Comsol Multiphysics

COMSOL Architecture

Model Structure

• Component: Independent geometry and physics
• Study: Solution sequence with one or more steps
• Physics Interfaces: Pre-configured PDE systems for specific phenomena
• Multiphysics Couplings: Automatic or manual coupling between physics
• Solver Configurations: Automatic or customized solver sequences

Solver Technology

• COMSOL Solver: Direct and iterative solvers
• Fully Coupled Approach: Simultaneous solution of all equations
• Segregated Solver: Physics-based segregation for large models
• Time-Stepping: Implicit methods for transient analysis
• Mesh Adaptation: Automatic refinement based on solution

Licensing Requirements

Commercial Licensing

COMSOL Multiphysics requires commercial licenses for all production use:

• Base Package:

  • COMSOL Multiphysics license (required for all users)
  • Includes basic PDE interfaces and computational tools
  • License managed via FlexNet License Server

• Module Licenses:

  • Individual modules purchased separately
  • Floating licenses (checked out during use)
  • Batch mode requires license checkout
  • HPC add-ons for parallel computing

• License Server Setup:

  # Set license server environment variable (Linux/Mac)
  export [email protected]
  
  # Or in Windows
  set [email protected]
  

• License Types:

  • Floating Network License: Shared among users
  • Node-Locked License: Tied to specific machine
  • HPC License: Additional cores for parallel computing
  • Batch License: For batch/automated simulations

Academic Licensing

• Classroom Kit: Limited to educational use
• Research licenses: Available for academic institutions
• Restrictions on commercial applications
• May have feature or model size limitations

Important Licensing Notes

• License Check: Always verify license availability before batch runs
• License Release: Properly close COMSOL to release licenses
• Batch Operations: Require special batch/HPC licenses
• Module Dependencies: Some modules require others (e.g., FSI needs CFD + Structural)
• API Access: Requires base license plus relevant modules

Key Modules for Pump Applications

CFD Module

Comprehensive computational fluid dynamics for single-phase and multiphase flows:

Capabilities:

• Turbulence Models:

  • RANS: k-epsilon, k-omega, SST
  • LES and DES for high-fidelity simulations
  • Wall functions and low-Re formulations

• Flow Types:

  • Incompressible and compressible
  • Laminar and turbulent
  • Steady-state and transient
  • Rotating machinery (frozen rotor, sliding mesh)

• Multiphase Flow:

  • Euler-Euler multiphase
  • Phase field methods
  • Level set tracking
  • Bubbly flow models

Pump-Specific Features:

• Rotating reference frames for impellers
• Mixing plane interfaces
• Cavitation modeling
• Pressure pulsation analysis

Structural Mechanics Module

Comprehensive structural analysis including linear and nonlinear behavior:

Capabilities:

• Analysis Types:

  • Static structural analysis
  • Eigenfrequency analysis (modal analysis)
  • Frequency response analysis
  • Transient dynamics
  • Prestressed analysis

• Material Models:

  • Linear elastic
  • Hyperelastic (rubber, polymers)
  • Plasticity (metals under high stress)
  • Composite materials
  • Contact and friction

• Dynamic Analysis:

  • Modal analysis for natural frequencies
  • Harmonic response
  • Time-dependent loading
  • Damping models (Rayleigh, modal)

Pump-Specific Features:

• Rotating machinery stress analysis
• Bolt preload and assembly stress
• Fatigue analysis
• Contact between impeller and casing

Fluid-Structure Interaction (FSI)

Couples CFD and Structural Mechanics for two-way interaction:

Coupling Approaches:

1. One-Way FSI:

   • Fluid loads applied to structure
   • Structure does not affect fluid
   • Faster computation
   • Suitable for rigid-like structures

2. Two-Way FSI:

   • Fully coupled fluid and structure
   • Deforming mesh methods
   • Accounts for large deformations
   • Required for flexible structures

3. Weak Coupling:

   • Sequential solution (staggered approach)
   • Fluid → Structure → Fluid iteration
   • Better for loosely coupled problems

4. Strong Coupling:

   • Fully implicit simultaneous solution
   • Better convergence for tightly coupled problems
   • Higher computational cost

FSI Features:

• Automatic mesh deformation (ALE method)
• Remeshing for large deformations
• Fluid loads computed automatically
• Pressure and viscous forces transferred
• Support for multiple solid bodies

Pump FSI Applications:

