Lagrange Multiplier Method for Contact

Category: Structural Analysis | Integrated 2026-04-06
CAE visualization for lagrange contact theory - technical simulation diagram
Contact using the Lagrange multiplier method

Lagrange Multiplier Method for Contact: Theoretical Foundations

What is the Lagrange Multiplier Method?

🧑‍🎓

Professor, how is the Lagrange multiplier method different from the penalty method?


🎓

The penalty method is like "pushing back penetration with a spring" (allowing slight penetration). The Lagrange multiplier method strictly satisfies the constraint condition (makes penetration completely zero).


$$ g_n = 0 \quad \text{(Normal direction gap at contact surface = 0)} $$

🎓

Introduces additional unknowns (Lagrange multipliers $\lambda$). $\lambda$ is the contact pressure itself.


🧑‍🎓

Penetration is completely zero... That's more accurate than the penalty method.


🎓

Penetration accuracy is high, but the system of equations becomes larger due to additional DOFs (Lagrange multipliers), leading to increased computational cost. Furthermore, the matrix may contain zero diagonals, which can worsen the condition number.


Penalty Method vs. Lagrange Multiplier Method

CharacteristicPenalty MethodLagrange Multiplier Method
PenetrationSmall but non-zeroCompletely zero
Additional DOFsNoneLagrange multipliers ($\lambda$)
Condition NumberWorsens with penalty stiffnessWorsens with zero diagonals
Parameter DependenceDepends on $k_p$No parameters
Computational CostLowHigh
Compatibility with Explicit MethodsGoodPoor (Implicit methods only)
🧑‍🎓

High accuracy but also high cost.


🎓

In Abaqus, the KINEMATIC option corresponds to the Lagrange multiplier method. The contact surfaces do not slip and "penetration" is zero. Suitable for precise bolt fastening or pressure vessel contact.


Summary

🎓

Key Points:


  • Penetration is completely zero — More rigorous than the penalty method
  • Requires additional DOFs (Lagrange multipliers) — Increases computational cost
  • Abaqus KINEMATIC — Lagrange multiplier method
  • Suitable for precise contact (bolt fastening, pressure vessels)
  • Cannot be used with explicit methods — Implicit methods only

Coffee Break Yomoyama Talk

Lagrange's Analytical Mechanics, 1788

In his 1788 publication "Mécanique analytique," Joseph-Louis Lagrange systematized a method for incorporating constraint conditions into variational principles using multipliers (later called Lagrange multipliers). It took about 180 years for this mathematical framework to be applied to contact problems. In the 1970s, B.M. Irons and others introduced multipliers into the Galerkin weak form for contact constraints, completing the FEM formulation that accurately satisfies the non-penetration condition.

Computational Methods for Lagrange Multiplier Method for Contact

Implementation of the Lagrange Multiplier Method

🎓

Extended system of equations:


$$ \begin{bmatrix} [K] & [C]^T \\ [C] & [0] \end{bmatrix} \begin{Bmatrix} \{u\} \\ \{\lambda\} \end{Bmatrix} = \begin{Bmatrix} \{F\} \\ \{0\} \end{Bmatrix} $$

$[C]$ is the constraint matrix, $\{\lambda\}$ are the Lagrange multipliers (contact pressure).


🧑‍🎓

The bottom-right is a zero matrix... that seems bad for the condition number.


🎓

It's a saddle point problem. Not an issue for direct solvers, but difficult to converge with iterative solvers.


Solver Settings

🎓
  • Abaqus: *SURFACE BEHAVIOR, KINEMATIC (Lagrange multiplier for normal direction)
  • Ansys: KEYOPT(2)=3 (Lagrange multiplier)
  • Nastran: SEGMENT contact (Lagrange-based)

  • Summary

    🎓
    • Extended system of equationsDOF + Lagrange multipliers
    • Abaqus KINEMATIC is the most widely used
    • Poor compatibility with iterative solvers — Direct solvers are safer

    • Coffee Break Yomoyama Talk

      Discretization of Mixed Variational Methods

      In contact FEM using the Lagrange multiplier method, a mixed variational method is used that couples displacement degrees of freedom with contact forces (Lagrange multipliers) as unknowns. The stiffness matrix after discretization becomes a saddle point problem, losing positive definiteness in the diagonal blocks. In the 1980s, the Brezzi-Babuška condition (inf-sup condition) was established as a criterion for appropriate selection of contact multiplier spaces, providing design guidelines for stable element pairs.

      Lagrange Multiplier Method for Contact in Practice

      Practical Use of the Lagrange Multiplier Method

      🎓

      Use only for problems where zero penetration is mandatory. For most problems, the penalty method is sufficient.


      Appropriate Use Cases

      SituationReason
      Sealing surfaces of pressure vesselsPenetration = leakage. Zero is mandatory.
      Precision bolt fasteningAccurate evaluation of seating pressure.
      Press-fit (interference fit)Precise control of interference amount.

      Practical Checklist

      🎓
      • [ ] Is the Lagrange multiplier method truly necessary? (Can't the penalty method suffice?)
      • [ ] Are you using a direct solver? (Iterative solvers are unstable.)
      • [ ] Is there contact chattering (oscillation between contact/non-contact)?
      • [ ] Are the contact pressures (Lagrange multipliers) physically reasonable?

      • Coffee Break Yomoyama Talk

        Press Die Contact Analysis

        In deep drawing press die analysis using ABAQUS Standard, introduced by Toyota around 2003, the Lagrange multiplier method was adopted for contact between the blank holder and the material. By strictly maintaining zero penetration, the prediction accuracy for sheet thickness reduction rate improved to within ±3%. Previously, with the penalty method, tuning contact stiffness took several days, but with the multiplier method, the number of trials was halved.

        Related Simulators

        Experience the theory through interactive simulators in this field

        All Simulators

        Related fields

        Thermal AnalysisManufacturing Process AnalysisV&V·Quality Assurance
        Rate this article
        Thank you for your feedback!
        Helpful
        More details
        Report error
        Helpful
        0
        More details
        0
        Report error
        0
        Written by NovaSolver Contributors
        Anonymous Engineers & AI — Sitemap
        About the Authors