We are happy to announce that our new residential project – named “EAST & WEST” – was just completed in Nagoya.
The principal feature of this project is that its spatial configuration has been almost automatically defined by a computational process, instead of an architect’s inspiration or rationalization.
More specifically, we used Processing© to produce a C-Language algorithm to define the layout of rooms or functional units in this residential piece. More details about the algorithmic process are as follows.
Layout defining process of EAST & WEST
The first step is to prepare a complete list of rooms and other functional units explicitly required by the client, such as:
- 1st Floor: Entrance hall, Living room, Dining & Kitchen, Grandmother’s room, Bath & Restroom
- 2nd Floor: Main Bedroom, Children’s room, Double-height interior void
So far, nothing is special about these 8 spatial units, but what comes next makes the planning process so unique.
The 3D image above (See image 'all particles') shows how we conceptualized each unit as consisting of small particles (the ‘balls’ in the image above). In other words, the “rooms” were broken down into an aggregation of colored particles, like molecules are made up of numerous atoms. In our modeling space, different particle colors represent each of eight spatial units, and the numbers of particles correspond to the different sizes of those units. The color-function relationship is as follows:
[Entrance Hall – Silver] [Grandmother’s room – Red] [Bath & Restroom – Blue]
[Living room – Purple] [Dining & Kitchen – Yellow] [Main Bedroom – Violet] [Children’s room – Cyan] [Interior void – White]
Starting point of the algorithmic process is to define a clear-cut model such as described above, to which a particular piece of code would be effectively applied.
In the initial state, the particles occupy random location in the framed three-dimensional box, which represents the maximum volume allowed by the budget and the building regulation. No matter where they are at the beginning, the particles are supposed to converge into a specific configuration as long as the code underlying the algorithmic process stays the same.
Before activating the process, we have added the following binding conditions as minimum interventions by human (i.e., architect’s) hand. There are only three of them:
1. Save large opening on the south and the west elevation (for daylighting)
2. Particles are not allowed to move across the floors (e.g., particles of first-floor rooms can not move to the second floor, because of the client’s requirement)
3. Save a balcony on the south.
What follows is the key of this automatic planning process. First operation is the rearrangement of the particles directly (i.e., without filtering through architect’s interpretation and deliberation) according to the client’s requirement, so that the particles, in analogical terms, ‘change their seats’ to gradually fit into a particular configuration as a whole. The ultimate output is supposed to be, so to say, the seat layout of particles best-fitted to the clients’ complex requests.
A few examples of the client’s requests are:
1. Save as much daylight as possible in the dining room.
2. Keep grandmother’s room close to the bathroom.
3. Keep the kitchen and the restroom away from Ki-mon (i.e., specific corner of the site and/or the building where water-using rooms should be avoided, according to traditional Feng-shui belief).
4. Keep the top of the living room high and open as a double-height interior void.
5. Keep the grandmother’s room and the children’s room away from each other, so that the noise from the latter does not disturb the former, etc.
Traditionally, it is the architect him/herself who, by the power of his/her experience and/or inspiration, organizes, prioritizes and optimizes the relationships of those piecemeal requests into a single consistent spatial relationship. However, EAST & WEST is characterized by the delegation of such ‘arranger’ role to a simple mechanistic (although in virtual sense) rule based on the degree of familiarity between particles (the elementary atoms of each spatial unit). In other words, the particles’ variable degrees of familiarity with each other becomes the single key to the pure materialization of the spatial relationship (that is, the result of the seat-changing process) directly induced from the client’s requests.
