MINLP Model and Solution Strategies for the Long-Term Planning and Development of Shale Gas Supply Chain Networks

DIEGO C. CAFARO E IGNACIO E. GROSSMANN

San Francisco, CA

Encuentro; AIChE Annual Meeting; 2013

AIChE (American Institute of Chemical Engineers)

Natural gas is an efficient energy source and the cleanest-burning fossil fuel. Natural gas extracted from dense shale rock formations has become the fastest-growing fuel in the U.S. and could become a significant new global energy source. Over the past decade, the combination of horizontal drilling and hydraulic fracturing has allowed access to large volumes of shale gas that were previously uneconomical to produce. The production of natural gas from shale formations has rejuvenated the natural gas and chemical industries in the U.S. The Energy Information Administration projects U.S. shale gas production to growth from 23% to almost 50% of the total gas production in the following 25 years. Shale gas is found in "plays" containing significant accumulations of natural gas, sharing similar geologic and geographic properties. A decade of production has come from the Barnett Shale play in Texas. Experience gained from Barnett Shale has improved the efficiency of shale gas development around the country. Today, one of the most productive plays is the Marcellus Shale in the eastern U.S., mainly in Pennsylvania. Regarding both economic and environmental impacts, the long-term planning and development of the shale gas supply chain network around each play is a very relevant problem. However, to the best of our knowledge, it has not been addressed before in the literature.   The raw gas extracted from shale formations is moved from wellbores to processing plants through gathering pipeline systems. The processing of shale gas consists of the separation of all the various hydrocarbons and fluids from the pure gas (methane) to produce what is known as "pipeline quality" dry natural gas. This means that before the natural gas can be transported by midstream distributors, it must be purified to meet the requirement for pipeline, industrial and commercial uses. The associated hydrocarbons (ethane, propane, butane, pentanes and natural gasoline) known as "Natural Gas Liquids" (NGLs) can be very valuable byproducts after the natural gas has been purified and fractionated. These NGLs are sold separately (usually moved through dedicated pipelines) and have a variety of different uses, including enhancing oil recovery in wells, providing raw materials for oil refineries or petrochemical plants, and as sources of energy. One of the most critical issues in the design and planning of the shale gas supply chain network is the sizing and location of new shale gas processing plants (as well as future expansions) due to their very high cost. On the other hand, the number of wells drilled in each location can dramatically influence costs and the ecological footprint of natural gas operations. The ability to drill multiple wells from a single location (or "pad") is seen as a major technological breakthrough driving natural gas development in the Marcellus Shale. The utilization of "multi-well pads" also has large environmental and socio-economic implications, as the landscape disruption of as many as 20 or more natural gas wells and associated pipeline infrastructure can be concentrated in a single location. Furthermore, the total amount of industrial activity can be compressed as these wells can be drilled in rapid-succession and the technology now exists to perform hydraulic fracturing stimulations on multiple wells simultaneously. Hence, another key decision tackled by this paper is the drilling strategy, i.e. how many wells to set up or add on existing well pads at every period. Another critical aspect in the shale gas production is water management. Shale gas production is a highly water-intensive process, with a typical well requiring around 5 million gallons of water to drill and fracture, depending on the basin and geological formation. The vast majority of this water is used during the fracturing process, with large volumes of water pumped into the well with sand and chemicals to facilitate the extraction of the gas. Although increasing volumes of water are being recycled and reused, freshwater is still required in high quantities for the drilling operations as flowback water is more likely to damage the equipment, reducing the chance of a successful well. The need for freshwater is a growing issue, especially in waterscarce regions and in areas with high cumulative demand for water, leading to pressure on sources and competition for water withdrawal permits. Therefore, a long-term planning model for the development of shale gas fields should also account for water availability, transportation, treatment, and final disposal. The goal of this paper is to develop a mixed-integer nonlinear programming (MINLP) model for the sustainable long-term planning and development of shale gas supply chains, which should optimally determine: (a) the number of wells to drill on new/existing pads; (b) the size and location of new gas processing plants (as well as future expansions); (c) the section, length and location of new pipelines for gathering raw gas, delivering dry gas, and moving NGLs; (d) the location and power of new gas compressors and pumps to be installed, according to the flow rate at every line; and (e) the amount of fresh water coming from available reservoirs, used for well drilling and fracturing; so as to maximize the economic results (NPV-based approach) and minimize the environmental impacts of the project, over a planning horizon comprising 20 years.   Since the proposed MINLP model is large and non-convex, usually intractable for commercial solvers (even for rather small case studies), we develop a decomposition approach for the global optimization of the problem, based on successively refining a piecewise linear approximation of the objective function. Results on realistic instances show interesting results, particularly the high relative importance of NGL selling incomes to the economic viability of the project, as well as the optimization of the pipeline usage by properly planning the well drilling strategy over the first 5 years, among others. By combining MILP relaxations and NLP models, we propose novel reformulations and approximation techniques, that lead to a solution strategy that provides very efficient solutions and tight bounds with modest CPU times.