[Generative Design] Beyond Randomness: How to Force "Convergence" in Optimization

 A Case Study on Tunnel Alignment and Geologic Constraints

https://www.linkedin.com/pulse/alignment-optimization-how-inputoutput-settings-drive-wonho-cho-kg0qc

Introduction: Generative Design is Not a Slot Machine There is a common misconception that Generative Design (GD) is simply a tool for creating thousands of random alternatives. However, the true power of GD lies in its ability to "Converge"—to self-evaluate, discard inferior options, and evolve toward a superior solution.

In this post, we analyze a civil engineering case study (Tunnel Alignment Optimization). We will explore how to transition from simple geometry generation to a sophisticated optimization engine that accounts for geological strata and construction costs.

1. From Simple Geometry to Real-World Data The initial phase of any automation script usually involves simple geometric rules: "Draw the shortest line from A to B while avoiding obstacles." However, real-world infrastructure requires more.

• Visualizing the Invisible: We must move beyond 2D plans. By cutting the terrain along the alignment path, we create a "Flattened Section" (Profile) that reveals the geological layers.

• Defining the Strata: We classified the ground into three solids:

  • Red (Soil/Sediment): High collapse risk – Avoid.

  • Green (Soft Rock): Ideal target zone – Maximize.

  • Grey (Hard Rock): Acceptable but costly.

2. The "Brain" of the Algorithm: Input/Output Tuning Once the geometry is set, the battle shifts to coordinating Inputs and Outputs. This is where "Optimization" truly happens.

• Precision Matters: In our tests, refined slider ranges (down to 0.01 units) allowed the algorithm to shave the total length significantly. In a real construction project, a difference of mere decimal points in alignment can translate to massive cost savings.

• Caching Data: Calculating intersections with complex terrain solids is heavy. Using the Data.Remember node to cache ground data is crucial to prevent the process from freezing.

3. Hard Constraints: "Don't Even Think About It" To force convergence, we applied Hard Constraints. Instead of just discouraging the path from entering the Red (Soil) zone, we programmed the logic to fundamentally invalidate any path that touches it.

• The Result: The algorithm stops wasting time evaluating unsafe paths. It immediately focuses its evolutionary energy entirely within the Green (Soft Rock) zone, drastically improving efficiency.

4. The "Geometry Gap" Problem A technical challenge arose during this process. We initially used Nurbs curves. However, a Nurbs curve can "belly out" between control points.

• The Glitch: The control points were safely inside the Green zone, but the curve arc strayed into the Red zone. The computer thought it was safe, but physics disagreed.

• The Fix: We converted the path to a Polycurve for intersection detection. This ensures the computer recognizes the exact path of the tunnel, eliminating false positives.

Conclusion: Teaching the Goal The 'Optimize' function is only as good as the rules you give it. By setting clear goals (Minimize Length, Maximize Soft Rock Score) and strict constraints (Soil Intrusion = 0), we transformed a random generator into a precision engineering tool.

Whether you set your Population size to 20 or 100 depends on your timeline, but the principle remains: You must design the logic to induce convergence, not just variety.