How Accurate Are Autonomous Valve CFD Results?

Validation studies against Physical Flow Loop Test data

For decades, Computational Fluid Dynamics (CFD) has been a powerful tool in the engineer's arsenal, allowing for the simulation of complex fluid behaviours. However, its power has always been gated by significant challenges: it is a notoriously complex, time-consuming, and expensive process that demands deep specialist knowledge to execute correctly.

A new approach promises to change this paradigm: Autonomous Valve CFD - an autonomous system designed specifically for valve flow simulation. But with any new technology, a critical question arises: Can an autonomous system deliver the accuracy and reliability of traditional methods?

This blog provides a formal validation of the performance and accuracy of the simulationHub Autonomous Valve CFD (AVC) software. The validation is conducted by systematically comparing simulation results generated by AVC against a range of established industry benchmarks. These benchmarks include data from experimental laboratory tests, and published academic research.

Credibility Built on Global Standards

The credibility of any simulation software rests on its adherence to industry-accepted standards and a transparent, repeatable computational process. Its methodology must be grounded in industry-wide standards to produce results that are trustworthy and comparable. This adherence to standards ensures users are speaking the same language as the rest of the industry.

Governing Industry Standards

The procedures for testing, simulation, and calculation within the AVC software are governed by a set of internationally recognized standards. The platform's methodology refers to and follows key international standards, including:

ANSI/ISA-75.01: Flow Equations for Sizing Control Valves

ANSI/ISA-75.02: Control Valve Capacity Test Procedures

IEC 60534-2-1: Flow capacity - Sizing equations for fluid flow under installed conditions

IEC 60534-2-3: Flow capacity - Test procedures

AWWA M49: Quarter-Turn Valves: Head Loss, Torque, and Cavitation Analysis

ASME V&V 20: Standard for Verification and Validation in Computational Fluid Dynamics and Heat Transfer

For the user, this foundation provides a crucial layer of trust. The data generated is not just numerically accurate—it is reliable, consistent, and meaningful within the broader engineering context.

Computational Fluid Dynamics (CFD) Simulation Process

The "autonomous" aspect of the AVC platform is more than a simple script. The system is designed to embed the best practices of a human CFD expert directly into its workflow.

Autonomous Meshing

The software automatically performs a series of sophisticated meshing operations, effectively mimicking an expert's decision-making process to accurately capture the complex flow physics inside a valve.

Volume mesh density boxes are automatically generated near the valve to resolve regions of flow separation

Surface mesh refinement is applied to precisely capture the geometric details of the disc and other valve feature edges

Prism layers are generated on all walls to capture the boundary layer phenomena

The system also includes automated mesh quality checks for critical parameters such as nonOrthoFaces, skewFaces, and wrongOrientedFaces. If the initial mesh does not meet predefined quality criteria, a corrective algorithm automatically refines low-quality regions to ensure all mesh elements meet stringent quality criteria before proceeding to the solver.

A special mention to Creative Fields Ltd cfMesh – AVC's meshing capabilities are powered by cfMesh, which ensures high-quality and accurate meshing for fluid dynamics simulations.

Solve

The solver and numerical schemes are configured with the industry-standard parameters to ensure reliable and consistent simulation outcomes. The flow conditions and pressure measurement locations are guided by the ANSI/ISA–75.02 standard:

Flow Conditions: Simulations adhere to a minimum valve Reynolds number of 10^5 and a maximum line velocity of 13.7 m/s (45 ft/s).
Pressure Measurement: In line with standard test procedures, pressure tapping points are located at 2D (two pipe diameters) upstream and 6D (six pipe diameters) downstream from the valve.
Inlet Condition: A parabolic velocity profile is applied at the inlet to simulate a fully developed flow, representing realistic pipe flow conditions.
Residual Monitors: Convergence is achieved when residuals drop below target thresholds: Pressure (1e-03); Velocity (U), Turbulent Kinetic Energy (k), and Turbulent Dissipation Rate (ε) (1e-04).
Additional Monitors: The simulation also monitors the stability of Pressure and Cv values to ensure a steady-state solution has been reached.
Quality Alert: A Y+ alert is triggered if more than 20% of the wall-adjacent cells fall within the buffer region of the boundary layer, signalling potential inaccuracies in wall shear prediction.

This structured methodology provides the foundation for the validation studies presented in the next section.

Validation Studies

The performance of the Autonomous Valve CFD software is systematically benchmarked against multiple independent data sources to quantify its accuracy and reliability. The results from a wide range of valve types and sizes are compared against physical flow loop test data and published academic findings.

AVC simulation data was benchmarked against experimental results obtained from two globally recognized facilities: the Utah Water Research Laboratory (UWRL) in the USA and the Fluid Control Research Institute (FCRI) in Kerala, India. The below graphs show a detailed validation of different butterfly valve types and sizes with data from different physical flow loop test sources.

Multiple such validation studies were conducted on a variety of common valve types, including Centric Butterfly, Double Offset Butterfly, Full Port Ball, Segmented Ball, and Globe Valves. As illustrated in the comprehensive data summary below, the simulation results demonstrate greater accuracy with the experimental benchmarks.

The vast majority of calculated data points fall within a tight +/- 5% deviation band, indicating a high degree of precision and consistency. This level of agreement across a wide range of valve types and valve size (Cv values spanning from less than 1 to over 300,000) demonstrates the robustness and scalability of the AVC’s autonomous simulation method.