Assessing Digital Twin Feasibility Before Building

Frame the study around a decision

Identify the decision that the twin should improve, its owner, action window, required evidence, and the cost of an incorrect recommendation. Then define the smallest digital representation needed to support that choice. This prevents the project from becoming an attempt to reproduce every component and signal.

Clarify whether the intended twin is descriptive, diagnostic, predictive, prescriptive, or connected to a governed control process. Each level requires stronger evidence, validation, integration, and oversight.

• Name the decision owner
• Set the physical and operational boundary
• Define the required maturity
• Record external dependencies

Establish the asset and data baseline

Inspect asset identifiers, hierarchies, sensors, calibration, sampling, units, timestamps, historians, maintenance events, operating modes, engineering records, and interfaces. A large data volume can still be unsuitable when important conditions are absent or source meanings disagree.

Asset identity deserves deliberate governance. Names that differ across a historian, maintenance system, and engineering model need a controlled mapping so events are not assigned to the wrong component.

• Profile missing and corrected values
• Check time synchronization
• Record maintenance and configuration changes
• Name the authoritative source for each state

Choose modelling and synchronization deliberately

Physics, statistics, machine learning, rules, discrete-event simulation, geometry, or a hybrid may be appropriate. Method selection should follow the decision and the available evidence. Document the model assumptions and the conditions in which they remain valid.

Define which signals update the twin, how often updates are needed, how late events are handled, how state is reconstructed after interruption, and what confidence is attached to estimated data. Real-time synchronization is useful only when the actual decision window requires it.

• Select fidelity by decision need
• Make uncertainty visible
• Define out-of-order event handling
• Test reconstruction after interruption

Prove validity and operating value

Use historical replay, controlled trials, expert review, sensitivity analysis, and accepted engineering calculations where suitable. Examine start-up, shutdown, maintenance, rare conditions, sensor failure, and equipment changes instead of relying on averages.

Integration feasibility includes the surrounding workflow. A recommendation has little value if it arrives after planning closes, uses the wrong asset identity, or cannot be explained to the person expected to act. Continuing data, calibration, platform, security, licence, and training costs belong in the assessment.

• Document the valid operating envelope
• Test failure as well as normal use
• Connect outputs to an actionable workflow
• Treat unverified benefits as pilot hypotheses

Digital twin feasibility review

• Name the decision and owner
• Define the physical boundary
• Choose the required twin maturity
• Reconcile asset identities
• Profile relevant data
• Select a justified model
• Match synchronization to the action window
• Specify the valid operating envelope
• Confirm workflow integration
• Include continuing maintenance cost

Digital twin modelling and validation references

Use these digital twin and manufacturing resources to examine model scope, synchronization, validation, and operating ownership for the selected asset or process.

1. ISO 23247 Digital Twin Framework for ManufacturingInternational Organization for Standardization. Standards family entry.
2. Digital Twin ConsortiumObject Management Group. Industry reference resources.
3. Digital Twins for Advanced ManufacturingNational Institute of Standards and Technology. Research programme resource.