• Impeller blade vibration under fluid forces
• Casing vibration and acoustic radiation
• Seal deflection under pressure
• Shaft deflection and critical speeds
• Cavitation-induced vibration

Additional Relevant Modules

Acoustics Module:

• Pressure pulsation analysis
• Noise radiation from pump casing
• Structure-borne noise
• Coupled acoustic-structure-fluid problems

Heat Transfer Module:

• Thermal loads in pumps
• Coupled thermal-structural analysis
• Conjugate heat transfer (fluid-solid)
• Thermal expansion effects

Optimization Module:

• Design optimization
• Topology optimization
• Shape optimization
• Parameter sweeps

Java API for Automation

COMSOL provides a comprehensive Java API for programmatic model building and automation:

API Structure

Core Components:

• Model object: Top-level container
• Component: Geometry and physics
• Physics interfaces: Add and configure physics
• Study: Define solution procedures
• Results: Post-processing and visualization

Basic Java API Workflow

import com.comsol.model.*;
import com.comsol.model.util.*;

public class PumpFSI {
    public static Model run() {
        // Create model
        Model model = ModelUtil.create("PumpFSI");

        // Create component
        model.component().create("comp1", true);

        // Create geometry
        model.component("comp1").geom().create("geom1", 3);

        // Import CAD geometry
        model.component("comp1").geom("geom1").create("imp1", "Import");
        model.component("comp1").geom("geom1").feature("imp1")
            .set("filename", "/path/to/pump_geometry.step");
        model.component("comp1").geom("geom1").run();

        // Add fluid physics (CFD)
        model.component("comp1").physics().create("spf", "LaminarFlow", "geom1");
        model.component("comp1").physics("spf").selection()
            .named("geom1_fluid_domain");

        // Add structural physics
        model.component("comp1").physics().create("solid", "SolidMechanics", "geom1");
        model.component("comp1").physics("solid").selection()
            .named("geom1_solid_domain");

        // Add FSI coupling
        model.component("comp1").multiphysics().create("fsi1", "FluidStructureInteraction", 3);
        model.component("comp1").multiphysics("fsi1")
            .selection().named("geom1_fsi_boundary");

        // Create mesh
        model.component("comp1").mesh().create("mesh1");
        model.component("comp1").mesh("mesh1").automatic(true);
        model.component("comp1").mesh("mesh1").run();

        // Create study
        model.study().create("std1");
        model.study("std1").create("time", "Transient");
        model.study("std1").feature("time").set("tlist", "range(0,0.01,1)");

        // Solve
        model.sol().create("sol1");
        model.sol("sol1").study("std1");
        model.sol("sol1").feature().create("st1", "StudyStep");
        model.sol("sol1").feature().create("v1", "Variables");
        model.sol("sol1").feature().create("t1", "Time");
        model.sol("sol1").attach("std1");
        model.sol("sol1").runAll();

        // Save model
        model.save("/path/to/pump_fsi_model.mph");

        return model;
    }

    public static void main(String[] args) {
        run();
    }
}

Running Java API Scripts

# Compile Java file
comsol compile PumpFSI.java

# Run with COMSOL
comsol batch -inputfile PumpFSI.class -outputfile results.mph

# Or run directly
java -cp /path/to/comsol/plugins/*:. PumpFSI

API Advantages

• Reproducibility: Scripts ensure consistent model building
• Parametric Studies: Easy to vary parameters
• Batch Processing: Run multiple cases automatically
• Integration: Connect with other tools and databases
• Version Control: Track model changes in source control

MATLAB LiveLink

COMSOL integrates seamlessly with MATLAB for enhanced scripting and data processing:

LiveLink Features

Model Control from MATLAB:

• Build and modify COMSOL models
• Run simulations from MATLAB scripts
• Extract results into MATLAB workspace
• Use MATLAB's data analysis tools

Installation:

• Requires separate LiveLink for MATLAB license
• Configure MATLAB path to COMSOL installation
• Start COMSOL server or use direct connection

Basic MATLAB LiveLink Usage

% Initialize COMSOL with MATLAB
import com.comsol.model.*
import com.comsol.model.util.*

% Start COMSOL server (if not already running)
mphstart

% Create or load model
model = mphload('pump_model.mph');

% Modify parameters
model.param.set('inlet_velocity', '5[m/s]');
model.param.set('outlet_pressure', '101325[Pa]');

% Run study
model.study('std1').run();

