🎓 Lesson 21
D5
Digital Twin Applications in Decommissioning Simulation & Monitoring
A digital twin is a virtual copy of a real decommissioning site that updates in real time using sensors and data, helping engineers test and monitor cleanup safely and efficiently.
🎯 Learning Objectives
- ✓ Explain how sensor-to-model synchronization enables predictive monitoring of soil stabilization during restoration
- ✓ Design a minimal viable digital twin architecture for an onshore wind farm decommissioning project
- ✓ Analyze time-series strain and settlement data from embedded sensors to validate model fidelity against field measurements
- ✓ Apply ISO 23247-1:2021 compliance criteria to evaluate digital twin readiness for regulatory reporting
📖 Why This Matters
Decommissioning renewable energy infrastructure—especially offshore wind foundations or legacy mining tailings facilities—carries high uncertainty in subsurface behavior, long-term stability, and ecological recovery. Traditional 'observe-and-react' methods risk cost overruns, regulatory noncompliance, and environmental harm. Digital twins shift the paradigm to 'simulate-before-act': enabling engineers to virtually rehearse demolition sequences, forecast erosion pathways under storm scenarios, and verify restoration milestones before ground disturbance begins—reducing rework by up to 35% (IEA 2023) and accelerating permit approval cycles.
📘 Core Principles
Digital twin maturity follows four progressive levels: (1) Descriptive (static 3D model + as-built data), (2) Informative (real-time telemetry integration), (3) Predictive (physics-based modeling + ML-driven forecasting), and (4) Prescriptive (autonomous decision triggers, e.g., adaptive dewatering control). In decommissioning, the twin must co-simulate coupled domains: structural integrity decay (concrete carbonation, steel corrosion), geotechnical response (pore pressure evolution, slope stability), hydrological flux (runoff, leachate migration), and ecological succession (vegetation regrowth rates). Critical enablers include IoT sensor networks (strain gauges, piezometers, GNSS beacons), open BIM/IFC interoperability, and ISO/IEC/IEEE 23247-1:2021–compliant data governance frameworks.
📐 Model Fidelity Index (MFI)
The Model Fidelity Index quantifies alignment between digital twin predictions and field measurements over time, serving as a KPI for twin reliability. It is calculated per monitored parameter (e.g., settlement, pH, turbidity) and aggregated for system-level trust assessment.
Model Fidelity Index (MFI)
MFI = 1 − (RMS_error / observed_range)Quantifies agreement between digital twin predictions and field measurements; used to assess twin readiness for operational decision-making.
Variables:
| Symbol | Name | Unit | Description |
|---|---|---|---|
| RMS_error | Root Mean Square prediction error | mm | Statistical measure of average deviation across all monitored points |
| observed_range | Range of observed field measurements | mm | Difference between maximum and minimum measured values in the dataset |
Typical Ranges:
Regulatory reporting twin: 0.85 – 0.95
Early-stage feasibility twin: 0.60 – 0.75
💡 Worked Example
Problem: Over 90 days, a digital twin predicted cumulative settlement at three sensor locations: A (2.1 mm), B (3.8 mm), C (1.9 mm). Observed values were A (2.3 mm), B (3.6 mm), C (2.0 mm). Calculate MFI using RMS error normalized to observed range.
1.
Step 1: Compute RMS error = √[(0.2)² + (−0.2)² + (0.1)²] / 3 = √[0.04 + 0.04 + 0.01]/3 = √0.09/3 = √0.03 ≈ 0.173 mm
2.
Step 2: Determine observed range = max(2.3, 3.6, 2.0) − min(2.3, 3.6, 2.0) = 3.6 − 2.0 = 1.6 mm
3.
Step 3: Apply MFI = 1 − (RMS error / observed range) = 1 − (0.173 / 1.6) = 1 − 0.108 = 0.892
Answer:
The result is 0.892, which falls within the acceptable fidelity range of ≥0.85 for regulatory-grade twin validation per ISO/IEC/IEEE 23247-2:2022 Annex D.
🏗️ Real-World Application
The Ørsted Hornsea Two Offshore Wind Farm (UK) deployed a Level 3 digital twin for turbine monopile removal and seabed restoration. Using real-time sonar mapping, corrosion sensors embedded in pile sleeves, and coupled hydrodynamic–sediment transport models, the twin simulated scour evolution under 100-year storm scenarios. When predicted localized sediment loss exceeded 0.8 m (threshold), the twin automatically triggered revised backfill scheduling—preventing unplanned habitat disruption and avoiding £2.4M in potential remediation penalties. Data was audited monthly against UK Marine Management Organisation (MMO) reporting requirements.