Faster Valve Cv prediction with improved initial flowrate prediction in Autonomous Valve CFD
Wednesday, March 25, 2020
Faster Valve Cv prediction with improved initial flowrate prediction in Autonomous Valve CFD
By
Chaitanya Rane
Blog Author - Chaitanya Rane
Written by Chaitanya Rane
Approximately
10 Minutes Reading
Approximately
10 Minutes Reading
Pune, India – The simulationHub research & product development team are here with an improved initial flowrate prediction model for the Autonomous Valve CFD app. This new improvement is first of many new upgrades that simulationHub is planning to bring in its apps serving multiple domains. This new feature will prove a substantial factor in reducing the simulation time for the Autonomous Valve CFD app and thus provide the essential Cv, Kv and Cdt flow performance coefficients much faster while ensuring its accuracy and validity.
Why is this important?
To give our readers an idea as to how this new feature affects the simulation time significantly, I would like to take you through the simulation cycle that runs behind every valve opening simulation in the Autonomous Valve CFD app.
Flow performance parameters evaluation process used in Autonomous Valve CFD app
In Autonomous Valve CFD, a virtual flow loop testing environment is created to predict the flow performance parameters. In doing so, the app has been developed with reference to the ANSI/ISA–75.02.01 standard and actual flow loop testing reports from UWRL, USA & FCRI, India. The flow performance values are calculated at a pressure drop of 1 bar (with tolerance). To achieve the said pressure drop, the app runs a simulation loop in which the inlet flowrate is predicted iteratively until a pressure drop of 1 bar is achieved across the valve. This loop runs 3 times and hence the app must run multiple simulations to achieve the said pressure drop. The simulationHub CFD platform runs all simulations to achieve the required convergence and thus provide a result with maximum accuracy. As this loop runs for 3 iterations, the time required for evaluating the flow performance parameters is directly affected by this.
In this process, the initial flowrate prediction plays a major role. If the initial flowrate is predicted accurately in the first attempt, the number of simulation iteration will reduce and so will the time to evaluate the flow performance parameters.
Improved initial flowrate prediction
simulationHub uses a Machine Learning (ML) model to predict the initial flowrate for the Autonomous Valve CFD app. This model is trained regularly to update its prediction algorithm and provide better predictions.
Here is the formula for Cv,
Cv formula
If we consider the ΔP value equal to 1 bar and value of SG (Specific Gravity) equal to 1 (considering water as fluid), then we get a relation that Cv = Q, where Q is the flowrate.
It is based on this logic, that the ML model is trained. So, by training the ML model for Cv values, the ML model can predict the correct value for Q and thus achieve a pressure drop of 1 bar across the valve.
simulationHub performed over 1000 CFD simulations for different types of valves. Using this data, the new ML model has been improved and the newly predicted flowrate values are showing much higher accuracy than before.
Here is an overview of how this model has been upgraded,
Flowrate prediction models comparision
Improvement in simulation time
Here is an analysis of the reduction in simulation time,
Time comparision chart
This new prediction model is showing significant reduction of time for the higher opening conditions. The data shows that the time has been reduced by nearly 30 minutes for the higher opening conditions. The model also shows an improved simulation run time for mid-range valve openings.
This new model has now been implemented and made available in the production version of the Autonomous Valve CFD app. Valve manufacturers can take advantage of this feature by running their valve simulations and obtaining the essential flow performance parameters faster than ever.
You can have free trial of the app and get a hands-on experience using the app.
Start free trial
Blog Author - Chaitanya Rane
Chaitanya Rane
Chaitanya is a CFD Support Engineer at simulationHub. He is interested in the fields of physics and mathematics and enjoys exploring the domains like CFD, FEA and industrial applications of engineering simulations. He has worked on simulationHub's CFD simulation apps like Autonomous Valve CFD, Pedestrian Comfort Analysis and Autonomous HVAC CFD. Chaitanya is also a blogging enthusiast and contributes to the technical content writing at simulationHub. He holds a Bachelor's degree in Mechanical Engineering from the University of Pune.
Blog Author - Chaitanya Rane
Chaitanya Rane
Chaitanya is a CFD Support Engineer at simulationHub. He is interested in the fields of physics and mathematics and enjoys exploring the domains like CFD, FEA and industrial applications of engineering simulations. He has worked on simulationHub's CFD simulation apps like Autonomous Valve CFD, Pedestrian Comfort Analysis and Autonomous HVAC CFD. Chaitanya is also a blogging enthusiast and contributes to the technical content writing at simulationHub. He holds a Bachelor's degree in Mechanical Engineering from the University of Pune.
Comments