Time granularity impact on propagation of disruptions in a system-of-systems simulation of infrastructure and business networks

TL;DR

This paper investigates how time granularity influences disruption propagation in system-of-systems simulations of infrastructure and business networks, proposing a model to select suitable granularity based on recovery time.

Contribution

It introduces a novel analysis of time granularity effects on disruption propagation and provides a model for choosing appropriate simulation time steps based on recovery time.

Findings

Time granularity significantly affects disruption propagation.

Recovery time is a critical factor influencing simulation accuracy.

Time granularity should be less than 1.13 times the expected recovery time.

Abstract

System-of-systems (SoS) approach is often used for simulating disruptions to business and infrastructure system networks allowing for integration of several models into one simulation. However, the integration is frequently challenging as each system is designed individually with different characteristics, such as time granularity. Understanding the impact of time granularity on propagation of disruptions between businesses and infrastructure systems and finding the appropriate granularity for the SoS simulation remain as major challenges. To tackle these, we explore how time granularity, recovery time, and disruption size affect the propagation of disruptions between constituent systems of an SoS simulation. To address this issue, we developed a High Level Architecture (HLA) simulation of 3 networks and performed a series of simulation experiments. Our results revealed that time granularity and especially recovery time have huge impact on propagation of disruptions. Consequently, we developed a model for selecting an appropriate time granularity for an SoS simulation based on expected recovery time. Our simulation experiments show that time granularity should be less than 1.13 of expected recovery time. We identified some areas for future research centered around extending the experimental factors space.