多目的最適化
Theory and Physics
Multi-objective Optimization
Professor, what is multi-objective optimization?
Simultaneously optimizing multiple objective functions. Example: "Minimize mass" AND "Maximize stiffness". Typically, these are trade-offs (reducing mass lowers stiffness).
Pareto Front
The set of optimal trade-off solutions is the Pareto front. Solutions on the Pareto front are optimal solutions where "no objective function can be improved without sacrificing another." The designer selects their preferred solution from the Pareto front.
Summary
Professor, what is multi-objective optimization?
Simultaneously optimizing multiple objective functions. Example: "Minimize mass" AND "Maximize stiffness". Typically, these are trade-offs (reducing mass lowers stiffness).
The set of optimal trade-off solutions is the Pareto front. Solutions on the Pareto front are optimal solutions where "no objective function can be improved without sacrificing another." The designer selects their preferred solution from the Pareto front.
The concept of Pareto optimality originates from a 19th-century economist
The concept of "Pareto optimality" was introduced by the Italian economist Vilfredo Pareto in 1906 in "Manuale di Economia Politica (Manual of Political Economy)". It refers to a state of resource allocation where "improving someone's situation worsens someone else's". This concept was adapted to multi-objective optimization by Kuhn-Tucker's extension in 1963 and by Schaffer (VEGA method) in 1985. Simultaneous optimization of automotive lightweighting and safety cannot be discussed without the concept of the Pareto front.
Physical Meaning of Each Term
- Inertia term (mass term): $\rho \ddot{u}$, i.e., "mass × acceleration". Have you ever experienced being thrown forward during sudden braking? That "feeling of being pulled" is precisely the inertial force. Heavier objects are harder to set in motion and harder to stop once moving. Buildings shake during earthquakes because the ground moves suddenly while the building's mass "gets left behind". In static analysis, this term is set to zero, assuming "forces are applied slowly enough that acceleration can be ignored". It absolutely cannot be omitted for impact loads or vibration problems.
- Stiffness term (elastic restoring force): $Ku$ or $\nabla \cdot \sigma$. When you pull a spring, you feel a "force trying to return it", right? That's Hooke's law $F=kx$, the essence of the stiffness term. Now a question—an iron rod and a rubber band, which stretches more under the same pulling force? Obviously, the rubber band. This "resistance to stretching" is the Young's modulus $E$, which determines stiffness. A common misconception: "High stiffness = strong" is incorrect. Stiffness is "resistance to deformation", strength is "resistance to failure"—they are different concepts.
- External force term (load term): Body force $f_b$ (gravity, etc.) and surface force $f_s$ (pressure, contact force, etc.). Think of it this way—the weight of a truck on a bridge is a "force acting on the entire volume" (body force), while the force of the tire pressing on the road is a "force acting only on the surface" (surface force). Wind pressure, water pressure, bolt tightening force... all are external forces. A common mistake here: getting the load direction wrong. Intending "tension" but it becomes "compression"—it sounds like a joke, but it actually happens when coordinate systems are rotated in 3D space.
- Damping term: Rayleigh damping $C\dot{u} = (\alpha M + \beta K)\dot{u}$. Try plucking a guitar string. Does the sound continue forever? No, it gradually fades away. That's because vibration energy is converted to heat by air resistance and internal friction in the string. Car shock absorbers work on the same principle—they intentionally absorb vibration energy to improve ride comfort. What if damping were zero? Buildings would continue shaking forever after an earthquake. Since that doesn't happen in reality, setting appropriate damping is crucial.
Assumptions and Applicability Limits
- Continuum assumption: Treats material as a continuous medium, ignoring microscopic heterogeneity.
- Small deformation assumption (for linear analysis): Deformation is sufficiently small compared to initial dimensions, and the stress-strain relationship is linear.
- Isotropic material (unless otherwise specified): Material properties are independent of direction (anisotropic materials require separate tensor definitions).
- Quasi-static assumption (for static analysis): Ignores inertial and damping forces, considering only the balance between external and internal forces.
- Non-applicable cases: For large deformation/large rotation problems, geometric nonlinearity is required. For nonlinear material behavior like plasticity or creep, constitutive law extensions are needed.
Dimensional Analysis and Unit Systems
| Variable | SI Unit | Notes / Conversion Memo |
|---|---|---|
| Displacement $u$ | m (meter) | When inputting in mm, unify loads and elastic modulus to MPa/N system. |
| Stress $\sigma$ | Pa (Pascal) = N/m² | MPa = 10⁶ Pa. Be careful of unit system inconsistency when comparing with yield stress. |
| Strain $\varepsilon$ | Dimensionless (m/m) | Note the distinction between engineering strain and logarithmic strain (for large deformation). |
| Elastic modulus $E$ | Pa | Steel: ~210 GPa, Aluminum: ~70 GPa. Note temperature dependence. |
| Density $\rho$ | kg/m³ | In mm system: tonne/mm³ (= 10⁻⁹ tonne/mm³ for steel). |
| Force $F$ | N (Newton) | Unify as N in mm system, N in m system. |
Numerical Methods and Implementation
Multi-objective Optimization Algorithms
Summary
NSGA-II is the de facto standard algorithm for multi-objective optimization
NSGA-II (Non-dominated Sorting Genetic Algorithm II) is a multi-objective evolutionary algorithm published by Kalyanmoy Deb (Indian Institute of Technology Kanpur) in 2002 in IEEE Transactions on Evolutionary Computation. With over 40,000 citations on Google Scholar (as of 2024), it ranks among the top in computational science. Its combination of computational cost O(MN²) and density preservation mechanism is excellent, and it is also standard in tools like optDesign and Cadence's AMS simulation tools.
