Application of a General-Purpose Network Optimizer to Oil and Gas Production

Abstract

Abstract A general purpose non-linear Sequential Quadratic Program (SQP) based network optimizer such as NETOPT can be used to solve a wide range of optimization problems in the Oil and Gas production industry. Optimization can be used as a strategic field planning or design tool to maximize revenue or profit from a field or minimize cost. In addition it is applicable in the field on a periodic basis as a operations tool to determine optimum control settings for changing field conditions. Optimization may be also used to determine control strategies to meet production targets. This paper discusses some of the typical problems where SQP based optimization can be readily applied to significantly reduce costs, boost production or profitability and make marginal projects viable. Specific application examples are briefly discussed to illustrate the use of a general purpose optimizer and to demonstrate the benefits that can be realized by applying such technology. Introduction Networks: Figure 1 illustrates a typical network. A network consists of points called nodes which are connected by node connecting elements called links. A link is made up of one or more pressure and/or temperature change device in series. A device may be a reservoir 'PR' completion, tubing, pipe, compressor pump, cooler, heater and other oil field equipment. A general network solver should have the ability to solve for any network structure and boundary condition. For e.g. the network may be a tree network or a complex looped network. Nodes may be sources, sinks or junctions. The source and sink nodes are also called boundary nodes. To set up the network equations correctly, either pressure (P) or flow (Q) must be specified (known variable) at the boundary nodes. The network simulator solves the network equations for the unknown node pressure (and temperature) or flow and unknown link flows in the network. The Network Optimization Problem: The optimizer adjusts the network decision variables such that the objective function is maximized or minimized while simultaneously satisfying the user defined constraints and the network equations. The network optimization problem may be mathematically expressed as x represents the network variables. and y represents the unknowns of the network problem. From the point of view of computer implementation there are two classes of variables in the optimization problem:The primitive or internal network variables andThe user defined external variables which are functions of the internal variables. The external variable definition allows the user to define any algebraic function for F or Ci. The non-linear constraint function Ci may be defined as a external variable. N(y) are the network equations which are in general non-linear. P. 943^