Blockchain-anchored privacy protection and smart contract design for engineering cost management education data

Abstract Engineering cost management education produces a peculiar hybrid: personal identifiers and academic records sit alongside pricing logic drawn from real bidding archives, so a leak crosses into commercial intelligence. Conventional role-based access control, designed for uniform academic records, does not cope well with this mixture. We describe a blockchain-anchored privacy framework built for the niche, pairing a three-layer main-chain / sidechain / subchain architecture with attribute-based smart contracts and lifecycle handling that spans generation, classification, sharing, and revocation. Data items are graded into four sensitivity tiers, encrypted under ciphertext-policy attribute-based encryption, dispersed across IPFS, and anchored on-chain through four chaincodes for identity, access, sharing, and audit; a Schnorr-style zero-knowledge construction gates cross-institutional disclosure of statements over committed values, with a Bulletproofs upgrade path noted for range assertions. Evaluation on a Hyperledger Fabric 2.5 testbed, benchmarked against same-environment single-chain Fabric and hardened role-based baselines rather than a deprecated public testnet, records around 870 TPS with 218 ms end-to-end latency in the twelve-peer configuration, reports Fabric-native execution costs per chaincode in place of the earlier gas-equivalent mapping, and includes a node-count sweep that traces the O(N²) growth of PBFT communication as the main-chain committee scales. A red-team campaign with an explicit threat model and raw event counts holds privacy leakage below 0.4% under replay, Sybil, and collusion attacks. A two-semester feasibility deployment at one applied university (142 undergraduates, 11 instructors, 4 mentors) reports associational rather than causal gains in sharing latency, traceability completeness, and student-reported data autonomy, with dispersion and confidence intervals attached to every indicator. Limitations in sample scope, attribute-authority centralisation, cross-chain atomicity, and long-horizon governance are stated openly, and future directions toward multi-authority key issuance, formal contract verification, range-proof upgrades, and post-quantum primitives are outlined.