Intelligent Design Technology for Spatial Grid Structures
Core Key Technology #1. A CAD-CAM integrated design system where parametric 3D modeling, finite element analysis, structural optimization, and manufacturing data export happen in a single continuous digital workflow. No manual drawing interpretation. No data re-entry between design and fabrication. This is the technology that enabled the 229-meter world-record bolted-sphere grid span — a structure that could not have been designed with traditional 2D methods within a practical timeframe.
What Is Intelligent Design Technology?
Traditional steel structure design follows a sequential, disconnected workflow: an engineer creates 2D drawings in CAD software, a separate analyst imports geometry into FEA software for structural verification, a detailer creates fabrication drawings from the design drawings, and a shop floor programmer manually programs CNC equipment from the fabrication drawings. At each handoff point — from design to analysis, from analysis to detailing, from detailing to manufacturing — data is re-entered, re-interpreted, or re-created. Errors accumulate. A dimension that is wrong in the original design propagates through the entire chain, or a correction made during analysis is not reflected in the manufacturing drawings. For simple structures (a portal frame warehouse, a standard beam-column building), this workflow is manageable because the number of members and connections is small. For spatial grid structures with tens of thousands of members and hundreds of node types, it is unworkable.
Permanent Steel's Intelligent Design Technology replaces this sequential workflow with an integrated digital environment. The system is built around a single parametric model that serves as the single source of truth throughout the project lifecycle. When a design parameter changes — a span adjustment, a loading revision, a material substitution — the change propagates automatically through the structural analysis, member sizing, node design, and manufacturing data export. There is no manual re-drawing, no re-importing geometry into analysis software, no re-programming CNC machines. The design model is the manufacturing model.
This is what we mean by "CAD-CAM integration" in the context of spatial grid structures: not merely using CAD software and CAM software (many companies do this), but a single parametric model that drives both the structural analysis and the CNC machine programming without any intermediate human interpretation of data. The design engineer who adjusts a member size is simultaneously adjusting the CNC cutting path for that member. The analyst who changes a node configuration is simultaneously updating the robotic welding program for that node.
Parametric 3D Modeling of Grid Structures
The foundation of intelligent design is parametric modeling — the ability to define a grid structure not by drawing individual members, but by defining the rules that generate the member layout. A spatial grid is described by parameters: span (x and y dimensions), support boundary conditions (fixed, pinned, roller), grid topology type (square-on-square, diagonal-on-square, tri-chord, hexagonal), module size (the spacing between nodes in each direction), member cross-section library (available pipe or tube sizes), material grade, loading cases (dead, live, wind, snow, seismic), and deflection limits. The system generates the complete 3D model — every node coordinate, every member length, every connection angle — from these parameters.
The power of parametric modeling becomes apparent when a design parameter changes. In a traditional workflow, changing the span of a grid from 120m to 130m means manually re-drawing every member, re-calculating every node coordinate, and re-checking every connection. In the parametric system, the engineer changes the span parameter from 120 to 130, and the system regenerates the entire model in seconds — all node coordinates are recalculated, all member lengths are updated, and the structural analysis is automatically re-run with the new geometry. This enables rapid design iteration: an engineer can evaluate dozens of span/module/topology combinations in the time it would take to manually draw one.
For complex grid geometries — curved roofs, domes, irregular boundary shapes, structures with penetrations for mechanical equipment — parametric modeling is not just an efficiency improvement; it is the only practical approach. The 229-meter coal storage grid, for example, has a curved roof profile with varying curvature, irregular support conditions, and approximately 100,000 individual members. Manually drawing this structure in 2D would be impractical; the parametric model generates it from a defined set of geometric rules and loading conditions, with full coordinate accuracy and automatic member-by-member structural verification.
Structural Optimization Algorithms
Designing a grid structure is fundamentally an optimization problem: minimize steel tonnage (cost) while satisfying strength, deflection, and stability constraints (performance). The traditional approach relies on the engineer's experience to select member sizes and grid topology — an approach that works but tends to produce over-designed structures because the engineer, uncertain about the exact load distribution in a complex three-dimensional grid, sizes members conservatively. Optimization algorithms replace this conservatism with mathematical precision.
Our design system integrates two levels of structural optimization:
Topology Optimization
The system evaluates different grid topologies — square-on-square, diagonal-on-square, tri-chord, hexagonal, and custom patterns — for the same span and loading conditions, and identifies which topology achieves the lowest steel tonnage while meeting all structural constraints. For a given span, a diagonal-on-square topology may be 12% lighter than a square-on-square topology because the diagonal members carry load more efficiently in certain directions. This difference, multiplied across a 100,000-member grid, translates to hundreds of tons of steel and significant cost savings. The optimization algorithm evaluates these alternatives automatically, without requiring the engineer to manually build and analyze each topology variant.
