Myth-busting Digital twins for infrastructure & industry: separating hype from reality

The global digital twin market reached $17.7 billion in 2024 and is forecast to explode to $259.32 billion by 2032, representing a compound annual growth rate exceeding 40%, according to Fortune Business Insights. Vendor marketing materials promise 25-30% energy savings, 35% reductions in maintenance costs, and payback periods under 18 months. Yet when infrastructure operators dig beneath the glossy case studies, they discover a more nuanced reality: while digital twins can deliver transformational value, the gap between vendor promises and implementation outcomes remains substantial. This article dissects the most persistent myths surrounding digital twins for infrastructure and industry, separating evidence-based capabilities from aspirational marketing.

Why It Matters

Infrastructure systems are responsible for approximately 79% of global greenhouse gas emissions when accounting for both operational energy and embodied carbon in construction materials. Buildings alone contribute nearly 40% of global CO2 emissions, with the built environment representing the largest single opportunity for decarbonisation. Digital twins offer a compelling proposition: real-time visibility into asset performance, predictive capabilities that prevent wasteful failures, and simulation environments that optimise operations before physical changes are made.

The adoption trajectory has accelerated significantly. A 2024 Hexagon survey found that 76% of executives are actively investing in digital twin technologies, up from approximately 50% in 2021. The manufacturing and energy sectors lead adoption, with process industries achieving the most mature implementations. Infrastructure operators face increasing regulatory pressure—the EU's Corporate Sustainability Reporting Directive (CSRD), the UK's Building Safety Act 2022, and emerging SEC climate disclosure requirements all demand the kind of granular operational data that digital twins can provide.

From a financial perspective, the business case appears compelling. AVEVA reports that power plant digital twin implementations achieve 28% reductions in fuel costs and 20% reductions in maintenance expenditure. Siemens claims 10-30% reductions in unplanned downtime for industrial customers. Yet these headline figures often come from ideal implementations with favourable conditions—sophisticated operators, comprehensive sensor coverage, and multi-year investment horizons that many organisations cannot replicate.

Key Concepts

Understanding digital twins requires distinguishing between marketing terminology and technical reality. Several foundational concepts are frequently conflated or misrepresented.

Digital Twin Maturity Spectrum: Not all digital twins are created equal. A static twin—essentially a 3D model enriched with asset metadata—provides visualisation value but limited operational insight. A connected twin integrates real-time sensor data to reflect current asset state. A predictive twin applies machine learning to forecast future conditions. A prescriptive twin recommends or automates interventions. Most infrastructure implementations today remain at static or connected levels, with genuinely predictive capabilities concentrated in controlled industrial environments.

IoT Sensor Integration: The value of any digital twin is fundamentally constrained by sensor coverage. A twin can only model what it can measure. Industrial facilities may have thousands of sensors per asset; older infrastructure often has minimal instrumentation. The cost of retrofitting comprehensive sensor networks frequently exceeds the digital twin platform investment itself.

Building Information Modelling (BIM) Foundation: BIM provides the geometric and attribute data foundation that digital twins extend into operations. ISO 19650 establishes information management standards, while Industry Foundation Classes (IFC) enable interoperability between design software. Digital twins consume BIM data but add the temporal dimension—transforming "as-designed" and "as-built" models into "as-operated" continuous monitoring.

Predictive Maintenance Algorithms: Machine learning models that predict asset failures require substantial training data—typically 2-3 years of operational history including failure events. Without sufficient failure data, predictive models cannot learn degradation patterns. This creates a chicken-and-egg problem: organisations implementing digital twins for predictive maintenance must often wait years before the predictive capabilities mature.

Simulation and Scenario Modelling: Digital twins enable "what-if" analysis—testing operational changes, design modifications, or extreme weather scenarios in a virtual environment. The accuracy of these simulations depends entirely on model fidelity and calibration against real-world performance data.

What's Working and What Isn't

What's Working

Predictive Maintenance in Process Industries: Digital twins have demonstrated clear ROI in environments with well-defined operating parameters, comprehensive sensor coverage, and sufficient historical data. Oil refineries, chemical plants, and power generation facilities report maintenance cost reductions of 15-25% and substantial reductions in unplanned downtime. These successes share common characteristics: mature sensor infrastructure, dedicated analytics teams, and multi-year commitment to model refinement.

Energy Performance Optimisation: Building energy modelling using calibrated digital twins can predict energy consumption within 10-15% accuracy—sufficient for meaningful optimisation. The City of Boston's building performance standards leverage digital twin approaches to identify retrofit opportunities across municipal portfolios. Commercial real estate operators report 10-20% energy savings from continuous commissioning enabled by digital twin insights.

