Riverside Primary School

The project was a two-storey, six-classroom primary school on a constrained Cambridge site, with the building envelope effectively set by the highways setback on one side and a mature tree line on the other. The brief arrived with a tight planning deadline and a hard cost ceiling, so the technical challenge was never the architecture — it was being able to prove, quickly and repeatedly, that every classroom met the acoustic, daylighting, and floor-area standards the local authority’s panel would check.

The core of the job was a Dynamo-driven parametric classroom module. A single Dynamo graph took a small schedule of inputs — room width, depth, floor-to-ceiling height, window wall percentage, and roof pitch — and rebuilt the classroom geometry, its window mullion pattern, its roof panels, and its acoustic zoning lines from those values. Twenty-four module variations were generated across the project, and every one carried the same shared parameters: area, teaching zone, circulation allowance, and glazing ratio. When the acoustic consultant changed the party-wall spec mid-programme, the update was a parameter edit in one schedule, not a redrawing of twenty-four rooms.

Spatial coordination reviews

The school was coordinated in Navisworks at stage 3 — deliberately before planning — against the landscape and MEP packages. The landscape architect’s model defined the fall of the external play deck against our ground-floor datum; the MEP model threaded ventilation ducts over the classrooms at a level we had reserved in the module. Clashes were run to a 50 mm tolerance and held in a weekly review that the whole design team attended, so that by the time the drawing set started there was not a single unresolved intersection in the model.

The daylighting parameters were enforced in the model itself, not in a separate spreadsheet. Every option carried its window-to-floor ratio, and the planning visualisations were rendered straight from the module.

Planning visualisations

For the planning submission the module was used to generate three massing options in days rather than weeks. Each option was rendered as a shaded elevation plus a site-context plan, with the tree line and the highways setback shown on every sheet. Because the options were real variations of the same model rather than a series of illustrations, the panel’s feedback translated directly into parameter changes — the winning option was the one that satisfied the daylight target with the shallowest building depth, which was Option B with the mullions adjusted to 450 mm centres.

---

The module survived all the way to stage 4, where the typical external wall junction detail — wall, cill, cavity, and roof — was developed once and placed throughout the set. The school was consented without appeal, and the graph that generated the rooms is archived with the project, available if the authority ever asks for a sixth classroom.