Operational Data Foundation Framework for Smart Manufacturing in SMEs: Field Implementation and Evaluation

Smart manufacturing depends on operational data that remain continuous, interpretable, and reusable in practice. In constrained small and medium-sized enterprise (SME) factories, however, the main bottleneck often lies not in later-stage analytics or AI applications, but in securing an operationally viable data foundation under real deployment conditions. A lifecycle-based analysis of smart manufacturing data pipelines, together with recurrent SME deployment constraints identified in prior studies, led this study to derive six recurring operational risks. On that basis, the study proposes an Operational Data Foundation Framework structured around core requirements of continuity, governance, diagnosability, operability, reprocessability, and evolvability. These requirements are further articulated through design principles and assessable operational invariants. The framework was instantiated in a real SME factory, where heterogeneous field sources were integrated into a coherent operational data foundation for smart manufacturing through constrained communication paths, durable edge-side capture, cloud-side stream processing, controlled data normalization, and monitoring and alerting functions. Requirement-based evidence from the field implementation showed that the system preserved stable semantics across the pipeline, made failures traceable to specific lifecycle segments, preserved historical records for later reprocessing, and remained manageable under constrained deployment conditions. A representative field case further demonstrated the framework's practical value: severe communication instability was diagnosed through lifecycle-segment discrepancy analysis and improved from approximately 33% to 95% packet reception after targeted intervention. The study contributes a field-grounded and assessable design logic for making smart manufacturing practically achievable in constrained SME factories.