Member Sizing Optimization
Within a selected topology, the system optimizes individual member cross-sections. Each member is assigned the smallest available cross-section that satisfies the demand-to-capacity ratio for axial load, bending, shear, and combined loading under all load cases. The system iterates: member sizes affect the stiffness distribution, which affects the load distribution, which may change which member sizes are optimal. The algorithm converges to an optimal member assignment that minimizes total steel weight while meeting all code requirements. For the 229m grid, this optimization process reduced steel tonnage by an estimated 15-20% compared to a uniform member sizing approach — a reduction of hundreds of tons.
Finite Element Analysis Integration
Structural analysis is not a separate step performed in separate software after design is complete — it is integrated into the parametric model. When the engineer changes any parameter, the FEA solver automatically re-analyzes the structure and reports whether the revised design satisfies all code requirements. This real-time feedback loop enables design decisions to be validated immediately rather than discovered days later when a separate analyst runs the model.
The integrated FEA performs the following analyses for every design iteration:
- Static analysis — axial forces, bending moments, shear forces, and torsion in every member under each load case (dead, live, wind, snow, temperature).
- Deflection analysis — vertical and horizontal deflections at every node, checked against code-specified limits (typically span/250 to span/400 for grid structures, depending on application and cladding type).
- Buckling analysis — linear and nonlinear buckling analysis for compression members, identifying critical buckling modes and ensuring adequate safety factors against member instability.
- Connection analysis — bolt shear and tension checks, weld capacity verification, and node sphere yielding checks for every node in the grid, not just "representative" nodes. In a grid with 5,000 nodes, all 5,000 are checked.
- Combined loading — demand-to-capacity ratios under load combinations per the applicable design code (Chinese GB 50017, Eurocode 3, AISC 360, or AS/NZS 4100), with automatic identification of members that exceed unity (DCR > 1.0).
The analysis speed is critical. For a grid with 100,000 members, a complete analysis under 10 load combinations involves solving a stiffness matrix of significant size. Our system's integration with the parametric model means that each design iteration — a topology change, a member resize, a span adjustment — triggers an automatic re-analysis within minutes, not hours. This enables iterative design refinement that would be impractical with a disconnected design-analysis workflow.
Wind Load & Seismic Analysis for Long-span Structures
Long-span steel structures are particularly sensitive to wind and seismic loading because of their low mass-to-span ratio and inherent flexibility. A 200-meter grid roof weighs a fraction of what a comparable reinforced concrete roof would weigh, which means wind uplift forces can represent a significant percentage of the structure's self-weight — in extreme cases, wind uplift can exceed the dead load, requiring specific attention to support connection design to prevent net upward reaction.
Wind Load Analysis
Our design system performs wind load analysis in accordance with the applicable wind loading code (Chinese GB 50009, ASCE 7, EN 1991-1-4, or AS/NZS 1170.2). For standard geometries, wind pressure coefficients are applied directly from code-specified tables. For unusual geometries — curved roofs, irregular shapes, structures in complex terrain — the system supports wind tunnel data input, where pressure coefficient distributions measured in a physical wind tunnel test are applied directly to the FEA model node-by-node. This is particularly important for long-span structures where the code-specified coefficients may not adequately capture the pressure distribution across a 200m roof surface.
The system also performs dynamic wind analysis (vortex shedding, galloping, flutter) for structures with long unsupported spans, where wind-induced vibrations could cause serviceability problems or fatigue damage. The natural frequency of the structure is computed from the FEA model, and if it falls in the range where dynamic wind effects are significant, a dynamic analysis is performed to ensure the structure's response is acceptable.
Seismic Analysis
Seismic design of long-span grid structures requires special attention because these structures behave differently from conventional buildings under earthquake loading. Their large horizontal extent means that different support points may experience different ground motions (asynchronous input), and their low mass means that seismic forces are often less critical than wind forces — but the seismic detailing requirements (ductility, energy dissipation capacity) must still be met. Our design system performs response spectrum analysis and, where required by code or project specification, time-history analysis using recorded or synthetic ground motion records. The seismic analysis checks member forces, node capacities, and support reactions under the design basis earthquake and the maximum considered earthquake, ensuring the structure meets both life safety and collapse prevention performance objectives.
Automatic Member Sizing & Node Design
In a bolted-sphere grid structure, the node is the most complex component. Each steel sphere must be precision-machined with threaded holes at specific angles to receive the connecting members. The sphere diameter, hole angle, thread specification, and bolt size are all determined by the forces converging at that node — which are different for every node in a non-uniform grid. Designing 5,000 unique nodes manually is impractical; the intelligent design system automates this entirely.
