Designing by Chance: Controlled Randomness as a Generative and Constructible Strategy for Architectural Façade Composition—A Case Study

Façade design has long oscillated between repetitive order and visual disorder, and computational tools now make controlled irregularity easy to generate but hard to build; comparatively few studies follow a stochastic envelope from algorithm to a completed building. This paper examines controlled randomness as a deliberate, constructible strategy for façade composition. A parametric Grasshopper workflow erodes a square grid of wall modules through a seeded random-reduction rule—a deterministic, rule-based procedure that uses no machine learning or artificial intelligence—whose intensity increases up the building, producing a depth-modulated, quarry-like relief realized with one ordinary product—40 × 20 cm concrete blocks in four thicknesses—set to a color-coded projection key. The method is documented through a completed mid-rise residential building in Aramoun, Mount Lebanon, and evaluated for aesthetics, environmental function, structure, and constructability. Across 200 seeds, box-counting fractal dimension and lacunarity show that the generated pattern occupies a narrow, reproducible band of complexity—stable to within about 0.5%—statistically distinct from both periodic order and free disorder. A module-by-module comparison of the as-built record against the model shows the designed pattern reproduced across all 2978 units without projection-class error, the only departures being sub-centimeter positional shifts in the base plane absorbed from a structural tolerance. Bounded by explicit rules and parameterized in a standard masonry block, randomness behaves as a controllable, buildable design parameter rather than noise.