Click the book icon to navigate table of contents
swipe left & right
to navigate
Click the book icon to navigate
table of contents
Click here for
previous articles
Click here for
next articles
Scroll Down for
Feature article
LOADING
NEXT ARTICLE

Architectural

Neural

Network

Image courtesy of Aaron Poupard and Alexandra de los Santos.

WORDS BY Lee Lambrou

In the brain, the ‘starting conditions’ are the specific neural signals arriving from the spine, or elsewhere in the brain. In the case of an ANN, they’re whatever we’d like them to be, from the results of a search algorithm to randomly generated numbers to words typed out manually by researchers.” – Graham Templeton, ‘Artificial neural networks are changing the world. What are they?’ [sic]

Kensington House started with a simple premise, a typical Aussie home extension, which inevitably is stage 2 of the home ownership dream, especially as the family expands. Many have these dreams but it usually comes crashing down in the reality of the budget required. Those who continue with the dream end up with a builder, a drafter, a trip to Ikea and an uninspired result.

This is an experiment in a model for architectural domestic building.

At a gathering, clients and friends, Victor and Alexandra outlined their plans for their little used carpark space and asked if I could help. Faced with the scale of the project yet its complexities and the potential cost of an architectural intervention, most architects would run a mile.

My architectural education took place in Adelaide at a time when the internet was about to arrive on the scene and 3D modelling as a tool had not yet infiltrated our profession. As an early adopter, through interest in 3D gaming design and technology, there was a lot of cynicism and push back by most at uni who saw the use of the computer in architecture as cheating and certainly not as a design tool. Once I had my degree and on a trip to Melbourne, I chanced upon the just completed Storey Hall which cemented my belief that technology had a lot to offer the architectural profession in practise and could make possible a rich original local architecture.

Since then much has changed. BIM and advanced 3D Modelling tools have finally, after a long tortured process, entered the mainstream, although maybe not as a design tool.

So after more thought it dawned on me that this little project could augment my interests and could be used as a research project. As a practitioner and designer, I’ve always been interested in technology and the feedback process between man and machine, as an enabler, like a cerebral version of Stelarc. An artificially augmented intelligence.

As a practitioner and designer, I’ve always been interested in technology and the feedback process between man and machine, as an enabler, like a cerebral version of Stelarc. An artificially augmented intelligence.

In a world of cultural saturation, we are, without choice, thrust into a post-everything paradigm and thus filters are of paramount importance. What better than an augmented, parametric filter and sorter. A parametric augmentation of the design process. Feeding in variables, history, art, anything into the mix and iterating and mixing and mashing. Like a game of chance, I like surprises.

Victor had renovated a house before and was an above average handyman, not to mention an experienced self-employed locksmith, so he was keen to do as much as he could. His wife Alexandra wanted something different and special, a personal one-off. She was also a bit nervous it might look commercial and non-residential, especially knowing I just worked on a desalination plant and a freeway.

They also had to live in the house with their child, toddler and baby during construction. So the owner builder method of procurement was the obvious way forward. It was clear that cost of labour would not be an issue, which could open up a world of materials, techniques and experimentation.

My goal was to develop an Ikea like manual which would enable Victor to put it all together. I treated the project like a piece of virtual joinery. Every component was to be modelled and interrogated in the computer.

The suburb and street has a heritage overlay and is dominated by quaint little renovated Victorian cottages and virtually no modern architectural expression.

The existing building has an interesting history, it was built in 1913, first housing delivery horses and carriages, then a boxing gym, a homing pigeon loft and then modestly converted into a townhouse.

The extension was to occur on the front carpark space and it quickly became apparent that the dimensions of the site along with planning restrictions would self-design the form. There would be no gratuitous form making.

So into the iterative mix went the decorative, pretty lacework of the cottages, and ideas of heritage as memory. Rather than a binary memory, more the idea of a sensorial memory, the kind of childhood memories one has, and the evolution or perhaps devolution of these memories.

I started with a childlike representation of the original stables. In the computer an iterative process of subdivisions, triangulations and tessellations begins. The cycle continues until all the iterations are exhausted or an output produced. The image evolved over time into just the pretty motif I had envisioned and it wasn’t just a dumb graphic, randomly selected on google, but a snapshot of a complex process.

It could be a long lost design for a Victorian doily, or a complex but intriguing network, an artificial neural network, a network of the cultural melting pot. But it would be, ultimately, an embodiment of the design thought process, facilitated by the ANN.

Having explored the use of rain screens in my day job, it seemed like a good idea to build as simple a building as possible, and wrap it in a kind of architectural fabric. This meant that Victor could easily construct the majority of the building himself using the good old staples, Colorbond, brick, timber and a bit of steel. This would not break the bank and would most likely not leak.

I modelled it all, element by element, screw by screw, in BIM. From the model, we extracted shop drawings for steel fabrication, quantities, etc. To my relief, the steel fit perfectly when delivered, the quantities were correct and slowly but surely the base building rose in Victor’s spare time. Victor spent many nights researching construction techniques and watching Youtube instructional videos.

