Risk-based Industrial Site Optimization

A key consideration in the design of a new industrial site is the spatial arrangement of units with respect to required safety distances. Safety distance requirements are intended to reduce the risk of hazardous events escalating from their origin to other units as well as to limit the damage to particularly vulnerable targets (e.g., control rooms), and are typically based on industrial standards (e.g., EIGA, CGA), which rely on credible worstcase scenarios, acceptable escalation frequencies or fixed values - the latter often without clear justification. This paper proposes an optimization algorithm that aims to minimize total site cost by balancing land use and interconnection costs (e.g., for pipes and cables), while ensuring compliance with required safety constraints. For each planned unit, the risk of escalation is represented using an F-D curve - a function analogous to the traditional F-N curve, but with escalation distance (D) on the x -axis. This is provided for each pair of unit types. The escalation distance is the distance at which a secondary hazardous event occurs for a given scenario, e.g., spread of fire from one tank to another. Due to the high degrees of freedom, the problem is numerically challenging since it has many local optimization minima. To address this challenge, optimization is performed using the bestlbin differential evolution algorithm which is specifically tailored to address non-monotonic problems. The resulting site layouts can be visualized in a 2D scatter plot, where each circle represents a unit and the extent of the scatter is the required lot size. The algorithm can be used both to generate early-stage input to concept development of siting projects and for theoretical investigations into optimal clustering, central placement, and proximity to lot boundaries for different unit types.

Keywords: Safety distance, Optimization, Process equipment, Domino effects, Facility siting.