280x
001868
2024-10-17

Esempio di validazione per la simulazione CFD utilizzando i dati sperimentali della galleria del vento dell'Università di Aquisgrana

La convalida delle simulazioni CFD con dati sperimentali migliora la precisione confrontando i risultati della simulazione con le condizioni del mondo reale. Questo processo identifica le discrepanze, consentendo modifiche per migliorare l'affidabilità del modello. In definitiva, crea fiducia nella capacità della simulazione di prevedere scenari di carico del vento.

Gli esempi di validazione di simulazioni fluidodinamiche computazionali (CFD) utilizzando lo studio sperimentale (Figura 1) sono un passaggio cruciale nella verifica dell'accuratezza dei modelli di simulazione. This process involves a detailed comparison between the results obtained from CFD simulations and those derived from real-world experiments. It ensures that the simulations can be reliably used to predict aerodynamic simulation in various applications, from engineering design to environmental analysis. Validating CFD models against experimental data helps identify discrepancies, allowing for adjustments to model parameters, turbulence models, or numerical methods. Ultimately, this iterative process builds confidence in the simulation's predictive capabilities and ensures that the CFD model can reflect real-world phenomena.

In the current example which is a collaboration between Dlubal Software and Aachen University, we outline the key steps for implementing CFD simulation in RWIND using experimental data. We greatly appreciate Prof. Frank Kemper and <nobr>Dipl.-Ing. Mirko</nobr> Friehe from Aachen University for providing the experimental wind tunnel data and their invaluable support throughout this project.

Image 1 and Image 2 show the experimental model, represented as a 3D rectangular building in the wind tunnel. In the main model, sensors are included to measure key parameters, such as wind pressure values and the wind pressure coefficient. The small blocks surrounding the model simulate roughness terrain to reflect the surrounding conditions accurately.


Producing a validation example for a CFD (Computational Fluid Dynamics) simulation in RWIND using experimental data from Aachen University involves a systematic process. Here’s a step-by-step guide:

Step 1: Defining Validation Objectives

  • Purpose: In this part, we establish why we are conducting this validation. Common goals are to verify the accuracy of RWIND results compared to physical experimental data.
  • Scope: The target results for validations include wind pressure values on defined sensors and base forces according to various wind directions.

Step 2: Collect Experimental Data from Aachen University

  • Data Acquisition: Gather all the necessary experimental data, such as wind speed, wind directions, pressure measurements, and any relevant boundary conditions.
  • Data Format: Ensure the data is in a format that RWIND can process, such as text files or spreadsheets, and confirm that it aligns with the units and scale required by RWIND. Here is a FAQ link about how to introduce experimental data to RWIND:
  • Quality Check: Review the data for completeness and accuracy. Ensure that the data covers the range of conditions you plan to simulate.

Step 3: Model Setup in RWIND

  • Geometry Import: Create or import the geometry of the structure being studied (for example, a building or bridge). This can be modeled directly in RWIND or imported from RFEM or CAD program (Image 3).
  • Boundary Conditions: Apply the same boundary conditions as those used in the experimental setup. This includes specifying wind velocity, turbulence intensity, and other required factors (Image 4).
  • Meshing: Generate a computational mesh suitable for your study (Image 5). This step involves discretizing the geometry into smaller elements that RWIND uses for calculations. Ensure that the mesh is fine enough in areas with high gradients (like around edges or surfaces with expected turbulent flow).

Step 4: Running Simulation

  • Initial Test Run: Start with a test run to identify any issues with the setup. Check for mesh quality, boundary conditions, and any convergence problems.
  • Full Simulation: Once the test run is successful, proceed with the full simulation. Monitor the simulation for convergence and stability, making adjustments if necessary.

Step 5: Post-Processing Results

  • Data Extraction: Export the simulation results, including resultant wind force, pressure distributions for defined measurements points from RWIND to compare with experimental data.
  • Visualization: Use RWIND’s post-processing tools to visualize the flow patterns and pressure distributions. Create plots, graphs, or 3D visualizations to better interpret the results.

Step 6: Compare Results with Experimental Data

  • Data Alignment: Ensure the simulation and experimental data are aligned in terms of spatial locations, units, and scales.
  • Statistical Analysis: Perform a statistical comparison between the simulation and experimental data. Calculate deviation metrics such as correlation coefficients to quantify the accuracy.

Here are the resultant base forces according to various wind directions, as analyzed in RWIND and compared to an experimental study (Image 6). The k-epsilon turbulence model was used for the base force calculations, considering both low and high levels of turbulence intensity. The results with higher turbulence intensity showed closer alignment with the experimental study, with a deviation of approximately 6%.

The second parameter is the wind pressure values calculated at measurement points in both numerical and the experimental study (Image 7). In RWIND, the standard k-epsilon and k-omega SST models were used to compare these wind pressure values with the experimental results. The statistical analysis indicates that the k-omega SST model provides a closer trend to the experimental results according to the correlation coefficient (R=0.98) and coefficient of determination (R2=0.96) in Image 8.

Step 8: Documentation and Reporting

Document the entire validation process, including setup, simulation parameters, comparison methodology, and results. Highlight any deviations from experimental data and potential reasons. Provide insights into the CFD model's accuracy and suggest improvements or further validation steps if necessary.


Autore

Il signor Kazemian è responsabile dello sviluppo del prodotto e del marketing per Dlubal Software, in particolare per il programma RWIND 2.



;