LSDO Lab
Methods

Efficient enforcement of geometric non-interference constraints

Context
Physically feasible configurations require components to remain inside enclosures and clear of one another as their shapes and positions change.
Gap
Discretized clearance conditions produce many constraints, and aggregating them can impair conditioning.
Research goal
Develop differentiable non-interference constraints that remain efficient as component count and geometric complexity increase.

Selected studies

References
Clearance enforcementSigned-distance surrogates

Enforcing component clearance with fewer constraints

Signed-distance fields and partial aggregation reduce constraint count while retaining minimum-clearance bounds between sampled boundary points. The box-packing and BWB multidisciplinary benchmarks converge faster than their fully aggregated formulations. Local aggregation lets boundary sampling be refined without increasing the number of optimization constraints.

Primary paper

Andrew H. Fletcher, Hollis A. Smith, John T. Hwang. Efficient and Robust Enforcement of Geometric Non-Interference Constraints for Large-Scale Multidisciplinary Design Optimization. AIAA AVIATION 2026 Forum, AIAA 2026-4501 (AFRL public release), 2026. Distribution Statement A: Approved for public release; distribution is unlimited. PA# AFRL-2026-2044

PDF
Packaging-constrained BWB optimization enforces geometric non-interference between the payload and aircraft geometry. Fletcher et al., 2026 [1]
Overlaid aircraft shapes and internal payload boxes compare designs with and without non-interference constraints for two payload arrangements.
Non-interference constraints keep payloads inside the optimized BWB; a volume constraint alone permits boundary penetration. Fig. 6, Fletcher et al., 2026 [1] Fig. 6
Two convergence histories plot optimality and feasibility against optimization iteration for the blended-wing-body problem, one for the proposed non-interference formulation and one for full aggregation.
Optimality and feasibility histories for the BWB multidisciplinary optimization; the proposed formulation converges in fewer iterations than full aggregation of the non-interference constraints. Fig. 7, Fletcher et al., 2026 [1]

Representing component boundaries with differentiable distance fields

Compactly supported radial basis functions combine a background distance field with a local correction, refined where residual errors remain large. Wing and fuselage tests demonstrate accurate boundary reconstruction and inexpensive field evaluation, although the wing's sharp trailing edge remains difficult to resolve.

Primary paper

Marius L. Ruh, John T. Hwang. Differentiable Signed Distance Function Surrogates for Multidisciplinary Design Optimization Using Radial Basis Functions. ASME IDETC/CIE 2026 · DETC2026-194018, 2026.

Cross-sectional signed-distance contours surround black wing and fuselage outlines at several spanwise locations.
Wing and fuselage signed-distance contours show the reconstructed boundaries and local irregularities near the wing trailing edge. Fig. 1, Ruh and Hwang, 2026 [2]contributions: Fig. 1

References

  1. Andrew H. Fletcher, Hollis A. Smith, John T. Hwang. Efficient and Robust Enforcement of Geometric Non-Interference Constraints for Large-Scale Multidisciplinary Design Optimization. AIAA AVIATION 2026 Forum, AIAA 2026-4501 (AFRL public release), 2026. Distribution Statement A: Approved for public release; distribution is unlimited. PA# AFRL-2026-2044
    PDF
  2. Marius L. Ruh, John T. Hwang. Differentiable Signed Distance Function Surrogates for Multidisciplinary Design Optimization Using Radial Basis Functions. ASME IDETC/CIE 2026 · DETC2026-194018, 2026. Manuscript

Research connections

Related to Efficient enforcement of geometric non-interference constraints

Methods

Applications

Software

Authors of related work