DESIGN
Bridging mechanical design and industrial AI
How valve manufacturers can prepare ball valves for evidence-based maintenance, digital twins and future industrial-AI workflows.
By Zahra Farrokhi, Mechanical Engineer, Batu Valve Türkiye
Figure 1. Ball-valve assemblies manufactured by Batu Valve Türkiye and fitted with electric actuators. Such configurations illustrate the mechanical foundation through which available actuator and plant data may support future condition assessment.
The industrial opportunity is not to put artificial intelligence( AI) inside every valve. It is to preserve the mechanical evidence, operating context and lifecycle baselines that allow selected critical ball valves to participate credibly in modern maintenance workflows.
Plants lack valve evidence
Plants already record pressure, flow, temperature, commands and valve position. Yet these signals do not automatically identify internal leakage, distinguish valve friction from actuator or coupling problems, or prove that an infrequently operated isolation valve will move on demand. Maintenance requires a decision on a named asset, not a generic prediction. The economic use case is therefore a selected population of critical actuated valves, for which design data, actuator history, process context and verified baselines remain connected [ 17, 23 ].
What owners are adopting now
Asset Performance Management is already being deployed across operating assets, including an Aker BP programme covering five North Sea assets [ 41 ]. Digital twins support monitoring, scenarios and maintenance, although data integration, validation, scale and workforce readiness remain major barriers [ 42-44 ]. Selective autonomous control has reached full-scale refinery and gas-treatment applications [ 45, 46 ], but this does not justify autonomy for isolation or safety functions. For valves, the nearer-term opportunity is time-series detection, digital-twin context and retrieval-grounded maintenance assistance with human approval.
Link each failure mode to defensible evidence
A credible use case begins with a failure mode and a decision. An indicator may justify investigation without proving root cause: increasing torque can reflect seat friction, packing load, corrosion, misalignment, differential pressure or actuator condition; partial stroke can demonstrate movement but not seat tightness. The engineering task is to state the evidence, its confounders and the confirmation required before action.
Make actuator data mechanically trustworthy
Torque, position, and stroke time are only valuable if the mechanical load path is repeatable. ISO 5211:2026 standardises partturn actuator attachments, while ISO 5640:2024 addresses mounting kits [ 3, 4 ]. The valve manufacturer should additionally control bracket stiffness, alignment, coupling fit, backlash, fasteners and shaft datum. Otherwise, a change in the recorded signature may originate in the adapter rather than the valve. ISA75.26
40 Valve World November 2026 www. valve-world. net