The design process

Halfway through the term, Visser asks each student to bring every drawing and model they have made for a single small exercise: a bench for Kestrel Point. Wren arrives with one beautiful drawing. Callum arrives with a shoebox of rough sketches, three crushed card models and a page of crossed-out notes. Visser spreads Callum's work across the table first. 'Look,' she says, pointing at a sketch near the bottom of the pile. 'Here, on Tuesday, you found the idea. Everything after this gets better.' Then she turns to Wren. 'Your drawing is lovely. But I cannot see how you got there, and neither can you. That means you cannot get there again.' Design is a process, not a flash of inspiration. This lesson describes how designers move from a problem to a proposal, how they generate and test ideas, and how they keep track of their thinking so they can learn from it.

Design is iterative

Many people imagine designers have a single brilliant idea and then draw it up. In practice, almost all design proceeds in loops. A designer studies the problem, proposes a response, tests it, learns something, and revises, over and over. Each loop is an iteration. Early iterations are fast and rough; later ones are slower and more precise. The British design researcher Bryan Lawson, in How Designers Think (first published in 1980), showed that designers do not first fully define a problem and then solve it; they develop their understanding of the problem by trying solutions. A proposal reveals what matters about the problem, which leads to a better proposal.

Understanding the problem

Every design begins with research and analysis. For a building, that means the program (what it must contain), the site (its size, slope, orientation, sun, wind, views, trees, neighbours, noise and access), the people who will use it and how, the budget, the codes and regulations that apply, and the wider context of the place. Designers record this as notes, photographs, measured drawings and diagrams. A site analysis drawing, for example, might show the path of the sun, the prevailing winds, the best views, the noisiest edge and the existing trees on a single plan.

Understanding also means asking questions that the client may not have asked. For a riverside shelter: who will use it in winter? Where will snow drift? What happens when the river floods in spring? Is there a lighting problem at night? These questions often lead to the best ideas.

Generating ideas

Once you understand the problem, generate many possible responses quickly, before choosing one. Architects call these options or schemes. A good rule is to produce at least three genuinely different schemes, not three variations of the first idea. Techniques that help include: drawing many small thumbnail sketches in a limited time; making quick models from cheap materials; starting from different priorities (the view, the wind, the path, the trees) and letting each one drive a scheme; borrowing a strategy from a precedent; and setting yourself a constraint, such as using only one material or only curved walls.

At this stage, quantity matters more than quality. The psychologist J. P. Guilford called this divergent thinking: producing many possibilities. It must be followed by convergent thinking: comparing and choosing. Most students converge too early, falling in love with their first idea before they have seen what else is possible.

The concept

Out of the generating phase, a designer chooses a concept: a central idea that guides the many decisions to come. A concept can be spatial (a single sheltering wall that wraps around a view), experiential (a sequence from darkness to light), material (a building made entirely of stacked timber), or about the site (a roof that continues the slope of the ground). A good concept is simple enough to state in one sentence and draw in one diagram, yet rich enough to answer questions throughout the design. When you are unsure how to design a detail, the concept should tell you.

Louis Kahn organized many of his buildings around the idea of served and servant spaces: large rooms for the main activities (served) and smaller rooms for stairs, washrooms and mechanical services (servant), each clearly expressed. His Richards Medical Research Laboratories at the University of Pennsylvania in Philadelphia (completed in the early 1960s) give each laboratory floor a clear open space served by separate brick towers for stairs and ducts. Once you know the concept, you can read the whole building.

Choosing and developing

To choose between schemes, compare each against the goals you set during analysis. A simple tool is a weighted matrix: list the criteria, give each a weight for how much it matters, score each scheme against each criterion and add up the weighted scores. The numbers do not make the decision for you, but they force you to be explicit about what matters and often reveal that a scheme you liked is weak on something important.

# A simple weighted design matrix for comparing three pavilion schemes.
criteria = {            # weight out of 10
    "frames the river view": 9,
    "shelter from north wind": 8,
    "sense of arrival": 7,
    "fits the trees": 6,
    "easy to build": 5,
}
scores = {              # each scheme scored 1 (poor) to 5 (excellent)
    "A: long bar":      [5, 3, 3, 4, 5],
    "B: L-shaped wall": [4, 5, 4, 4, 4],
    "C: round room":    [3, 4, 5, 2, 2],
}
weights = list(criteria.values())
best = max(weights) * 5 * len(weights)
for name, s in scores.items():
    total = sum(w * x for w, x in zip(weights, s))
    print(f"{name:<18} {total:>4} of {best}  ({total / best:.0%})")
print("A matrix organizes judgment; it does not replace it.")