It was time to realise the screens. They had to be lightweight, 35% open (as specified by council), filter the light, not block the impressive city views but provide privacy from the street and embody the pretty motif. A seemingly impossible task.

I chanced upon a sample of expanded mesh which seemed perfect. It was lightweight, strong, efficient and clever in its use of material density. It looked like a bunch of little sun screens woven together like fabric or fish scales. Even though I’d never seen it done before, I didn’t think it would be a problem laser cutting a pattern into it.

I had also just explored the use of polycarbonate on another project. Another great material for introducing light with privacy. And no need to use a glazier. Two perfect candidates.

A.N.N Diagram, Rendering by Lee Lambrou.

Image courtesy of Aaron Poupard and Alexandra de los Santos.

Image courtesy of Aaron Poupard and Alexandra de los Santos.

Patterns Process, Rendering by Lee Lambrou.

We also wanted to use one type of light coloured timber for everything inside and outside. Floors, walls, window frames. We wanted it to look like it was carved out of one block of timber. I came across a miracle timber called Accoya which was the perfect candidate. It’s sustainable and incredibly stable and durable. It’s also soft and easy to work with. We tracked down a local supplier and luckily it was affordable, although it only came in large, thick sizes. Victor found a joiner who had it milled into floorboards, wall panelling, window frame members and we also used it for the external mesh screen framing. Aluminium was way too complex and expensive. Another advantage of the timber is that Victor could easily do it himself.

I used the BIM model to generate an array of cutting sheets for the window framing using custom timber profiles as specified by us in the milling process. It was then fabricated onsite and installed without issue. Same for the joinery.

While on a tram, I noticed how from the outside, the printed advertising film on the glass was opaque, but sitting next to it inside, appeared totally transparent, a combination of the dimmer light conditions inside and the little perforations in the print.

Upon research of local mesh suppliers, I modelled various samples and ran them through sun studies and tested, through computer rendering what looking through them would be like. None of the samples were right. They were mostly too large and the small samples were too heavy and dense. I then spoke to the suppliers about customization of the sheets. Even if I wanted to pay the exorbitant rate for supposed retooling, they would not do such a small order. The response was similar with the polycarbonate, made overseas, which comes in very few sheet widths and needs an expensive proprietary framing system.

A barrier to affordable small residential projects is the lack of access to a large range of affordable or bespoke materials, due to the small quantities required and hence lack of customisation options. My experience in speaking with local suppliers invariably ends with perceptions of being a time waster.

The internet has opened up the complex network of global trade. Sites like Alibaba offer a window into this trade network. Within this vast network, it was not difficult to find a supplier who was more than willing to do a short custom run of expanded mesh for me. They’d never used their product for architectural cladding and were genuinely excited by this potential market.

After much testing and modelling in the computer, I arrived at the perfect combination of size/openness and thickness and they made a sample which I then tested and tweaked further. The price was much cheaper than even the standard local product. The same applied to the polycarbonate which was fabricated to very large widths which eliminated any complex jointing and framing. We had to buy than more than we needed but the rest was sold on ebay to very happy buyers.

It was the global economy at work, but open to everyone and not just the construction giants.

The next challenge was the laser cutting of the expanded mesh. I did try getting this done through Alibaba, but although it was a little cheaper, I felt I needed the first hand customer service and skills of the locals. In the end I couldn’t laser cut it because of the existing perforations, so waterjet cutting was recommended to me. We found a helpful local waterjet cutter. I used parametric CAD tools which quickly adjusted the scale/pitch of the pattern and culled the part of the pattern that was too small to cut. I also biased the openness of the motif over the windows to allow more light in. The total distance travelled by the machine was calculated and I could tweak the pattern to be more efficient. After a few trial runs and fails; ie. the sheets falling apart, we had successful, structurally sound custom double perforated sheets.

I also discovered that by turning the sheets upside down, the view from below was blocked and there was more view to the sky and city. Along with the tram window phenomenon, this helped solve the street privacy issue.

Victor, with his lock smithing skills, knew his way around a door and hinge. He rose to the challenge and helped devise a simple system of shutters using standard hinges and tracks, and managed to conceal them. They could then be easily retracted for glass cleaning and maintenance, while not ruining the façade.

3 years later and its finally complete. No nasty surprises nor unexpected cost blowouts and no leaks.

Our research project has been a success. The result is bespoke, architectural, yet simple.

The effort on my part was more onerous and potentially risky, but luckily I had very trusting and ambitious clients. I feel taking responsibility for much of the construction process is good for a profession in which, at times, our responsibilities are diffused and eroded into a drafting service.

Technological advances in computer software, hardware, materials and trade have increased the scope for an architect to cut out most of the middlemen and complications involved in a traditional small project; and with a willing participant deliver a truly bespoke architectural solution. Having worked on many different sized projects, this process was by far the most architecturally pure and fulfilling. The future is looking very promising indeed, especially when current computing technology is coupled with daily developments in materials technology and the 3D printing revolution.