AI and machine learning offer new framework for managing urban plastic waste

Apr 09, 2026

A new AI-enhanced framework combines machine learning validation with life-cycle assessment to cut plastic waste emissions by up to 96.3% in urban settings.

(Nanowerk News) A new evaluation framework uses artificial intelligence to assess and optimize how cities manage plastic waste, targeting both carbon emissions and cost-effectiveness. Published in Engineering (“AI-Enhanced Assessment Framework for City-Scale Management of Municipal Living Plastic Waste Towards Zero-Waste Cities”), the study addresses a persistent obstacle in urban waste management: incomplete data that limits how well models transfer between cities. By integrating machine learning for data validation and gap-filling, the framework provides quantitative guidance for municipalities pursuing zero-waste goals.

Key Findings

Managing plastic waste at the city scale requires simultaneous optimization across collection, recycling, and treatment. Waste composition, population density, economic activity, and available disposal pathways — incineration, landfilling, and recycling — all shape both life-cycle carbon emissions and processing costs. These variables interact in ways that make evaluation difficult, particularly when measurement inconsistencies and missing data undermine model reliability. The researchers built their baseline material flows from systematic field measurements and differential scanning calorimetry characterization. To correct for potential biases, they applied an artificial neural network model that draws on multiple independent data sources. This model performs cross-validation, missing-data imputation, and explicit uncertainty propagation, strengthening the credibility of the environmental and economic assessments that follow. The framework was then applied to evaluate several intervention scenarios. Source reduction and substitution with bio-based materials deliver meaningful emissions cuts in the near to mid term. High-quality recycling pathways, however, produce the largest single mitigation effect. When all strategies are combined, annual greenhouse gas emissions fall by 96.3% relative to baseline levels by 2060. Economic and environmental performance of municipal living plastic waste management. (Image: Reproduced from DOI:10.1016/j.eng.2026.03.009, CC BY) The economic analysis reveals important trade-offs between recycling technologies. Mechanical recycling holds a clear advantage over chemical recycling for near-term deployment, owing to lower costs and greater process maturity. It produces an emission intensity of approximately 108 kg CO2-eq per tonne while generating economic returns of approximately 613.9 CNY per tonne. Across the full modeled trajectory, the optimal pathway accumulates emission reductions of 22.22 Mt CO2-eq alongside economic benefits of approximately 197.7 billion CNY. The authors recommend that policymakers treat source reduction and design for circularity as long-term structural constraints rather than secondary measures. Relying too heavily on high recycling rates alone risks weakening long-term climate performance. Near-term priorities should center on expanding mechanical recycling infrastructure, while chemical recycling advances through targeted demonstration projects. Beyond individual city applications, the framework offers a replicable methodology for urban planners and researchers operating under data-constrained conditions. Its combination of field-validated baselines, machine learning verification, and scenario modeling provides a structured basis for policy prioritization, facility siting, and budget allocation in zero-waste city initiatives.