% Extract results
pressure = mpheval(model, 'p', 'dataset', 'dset1');
velocity = mpheval(model, 'u', 'dataset', 'dset1');

% Process in MATLAB
mean_pressure = mean(pressure.d1);
max_velocity = max(sqrt(velocity.d1.^2 + velocity.d2.^2 + velocity.d3.^2));

% Plot using MATLAB
figure;
plot(pressure.p, pressure.d1);
xlabel('Position');
ylabel('Pressure [Pa]');
title('Pressure Distribution');

% Save results
save('pump_results.mat', 'pressure', 'velocity');

% Close COMSOL
ModelUtil.remove('model');

Parametric Study with MATLAB

% Parametric study of inlet velocity effects
velocities = 1:1:10;  % m/s
results = struct();

for i = 1:length(velocities)
    fprintf('Running case %d: velocity = %.1f m/s\n', i, velocities(i));

    % Set parameter
    model.param.set('inlet_velocity', sprintf('%f[m/s]', velocities(i)));

    % Solve
    model.study('std1').run();

    % Extract force on impeller
    force = mphint2(model, 'spf.Fp_x', 'surface', 'selection', 5);
    results(i).velocity = velocities(i);
    results(i).force = force;

    % Extract vibration amplitude
    displacement = mphmax(model, 'sqrt(u^2+v^2+w^2)', 'volume', 'selection', 3);
    results(i).max_displacement = displacement;
end

% Plot results
figure;
subplot(2,1,1);
plot([results.velocity], [results.force], '-o');
xlabel('Inlet Velocity [m/s]');
ylabel('Force on Impeller [N]');
grid on;

subplot(2,1,2);
plot([results.velocity], [results.max_displacement]*1e6, '-o');
xlabel('Inlet Velocity [m/s]');
ylabel('Max Displacement [μm]');
grid on;

% Save results
save('parametric_results.mat', 'results');

Common Workflows

Workflow 1: Pump Casing Vibration Under Fluid Loads

Application: Analyze vibration of pump casing due to pressure pulsations from fluid flow.

Approach: One-way FSI (fluid loads mapped to structure)

Steps:

1. Geometry Setup:

   • Import pump casing geometry
   • Create fluid domain inside casing
   • Define FSI boundary (fluid-structure interface)

2. CFD Setup:

   • Define inlet and outlet boundaries
   • Set up turbulence model (k-epsilon or k-omega SST)
   • Configure transient solver
   • Apply rotating reference frame if analyzing flow with rotating impeller

3. Structural Setup:

   • Define material properties (steel, cast iron)
   • Apply boundary conditions (fixed support at mounting points)
   • Set up transient structural analysis
   • Include damping if known

4. FSI Coupling:

   • Map fluid pressure and shear stress to structure
   • One-way coupling (structure assumed rigid relative to fluid)
   • Time synchronization between fluid and structural solvers

5. Solution:

   • Solve transient CFD first
   • Extract time-varying pressure loads
   • Apply loads to structural model
   • Solve structural dynamics

6. Post-Processing:

   • Extract displacement at critical points
   • FFT analysis for frequency content
   • Compare with natural frequencies
   • Stress concentration analysis

Workflow 2: Impeller FSI Analysis

Application: Coupled analysis of impeller blade deformation under fluid forces.

Approach: Two-way FSI with moving mesh

Steps:

1. Geometry and Mesh:

   • Import impeller geometry
   • Create fluid domain around impeller
   • Define FSI interface at blade surfaces
   • Mesh solid impeller and fluid domain
   • Fine mesh at FSI boundary

2. Fluid Physics:

   • Turbulent flow (k-omega SST recommended)
   • Rotating reference frame for impeller
   • Inlet velocity or mass flow boundary
   • Outlet pressure boundary
   • Moving mesh (ALE) enabled

3. Structural Physics:

   • Linear elastic material initially
   • Fixed at shaft connection
   • Prestress from centrifugal loading
   • Include rotation effects

4. Two-Way FSI Coupling:

   • Fluid applies loads to structure
   • Structure deformation moves mesh
   • Iterative coupling within each time step
   • Convergence criteria for FSI iteration

5. Prestressed FSI:

   • First: Stationary structural analysis with centrifugal load
   • Second: Use prestressed state as initial condition for FSI
   • Improves convergence and accuracy

6. Solution Strategy:

   • Start with coarse time steps for initial transient
   • Refine time stepping for periodic solution
   • Monitor residuals and FSI iterations
   • Expect longer solve times for strong coupling

7. Post-Processing:

   • Blade tip displacement vs. time
   • Stress distribution on blades
   • Effect of deformation on flow field
   • Comparison of rigid vs. flexible results

Workflow 3: Coupled Thermal-Flow Analysis

Application: Thermal effects in pumps handling hot fluids.

