Revolutionizing manufacturing processes: How the integration of Stochos enhances genAI applications in CFD predictions for a benchmark case.
Stochos has emerged as a revolutionary tool in the realm of generative artificial intelligence (genAI), particularly in the application of Computational Fluid Dynamics (CFD) predictions. This advanced technology leverages stochastic processes to enhance the accuracy and efficiency of predictive models, making it a game-changer in various industrial applications.
The integration of Stochos into CFD workflows allows engineers to account for uncertainties and variabilities in fluid behavior, leading to more robust and reliable simulations. This is crucial in complex engineering applications where even minor deviations can have significant impacts on performance and safety.
Despite the immense potential, integrating genAI and machine learning into CFD predictions presents several challenges. Ensuring data quality and consistency is paramount, as inaccurate data can lead to erroneous predictions.
Additionally, the complexity of fluid dynamics and the need for high computational power can be barriers to widespread adoption. Overcoming these challenges requires continuous advancements in computational methods and technology infrastructure.
To address these challenges, a robust engineering solution involves the development of advanced algorithms and computational models that can handle the complexities of fluid dynamics. Investing in high-performance computing infrastructure is also crucial to support the intensive computational demands of CFD simulations.
Previously, the Stochos application of machine learning and artificial intelligence was demonstrated to predict the scalar quantities and flow field properties for a benchmark case1, 2. In this blog, the application will be extended to genAI.
A compressed tube with a valve was utilized. The typical CFD workflow is demonstrated in Figure 1. This benchmark study was for a medical device application, therefore the operating fluid is set to blood properties. The variables considered are pressure, velocity, and hemolysis index (HI)
Figure 1. The CFD workflow of the benchmark application
To feed the Stochos model, the 12 training conditions and four test conditions were considered as before2 (Figure 2).
Figure 2. Training and test conditions
For all these conditions, the CFD workflow was followed, and field properties were predicted. The data were extracted for the entire flow field and used to feed the Stochos model. The model uses regression analysis and makes a model fit with the train data. Then, the model predicts using the test data. Furthermore, the model creates a "cfx" model which is loaded in genAI application with dimgp function. The genAI model predicts the field properties of the benchmark case (velocity, pressure, hemolysis index) for the tube of the range of the compression at a specified flow rate. The x_global_feat [:] = 2.0 indicates the flow rate of 2 L/min. The instances of different levels of compression were saved as individual images, which are then converted to a Gif file (Figure 3).
Figure 3. The basic steps of the genAI model
The genAI created velocity predictions in m/s are shown in Figure 4 for 2 L/min flow at different viewpoints of the tube for the range of compression (from fully open to 80% occluded).
Figure 4. The genAI generated velocity field at 2 L/min flow.
Similarly, Figure 5 demonstrates the pressure predictions in Pascal for 2 L/min flow.
Figure 5. The genAI generated pressure field at 2 L/min flow.
For the hemolysis, the impact of the flowrate and the tube compression are shown in Figure 6.
Flow: 1L/min
Flow: 2L/min
Flow: 3L/min
Flow: 4L/min
Figure 6. The genAI generated hemolysis index fields for different flow rates
As may be clear from all the hemolysis index Gif animations, the direction of flow is from the darker to the warmer color side of the tube.
The benefits of integrating Stochos and machine learning into CFD predictions are manifold. Enhanced accuracy and reliability of simulations lead to better design decisions, reduced costs, and improved performance of engineering systems.
Moreover, the ability to model uncertainties and variabilities provides a more comprehensive understanding of fluid behavior, which is critical for ensuring safety and efficiency in industrial applications.
The above predictions and Gif generations took less than a minute at each case with a minimum r2 of 0.8 to maximum of 0.99 during the model training. Considering the size of the training and the test data, the predictions are significantly pleasing.
We offer support, mentoring, and consulting services to enhance the performance and reliability of your mining equipment and systems. Trust our proven track record to accelerate projects, optimize performance, and deliver high-quality, cost-effective results for both new and existing systems. For more information, please visit https://ozeninc.com.
1 Ozen Engineering Blog: Machine Learning on CFD Predictions: Use of Stochos on a Benchmark Application
2 Ozen Engineering Blog: Machine Learning on CFD Predictions Part-2: Use of Stochos on a Benchmark Application to Predict Flow Field