Truss Problem Types

Description

Truss topology optimization finds optimal layouts of bar elements. Design variables are cross-sectional areas (0 = remove bar, A_max = full bar). This tutorial covers loading truss problems from JSON files and programmatic definition.

Setup

using TopOpt

Loading Truss from JSON

Truss data (nodes, connectivity, materials, supports, loads) is stored in JSON files. Download the sample file: tim_2d.json.

path_to_file = joinpath(@__DIR__, "tim_2d.json")
mats = TrussFEAMaterial(10.0, 0.3)    # Young's modulus, Poisson's ratio
crossecs = TrussFEACrossSec(800.0)    # Cross-sectional area
node_points, elements, _, _, fixities, load_cases = load_truss_json(path_to_file)
loads = load_cases["0"]
problem = TrussProblem(Val{:Linear}, node_points, elements, loads, fixities, mats, crossecs)

The load_truss_json function parses all required data structures.

JSON File Structure

{
  "dimension": 2,
  "node_num": 10,
  "element_num": 20,
  "nodes": [
    {"node_ind": 0, "points": [0.0, 0.0]},
    {"node_ind": 1, "points": [1.0, 0.0]}
  ],
  "elements": [
    {"end_node_inds": [0, 1], "elem_tag": "steel_set1"}
  ],
  "supports": [
    {"node_ind": 0, "condition": [true, true]}
  ],
  "loadcases": {
    "0": {
      "lc_ind": 0,
      "ploads": [
        {"node_ind": 5, "force": [0.0, -1.0], "loadcase": 0}
      ]
    }
  },
  "materials": [
    {"name": "steel", "elem_tags": ["steel_set1"], "E": 210e9}
  ],
  "cross_secs": [
    {"name": "beam", "elem_tags": ["steel_set1"], "A": 0.01}
  ]
}

Critical: All indices are zero-based (Python convention), not Julia’s one-based indexing. The parser handles conversion.

Required JSON Fields

• "dimension": model dimension (2 or 3)
• "node_num": total number of nodes
• "element_num": total number of elements
• "nodes": list of {node_ind, points} dictionaries
• "elements": list of {end_node_inds, elem_tag} dictionaries
• "supports": list of {node_ind, condition} dictionaries
• "loadcases": dictionary mapping load case indices to load dictionaries
• "materials": list of {name, elem_tags, E, density} dictionaries
• "cross_secs": list of {name, elem_tags, A} dictionaries

Optional JSON Fields

• "unit": units string (e.g., “meter”, “mm”)
• "model_type": “truss”
• "generate_time": generation timestamp
• "TO_model_type": “ground_mesh”
• "model_name": model name
• "_info": additional information

Programmatic Truss Definition

For simple trusses, define directly:

using StaticArrays

# 2D cantilever truss
node_points = Dict(
    1 => SVector(0.0, 0.0),   # node 1: fixed
    2 => SVector(0.0, 1.0),   # node 2: fixed
    3 => SVector(1.0, 1.0),   # node 3
    4 => SVector(1.0, 0.0),   # node 4: load applied
)

elements = Dict(
    1 => (1, 2), 2 => (1, 3), 3 => (1, 4),
    4 => (2, 3), 5 => (2, 4), 6 => (3, 4),
)

mats = fill(TrussFEAMaterial(1.0, 0.3), length(elements))
crosssecs = fill(TrussFEACrossSec(1.0), length(elements))

# Fix DOFs at nodes 1 and 2 (x and y)
fixities = Dict(
    1 => SVector(true, true),
    2 => SVector(true, true),
)

# Downward force at node 4
loads = Dict(4 => SVector(0.0, -1.0))

problem = TrussProblem(Val{:Linear}, node_points, elements, loads, fixities, mats, crosssecs)

For 3D trusses, use 3-element coordinate vectors and 3 DOFs per node.