gyeolgu · the way of joining

Approach

Five joints, four seasons, one blower door — how a frame with no nails passes a passive-house audit.

People assume the hard part is the physics. It is not. Passive-house physics is forty years old and generously documented; a careful builder anywhere can hit the numbers with tape, membranes, and patience. The hard part is refusing the tape — making a frame of interlocking pine behave like a sealed envelope without betraying what the frame is.

First, the survey

We begin every project with a season on site before a line is drawn. Sun angles are measured, not looked up; the February shadow of a neighbour's persimmon tree has redrawn more than one of our plans. We record where wind enters the valley, where water sits after rain, and where the clients actually stand when they talk to each other. The courtyard is positioned from this evidence, and the house wraps it.

Second, the joint

The studio keeps a working vocabulary of five structural joints, all cut in geumgang pine, all reversible. Each joint is mocked up at full scale in the workshop and racked to failure before it is trusted in a frame — the test rig is a retired persimmon press and a hydraulic ram, and it has settled arguments no textbook could. The rule the carpenters hold us to: any joint we specify, they must be able to take apart in fifty years with hand tools.

Third, the envelope

Airtightness lives in its own layer, so the frame never has to pretend to be a membrane. Inside the pine structure we run a continuous sheathing wrapped in taped hanji-faced board, lapped and clamped at every penetration; the joints themselves stay dry, visible, and serviceable. Insulation is wood fibre, thick enough that the giwa roof carries a metre of winter sun without passing it on. The full palette is catalogued in the material library.

Fourth, the proof

Every house is measured twice by a blower door — once with the envelope open for inspection, once finished — and monitored through its first full year. Temperature, humidity, and heat demand are logged in every room and in the madang itself. The dataset goes to the owners with the keys. If the house does not perform to the model, we return until it does; so far that clause has cost us three site visits and one deeply educational chimney.