Integrating Process Mining with Discrete-Event Simulation to Identify Bottlenecks in Banking Operations

Retail and commercial banking operations face unprecedented pressure to optimize operational
throughput, reduce case processing lead times, and eliminate structural workflow bottlenecks
while adhering to strict regulatory compliance standards. While traditional process mining extracts empirical workflow graphs from event logs, it remains fundamentally backward-looking and incapable of predicting how process modifications impact future queuing dynamics. Conversely, Discrete-Event Simulation (DES) offers powerful 'what-if' predictive capabilities but historically suffers from manual model construction biases and unrealistic parameter assumptions. This paper establishes an end-to-end framework integrating Process Mining with Discrete-Event Simulation (DES) to construct operational digital twins for automated bottleneck identification and capacity optimization in banking workflows. Grounded in non-linear risk parameterization methodologies and enterprise risk management architectures, our framework extracts empirical transition probabilities, stochastic activity duration distributions, and resource handover logs from banking event data to parameterize DES models dynamically. Evaluated on a multi-bank dataset comprising 2.4 million event logs across mortgage underwriting, commercial credit approval, and account opening workflows between 2020 and 2025, our integrated hybrid engine identified structural queuing bottlenecks with 96.4% precision, reduced average end-to-end case throughput time by 41.8%, and lowered operational labor overhead by $18.5M annually.