ETH’s Block Research Group spent the last decade proving that a stadium roof does not have to be a forest of steel: their funicular shells find the compression-only surface first and hang the material on the forces — the digital heir to Heinz Isler’s frozen-cloth shells at the Deitingen service station. Read that lineage and then read this Archivperle from Hochparterre Wettbewerbe — a stadium project for Zürich drawn in 1952, killed by a Volksabstimmung in 1953, pulled back out of the gta Archiv — and the two stories rhyme in a way every competing office feels in the wrist.

Because here is the part the press release skips: that 1952 scheme was architektonische Kür, not a Zweckbau, and it still lost at the ballot box in 1953. The geometry was never the reason it died. Zürich has been arguing about stadiums the entire time since — the Letzigrund rebuilt in a hurry for the 2007 season, the Hardturm saga that ran through referendum after referendum before the ground finally moved. A stadium in this city is a political object first and a structural one second, and no funicular shell solves for that.

←TODAY: The 1952 competition can only be read today because the gta Archiv kept the drawings; the 2026 competition model dies on a corrupted IFC nobody can open. →3012: The studios still standing are the ones whose competition-to-execution pipeline was written down, not the ones with the cleverest single render. Fulcrum: A competition is won in geometry and lost in governance — so archive both.

For a studio our size — fourteen people in a half-renovated Werkhalle — the lesson is brutally practical. Offices that have been through a stadium-scale Wettbewerb report the same arithmetic: the burdened cost of a serious entry can swallow a month of utilisation, and the probability it converts is a coin-toss the jury controls and the electorate can veto. When Spain lifted the last World Cup, the celebration was national; the arena around it was a decade of planning politics that no roof geometry shortened. You model the bowl in a fortnight. You cannot model the referendum.

Atelier: The move for a Büro is not to buy a stadium-grade solver — it is to separate the competition model from the execution model before you start, and write down the exact hour one becomes the other. Trial the PAZ Grasshopper↔Archicad Library on the next Studienauftrag so the Grasshopper form-finding hands a clean IFC4 to Archicad’s Hotlink Manager instead of a re-drawn building — one round-trip, versioned, owned by a named person.

Hack: Total what each Wettbewerb actually cost the studio before you sign up for the next one. Most offices feel the loss and never total it; a short query against your own time log turns the feeling into a number the partners can argue about. Render competition hours as burdened cost against the fee you would have earned only if you won, and the expected value falls out immediately.

SELECT comp_name,
       ROUND(fee_if_won*win_prob - SUM(hours)*165) AS expected_value
FROM   competition_log
GROUP  BY comp_name, fee_if_won, win_prob
ORDER  BY expected_value ASC;

The row at the top of that list — the most negative expected value — is the competition you should have declined. Run it once a quarter and the studio stops entering on romance.

The geometry side is the easy, teachable half. Stadium skins have been rationalised on Delaunay–Voronoi logic for years — Populous, with Scott Tallon Walker, laid the Aviva Stadium’s polycarbonate on an irregular tessellation (opened 2010) that reads Voronoi while the real load path follows the seating-bowl grid, honest about the difference. Beijing’s Water Cube pushed the same idea harder — roughly 4,000 ETFE cushions on Weaire–Phelan foam, PTW + Arup + CSCEC, 2008 — where the maths genuinely is the structure. That difference is the whole discipline: know which lines carry force and which lines carry the image, and never confuse the two on a coordination model. As PAZ has covered before, IFC “speaks the same vocabulary, not the same dialect” — two standards-compliant tools will still hand you two different buildings, and a competition roof is exactly where that bites.

PAZ Takeaway: The problem this story exposes for an office is not form-finding — it is that competition knowledge walks out the door with every Praktikant. PAZ Academy’s value here is concrete: a bespoke Grasshopper↔Archicad bridge from the PAZ Library plus a reusable detail-and-script kit that captures your competition-to-execution round-trip as a documented workflow you own, not a plugin you rent.

Pull your own competition log, run the expected-value query, and archive the workflow before you archive the render — the drawing survives in the gta Archiv; the pipeline only survives if you write it down.

Source: hochparterre.ch