil. It’s not bad. Habitable rooms face outwards, non-habitable rooms face inwards. All are naturally ventilated as natural ventilation is all there is. This would be mostly induced by convection, but some due to general turbulence or a weakened Venturi effect.  The configuration itself is basically semi-detached houses on top of each other, arranged to share vertical access as efficiently as possible. It’s large surface area per apartment is expensive to build, but it’s what gives it its advantages. 

garnet floor plan

There’s scope to increase surface area even more by, say, six cylinders accessed from a central seventh cylinder – “a honeycomb that’s all honey!”. But then you get six apartments, not eight – bummer. (Maybe go for 12?) Pune, India This is Sobha Garnet. It’s no longer a cluster but it still uses the same passive design principles – less effectively but with less surface area. Differential air pressure front and back would create stronger cross ventilation. Variations don’t improve upon it, but introduce the idea of corridor access and lightwells/ventilation shafts.

Floor plans new size

Chennai, India This is Orchid Pearl. Spaces between the buildings have atrophied into lightwells but all rooms are still naturally ventilated. As soon as there’s corridor access to everything, those corridors can be made any length. And they are. As long as the light wells are open at the bottom, we can expect the Venturi effect to create updrafts ventilating non-habitable rooms and cross-ventilating any adjacent habitable rooms.


Penang, Malaysia This is Camellia Park. All rooms are naturally ventilated but now, shrunken shafts ventilate and might illuminate a bedroom, bathroom, kitchen, and laundry drying area.

Despite their various shortcomings, it’s still possible to live in all these apartments without mechanical ventilation or, if need be, elevators. 


Kuala Lumpur, Malaysia This is Cempaka Sdn Bhd.  Here it starts to get nasty! Mechanical ventilation is now essential for all apartments. Studio apartment sizes range from 38.6 sqm. to 52.6 sqm. (416 sq.ft to 566 sq.ft). Which isn’t bad. But it really is time for these small apartments to stop pretending they’re larger than they are. In Malaysia, people prefer to greet their visitors in a space with no natural light rather than let them glimpse where they sleep.

Tokyo, Japan This is Roppongi Hills. Roppongi Hills is a big, urbany, mixed-use development much hailed as the shape of future living in Tokyo.