Lightweight Blockchain Data Model for Provenance Tracking in Additive-Manufacturing Supply Chains: A Quantitative Performance Analysis

Provenance in additive manufacturing (AM) supply chains requires tamper-evident records without severely degrading ledger performance. We present a lightweight hybrid data model that anchors only an average of 270 B of metadata comprising event type, stakeholder MSP ID, timestamp, and a SHA-256 digest on Hyperledger Fabric, while persisting the complete artefact in a MinIO S3-compatible object store addressed through a long-lived presigned URL. A Docker-Compose test-bed (Fabric v3.1.0, four organizations, two peers for each organization, one Raft orderer) demonstrates the scalability limits of on-chain payloads and the gains of decoupling data and provenance. Under an intense write workload, the lightweight scheme sustains a median end-to-end latency of 47 ms and approximately 471 transactions per second (TPS), while peer CPU remains below 35 % and memory remains stable. Embedding the raw artefact instead increases median latency by over an order of magnitude and reduces throughput to below 10 TPS at a 1 MB payload, with larger transactions unable to commit due to Fabric’s 2 MB block limit. By reducing the AM lifecycle’s on-chain footprint from 7.23 MB to 1.6 kB ( > 4,000×), the study shows that separating bulk artefacts from their cryptographic proofs is a prerequisite for maintaining low latency, high throughput, and manageable ledger growth in cross-organisational provenance systems.