Design Optimisation and Clash Detection: During design and construction phases, digital twins prevent costly on-site modifications. Bentley Systems reports that projects using their iTwin platform identify 60-80% of potential clashes before construction begins. This application succeeds because it addresses a well-understood problem with clear financial metrics.

Asset Lifecycle Management: Integrating operational data with asset registers enables better capital planning decisions. Organisations can model degradation curves, optimise replacement timing, and allocate maintenance resources more effectively. Water utilities using digital twins for distribution network management report reductions in non-revenue water losses of 5-15%.

What's Not Working

Data Quality and Integration Challenges: The most persistent barrier to digital twin value is data quality. Most organisations operate with fragmented data across multiple systems—SCADA, CMMS, BMS, and ERP platforms that were never designed for integration. The effort required to establish reliable data pipelines frequently exceeds platform implementation costs. Approximately 60-70% of digital twin project timelines are consumed by data preparation rather than analytics development.

Skills Gaps and Organisational Readiness: Successful digital twin implementations require cross-functional expertise spanning data engineering, domain knowledge, analytics, and change management. Most organisations lack this combination internally and struggle to retain specialised talent. Vendor promises of "turnkey" solutions understate the ongoing expertise required for model calibration, data quality monitoring, and insight operationalisation.

Integration Complexity and Vendor Lock-in: Despite industry rhetoric about open standards, most digital twin platforms remain effectively siloed. Data exported from one platform rarely transfers cleanly to another. Organisations with heterogeneous technology stacks face significant integration challenges that erode the promised benefits of unified visibility.

Unrealistic ROI Expectations: Vendor case studies typically showcase best-case implementations with favourable conditions. Organisations expecting similar results often discover that their data maturity, sensor coverage, or organisational readiness falls short. The 25-30% savings figures commonly cited represent exceptional outcomes, not typical results.

Key Players

Established Leaders

Siemens Digital Industries offers the Xcelerator portfolio integrating automation, industrial software, and digital twin capabilities. Their strength lies in connecting operational technology with information technology across manufacturing and process industries. The Siemensstadt Square development in Berlin demonstrates urban-scale digital twin implementation.

Bentley Systems focuses specifically on infrastructure engineering through the iTwin Platform. With roots in civil engineering software, Bentley excels at transportation, water, and utility infrastructure applications. Their partnership with Microsoft Azure provides cloud infrastructure for large-scale deployments.

GE Digital (now part of GE Vernova) specialises in industrial asset performance management through the Proficy suite and Predix platform. Their strength lies in rotating equipment monitoring for power generation, aviation, and oil and gas sectors.

Autodesk provides construction-focused digital twin capabilities through Tandem and Autodesk Construction Cloud. Their October 2025 partnership with Eaton introduced cloud-based energy twins targeting buildings and data centres.

NVIDIA Omniverse has emerged as a critical enabling platform, providing the real-time visualisation and simulation infrastructure that powers many digital twin implementations. Their Universal Scene Description (USD) format is becoming an industry standard for 3D data interchange.

Emerging Startups

Cityzenith (Chicago) has differentiated through urban-scale digital twins specifically targeting carbon emissions tracking. Their SmartWorldOS platform includes sustainability-focused applications deployed for India's planned capital city of Amravati.

Neara (Australia/US) creates digital twins for electrical utility infrastructure, enabling power companies to model network response to extreme weather events—increasingly critical for climate resilience planning.

Akselos (Switzerland) applies reduced-order modelling to create computationally efficient twins for large structures including offshore wind foundations.

Digital Twin KPI Benchmarks by Sector

Sector	KPI	Typical Baseline	Realistic Target	Best-in-Class
Commercial Buildings	Energy Use Intensity (kWh/m²/yr)	200-350	150-200	<100
Manufacturing	Overall Equipment Effectiveness (%)	55-65%	75-85%	>90%
Water Utilities	Non-Revenue Water (%)	20-40%	12-18%	<8%
Power Generation	Heat Rate Deviation (%)	3-6%	1.5-3%	<1%
Transportation Infrastructure	Unplanned Downtime (hrs/yr)	100-250	40-80	<25
Oil and Gas	Maintenance Cost per Barrel ($)	$4-8	$2.50-4	<$2

Myths vs Reality

Myth 1: Digital twins provide immediate ROI Reality: Meaningful returns typically require 18-36 months. Predictive maintenance value depends on accumulating sufficient failure data. Energy optimisation requires baseline establishment and model calibration. Organisations expecting payback within 12 months are frequently disappointed.

Myth 2: Digital twins eliminate the need for domain expertise Reality: Digital twins augment rather than replace human expertise. The most sophisticated platforms still require knowledgeable operators to interpret outputs, validate recommendations, and make judgment calls that algorithms cannot. Organisations that reduce operational staffing based on digital twin implementation often discover degraded rather than improved performance.

