Anda di halaman 1dari 9

Multiphase Pipeline Simulation

Corelium Software Inc, Jan. 2012

Summary
In 2011, Corelium began development of a simulation engine for multiphase pipeline networks, targeted at oil and gas production. The engine is based on Coreliums advanced parallel simulation platform. The latest version of the engine incorporates a full Multi-Fluid liquid-droplet-gas model. This document describes the results of the application of the engine to a gas-production example problem published by the Pipeline Simulation Interest Group (PSIG) (http://www.psig.org/papers/2000/0403.pdf).

Global Gas Production


The world energy market today is based on the production and delivery of hydrocarbons. A large part of that market is natural gas, which is the cleanest hydrocarbon source we have, with the potential for a few hundred years global supply. With the recent advances in economic production of shale gas, the gas market will continue to grow. Gas reservoirs and fields are typically located far from their market and delivering the gas to market requires a significant capital investment. Reservoirs are underground or under a seabed which might be then thousands of meters under the water. A full production system for a large gas field requires billions of dollars in capital and advanced simulation can help optimise the production system design and operation.

The Need for Pipeline Simulation


Pipelines are among the largest physical components in a production system, one of the most expensive and have the most significant effect on how the system operates. In gas production there can be light oil or condensate produced from the reservoir in addition to gas. This creates multiphase flow in the production wells and pipelines where liquid and gas travel together in a variety of different flow patterns. Understanding and predicting this multiphase flow is critical to the design and operation of the production facility. Flow Assurance is the practice of ensuring consistent, reliable delivery of the product gas from the reservoir to the point of sale. Simulation of the flow through the pipelines under operating conditions is a key enabler for Flow Assurance.

Simulation Model
Pipeline flow models range from empirical methods that use correlations to predict behaviour to mechanistic models that attempt to describe the physical phenomena actually occurring in fluid flow. The most advanced mechanistic models are based on first-principles multi-fluid models that describe each flowing phase separately and account for their interaction.

coreliuminc.com

info@coreliuminc.com

Multi-fluid models for pipelines evolved from the nuclear industry where it was important to simulate the flows through the cooling system during a loss-of-coolant accident. The original landmark paper in multi-fluid modelling for pipelines was published in 1991: Bendiksen, K.H. (IFE), Malnes, D. (IFE), Moe, R. (IFE), and Nuland, S. (IFE): The Dynamic Two-Fluid Model OLGA: Theory and Application, SPE Production Engineering, May 1991, pp. 171-180 The model used in Coreliums pipeline engine is similar to the OLGA model and is described in Appendix A.

Example Problem
The test problem is taken from a paper published by the Pipeline Simulation Interest Group (PSIG) in 2004 (http://www.psig.org/papers/2000/0403.pdf). The problem is based on three offshore platforms that recover gas from three different gas reservoirs. Two of the reservoirs produce gas with some condensate liquid and one produces dry gas. The three platforms feed undersea pipelines that deliver their gas to a fourth host platform that collects the gas into a larger fourth export line that delivers the gas to shore. There are limits on the line and platform pressures. The layout and major parameters from the paper are shown below in Figure 1. The platform/reservoir gas compositions and other parameters are provided in Appendix B.
Platform A Dry Gas 183 m water 5.7 MM Nm3/d (200 MMSCF/d) gas 13.8 MPa max

32

km ,0

.3

di a
Host Platform 160 km, 0.8 m dia Shore 6.9 MPa min

Platform B 48 km, 0.4 m dia Wet Gas 305 m water 8.5 MM Nm3/d (300 MMSCF/d) gas a di 121 std m3/d (760 bbl/d) condensate m .6 13.8 MPa max ,0 m k 80

Export Line Wet Gas 152 m water 22.7 28.3 MM Nm3/d (800 - 1000 MMSCF/d) gas 981 1561 std m3/d (6190 9810 bbl/d) condensate 10.3 MPa max

Platform C Wettest Gas 1520 m water 8.5 14.2 MM Nm3/d (300 - 500 MMSCF/d) gas 860 1440 std m3/d (5430 9050 bbl/d) condensate 13.8 MPa max

Figure 1 The system is designed to bring 28.3 Nm3/d of gas to shore (1000 MMscf/d), at a minimum shore pressure of 6.9 MPa (1000 psi). The export line and host platform have a maximum pressure restriction of 10.3 MPa (1500 psi). The three remote platforms have a maximum pressure restriction of 13.8 MPa (2000 psi). Platform A supplies

coreliuminc.com

info@coreliuminc.com

dry gas through a 32 km line to the host platform. Platform B supplies gas with some liquid condensate through a 48 km line. Platform C supplies gas with a substantial amount of liquid condensate through an 80 km line. The PSIG paper describes a variety of steady-state and transient scenarios.

