Residential Typology of Tomorrow, Today

How do we harness Computational Design in transforming the design process and re-interpreting the creation of space? How do we leverage on the modular system of building in residential design, challenge conventional wisdom and ride on emergent trends in technology as design and implementation enablers? 

On 1 December 2017, Elon Musk, Chief Executive of Tesla – maker of the Model S electric car – promised the world’s largest lithium-ion battery system for South Australia within 100 days. His battery system will be contingency to their conventional electrical grid which had been struggling to cope with current energy demands. He was so confident of the technological capability of the battery system that he promised South Australia will get the system free if he missed the deadline. Many had considered Elon’s twitter declaration hubris and a publicity stunt. However, Tesla did achieve its target. In fact, its battery system has exceeded expectations in its almost instantaneous response to smooth out outages far faster than conventional coal-fired back-up power systems. Credit should be given to Tesla for the confidence in its own research and development; and Elon Musk’s bold move to implement an untested solution to address a prevalent real-life problem.
Similarly, innovative solutions, which upend the conventional methods of problem-solving, can also be selectively applied to challenges in the built environment. Between now and 2050, it has been estimated that a million people per week will move to a city somewhere in the world[1]. There will be unprecedented stress on the building industry to provide residential housing in sufficient quantity while maintaining a high quality of living standards. In a land-scarce city-state like Singapore, the projected density is approximately 10,000 people per square kilometre of land area by 2030[2].
So, can the housing issues of tomorrow be addressed using today’s accepted solutions? For a city to cope with today’s rate of urban migration, it has to build faster, better and smarter; all within an increasingly limited land space. As architects, we are acutely aware of the challenges ahead. We understand the workings of residential design, which is essentially about the design and assembly of different component modules. The modules within a residential unit are the bedrooms, bathrooms, etc. The residential unit modules are stacked up to form a tower module. The tower is given a façade skin and a pre-determined number of them are placed on a given site. The question we face is: how do we leverage on this modular system of building, challenge conventional wisdom and ride on emergent trends in technology as design and implementation enablers? 


The technological magic wand

Computational Design uses algorithmic scripting to translate a set of rules as design intent into a design response. Computation Design can augment the conventional design process to create well balanced, environmentally sensitive buildings.
At a macro level, maximising quality of external views from the bedrooms of a residential tower and minimising solar radiation on the façade to determine the optimum tower orientation are examples of rules that can be incorporated into a parametric script. As the script runs iteratively, it identifies all the possible design options – each with a slightly different weightage assigned to the given set of rules.
In addition, Computational Design can also be harnessed at a detailed level to study how the tower’s external façades can be developed in response to the environment. There can be a mix of different glass types – e.g. single and double-glazed glass systems – and different depths of external horizontal sun-shading ledges specified for a residential tower.
Let’s break this process down further.

Automating design feasibility: The tech-vantage

The beauty of Computational Design is in the strength of the computer to perform complex calculations at speed. It is also useful for performing mundane tasks like populating a module across an entire residential block. This computational workflow reduces the cognitive load on a designer, allowing one to receive instantaneous design feedback. There is great potential, and the gap to bridge is between the designer’s input and his desired output.

Imagine a project with three unit types; each with a unique façade element. It takes minutes to think up a unit mix for the block, which will then affect the façade design. However, to produce five different block options, it might take a designer a few days just to produce three-dimensional models to study their different aesthetic properties. If a parametric model was used, the process of producing study models would be automated, thereby reducing the design feedback time. Potential interactions with the parametric model could include tweaking the percentage unit mix or block placement within a site.
Designers are in turn empowered to study how the generated block performs aesthetically and the success with which it fits into the site synergy. This shows that by dedicating repetitive work to a machine, human designers are allowed more cognitive space to focus on qualitative aspects of design. In this case, the unit type input goes on to produce an interactive parametric residential block model output.

Shortlisted options generated by the parametric script that offers the best cost-effective compromise between view optimisation and energy efficiency can be quickly identified for further development. This can in theory allow for a very productive period at the schematic and design development stages of the design process. The potential of Computational Design in the residential typology lies in the premise that it is essentially designing modules of different scales based on rules within a script.

Measuring site feasibility: Performance data

Besides automating current workflows, computation design can also be supported or driven by performance data like structural information or sustainable design considerations. These performance data are derived through building physics simulations, which mimic real life behaviours. Used widely within other engineering design fields, concepts like fluid flow, ray tracing and finite element analysis can be adapted to the building scale to study overall performance of a floor plan.