Myth 3: More data automatically means better insights Reality: Data quality matters more than data quantity. A digital twin fed by unreliable sensors or inconsistent data produces unreliable outputs—often with false precision that creates worse outcomes than no model at all. The principle of "garbage in, garbage out" applies forcefully to digital twin implementations.

Myth 4: Digital twins can accurately model Scope 3 emissions Reality: While twins excel at Scope 1 and 2 emissions from direct operations and purchased energy, Scope 3 supply chain emissions remain deeply problematic. Material carbon intensity data is inconsistent, supplier-specific information is often unavailable, and emission boundaries are inherently fuzzy. Platforms claiming precise Scope 3 accounting are overstating current capabilities.

Myth 5: Open standards ensure interoperability Reality: Despite progress on standards like IFC, CityGML, and USD, practical interoperability remains limited. Most implementations require significant custom integration work. Organisations should plan for vendor lock-in risk and budget for data transformation efforts when platforms change.

Myth 6: Digital twins work equally well for new and existing assets Reality: Digital twins for greenfield construction benefit from purpose-designed sensor networks and clean data foundations. Brownfield implementations face fundamental challenges: incomplete documentation, concealed systems, unknown degradation states, and retrofit sensor costs that often exceed platform investment.

Action Checklist

• Conduct a data maturity assessment before platform selection—identify all operational data sources, assess quality levels, and map integration requirements
• Define 2-3 specific use cases with measurable success criteria achievable within 18 months rather than pursuing enterprise-wide implementation
• Budget 40-60% of project timeline for data preparation, integration, and quality remediation
• Establish realistic baseline metrics before implementation to enable credible ROI measurement
• Plan for ongoing model calibration and validation—allocate permanent staff resources, not just implementation project budget
• Evaluate vendor lock-in risk explicitly—document data export capabilities, standard compliance, and exit strategies before procurement

FAQ

Q: What is the minimum sensor coverage required for effective digital twin implementation? A: This varies by use case. Energy monitoring requires sub-metering at major load centres (typically 15-25 sensors per building). Predictive maintenance for rotating equipment typically needs 8-12 sensors per asset covering vibration, temperature, pressure, and flow. Process industries may require 50-200 sensors per production line. The key principle is measuring the variables that drive the outcomes you are trying to predict or optimise.

Q: How should organisations prioritise digital twin investments when budgets are constrained? A: Start with use cases that address existing pain points with clear financial impact—typically unplanned downtime reduction or energy cost optimisation. Avoid "boil the ocean" approaches that attempt comprehensive digital twins before proving value. Prioritise assets with sufficient sensor coverage and accessible historical data over assets requiring major instrumentation investment.

Q: What data governance considerations should inform digital twin strategy? A: Define data ownership, access controls, and retention policies before implementation. Consider regulatory requirements for operational data storage, particularly in regulated industries. Establish protocols for model validation and auditability—especially if digital twin outputs will inform compliance reporting or safety decisions.

Q: How do digital twins integrate with existing enterprise systems? A: Most implementations require middleware or integration platforms to connect OT systems (SCADA, BMS, PLC) with IT systems (ERP, CMMS, analytics platforms). API-based integration is increasingly standard, but legacy systems often require custom adapters. Plan for ongoing integration maintenance as source systems evolve.

Q: What is the realistic accuracy of predictive maintenance enabled by digital twins? A: Mature implementations in controlled environments (rotating equipment in process industries) can predict failures 1-4 weeks in advance with 70-85% accuracy. Accuracy depends heavily on data quality, model calibration, and sufficient training data including failure events. Most organisations should expect 18-36 months before predictive models achieve reliable accuracy.

Sources

• Fortune Business Insights. "Digital Twin Market Size, Share & COVID-19 Impact Analysis, By Type, By Application, By End-Use Industry, and Regional Forecast, 2024-2032." Fortune Business Insights, January 2025.
• Hexagon. "Executive Survey: Digital Twin Adoption and Investment Trends 2024." Hexagon AB, 2024.
• AVEVA. "Reimagining Digital Insight for 2024: How AI and the Digital Twin Will Accelerate Industry 5.0." AVEVA Group plc, 2024.
• International Energy Agency. "Buildings Sector Energy Consumption and Emissions Tracking." IEA, 2024.
• Bentley Systems. "iTwin Platform: Infrastructure Digital Twins White Paper." Bentley Systems, Incorporated, 2024.
• McKinsey & Company. "Digital Twins: The Art of the Possible in Product Development and Beyond." McKinsey Digital, 2023.
• Siemens. "Digital Twin Technology: Bridging the Physical and Digital Worlds." Siemens AG, 2024.
• NVIDIA. "Omniverse Platform: Building the Industrial Metaverse." NVIDIA Corporation, 2024.