Steady-State Simulation
The steady-state flowrates are given in Table 1 below. This is a base case for the transient simulation described in the next section. Platform A B C Host/Export Gas rate (MM Nm3/d) 5.7 8.5 8.5 22.7 Table 1 Liquid Rate (std m3/d) 0 121 860 981

Figure 2 below shows the topology and pressure profile of Line A. Line A travels a short distance along the platform deck, which is almost 50 m above sea level, then drops almost 200 m to the sea floor and follows a reasonably linear incline along the seabed to the host platform. The line pressure peaks at 10,700 kPa (1552 psi) at the seabed and decreases to 8825 kPa (1280 psi) at the host platform deck.

Figure 2

coreliuminc.com

info@coreliuminc.com

Figure 3 below shows the topology and pressure profile of Line B. Line B travels a short distance along the platform deck, then drops 350 m to the sea floor in deeper water than Line A. The seabed is more irregular with a steeper incline as the line approaches the host platform. The line pressure peaks at 10,800 kPa (1566 psi) at the seabed and decreases to 8825 kPa (1280 psi) at the host platform deck.

Figure 3

Figure 4 below shows the topology and pressure profile of Line C. Line C is in the deepest water with a depth of just over 1500 m and an undulating seabed. The increased depth requires a higher pressure to drive gas to the host platform. The line pressure peaks at 12,000 kPa (1740 psi) at the seabed and decreases to 8,825 kPa (1280 psi) at the host platform deck.

Figure 4

coreliuminc.com

info@coreliuminc.com

Figure 5 below shows the topology and pressure profile of the Export line from the Host platform. The Export Line delivers the gas and condensate to shore. It travels along the seabed for 40 km before climbing a shelf that leads to shallower water closer to shore, and the delivery point is at sea level. The line pressure peaks at 8,900 kPa (1291 psi) at the seabed and decreases to 6,900 kPa (1,000 psi) at the delivery point.

Figure 5

The production system is operating within the pressure constraints for each platform and the shore delivery.

Transient Simulation
Transient simulation of production systems is an important tool and is fundamental to Flow Assurance modelling. Several transient scenarios are described in the original paper, including the scenario described here. The scenario involves increasing the flow from Platform C to its full capacity. The gas rate increases from 8.5 to 14.2 MM Nm3/d (300 to 500 MMscf/d) and the liquid rate increases from 860 to 1440 std m3/d (5430 to 9050 bbl/d), an increase of 67%. The flow rate changes are immediately stepped to their new rates and not ramped over time, which is a severe change. The previous steady-state simulation provides the initial conditions.

coreliuminc.com

info@coreliuminc.com

Figure 6 below shows the major transient responses of the network over two days of simulated time.

Figure 6 These graphs show the response of the system to the step increase from Platform C. In the first graph, the step increases from Platform C can be seen on the red and blue lines. Gas mass flow is read on the left Y axis and liquid mass flow on the right. The shore gas mass flow is in green and the shore liquid mass rate is in purple. The shore delivery rates start to respond reasonably quickly, in line with the pressure increase that travels through the network. The liquid rate at the shore increases directly in line with the gas rate. This requires some explanation. The B, C and Export pipelines are operating in the annular flow regime because of the relatively small amount of liquid in the gas. Some liquid is transported in a thin film around the inside of the pipe, and the rest of the liquid is

coreliuminc.com

info@coreliuminc.com

entrained as droplets flowing at almost the same speed as the gas. Consequently, as soon as the gas velocity increases, the liquid droplets are accelerated and the liquid rate increases with the gas. About 19 hours after the step change, a surge of liquid arrives at the shore. The peak volumetric rate is approximately 3,000 std m3/d (18,870 bbl/d). The surge lasts for about 8 hours and then subsides to the new base liquid rate. The surge contains an extra 14 tonnes of liquid which must be dampened out by an onshore slug catcher. A similar but smoother and slower response is shown in the PSIG paper results. The difference is probably because in this simulation there is no slug catcher and control system simulated on the host platform, and the four lines are connected with a node that has zero volume. The PSIG paper does not describe whether a slug catcher on the host platform was incorporated into the simulation. The second graph shows the pressure responses at each platform with the shore pressure specified at 6,900 kPa. Platform C responds the fastest, with Platforms A, B and the Host responding more slowly. The peak pressures coincide with the liquid surge at the shore and then gradually decline to the new steady-state. With the increased flow from Platform C the system is still within the pressure constraints.

