Discover how the synergy of GenAI and Stochos is revolutionizing web app creation, making it smarter and more efficient than ever before.
Challenges
In the rapidly evolving landscape of web app development, companies face numerous challenges. These include the need for faster development cycles, higher quality standards, and the integration of increasingly complex functionalities. Additionally, the pressure to innovate continuously while maintaining cost-efficiency adds another layer of difficulty.
Traditional methods often fall short in addressing these multifaceted demands. Manual coding, extensive testing phases, and iterative design processes can be time-consuming and error-prone, leading to delays and increased costs. The need for a more intelligent and streamlined approach is evident.
Engineering Solution
Integrating GenAI and Stochos to optimize the web app development process addresses these challenges and has leveraged advanced technologies to develop robust solutions. By harnessing the power of AI and sophisticated algorithms, we can automate repetitive tasks, enhance accuracy, and significantly reduce development time.
Case Study:
Previously, a 2D benchmark case of a simplified shower head (Figure 1) was utilized to demonstrate a Probabilistic Inference for Bayesian Optimization (PI-BO) application in optiSLang1.
Figure 1. 2D Benchmark Case
The study explored the optimal design (i.e. the diameters of internal holes) to satisfy two output conditions of maximum pressure drop and minimum uniformity. Out of 20 designs, a Pareto Front solution suggested, design 19 as the optimal one (Figure 2).
Figure 2. PI-BO Optimization Solution
Later, the study extended to a geometry prediction app generation using stand-alone Stochos. The app utilized a trained model with provided design cases and created a visualization to those cases regarding the distribution of pressure and velocities (Figure 3)2. The predicted geometries and pressure contours agreed well the CFD predictions.
Figure 3. The Comparison of Stochos App Predicted Geometry and Pressure Field to the Fluent Simulation for A Design
The current study enhances the previous app with the utilization of the "design objectives" to create the optimized design and associated pressure distribution. The work frame to generate a web app is summarized in Figure 4.
Figure 4. Schematic Work Frame of the Web App Generation
There are two scripts associated with the app generation:
The Train Model script reads multiple geometric and scalar data, preprocesses it, trains a 3D regression model on the geometries and values, and saves the resulting model to disk.Deep Infinite Mixture of Gaussian Processes (DIM-GP) regression model from the Stochos framework to fit the geometric (point cloud) data and associate it with regression targets, including both shape (geometry) and scalar optimization parameters. Figure 5, shows the corresponding section of the code. This script was run in Anaconda Spyder platform and press_vel_model was generated.
Figure 5. The Train Model Script Code
The App Generation Script utilizes the previously generated trained model as well as the scalar data pack. The code creates an interactive Streamlit-based web dashboard for Geometry Prediction using AI models. The code eets up sliders (using Streamlit) for users to specify physical parameters (like "PressureDrop [Pa]" and "Uniformity [Pa]") within the real data range.
When users click "Update Mesh and Prediction" button, it uses the generative model to create new geometry ("mesh") conditioned on user-specified scalars, adjusts the mesh, predicts field values over the mesh using the regression model, smooths the predictions (optional), and visualizes the mesh and prediction using PyVista, embedding the 3D plot in the app.
Figure 6 shows the predicted design features and pressure distribution with the scale shown in Figure 3 for high pressure drop and low uniformity objectives.
Figure 6. Web App Generated Design for High pressure Drop and Low Uniformity
Figure 7 shows the predicted design with pressure distribution for low pressure drop and low uniformity.
Figure 7. Web App Generated Design for Low pressure Drop and Low Uniformity
Benefits
The integration of GenAI and Stochos offers numerous benefits, including significantly reduced development times, enhanced accuracy, and improved overall quality of web apps. By automating repetitive tasks, our solution allows developers to focus on more critical aspects of the project, fostering creativity and innovation.
Moreover, the predictive capabilities of Stochos enable better project planning and resource management, reducing the risk of delays and cost overruns. Clients benefit from faster time-to-market, higher satisfaction rates, and a competitive edge in their respective industries.
The details of this application can be found in the video below:
The video can be reached from Ozen Engineering YouTube channel with the following link: Web App generation by Stochos and GenAI
Ozen Engineering Expertise
Ozen Engineering Inc. leverages it's extensive consulting expertise in CFD, FEA, thermal, optics, photonics, and electromagnetic simulations to achieve exceptional results across various engineering projects, addressing complex challenges like multiphase flows, erosion modeling, and channel flows using Ansys software.
We offer support, mentoring, and consulting services to enhance the performance and reliability of your hydraulic systems. Trust our proven track record to accelerate projects, optimize performance, and deliver high-quality, cost-effective results for both new and existing water control systems. For more information, please visit https://ozeninc.com.
References:
1 Ozen Engineering Blog: Probabilistic Inference for Bayesian Optimization Application on a 2D Benchmark Case
2 Ozen Engineering Blog: 3D Geometry Prediction using Data from Probabilistic Optimization: A Stochos App Application