A superstructure‐based mixed‐integer programming approach to optimal design of pipeline network for large‐scale CO₂ transport

Abstract

Pipeline transport is the major means for large-scale and long-distance CO 2 transport in a CO 2 capture and sequestration (CCS) project. But optimal design of the pipeline network remains a challenging problem, especially when considering allocation of intermediate sites, like pump stations, and selection of pipeline routes. A superstructure-based mixed-integer programming approach for optimal design of the pipeline network, targeting on minimizing the overall cost in a CCS project is presented. A decomposition algorithm to solve the computational difficulty caused by the large size and nonlinear nature of a real-life design problem is also presented. To illustrate the capability of our models. A real-life case study in North China, with 45 emissions sources and four storage sinks, is provided. The result shows that our model and decomposition algorithm is a practical and cost-effective method for pipeline networks design.

Superstructure-Based Optimal Design of Pipeline Network for CO2 Transport in Large-Scale Carbon Capture and Sequestration

“…They will also lead to pump cavitation and thus negatively affect the lifespan of the booster pumps (Poullikkas, 1992). Moreover, the volume of CO 2 will explode with rapid vaporization of CO 2 flows and accidents may thus happen (Zhou et al, 2014). Thus, the thermophysical properties of CO 2 flows should be well controlled based on the concerns about the stability and reliability of CCS projects.…”

“…The methodologies of these pipeline network studies can be divided into two categories: one pertains to metaheuristics, like genetic algorithm (Andre, 2009;Fimbres Weihs and Wiley, 2012;Boyd et al, 1994;Abebe and Solomatine, 1998); the other pertains to superstructure-based modeling and continuous relaxation (Zheng et al, 2009;Middleton and Bielicki, 2009;Zhou et al, 2014). Metaheuristics methods can usually provide good results with relatively less computational effort, especially in large-scale problems where a global or local optimum would otherwise take rather long computational time to obtain.…”

“…The problem of pressure drop is generally not considered, or only superficially, and booster pump stations are never employed in these models. A typical sourceesink match model can be outlined as follows (Zhou et al, 2014): where U is the objective function, h and g are constraints, x is the vector of continuous real variables, and Y is the vector of binary variables. Sourceesink match models have been applied in many early trials proposed to tackle the optimization problem of pipeline network for CO 2 , transport.…”

“…To be specific, the model should integrally solve the following problems (Middleton and Bielicki, 2009;Zhou et al, 2014): …”

Analysis of GHG Emission Reduction in South Korea Using a CO2 Transportation Network Optimization Model

“…pipeline length between CO 2 -emission sources i and i' (i = i') m r ratio of NPV -RC 1 right-of-way cost factor 1 -RC 2 right-of-way cost factor 2 -RC 3 right-of-way cost factor 3 -RC 4 right-of-way cost factor 4 -T …”