For each node, the system extracts the member forces from the FEA results, determines the required bolt diameter for each connecting member (based on the axial force and the bolt's tensile and shear capacity), selects the minimum sphere diameter that can accommodate all bolt holes without interference, and generates the CNC machining instructions for the sphere — including hole angles, depths, thread specifications, and machining tolerance requirements. The same automation applies to welded-sphere nodes, where the system generates the weld groove geometry and welding procedure specifications for each member-to-sphere connection.
This automatic node design is the critical link between design and manufacturing. The output of the node design module is not a drawing that a machinist interprets — it is a CNC program that directly drives the sphere machining center. The design engineer who changes a member size triggers an automatic node redesign, which triggers an automatic CNC program update, with no manual steps in between. This is the essence of CAD-CAM integration: the design model is the manufacturing instruction set.
BIM Integration & Design-to-Manufacturing Data Flow
The intelligent design system is not an island — it integrates with the broader project information environment through Building Information Modeling (BIM) compatibility. The parametric grid model can be exported in IFC (Industry Foundation Classes) format for coordination with architectural, mechanical, electrical, and plumbing models. Clash detection between the grid structure and building services (HVAC ducts, sprinkler pipes, electrical conduits) can be performed in the BIM environment before fabrication begins, avoiding costly on-site modifications.
The data flow from design to manufacturing follows a defined pipeline:
- Parametric model — the single source of truth containing all geometric, structural, and material data.
- FEA verification — automatic structural analysis confirming all members and nodes satisfy code requirements.
- Optimization — member sizing and topology optimization to minimize steel tonnage.
- Manufacturing data export — the model generates CNC cutting programs for member fabrication, CNC machining programs for sphere drilling, welding procedure specifications for welded connections, and digital pre-assembly data for fit-up verification.
- Production management — the intelligent manufacturing system receives the data and tracks each member through the production process.
- Construction sequencing — the model generates installation sequence data for the intelligent construction system, including jacking point locations, lifting weight calculations, and prestress sequence.
At no point in this pipeline does a human re-enter data from one system to another. The parametric model is the data source; each downstream system consumes the data it needs automatically. This eliminates the most common source of fabrication errors: misinterpretation of design drawings by shop floor personnel. The CNC machine receives the same data the design engineer created — not a drawing that someone else interpreted from the design engineer's work.
The data pipeline also enables a level of quality control that is impossible in a paper-based workflow. Because every fabricated member is linked to its design model identifier, and the design model records every parameter used in its design (material grade, cross-section, node connection type, loading case), any quality question can be traced to its design origin. If a member fails ultrasonic weld testing, the system can immediately identify the welding procedure, welder, material batch, and design joint configuration involved — and check whether other members with the same configuration or same material batch might have the same issue. This is not traditional quality control, which inspects after fabrication; it is quality assurance by design, where the digital data pipeline prevents the most common error categories before they occur and enables rapid root-cause analysis when issues are identified.
For international clients, the practical implication of this data pipeline is that we can provide a complete digital package with every delivered structure: the design model, the manufacturing data (CNC programs, welding procedures, inspection records), the material certificates, and the as-built verification survey. This package serves as the as-built documentation for the structure — not a set of drawings that may or may not reflect what was actually fabricated, but a verified digital record of every member, every connection, and every quality event in the fabrication process. For projects requiring regulatory documentation, insurance certification, or long-term structural health monitoring, this digital package is a significant asset.
The 229m Record Span: A Design Validation
The ultimate validation of intelligent design technology is the structures it produces. The 229-meter bolted-sphere grid coal storage facility holds the world record for the longest clear span of this grid type. Designing this structure required:
- A parametric model of approximately 100,000 members and 5,000+ nodes, generated from geometric rules defining the curved roof profile, support boundary conditions, and grid topology.
- FEA under 12 load combinations including dead, live, wind (with site-specific pressure coefficients), snow, temperature differential, and seismic loading, with full member-by-member and node-by-node verification.
- Topology optimization to select the grid pattern that minimized steel tonnage for the 229m span while meeting all deflection and strength constraints.
- Member sizing optimization across the available pipe cross-section library, with convergence to a design where each member is sized to its specific force demand.
- Automatic node design for 5,000+ bolted spheres, each with unique hole angles and bolt specifications, with direct CNC machining program output.
- Digital pre-assembly verification of critical node intersections to ensure fit-up before physical fabrication.
The result: a structure that spans 229 meters without interior columns, using a fraction of the steel tonnage that a conventional steel framing approach would require for the same clear span. The design was completed in a fraction of the time that a traditional 2D workflow would have required, and every member was fabricated directly from CNC programs generated by the design model — zero manual drawing interpretation.
This is the practical meaning of intelligent design technology. It is not software marketing; it is the engineering infrastructure that enables structures at the boundary of what is structurally possible.
Have a Structure That Demands Intelligent Design?
If your project involves spans, geometries, or loading conditions that exceed standard design capabilities, our parametric design team can evaluate the structural feasibility and provide a detailed engineering proposal.