While it was previously demanding to juggle between spatial design, code compliance and performance simultaneously, a computation design approach can help take over some of the quantitative aspects of design. Performance data being visualised alongside spatial layouts can also facilitate design collaboration as each stakeholder’s concerns are weighed and visualised within the same parametric environment. A structural engineer’s advice to shift a pillar can be tested immediately against the daylight availability of a room. At the same time, the design architect can also examine how both performance come together to impact his unit layout. In short, this process takes in spatial and performance data to generate a parametric model output, serving as a collaborative platform between previously segregated departments of the building industry.

From digital to actual: Prefabricated Prefinished Volumetric Construction

Modular design also underpins the concept of Prefabricated Prefinished Volumetric Construction (PPVC). PPVC is a construction method in which the different components of a typical residential unit are rationalised into modules complete with internal finishes, fixtures and fittings. These modules are manufactured at off site factories and transported to site for assembly in a tetris-like manner by tower cranes. Aside from improved quality control from the off-site production, PPVC also results in better productivity on site as the bulk of work is already done prior to installation; and an improved construction site environment with less noise and dust pollution to the existing surrounding developments.
The use of PPVC is already applicable for selective land parcels under the Government Land Sales (GLS) programme on 1 December 2015 onwards. The minimum level of use of PPVC shall be 65% of the total super-structural floor area of the building or the component of the building that is to be used for residential or private dwelling purposes.

Digital-abled: The man-machine collaboration

If Computation Design is driven by a numeric input and PPVC functions like a Lego set, are architects throwing aesthetics and bespoke design out the window? No, far from it.
Residential spaces are often imagined to be a space for rest and refuge, a space of individuality, promoting a sense of security through arbitrarily set boundaries. It becomes ironic to push forth computation as the answer to repetitive module population. But this has to be seen in the context of the industry desire to provide housing at speed. It has resulted in design uniformity, unit layouts with typical spatial constructs. These designs are led by building code compliance, which are in turn informed by quantifiable aspects of design. Quantities include façade distance to encourage privacy or having minimum unit separation to encourage cross ventilation. Common strategies to implement computation workflows is a literal translation of these code compliances into inputs and generating an output which retains uniform spatial constructs.

However, such literal code compliance lead design  tilts the man-machine collaboration towards unfulfilled potential. It would be a more fulfilling challenge to first identify base elements which a designer perceives as main influencers of a residential design. These could include maximising sun exposure in cold climates, encouraging natural ventilation in tropical areas or reducing insolation gains in arid countries. Armed with these elementary concepts, a parametric model can then be built based on site constraints. Thereafter, the machine understands its design constraints and can perform design search for the best performing results. With this base results, the architect can leverage on the options and create unique designs far beyond the project’s modular underpinnings, safe in the knowledge that the performative aspects have already been dealt with by the parametric model.  
By designing strategies based on available data, computation design can become a tool that explores possibilities, identifying geometrical trends and ultimately augment design decisions. A successful integration of computation into architecture design would become a process whereby inputs and outputs are heavily influenced by both man and machine. If the designer does not define a design objective like maximising solar exposure, the computer can never perform simulations to verify it. If the computer is left to produce building designs, most of the massing might be useless geometry which do not support spatial layouts. When this balance is properly addressed, new residential typologies can then emerge as a man-machine output.

The potential synergetic benefits of Computational Design applied in conjunction with the modularity of PPVC in the residential typology are significant as they have positive impact on all stages of the development – from design through to construction implementation. Despite prevailing scepticism, these two emergent trends are slowly being recognised by the building industry. A firm understanding will bring us within reach of sustainability, productivity and design goals of tomorrow in the industry.  

[1] Source: UNHabitat for a better urban future
[2] Source: The World Bank

Alvin Liau
Alvin Liau, Associate Director at DPA, is a member of the Residential Typology Research group as part of the team researching on design trends and awards. He has an interest in harnessing modern technology to augment the traditional design process. His current portfolio includes mid to high end residential projects.

Ling Ban Liang
Ling Ban Liang, an ESD analyst from DPSD, has a deep interest in computational design processes and data visualisation which stems from his academic research on the applications of multi-objective climate optimisation as a design tool. He also applies his technical knowledge to daily operations, optimising existing simulation practices.

Residential Typology Group
The residential typology research group conducts in-depth research in residential design trends with critical analysis from design, code and technical perspectives. Thus, DPians are able to refine their ideas based on sound theory.