Digital Twins for Operational Decision-Making: Integrating Simulation and Management Analytics

Abstract

Digital twins connect operational data, computational representations and managerial decisions

through a continuously updated feedback loop. This study develops an integrated decision

architecture combining discrete-event simulation, short-horizon predictive analytics, constrained

optimization and human governance. A hypothetical multi-stage production and fulfillment

system illustrates how the architecture can compare dispatching, maintenance and staffing

policies under uncertainty. The method specifies state synchronization, prediction, scenario

generation, optimization, implementation and audit stages, with explicit performance and

feasibility measures. In a fully disclosed synthetic example, the integrated policy reduces average

lead time from 18.4 to 13.2 hours and total modeled cost from 148.0 to 128.4 thousand currency

units per month relative to a static baseline. These values are generated assumptions and

calculations, not observations from an industrial deployment. Sensitivity analysis identifies

model drift, data latency and intervention cost as critical limitations. The contribution is a

transparent decision framework and an evaluable research protocol, rather than an empirical

claim of demonstrated industrial performance.