SUSTAINABLE GREEN CLOUD INFRASTRUCTURE FOR SMART MANUFACTURING USING CARBON-AWARE RESOURCE MANAGEMENT

Abstract

The rapid expansion of smart manufacturing has increased dependence on cloud computing, Industrial Internet of Things platforms, artificial intelligence services, digital twins, real-time analytics, and large-scale industrial data processing. Although these technologies improve production intelligence, flexibility, automation, and operational efficiency, they also increase computational energy consumption and the associated carbon footprint of manufacturing information infrastructure. Conventional cloud resource management primarily focuses on performance, cost, availability, and service-level objectives without adequately considering the temporal and geographical variation of electricity-grid carbon intensity. This paper proposes a sustainable green cloud infrastructure for smart manufacturing using carbon-aware resource management. The proposed framework integrates industrial workload monitoring, cloud resource telemetry, renewable energy availability, regional carbon-intensity information, workload classification, intelligent resource scheduling, containerized service orchestration, and sustainability-aware decision support. Manufacturing workloads are categorized according to urgency, latency sensitivity, computational demand, migration capability, and operational criticality. Real-time control and safety-critical workloads are maintained close to production systems, whereas delay-tolerant analytics, model training, digital twin synchronization, historical processing, and batch workloads are scheduled according to lower-carbon execution opportunities. The framework employs carbon-aware temporal shifting, geographical workload placement, energy-efficient resource consolidation, dynamic scaling, idle-resource reduction, and renewableenergy-aware scheduling while preserving manufacturing performance requirements. A sustainability monitoring layer continuously evaluates energy consumption, estimated carbon emissions, resource utilization, workload completion, and service-level compliance. The representative evaluation indicates that the proposed framework can reduce cloud energy consumption, decrease carbon emissions, improve renewable energy utilization, and maintain acceptable industrial application performance compared with conventional performance-only and cost-oriented scheduling approaches. The study demonstrates that carbonaware cloud management can become an important architectural component of sustainable smart manufacturing and provides a scalable foundation for integrating environmental objectives with cloud-native industrial computing.