For better understanding, let us redefine the client’s requests described above in terms of the virtual mechanical rules of attraction/repelling between particles, as follows:
1. DINING particles are attracted to the OPENING of the SOUTH ELEVATION.
2. GRANDMATHER and BATHROOM particles are attracted to each other.
3. KITCHEN and BATHROOM particles are repelled from Ki-mon (*see above) area.
4. LIVING ROOM and INTERIOR VOID particles are attracted to each other.
5. GRANDMOTHER and CHILDREN particles are repelled from each other, etc.
To put it simply, we assume a quasi-personal relationship between the room particles: that is, those who like each other naturally get close to each other, and vice versa. In our model, the degree of familiarity between particles is quantified on 5-point scale as following:
[5 pt.] = Strong attraction
[4 pt.] = Moderate attraction
[3 pt.] = Neutral
[2 pt.] = Moderate repulsion
[1 pt.] = Strong repulsion
According to this model, some couple of particles have adhesive or ‘soft’ relationship when they get close to each other (like two pieces of chewing gum), while some others have rather repulsive, or ‘hard’ relationship like two billiard balls bounce back as they hit each other. The essence of the algorithm adopted for this project resides in this familiarity rule set as the motivating power for the particles to settle in a most comfortable state through their ‘seat-changing’ game.
In the actual C-language coding of this algorithm, the type (attraction/repulsion) and the strength of the relationship between particles are defined as unique parameters in Processing©, and that’s it – the only thing left to do (for human hands) is to press the return key to run the code, and wait for a few minutes until the particles finish their game and settle down to a optimum state wherein the client’s requirements are fulfilled with the best balance as a whole.
Let us explain with an example. At the initial stage shown in the 3D image above, the INTERIOR VOID particles (white balls) are not placed above the LIVING particles (purple balls) – which is obviously against the client’s requirement No.4 (see above). Possible step toward the right solution would be, for instance, that a purple ball exchanges its seat with the adjacent red ball (represents GRANDMOTHER’S ROOM) so that, at least in this particular locality, the situation moves a step closer to the desirable state. Our assumption is that the accumulation of such step-by-step process would ultimately bring a holistic spatial configuration where the required conditions are evenly met wherein, for example, a large portion of the interior void properly sits above the living room and not above the grandmother’s room. It is obvious – with 8 particles and 28 relationships (or parameters, in terms of algorithmic processing) between each of them – the actual calculation of this dynamic process would become so complex to deal with for human, and this is where the algorithmic computation comes in. Once an appropriate code is written, a single laptop can output, so to say, a win-win seat-changing solution for all of the 8 particles almost in a blink.
In the 3D image below (See image 'Algorithmic process'), the blocks on the left represents the rooms in the first floor, and those on the right represents the ones in the second floor, respectively. Starting from the initial arbitrary state (Left), the particles begin to exchange their seats as the code runs (Middle) and gradually converge to a stable state where they no longer need to continue their seat-changing game (right).
”Architecturization” Process
The image above (See image 'maya_studys') shows the eventual result of the seat-changing game. Although it may look like a raw stone, it actually represents the optimum arrangement where all the particles have sit in the most comfortable state as a whole. This result is the direct output of the algorithmic computation process.
The next step is what we call the “architecturalization” of this raw stone – that is, the process by which this lumpy body is brought into a livable architectural space. We came up with a number of different methods for this process (a few examples of which are shown in the images below), from which we eventually choose to interpret the emergent compartment (i.e., rooms) as spaces surrounded by undulating ribbons of equal height (the second left image in the middle row).
More precisely, the sub-rule we applied was to trace the row stone-like volume with wall strips with equal 800mm height layered vertically. We have subdivided a floor into three layers, so the entire building turned out to have 6 layers in all (three strips per floor). The result is a sextuple layer of winding wall strips.
What should be emphasized here is that this architectural form was more like a natural substance generated through a quasi-natural process rather than a product produced through an intentional design process by an architect.
In this emergent form, the interstices of various sizes between layered strips came to function as architectural openings that activate light, air and/or visual intercourse between the inside and outside of this architecturalized space.
Next, we bring the 3D image into an actual model to check and fine-tune the details, the results of which are shown below (the first floor on the left and the second floor on the right, respectively). (See image 'Screenshot')
Finally, we have translated the spatial arrangement into 2D plan drawings. The overlapping lines correspond to the triple wall layers for each floor. (See image 'Plans')
Designed by NORISADA MAEDA ATELIER + Osamu Murayama
Structural Engineering by UMEZAWA STRUCTURAL ENGINEERS
↧