A: long bar         139 of 225  (62%)
B: L-shaped wall    148 of 225  (66%)
C: round room       116 of 225  (52%)
A matrix organizes judgment; it does not replace it.

In this example scheme B, the L-shaped wall, comes out ahead because it does well on the two most important criteria, the river view and shelter from the wind. Scheme C, the round room, scores well on arrival but poorly on fit with the trees and ease of building. A designer might combine the best of both: B's sheltering wall with a curved end that gives some of C's sense of arrival. Choosing often means combining.

Developing a scheme means testing it at larger scales and in more detail. Draw it to scale in plan and section. Build a larger model. Check it against the human body: can you sit there, reach that, walk through here? Check it against the sun and the wind. Each test may send you back to revise the concept, and that is normal.

Constraints are useful

Students often see constraints, such as the budget, the code, the site boundary or a client's demand, as obstacles to good design. Experienced designers see them as sources of ideas. A steep slope suggests a building that steps down; a tiny budget suggests a single cheap material used cleverly; a strict height limit suggests a sunken room. The American designers Charles and Ray Eames often said that design depends largely on constraints. When you feel stuck, look harder at your constraints, not away from them.

Keeping a process record

Keep everything: every sketch, every model, every note, dated and in order. This is your process record. It lets you go back to an idea you abandoned too early, explain your design to others and learn how you work. In studio reviews, instructors often ask to see the process, not just the result, because the process is what can be taught and improved. Many students photograph each model at the end of every working session and keep a simple log of what they tried and why.

Exercise 9: three schemes and a concept

This exercise takes about six hours and prepares directly for the final project.

• Write a one-page site analysis of a small spot at your site, with a diagram showing sun, wind, views, noise, paths and trees.
• Draw twenty thumbnail sketches, no larger than 5 cm (2 in) square, of possible small shelters for the spot, in no more than forty minutes.
• Choose three that are genuinely different and develop each as a quick model and a plan and section sketch.
• Compare them with a weighted matrix and choose one, or combine two.
• Write the concept of your chosen scheme in one sentence and draw it as one diagram.
• Deliverable: the analysis, the thumbnails, the three schemes, the matrix and the concept statement and diagram.

Self-critique checklist

• Are my three schemes truly different, or variations of one idea?
• Did I choose based on the goals from my analysis?
• Can I state my concept in one sentence that someone else understands?
• Is my whole process recorded and dated?

Diagrams as thinking tools

Between analysis and design sits the diagram: a simple drawing that captures relationships rather than appearances. A bubble diagram shows the parts of a program as circles sized by area and joined by lines where they need to be close. An adjacency matrix lists the spaces down the side and across the top and marks which pairs must touch, which should be near and which should be apart. A circulation diagram traces how people, goods or services move. A section diagram shows how light, air or views pass through a building. None of these is a plan, but each helps you find one.

Diagrams are powerful because they leave out almost everything. By removing walls, materials and details, they make the essential decisions visible and easy to change. A good habit is to draw the same idea in several diagram types: if the bubble diagram, the circulation diagram and the section diagram all point toward the same arrangement, you are probably onto something.

Working with clients and users

In practice the problem is rarely defined by the architect alone. Clients bring needs, hopes and budgets; future users bring knowledge of how they actually live and work, which may differ from what the client imagines. Good designers listen carefully, ask open questions and observe how people use existing spaces. Some projects use participatory design, in which future users take part in workshops, drawing, model-making and walking the site with the design team. The Chilean architect Alejandro Aravena's Quinta Monroy housing in Iquique (2004) is a well-known example: families were involved in the design, and each house was built as half of a good home, with space and structure for the family to complete the other half themselves over time.

In studio, your client is often imaginary, but users are real. For any project, find people who resemble the future users and talk to them. Ask a parent with a stroller how they use a park path, or a cyclist where they stop to rest. Their answers will shape your analysis and often reveal problems you would not have found alone.

When you are stuck

Every designer gets stuck. When it happens, change something: the medium (switch from drawing to models), the scale (zoom out to the site or in to a detail), the place (go back to the site), or the question (ask what the user would do, or what the opposite of your idea would look like). Explaining your problem aloud to a classmate also helps, because it forces you to state what you are trying to do. Above all, keep making things. Ideas come more often from working than from waiting.