Work

A closed-loop data-driven framework for value stream mapping and future-state design

Year: 2026
Type: article
Abstract: Value Stream Mapping (VSM) is widely used in Lean Manufacturing to identify waste and support continuous improvement, but conventional VSM relies on manual data collection, static process representations, and expert interpretation. This paper proposes a data-driven VSM/VSD decision-support framework that establishes a traceable path from operational event data to current-state reconstruction, process diagnosis, simulation-based evaluation, and future-state redesign. Process Mining reconstructs observed process behavior and supports quantitative current-state VSM generation; Complex Network Analysis characterizes structural dependencies; Explainable Machine Learning integrates operational and network-derived information to identify and interpret process criticalities; and Digital Twin simulation evaluates candidate improvement interventions before they are translated into Future-State Value Stream Designs. The framework was evaluated using a controlled synthetic manufacturing event log and three heterogeneous real-world event logs under a leakage-controlled temporal protocol. Network-derived features provided complementary predictive information beyond conventional operational variables, with the strongest gains observed for rework prediction: Random Forest F1-score increased from 0.2243 to 1.0000 for BPI2017 and from 0.4658 to 0.7451 for BPI2019. In the cumulative simulation-based evaluation, the complete framework reduced Lead Time by 32.40%, Work-In-Process by 45.09%, Defect Rate by 27.63%, and Rework Rate by 45.52%, while increasing Throughput by 71.73% relative to the baseline. The results support a reproducible transition from descriptive VSM toward evidence-based, quantitatively evaluated Future-State Value Stream Design.
Language: English
Open Access status: green
OpenAlex ID: W7212773969
FWCI:
Cites: 19
Cited by:
Related to: 10