Linear Elements (1st-order elements)
Linear interpolation between nodes. Low computational cost but low stress accuracy. Beware of shear locking (mitigated by reduced integration or B-bar method).
Quadratic Elements (with mid-side nodes)
Can represent curved deformation. Stress accuracy improves significantly, but degrees of freedom increase by about 2-3 times. Recommended when stress evaluation is critical.
Full Integration vs Reduced Integration
Full Integration: Risk of over-constraint (locking). Reduced Integration: Risk of hourglass modes (zero-energy modes). Choose appropriately for the situation.
Adaptive Mesh
Automatic refinement based on error indicators (e.g., ZZ estimator). Efficiently improves accuracy in stress concentration areas. Includes h-method (element subdivision) and p-method (order increase).
Newton-Raphson Method
Standard method for nonlinear analysis. Updates the tangent stiffness matrix each iteration. Provides quadratic convergence within the convergence radius but has high computational cost.
Modified Newton-Raphson Method
Updates the tangent stiffness matrix at the initial value or every few iterations. Lower cost per iteration but linear convergence speed.
Convergence Criteria
Force residual norm: $||R|| / ||F_{ext}|| < \epsilon$ (typically $\epsilon = 10^{-3}$ to $10^{-6}$). Displacement increment norm: $||\Delta u|| / ||u|| < \epsilon$. Energy norm: $\Delta u \cdot R < \epsilon$
Load Increment Method
Applies the full load not all at once but in small increments. The arc-length method (Riks method) can trace beyond limit points on the load-displacement curve.
Analogy: Direct Method vs Iterative Method
The direct method is like "solving simultaneous equations accurately with pen and paper"—reliable but takes too long for large-scale problems. The iterative method is like "repeatedly guessing to approach the correct answer"—starts with a rough answer but improves accuracy with each iteration. It's the same principle as looking up a word in a dictionary: it's more efficient to open it at an estimated location and adjust forward/backward (iterative method) than to search sequentially from the first page (direct method).
Relationship Between Mesh Order and Accuracy
1st-order elements are like "approximating a curve with a ruler"—represented by straight line segments, so accuracy is limited. 2nd-order elements are like "flexible curves"—can represent curved changes, dramatically improving accuracy even at the same mesh density. However, computational cost per element increases, so judgment should be based on total cost-effectiveness.
Practical Guide
Multi-objective Optimization in Practice
Automotive lightweighting (mass) + crash safety (injury value), aircraft fuel efficiency (weight) + strength.
Practical Checklist
Formula E aero optimization is a 3-objective simultaneous optimization
In aerodynamic design for Formula E cars, simultaneous 3-objective optimization of "maximize downforce, minimize drag, uniformize sidewash" is standard. In the 2019 season vehicle development by Mahindra Racing (a Formula E team), a multi-objective CFD optimization based on NSGA-III coupled with SIMOPTICAL and OpenFOAM ran 200 generations and 1000 evaluation points, reportedly improving aero efficiency by 7% compared to the previous season, as noted in the technical report.
Analogy of Analysis Flow
The analysis flow is actually very similar to cooking. First, buy ingredients (prepare CAD model), do the prep work (mesh generation), apply heat (solver execution), and finally plate it (post-processing for visualization). Here's an important question—which step in cooking is most prone to failure? Actually, it's the "prep work". If mesh quality is poor, the results will be a mess no matter how excellent the solver is.
Pitfalls Beginners Easily Fall Into
Are you checking mesh convergence? Do you think "the calculation ran = the result is correct"? This is actually the most common trap for CAE beginners. The solver will always return "some answer" for the given mesh. But if the mesh is too coarse, that answer is far from reality. Confirm that results stabilize across at least three mesh density levels—neglecting this leads to the dangerous assumption that "the computer gave the answer, so it must be correct."
Thinking About Boundary Conditions
Setting boundary conditions is like "writing the problem statement" for an exam. If the problem statement is wrong? No matter how accurately you calculate, the answer will be wrong. "Is this surface truly fully fixed?" "Is this load truly uniformly distributed?"—Correctly modeling real-world constraint conditions is actually the most critical step in the entire analysis.
Software Comparison
Tools
modeFRONTIER is the multi-objective optimization standard in the European automotive industry
ESTECO's (founded 1999, Trieste, Italy) modeFRONTIER holds a position close to the de facto standard for multi-objective optimization tools in the European automotive industry. Volkswagen, Porsche, and Audi have adopted modeFRONTIER as common infrastructure, deploying multi-code coupled optimization with Nastran, ABAQUS, and StarCCM. Competition with Altair, which acquired HEEDS in 2022, has intensified, but many users evaluate the technical depth originating from European academia as a strength of modeFRONTIER.
The Three Most Important Questions for Selection
- "What problem are you solving?": Does it support the physical models/element types needed for multi-objective optimization? For example, presence of LES support for fluids, contact/large deformation capability for structures can make a difference.
- "Who will use it?": For beginner teams, tools with rich GUI are suitable; for experienced users, flexible script-driven tools are better. Similar to the difference between automatic (GUI) and manual (script) transmission cars.
- "How far will you expand?": Selection considering future analysis scale expansion (HPC support), deployment to other departments, and integration with other tools leads to long-term cost reduction.
Advanced Technology
Advanced Multi-objective Optimization
Multi-objective Bayesian optimization reduces CFD evaluation cost by 90%
Evolutionary algorithms are strong for multi-objective optimization, but if one evaluation (CFD simulation) takes several hours, hundreds to thousands of evaluations are not realistic. Multi-objective Bayesian optimization (MESMO, MOTBO, etc.) using Gaussian Process surrogates...
Related Topics
なった
詳しく
報告