The missing layer between environmental analysis and architectural design

The usual sequence

On most projects I've worked on, the order goes like this. The architect develops a scheme until it's good enough to show. Then someone runs the daylight study, the overheating check, the energy estimate. The results come back as a report: coloured floor plans and a table of pass and fail.

By then the massing is fixed, the plan depth is fixed, and the window strategy has already been discussed with the client. The analysis can only say two things: fine, or fix this. Fixing means patching. A bigger window here, external shading there, a note that the north-facing flats are acceptable.

Nobody in that chain is doing a bad job. The analyst is accurate and the architect is answering the brief. The sequence is the problem.

Two languages

Environmental analysis speaks in metrics: daylight factors, hours above 26 °C, kWh per square metre per year. Architecture speaks in moves: deeper or shallower, taller or wider, a courtyard, a gallery access, a roof that lifts.

The missing layer is the translation between the two. Which moves change which metrics, in which direction, at this stage of this project. That knowledge exists. It lives in the heads of a few experienced people in each office, and in the environmental design principles I learned at the AA in London: orientation, plan depth, thermal mass, the relationship between a window and the room behind it. It rarely lives in the workflow.

Why it stays missing

• Setup cost. A proper analysis takes time to set up, so it waits until the design is stable enough to justify it.
• Reports instead of feedback. A PDF can't answer “what if we move the core?”
• Separate rooms. The people who understand the physics and the people who make the design moves are rarely in the same meeting.
• Compliance at the end. Regulation rewards passing a check late rather than designing well early. I think the rules themselves are good. The friction is in how compliance is done.

What the layer looks like

I don't think the answer is another simulation engine. Good engines exist. The layer I mean has three parts.

1. Principles, written down. For this climate and this building type: the plan depths that work, the glazing ratios where overheating starts, the orientations to avoid. Checked against the office's own past projects.
2. Fast, rough feedback while sketching. A direction rather than a certified result. Is this move making daylight better or worse? Will this façade overheat in the summers the building will actually see?
3. Someone, or something, that translates. Reading the results and turning them into design options the architect can act on the same day.

The third part is where AI becomes useful, in a modest way. It won't design the building. It can set up an analysis in minutes instead of days, read the output, and write the two paragraphs the architect actually needs.

Comfort as a starting point

The bigger shift is what we treat as the starting point. Visual, thermal and acoustic comfort is usually something we check. I think it should be something we design from. When daylight and thermal comfort shape the plan and section from the first sketch, the architecture follows from the performance instead of fighting it.

What I'm testing

Over the coming months I'll build small pieces of this layer and write up what works and what fails:

• a one-page climate brief for a new site;
• an early-stage daylight and overheating check that runs while sketching;
• an agent that turns simulation output into design notes.

If your practice has a version of this problem, I'd like to hear about it. Where does environmental analysis arrive too late in your projects? Email me at hello@franzforsberg.com.