CarAT: Carbon Atom Tracing across Industrial Chemical Value Chains via Chemistry Language Models

2025-08-13

AI Summary

Industrial chemical value chains face challenges in dynamically, transparently, and auditably tracking biogenic carbon content (BCC) across complex material flows. Method: This paper proposes the first scalable carbon flow modeling framework integrating chemical language models (CLMs) with linear programming. The CLM enables reaction-level atom mapping, while ERP system data and linear programming jointly enable carbon-atom-resolved tracing and dynamic BCC quantification. Contribution/Results: Unlike conventional life cycle assessment, the framework supports real-time, scenario-aware, and verifiable carbon attribute flow analysis. Applied to a 27-node toluene diisocyanate (TDI) value chain, it successfully quantifies BCC variations under diverse feedstock blends and delivers interpretable, auditable results via Sankey diagram visualization. The approach provides high-precision, automated, and audit-ready technical support for sustainability reporting and carbon neutrality planning in the chemical industry.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Sustainability and carbon-aware systems for Web, mobile, and WoT
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

The chemical industry is increasingly prioritising sustainability, with a focus on reducing carbon footprints to achieve net zero. By 2026, the Together for Sustainability (TfS) consortium will require reporting of biogenic carbon content (BCC) in chemical products, posing a challenge as BCC depends on feedstocks, value chain configuration, and process-specific variables. While carbon-14 isotope analysis can measure BCC, it is impractical for continuous industrial monitoring. This work presents CarAT (Carbon Atom Tracker), an automated methodology for calculating BCC across industrial value chains, enabling dynamic and accurate sustainability reporting. The approach leverages existing Enterprise Resource Planning data in three stages: (1) preparing value chain data, (2) performing atom mapping in chemical reactions using chemistry language models, and (3) applying a linear program to calculate BCC given known inlet compositions. The methodology is validated on a 27-node industrial toluene diisocyanate value chain. Three scenarios are analysed: a base case with fossil feedstocks, a case incorporating a renewable feedstock, and a butanediol value chain with a recycle stream. Results are visualised with Sankey diagrams showing the flow of carbon attributes across the value chain. The key contribution is a scalable, automated method for real-time BCC calculation under changing industrial conditions. CarAT supports compliance with upcoming reporting mandates and advances carbon neutrality goals by enabling systematic fossil-to-biogenic substitution. Through transparent, auditable tracking of carbon sources in production networks, it empowers data-driven decisions to accelerate the transition to sustainable manufacturing.

Problem

• Automated tracking of biogenic carbon content in chemical value chains
• Dynamic sustainability reporting for industrial carbon footprints
• Real-time compliance with upcoming biogenic carbon reporting mandates

Innovation

• Uses chemistry language models for atom mapping
• Leverages ERP data for dynamic BCC calculation
• Applies linear programming to track carbon flow