Fluid-Structure Interaction Chpater 2: One-way physics coupling

This chapter introduces one-way (weak) fluid–structure interaction as a practical and widely used approximation for coupled problems where structural response does not significantly alter the flow field. The chapter explains the mathematical and numerical meaning of one-way coupling, contrasts monolithic and partitioned solution strategies, and details how data is transferred from CFD to structural or thermal solvers. Emphasis is placed on physical validity, numerical assumptions, and industrial applicability.

 

Context and Motivation

In many engineering problems:

  • Fluids generate loads (pressure, shear, heat)

  • Structures respond elastically or thermally

  • The response does not modify the flow in any meaningful way

Examples include:

  • Wind loading on stiff structures

  • Pressure loads on pipes and valves

  • Thermal expansion due to hot gases

  • Early-stage design verification

In such cases, fully coupled FSI is unnecessary and inefficient.
One-way coupling offers a robust, low-cost alternative.


What “One-Way Coupling” Really Means

3.1 Definition

One-way coupling is a partitioned multi-physics approach where:

  1. The source physics (typically CFD) is solved first

  2. Selected quantities are transferred to a receiver physics

  3. The receiver is solved without feedback to the source

In FSI terms:

  • Fluid → structure

  • No structure → fluid feedback

3.2 Physical Interpretation

Physically, this assumes:

  • Structural deformation is small

  • Flow topology remains unchanged

  • Inertial and added-mass effects are negligible

A useful rule:

If the structure “feels” the flow, but the flow does not “notice” the structure moving, one-way coupling is valid.


One-Way Coupling in the FSI Landscape

4.1 Relation to Coupling Strength

  • One-way coupling ⟶ weak coupling

  • Two-way coupling ⟶ strong coupling

Weak coupling corresponds to:

  • Low feedback sensitivity

  • Off-diagonal coupling terms being negligible

  • Stable explicit execution

4.2 Comparison with Two-Way FSI

                  Aspect                                     One-Way                                     Two-Way                  
Feedback None Mutual
Stability Very high Conditional
Cost Low High
Setup complexity Low High
Typical use Industrial design Research / dynamic instabilities

Solution Strategies: Monolithic vs Partitioned

5.1 Why Partitioned Approaches Dominate

Although monolithic formulations are mathematically elegant, they are:

  • Hard to develop

  • Computationally expensive

  • Impractical for industrial workflows

Partitioned approaches:

  • Reuse validated CFD and FEA solvers

  • Allow independent meshing and modeling

  • Are the industry standard (e.g., ANSYS Workbench)

One-way coupling is always partitioned.

5.2 Sequential Execution

In one-way coupling:

  • Solvers are executed sequentially

  • No sub-iterations are required

  • Coupling is explicit in time

This avoids added-mass instabilities entirely.


Transferred Quantities in One-Way FSI

6.1 Mechanical Loads

Commonly transferred from CFD to structural solvers:

  • Pressure

  • Integrated forces

  • Shear stresses (less common)

Pressure transfer is the dominant case .

6.2 Thermal Loads

Thermal one-way coupling includes:

  • Surface temperature

  • Heat transfer coefficient (HTC)

  • Volumetric temperature fields

Important distinction:

  • If conjugate heat transfer (CHT) is solved in CFD → transfer solid temperature

  • Otherwise → transfer HTC, not fluid temperature


Data Transfer and Mapping

7.1 Matching vs Non-Matching Meshes

In practice:

  • Fluid and structural meshes rarely match

  • Data mapping is required at interfaces

Key requirements:

  • Conservation of global forces

  • Preservation of rigid-body motion

  • Avoidance of spurious oscillations

7.2 Interpolation Nature

Most industrial tools:

  • Use linear interpolation

  • Are not conservative

  • Trade exact conservation for robustness

This is acceptable in one-way coupling because:

  • No feedback amplification exists

  • Errors do not propagate back to the fluid solution


Time Treatment in One-Way Coupling

8.1 Steady CFD → Static Structure

Most common case:

  • Steady flow

  • Static structural response

Used for:

  • Wind loading

  • Pressure vessel verification

8.2 Transient CFD → Transient Structure

More advanced cases:

  • Sloshing loads

  • Pulsating pressure

  • Thermal cycling

Here:

  • Time-accurate data is transferred

  • Structural solver integrates independently in time


Typical One-Way FSI Applications

9.1 Mechanical

  • Chimneys and towers under wind

  • Valves, elbows, T-junctions

  • Offshore decks under wave impact

9.2 Thermal

  • Exhaust systems

  • Gas turbine blades

  • Storage tanks under thermal cycling

  • EGR and cooling systems


Engineering Intuition

  • One-way coupling is a modeling decision, not a solver limitation

  • It is valid when feedback is physically negligible

  • It is often the correct engineering choice

  • Over-coupling wastes time and introduces unnecessary instability

Good practice:

Start with one-way coupling. Escalate only if physics demands it.


Limitations and Assumptions

One-way coupling fails when:

  • Structural deformation alters flow paths

  • Added-mass effects dominate

  • Aeroelastic instabilities appear

  • Large rigid-body motion exists

In these cases, two-way FSI is mandatory.


Study Priorities (Chapter 2)

If short on time:

  1. Physical meaning of one-way coupling

  2. Validity assumptions

  3. Types of transferred quantities

  4. Difference between pressure vs HTC transfer

  5. Industrial use cases


Key Takeaways

  • One-way coupling is the most widely used FSI approach in industry.

  • It assumes no feedback from structure to fluid.

  • Partitioned solvers make it robust and efficient.

  • Data transfer accuracy matters, but stability is guaranteed.

  • It is the natural starting point for any FSI analysis.

Previous
Previous

Citizen Data Scientist, Module V: Unsupervised Learning: Discovering hidden patterns

Next
Next

Citizen Data Scientist, Module IV: Applying Data Science in Practice: Feature Engineering, Scaling, and Selection