Approach: Conjugate heat transfer with thermal expansion

Steps:

1. Multi-Domain Setup:

   • Fluid domain (hot liquid)
   • Solid domains (casing, impeller)
   • Define fluid-solid boundaries

2. Coupled Physics:

   • CFD for fluid flow
   • Heat Transfer in Fluids (energy equation)
   • Heat Transfer in Solids (conduction)
   • Thermal Stress in solids
   • Automatic temperature coupling at boundaries

3. Boundary Conditions:

   • Inlet: Temperature and velocity
   • Outlet: Pressure outflow
   • External surfaces: Convection or radiation
   • Solid boundaries: Fixed temperature or insulation

4. Material Properties:

   • Temperature-dependent viscosity
   • Thermal expansion coefficients
   • Thermal conductivity

5. Solution Sequence:

   • Steady-state thermal-flow first
   • Use as initial condition for thermal stress
   • Or fully coupled transient simulation

6. Post-Processing:

   • Temperature distribution
   • Thermal stresses
   • Thermal expansion displacements
   • Effect on clearances and fits

Workflow 4: Modal Analysis with Fluid Loading

Application: Natural frequencies of pump casing in contact with fluid.

Approach: Eigenfrequency analysis with fluid-structure coupling

Steps:

1. Geometry:

   • Pump casing structure
   • Surrounding or contained fluid
   • FSI boundaries

2. Physics Setup:

   • Pressure Acoustics for fluid
   • Solid Mechanics for structure
   • Acoustic-Structure Boundary coupling

3. Boundary Conditions:

   • Structural: Fixed supports
   • Acoustic: Sound hard walls at rigid boundaries
   • FSI: Acceleration coupling

4. Study:

   • Eigenfrequency study
   • Search for eigenfrequencies in range of interest
   • Acoustic-structure coupling included automatically

5. Results:

   • Natural frequencies with fluid loading
   • Mode shapes
   • Comparison with in-air and in-vacuum modes
   • Fluid added mass effect

Batch Execution

Command Line Execution

COMSOL can run in batch mode without GUI:

# Run existing model file
comsol batch -inputfile pump_model.mph -outputfile results.mph

# Run Java method file
comsol batch -inputfile PumpFSI.java -outputfile results.mph

# Run with specific study
comsol batch -inputfile pump_model.mph -outputfile results.mph -study std1

# Run with parameter override
comsol batch -inputfile pump_model.mph -outputfile results.mph \
  -pname inlet_velocity -plist 5,10,15,20

# Run MATLAB script
comsol batch -inputfile pump_analysis.m -outputfile results.mph

# Specify number of processors
comsol batch -np 8 -inputfile pump_model.mph -outputfile results.mph

Batch Script Example (Linux)

#!/bin/bash
# batch_comsol.sh

# Set license server
export [email protected]

# Set number of processors
NPROCS=8

# Input/output files
INPUT_MODEL="pump_fsi_base.mph"
OUTPUT_DIR="results"

# Create output directory
mkdir -p ${OUTPUT_DIR}

# Parameter sweep: inlet velocities
VELOCITIES="2 4 6 8 10"

for VEL in ${VELOCITIES}; do
    echo "Running simulation with inlet velocity = ${VEL} m/s"

    OUTPUT_FILE="${OUTPUT_DIR}/pump_fsi_v${VEL}.mph"
    LOG_FILE="${OUTPUT_DIR}/pump_fsi_v${VEL}.log"

    # Run COMSOL in batch mode
    comsol batch -np ${NPROCS} \
        -inputfile ${INPUT_MODEL} \
        -outputfile ${OUTPUT_FILE} \
        -pname inlet_velocity \
        -plist ${VEL} \
        > ${LOG_FILE} 2>&1

    # Check exit status
    if [ $? -eq 0 ]; then
        echo "  Simulation completed successfully"
    else
        echo "  ERROR: Simulation failed - check ${LOG_FILE}"
        exit 1
    fi
done

echo "All simulations completed"