Simulation Performance
The production system model contains 15,000 cells and 360,000 variables and is solved simultaneously as one equation-based flow network. This is a relatively small model, but still able to gain useful speedup from parallelism. Using 8 cores the simulation runs at around 50 times real-time. Figure 7 below shows the speedup obtained from using multiple cores in the transient simulation.
4 3.5 3 Speedup 2.5 2 1.5 1 0 2 4 Cores 6 8

Figure 7 From 1 to 2 cores, the speedup is 2.0, at 4 cores the speedup is 2.8 and at 8 cores it is almost 3.5. This is a reflection of the linear solver used in the transient engine and further work with different algorithms that provide better scale-up is under way.

coreliuminc.com

info@coreliuminc.com

Appendix A Mathematical Model


A detailed description of all the flow patterns encountered in pipelines is outside the scope of this document. Good descriptions are available in many reference books. Briefly, in pipelines with small amounts of liquid, the liquid is often entrained in the gas as droplets, with the remainder traveling around the inside edge of the pipe in a layer. Lines with significant amounts of liquid can form slugs, where liquid and gas travel partially separated as lumps that can be several meters long, or as long, undulating waves. Lines with mostly liquid can form bubble flow, where the gas travels as bubbles inside the liquid. These flow regimes depend on the liquid and gas properties, the temperature and pressure, the amounts of liquid and gas, how fast the fluids are moving and the angle of the line. These parameters also affect how the flow transitions between these regimes. There are two main categories of mechanistic model used to simulate these flow regimes. The simpler category is the Drift-Flux model, which considers the flowing mixture as a whole, with slip between individual phases. It requires some assumptions to be made across the various phases and is not as accurate or as flexible as a true Multi-Fluid model. Multi-Fluid models consider each flowing phase separately and accounts for the interactions between them. The model used in Coreliums engine is very similar to the OLGA model for gas, droplet and liquid flow. The fundamental equations are shown below. A complete derivation is outside the scope of this document. The equations are a set of nonlinear, hyperbolic partial differential equations, with very nonlinear source terms.

Equation (1) is the gas mass balance. Equation (2) is the droplet mass balance. Equation (3) is the liquid mass balance. Equation (4) is a combined gas and droplet momentum balance. Equation (5) is the liquid momentum balance. Equations (6) and (7) describe the energy balance. The engine solves the equations implicitly in time and with a first-order upwind-difference discretisation in space. The energy balance is not solved at this stage. A future version will incorporate complete component and energy balances to enable compositional and temperature tracking through a network.

coreliuminc.com

info@coreliuminc.com

Appendix B Example Problem Parameters


The fluid properties were generated with a thermodynamic package using the Peng-Robinson equation of state. The compositions from the PSIG paper are given below in Table 2. Mole Fractions Component nitrogen CO2 methane ethane propane i-butane n-butane i-pentane n-pentane n-hexane n-heptane octane nonane undecane Platform A dry 0.0010 0.0010 0.9950 0.0025 0.0003 0.0002 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Platform B wet 0.0019 0.0005 0.9859 0.0041 0.0021 0.0004 0.0011 0.0004 0.0004 0.0005 0.0004 0.0003 0.0003 0.0017 Platform C wettest 0.0019 0.0005 0.9786 0.0041 0.0021 0.0004 0.0011 0.0004 0.0004 0.0005 0.0012 0.0010 0.0009 0.0070 Host/Export wet 0.0017 0.0006 0.9844 0.0037 0.0017 0.0004 0.0009 0.0003 0.0003 0.0004 0.0007 0.0006 0.0005 0.0039

Table 2

The pipeline diameters were taken directly from the PSIG paper. Line A is 0.305 m (12) diameter, Line B is 0.405 m (16) diameter, Line C is 0.610 m (24) diameter and the Export line is 0.813 m (32) diameter.

coreliuminc.com

info@coreliuminc.com

Anda mungkin juga menyukai