Network Modeling of the U.S. Petrochemical Industry under Raw Material and Hurricane Harvey Disruptions

Abstract

A geographically resolved network model of the U.S. chemical industry in 2017 is constructed, and optimal chemical flows between units are calculated using linear programming. A baseline solution and three disruption scenarios (primary raw material disruptions, reported Hurricane Harvey ethylene cracker disruptions, and assumed capacity disruptions based on the Hurricane Harvey geographic storm track) are studied to determine how the structure of the industry is modified to adapt to widespread and geographically specific disruptions. The calculated impacts of the assumed Hurricane Harvey disruption include 170 chemical units in 26 states that change production level as a result of supply chain disruptions during the storm. The systemic impact for the assumed Hurricane Harvey disruption is 19.3 million tonnes of gross chemical production. The day with the largest impact on gross chemical production shows a reduction from baseline operations of 1.3 million tonnes (42% of baseline). This model can be used for analysis of future disruption scenarios and to test resilience strategies, including impacts of new manufacturing configurations or technologies.

Geospatial Network Approach for Assessing Economic Potential of Ethylene-to-Fuel Technology in the Marcellus Shale Region

“…This model is quite detailed and incorporates production and transportation costs on national, regional, or statewide scales. DeRosa et al 14 developed a geospatially resolved LP model of the entire U.S. refining and petrochemical industry, based on the 2017 ICIS Supply and Demand Database, 45 to study the impacts on the industry of disruptions caused by Hurricane Harvey. In this case, a cost-based objective function was not used, the goal instead being to maximize production in the face of disruptions.…”

“…Following DeRosa et al, 14 the locations of existing manufacturing processes in the United States, as well as the main product and capacity of each plant, were obtained from the 2017 ICIS Supply and Demand Database. 45 Data on the manufacturing process costs (adjusted for inflation 46 ) and the stoichiometries of the technologies used were obtained from the IHS 2012 Process Economic Program Yearbook.…”

“…Model Formulation. As noted above, the industry network model is a modification of the directed graph model of DeRosa et al 14 The formulation of this model, as modified, is summarized here. The set of all nodes is denoted as N. The set of supply nodes is denoted by S, the set of existing chemical manufacturing unit nodes by U, the set of sink nodes by K, and the set of potential plant nodes by Z.…”

“…The capacity constraint on manufacturing unit nodes requires that the production of the main product in an existing unit node n ∈ U not exceed a specified unit capacity (14) and, as discussed above, for a new manufacturing node n ∈ Z, the capacity constraint is…”

“…Motivated by this, we have developed a spatially resolved network model of the refining and chemical manufacturing industry to investigate the potential adoption and impact of an ethylene to gasoline technology. Based on the model formulated by DeRosa et al., the spatially resolved network model is a mixed-integer linear program that aims to minimize total industry production costs and transportation costs. We use the model to address the following key questions: What is the net production cost threshold for market penetration of an ethylene to gasoline oligomerization process in the Marcellus shale region?…”