Python Automation with JPype

Use JPype to call COMSOL Java API from Python:

import jpype
import jpype.imports
from jpype.types import *
import numpy as np

# Start JVM with COMSOL
comsol_root = '/usr/local/comsol56/multiphysics'
jvm_path = jpype.getDefaultJVMPath()

jpype.startJVM(
    jvm_path,
    f"-Djava.library.path={comsol_root}/lib/glnxa64",
    f"-Dcs.lic.path={comsol_root}/license",
    classpath=f"{comsol_root}/plugins/*"
)

# Import COMSOL packages
from com.comsol.model.util import ModelUtil

# Load model
model = ModelUtil.load('/path/to/pump_model.mph')

# Parametric study
velocities = np.linspace(2, 10, 5)
results = []

for vel in velocities:
    print(f"Running velocity = {vel:.1f} m/s")

    # Set parameter
    model.param().set('inlet_velocity', f'{vel}[m/s]')

    # Run study
    model.study('std1').run()

    # Extract results (example)
    # Add actual result extraction here

    # Save case
    model.save(f'results/pump_v{vel:.1f}.mph')

    results.append({
        'velocity': vel,
        # Add extracted results here
    })

# Cleanup
ModelUtil.remove('model')
jpype.shutdownJVM()

print("Parametric study completed")

Best Practices

Model Setup

1. Geometry: Clean, defeatured geometry suitable for meshing
2. Selections: Use named selections for boundary conditions
3. Parameters: Use global parameters for easy modification
4. Units: Consistent unit system throughout model
5. Symmetry: Exploit symmetry when possible

Meshing

1. Mesh Quality: Check mesh quality metrics (skewness, growth rate)
2. Boundary Layers: Critical for turbulent flow
3. FSI Interface: Fine mesh at fluid-structure boundary
4. Mesh Independence: Verify solution mesh convergence
5. Adaptive Meshing: Use for complex flows

Physics Setup

1. Appropriate Models: Select physics models suitable for application
2. Material Properties: Verify material data accuracy
3. Boundary Conditions: Ensure physical consistency
4. Initial Conditions: Provide good initial guess
5. Coupling Settings: Check FSI coupling parameters

Solver Strategy

1. Solver Selection: Automatic vs. manual solver configuration
2. Convergence Criteria: Appropriate tolerances
3. Time Stepping: Adaptive time stepping for transient
4. Linearization: Check nonlinear solver settings
5. Parallel Computing: Use multiple cores effectively

Automation

1. Version Control: Track model files and scripts
2. Documentation: Comment code and document workflow
3. Error Handling: Implement checks and error recovery
4. Testing: Validate automation on simple cases first
5. License Management: Check license availability in batch jobs

Post-Processing

1. Data Extraction: Export only necessary data
2. Visualization: Standardized plots for comparison
3. Validation: Compare with experimental or analytical results
4. Reporting: Automated report generation
5. Data Management: Organize results systematically

Troubleshooting

Common Issues

Licensing Problems:

• Verify license server connectivity
• Check module availability
• Ensure batch licenses for automated runs

Meshing Failures:

• Simplify geometry (remove small features)
• Adjust mesh size parameters
• Use different mesh sequence

Convergence Issues:

• Reduce load step or time step
• Improve initial conditions
• Adjust solver tolerances
• Check for poor mesh quality
• Review boundary conditions

FSI-Specific Issues:

• Mesh deformation failures: Remesh or adjust smoothing
• Slow FSI convergence: Adjust relaxation factors
• Large deformations: Consider remeshing strategies
• Time step too large: Reduce for stability

Performance Issues:

• Use coarser mesh initially
• Simplify physics models
• Exploit symmetry
• Increase parallel cores
• Use segregated solver for large models

Additional Resources

Documentation

• COMSOL Multiphysics User's Guide: Core functionality
• CFD Module User's Guide: Fluid flow physics
• Structural Mechanics Module User's Guide: Solid mechanics
• COMSOL API Reference: Java API documentation
• LiveLink for MATLAB User's Guide: MATLAB integration

Online Resources

• COMSOL Website: https://www.comsol.com
• COMSOL Blog: Application examples and tips
• COMSOL Support Center: Knowledge base and downloads
• Application Gallery: Example models and tutorials
• COMSOL Forums: Community discussions

Training

• COMSOL Training Courses: Official training programs
• Webinar Archive: Recorded training sessions
• Learning Center: Video tutorials and guides

Support

• Technical Support: [email protected]
• Regional Support: Contact local COMSOL office
• Application Engineers: Assistance with complex problems