The simulation-based analysis of business processes (BPs) is a key activity at various phases of the BP lifecycle, from the design phase, to predict the process behavior, down to the execution and improvement phases, to recover from possible performance downgrades and/or improve the process performance. The BP analysis is usually carried out taking as input the BP description in a given BP modeling language. This paper specifically addresses BPs described in BPMN (Business Process Model & Notation) and introduces an approach that exploits both model-driven principles and the DEVS (Discrete Event System Specification) formalism to first annotate the BPMN model with the allocation of task resources described in terms of performance and reliability properties and then transform the annotated BPMN model into a DEVS-based model, which can be eventually executed to get the analysis results of interest. The BPMN annotation is carried out by use of PyBPMN, a lightweight BPMN extension that allows business analysts to specify the allocation of task resources and their properties in terms of both time-related attributes and reliability attributes. The paper overviews the proposed approach and gives the details of the DEVS components that are used to model the behavior of the corresponding BPMN primitives.