On the Importance of Geometric Nonlinearity and Temperature-Dependent Properties in Multi-Material Thermo-Mechanical Topology Optimization

This paper investigates two common simplifying assumptions in thermo-mechanical topology optimization: small-strain linear elasticity and temperature-independent material properties. The authors develop a physics-informed, simultaneous analysis-and-design framework called m-PIGP that incorporates (i) a finite-strain quadratic-Hencky (logarithmic-strain) constitutive model whose isotropic thermal eigenstrain admits an exact additive split in log-strain space, and (ii) temperature-dependent conductivity, thermal expansion, and elastic moduli for a titanium–copper–steel material system.

The framework optimizes a thermal actuator and a thermal gripper at three design temperatures (673 K, 873 K, 1073 K) under both a baseline model (linear elasticity with properties anchored at the design temperature) and the full physics (Hencky kinematics with temperature-dependent properties), subject to mass and manufacturability constraints. Every converged design is re-evaluated by verified nonlinear finite element solvers in the full factorial of constitutive law and property model—720 solves in total.

The central finding is that the constitutive law is the decisive modeling choice. The paper explains: These devices work as linkages where linear kinematics mistakes rotation for compressive strain; its error therefore grows with the design temperature and concentrates on the very layouts that exploit rotation best. Specifically, the constitutive-law error grows from 2–3% of stroke at 673 K to 8–11% at 1073 K, and for layouts relying most heavily on rotation, this error can consume up to a third of the stroke and invert the ranking of the best designs at high temperatures.

In contrast, the constant-property model that is anchored at the design temperature alters the results by less than one percent. The property effect never exceeds 0.31% of the reference stroke, and the interaction between the two effects stays below 0.25%.

The paper also reveals a deceptive self-assessment bias: Because a linear optimizer also steers away from the rotation-rich mechanisms that would expose this bias, the model can deceptively appear trustworthy when validated against its own designs. The baseline families misjudge their own optimal designs by only about one percent, creating a false sense of reliability.

Designing with the full physics yields consistently stronger and more temperature-robust devices at a modest increase in design-time cost—a factor of about 1.4 in computational time, returning 4–12% more best-design stroke. Temperature transferability analysis shows that Hencky-designed layouts optimized at moderate temperatures (873 K) transfer best along the nearly affine stroke–temperature response, dominating performance even at off-design operating temperatures.

The paper concludes with practical guidance: anchor constant properties at the design temperature and adopt finite-strain kinematics whenever the mechanism relies on rotation, and audit design tools by independent higher-fidelity re-evaluation rather than self-assessment.

Improvements to AI systems:

1. Physics-aware design optimization with finite-strain kinematics: Enhance topology optimization AI to use logarithmic-strain (Hencky) constitutive models instead of small-strain linear elasticity, enabling correct handling of rotation-dominated mechanisms (e.g., compliant linkages) and preventing stroke overestimation or underestimation by up to 33%.

2. Temperature-adaptive material property modeling: Integrate temperature-dependent conductivity, thermal expansion, and elastic moduli directly into the optimization loop, rather than assuming constant properties, to eliminate up to 0.31% stroke error and ensure accurate performance across 673–1073 K operating ranges.

3. Cross-fidelity validation and audit module: Add an automated, independent high-fidelity re-evaluation step (e.g., verified nonlinear FEM) after every optimization run, replacing self-assessment with external verification to expose deceptive model bias where linear optimizers validate their own flawed designs.

4. Transferability-aware design selection: Implement a meta-optimizer that selects designs based on robustness across off-design temperatures, using affine stroke–temperature response curves to prioritize layouts that maintain performance when operating conditions deviate from the design point.

5. Interaction-effect decomposition for design decisions: Incorporate a diagnostic tool that separates constitutive-law errors from property-model errors and their interaction, allowing AI to identify which modeling assumption dominates failure (e.g., kinematics vs. material data) and adjust design constraints accordingly.

What the improved AI system can do:

• Design thermal actuators and grippers that deliver 4–12% more stroke at high temperatures (1073 K) with 1.4× computational cost, while maintaining performance across a wide temperature range.

• Automatically flag and correct designs that rely on rotation, preventing catastrophic stroke loss (up to one-third) that linear models miss.

• Provide trustworthy performance predictions with verified error bounds, eliminating false confidence from self-validated optimizers.

• Recommend optimal design temperatures (e.g., 873 K) for maximum transferability to off-design conditions, reducing the need for re-optimization.

• Generate manufacturable, mass-constrained designs with temperature-robust behavior, suitable for aerospace, automotive, and energy applications where thermal loads vary.

Thermo-mechanical compliant devices are commonly designed with small-strain linear elasticity and temperature-independent material properties, even though they might operate hundreds of kelvin above ambient where both assumptions are questionable. In this work, we quantify the effect and cost of each assumption in multi-material topology optimization of thermally actuated compliant devices. To this end, we introduce a physics-informed, simultaneous analysis-and-design framework with (i) a finite-strain quadratic-Hencky (logarithmic-strain) constitutive model whose isotropic thermal eigenstrain admits an exact additive split in log-strain space, and (ii) temperature-dependent conductivity, thermal expansion, and elastic moduli for a titanium--copper--steel material system. We optimize a thermal actuator and a thermal gripper at three design temperatures under both a baseline model and the full physics, subject to mass and manufacturability constraints. Every converged design is re-evaluated by verified nonlinear finite element solvers in the full factorial of constitutive law and property model. The comparison between the two factors reveals that the constitutive law is the decisive modeling choice: These devices work as linkages where linear kinematics mistakes rotation for compressive strain; its error therefore grows with the design temperature and concentrates on the very layouts that exploit rotation best. Because a linear optimizer also steers away from the rotation-rich mechanisms that would expose this bias, the model can deceptively appear trustworthy when validated against its own designs. Designing with the full physics yields consistently stronger and more temperature-robust devices at a modest increase in design-time cost.