Digital twins: a faithful model to decide faster
A process digital twin is a mathematical model linked to real plant data, used to understand, predict and optimise operation. The ISO 23247 series of standards describes its reference framework for manufacturing.
Three maturity levels
- Offline model: steady-state simulation to size equipment, compare variants and establish mass and energy balances.
- Calibrated dynamic model: reproduces transients (start-ups, load changes) and trains operators on a simulator.
- Connected twin: continuously recalibrated on measurements, it feeds drift detection, forecasting and real-time optimisation.
High-value uses
- Design and revamping: test scenarios, including integrating a heat pump or a fuel switch, before committing to works.
- Training and safety: replay incidents and validate start-up procedures without risk.
- Energy control: feed predictive control and heat recovery calculations.
- Diagnosis: compare measured and expected behaviour to spot fouling, leaks or performance loss.
Architecture, limits and success conditions
On architecture, separate the data layer (historian, OPC UA), the model layer and the presentation layer (dashboards, operator advice). This separation avoids dependence on a single vendor and eases cybersecurity audits.
A twin is only valuable if it stays faithful. That requires periodic recalibration, enough instrumentation and a model owner, as for any equipment. The hidden cost is maintenance: physical properties, fouling state and raw-material quality all evolve. Hybrid approaches combining physics and industrial AI reduce calibration effort. To start, a limited scope, one unit or one utility, delivers a first measurable return within months before extension to the whole plant.
Indicators to judge a project
A twin is justified by measurable results: gap between simulated and measured values, reduction in specific consumption, stops avoided, engineering hours saved. Set these indicators before the project, then review them each quarter with operations and maintenance.
Key takeaway: a twin is a living asset, not a one-off study. Budget for its upkeep from the beginning, assign ownership, and extend its scope only when the first use case has demonstrably paid back.
Going further
- Industrial AI: use cases and governance
Predictive maintenance, soft sensors, set-point optimisation and operator assistants: deploying AI in plants with a sound method.
- Model predictive control (MPC) of energy processes
Principle of model predictive control, energy gains, prerequisites, cybersecurity and implementation on utilities and processes.
- Pinch analysis and heat integration
Pinch method: composite curves, ΔTmin, golden rules and retrofit to reduce a process’s heating and cooling consumption.
- Process engineering
Design, optimisation and operation of industrial processes: intensification, electrification, hydrogen, CO₂ capture and digital twins.
- Low-carbon hydrogen for industry
Alkaline, PEM and SOEC electrolysis, RED III framework, costs and integration: when low-carbon hydrogen makes sense for an industrial site.
- Carbon capture and storage (CCUS)
Amine capture, transport, geological storage and the European framework: when CCUS is justified for industrial process emissions.
- Energy storage for industry
LFP and sodium-ion batteries, flow batteries, thermal and hydrogen: choosing storage by power, duration and efficiency.
- Emerging technologies for business
Industrial AI, digital twins, hydrogen, carbon capture and storage: technologies to assess for competitiveness and sustainability.