Uncertainity

Main Sources of Uncertainty in CFD Modeling

Uncertainty in Computational Fluid Dynamics (CFD) modeling arises from potential deficiencies in the model due to limited knowledge or inaccuracies in the simulation process. The main sources of uncertainty include:

1. Input Uncertainty:

   • Domain Geometry: The geometry of the simulation domain might not perfectly represent the real-world object due to manufacturing tolerances, conversion errors from CAD models to CFD-compatible formats, and the discrete representation of boundaries. These discrepancies can lead to variations in the flow field that affect simulation accuracy.
   • Boundary Conditions: The accuracy of boundary conditions like velocity, temperature, and pressure is often limited by the availability and precision of data. Simplified or assumed conditions, such as uniform pressure at an inlet, can introduce errors.
   • Fluid Properties: Assumptions of constant fluid properties, such as density or viscosity, can lead to inaccuracies if these properties vary significantly with flow parameters like temperature or pressure.

2. Physical Model Uncertainty:

   • Inadequate Representation of Physical Processes: CFD models often rely on submodels (e.g., turbulence models) that are calibrated for specific conditions. Applying these submodels to more complex or different flows can result in inaccuracies if the model does not adequately represent the real physical processes.
   • Simplifying Assumptions: Simplifications, such as assuming steady or incompressible flow, can reduce computational cost but might introduce errors if these assumptions do not accurately represent the real flow conditions.

Impact of Uncertainties on Simulation Results

These uncertainties can lead to discrepancies between the simulation results and the real-world flow. For instance:

• Input Uncertainty: Errors in domain geometry can cause incorrect predictions of flow patterns, while inaccurate boundary conditions can lead to incorrect flow velocities, pressures, or temperatures in the simulated domain. Incorrect fluid properties might result in wrong predictions of heat transfer or fluid behavior.
• Physical Model Uncertainty: Inadequate turbulence modeling can lead to incorrect predictions of flow separation, reattachment, or mixing. Simplifying assumptions can result in the loss of critical flow features, such as vortex shedding or shock waves.

Minimizing Uncertainties

To minimize these uncertainties, the following approaches can be taken:

1. Improving Input Accuracy:

   • Refine Geometry and Boundary Conditions: Ensure the domain geometry is as accurate as possible and use precise boundary conditions by obtaining high-quality experimental data or measurements.
   • Advanced Meshing Techniques: Use finer meshes near boundaries and in regions of interest to better capture the real geometry and flow details.
   • Accurate Fluid Properties: Use variable fluid properties that depend on local flow conditions rather than assuming constant values.

2. Enhancing Physical Models:

   • Use of Advanced Submodels: Select the most appropriate turbulence or combustion models for the specific flow conditions, even if they are computationally more expensive. For example, large eddy simulation (LES) may provide more accurate results for complex flows than the standard k–ε model.
   • Validation with Experimental Data: Regularly validate CFD models with experimental data to ensure the selected models and assumptions are accurate for the specific application.

3. Careful Application of Simplifications:

   • Evaluate Simplifications: Before applying simplifications like assuming steady or symmetrical flow, analyze the flow conditions to determine if these assumptions are valid.
   • Sensitivity Analysis: Perform sensitivity analyses to assess the impact of various assumptions and inputs on the simulation results, helping to identify and mitigate significant sources of uncertainty.