Digital Twins for Biofuel Integration and Optimization in Internal Combustion Engines: A Critical Review and Research Roadmap

Digital twins (DTs) are increasingly proposed as a bridge between high-fidelity engine physics, real-time sensing, artificial intelligence (AI), and closed-loop optimization, yet their role in biofuel-powered internal combustion engines (ICEs) remains fragmented. This critical review evaluates how condition monitoring, AI-based diagnostics, fast-running engine models, and DT implementations can be integrated into a fuel-aware framework for biofuel ICEs. A structured search of original research was conducted across major publisher databases and scholarly discovery services, including ScienceDirect, Springer Nature, Wiley/IET, SAE Mobilus, RSC Publishing and MDPI, with the search updated to 19 August 2026. Review-only papers were excluded from the evidential corpus; the synthesis therefore relies on primary experimental, computational, or validated digital-twin studies. Three contributions distinguish this review. First, a five-layer taxonomy is proposed that separates sensing, state estimation, twin-model fidelity, intelligence, and actuation, allowing DT maturity to be compared independently of terminology. Second, conventional-fuel and biofuel deployments are compared through a fuel-to-fault pathway linking viscosity, volatility, oxygen content, ignition quality, oxidation stability and material compatibility to measurable signatures, degradations and control actions. Third, a BioTwin-ICE architecture is proposed in which a physics core is continuously corrected by data-driven residual models and uncertainty-aware parameter adaptation. The literature shows that engine DTs have progressed from real-time 0D/1D virtual engines to hybrid combustion prediction, fault isolation and optimization, whereas direct biofuel-specific DT demonstrations remain limited. The principal research opportunity is therefore not another generic engine surrogate, but validated fuel-aware twins that remain accurate under feedstock variation, ageing, transient operation and component degradation while satisfying real-time and safety constraints. Actionable research questions and validation metrics are provided to guide this transition.