About This Project
This project develops a boundary to core and core to boundary shape grammar for office tower planning, aimed at systematically exploring core configurations under competing demands for net to gross efficiency, daylight, and flexible floor plates, alongside post 911 code-driven core complexity. It organizes core generation as a rule sequence that sets the core boundary, places corridors, establishes a grid, and then places passenger elevators, service elevators, stairs, and restrooms using edge-based logic and adjacency constraints. The workflow also supports controlled variation through rule-based find and replace, such as converting selected elevator elements into storage or other functions when service requirements shift.
SHAPE GRAMMAR
George Stiny (2022), Shapes of Imagination: Calculating in Coleridge's Magical Realm. The MIT Press. https://doi.org/10.7551/mitpress/14469.001.0001
George Stiny defines a shape grammar as a way to calculate visually, entirely in terms of looking and what I see. This assumption distinguishes shape grammars from symbolic computing, where calculation depends on indivisible primitives and counting. In shape grammars, shapes are not treated as fixed assemblies of predefined parts. Instead, parts emerge through the act of applying rules, and what counts as a part can change as the calculation continues. Ambiguity is therefore not an error to eliminate but a productive condition that enables seeing new structures in the same drawing over time.
A shape grammar rule is a pair of shapes, often written as A to B, that can be applied whenever a transformed copy of A can be embedded as a part of a current shape C. Following a rule means identifying an instance of the left-hand shape through embedding, then carrying out a corresponding replacement on the drawing through erasing and drawing. Stiny formalizes this as C prime equals C minus t of A plus t of B, which highlights the two-stage embed and fuse cycle that links analysis and synthesis within one recursive operation.
SHAPE MACHINE
Shape Computation Lab, Georgia Institute of Technology. https://shape.gatech.edu/index.html
Shape Machine is a shape rewrite system developed at the Georgia Tech Shape Computation Lab that brings shape grammar operations directly into CAD workflows. It provides vector-based search and replace within the actual geometry of a model, so a designer can find a target subshape and replace it with new geometry that fuses cleanly into the drawing. Shape Machine operates within Rhino and supports rapid iteration, which makes it well-suited for systematic studies where many variants must remain consistent in a shared design language. It is implemented as a Rhino plug-in, and its published prototype emphasizes robust shape representation, recognition, and modification, including subshape recognition in vector graphics.
Conceptually, Shape Machine mirrors the rule logic described in shape grammars. A rule application involves finding a transformation that embeds the left-hand shape into a current design, then replacing it with the corresponding transformed right-hand shape to produce a new design. Beyond single operations, Shape Machine supports rule sequences and DrawScript, a Turing-complete visual programming language in which shapes stand in for lines of code, with optional Python integration for hybrid visual and textual workflows.
The precedent study focused on Richard Meier and Partners tower cores across both office and residential projects, using a comparative layout method that starts from the simplest core and scales up in complexity. The study maps how adding rows and columns within a core boundary enables more elevators, introduces service elevator requirements, increases stair count for code compliance, and accommodates vertically connected systems such as restrooms and wet infrastructure.
To keep comparisons consistent, each precedent was abstracted into a shared diagram set that tracks core-making decisions in order, including defining the floor plate and core boundary, establishing corridors, and organizing a grid before evaluating how elevator banks, service elevators, emergency stairs, and restrooms attach to circulation. The sample set includes Mitikah Office Tower, The Rothschild Tower, Reforma Towers, and Kiwoom Finance Square Headquarters, selected to capture a wide range of core organizations and program shifts across floors.
Key takeaways
- Elevator bank typologies repeat in legible families, especially one-side versus two-side arrangements and expandable bank counts that scale with core depth.
- Service elevators and stairs tend to be governed by edge and adjacency logic, where placement is constrained by circulation hierarchy and avoidance of primary public zones.
- Restroom and wet-spine integration becomes a major driver of core thickness and corridor structure as the layout scales up and shifts between floor types.
The rule set translates office tower core planning into a repeatable sequence that starts from a core boundary and progressively assigns circulation and vertical systems. It is designed to generate consistent alternatives while keeping key constraints legible, such as corridor hierarchy, elevator banking, and adjacency rules for service and wet functions.
Sequence
- Establish the core zone as a derived boundary, typically set 20-30% t inside the facade or floor plate edge.
- Draw the primary corridor lines that structure circulation and set up later placement rules.
- Populate the core boundary with a grid that becomes the spatial scaffold for shafts and rooms.
- Place passenger elevator shafts along the corridor with one-sided or two-sided configurations and bank logic for vertical zoning.
- Use edge detection logic so service elevators avoid opening directly into primary public areas when possible.
- Prioritize corner placement, then place stairs along feasible core edges when corner space is insufficient.
- Place restrooms with a preference to open onto secondary corridors, then relax constraints toward public adjacency if required.
- The system supports find and replace changes, including switching corridor sidedness for elevators and converting selected elevator segments into storage or other functions when service patterns change.
The rule set includes controls for detail level and for key configuration choices, including one-sided versus two-sided elevator layouts, corridor occupancy, stair count and placement tendencies, and restroom quantity.
Floor-to-floor Change
The precedent layouts show that tower cores are not static. Ground floors and upper floors often reorganize circulation and support space, and the same building can shift its core logic across different floor sections. This is documented directly in the precedent sheet where cores are compared by ground versus upper floor conditions and by explicit “change of floor section” diagrams.
In the rule set, this floor-to-floor variability is handled by separating what should remain stable from what can adapt. The passenger elevator shaft set can be treated as fixed across floors, while elevator banks are organized by vertical zones, which matches how towers typically segment service by building height. At the same time, the grammar allows core complexity to increase with additional grid rows and columns, enabling more elevators, service elevator additions, more stairs, and wet stacks that require vertical connections.
This set of scripts supports sectional transitions without rebuilding the whole core from scratch. It can reposition corridors, switch corridor hierarchy, and adjust service relationships as the tower shifts between floor sections. When a zone no longer needs certain elevator service, the same find and replace logic can convert targeted elevator segments into storage or other functions, which is a direct mechanism for handling transfer floors or program shifts between tower sections.