trans chapter3

PIPENET VISION TRAINING MANUAL TRANSIENT: CHAPTER 3 PAGE 1 OF 27 REVISION 2.1, SEP 2010

- 1 -

PIPENET VISION TRANSIENT MODULE

CHAPTER 3

SUBSEA PIPELINE SYSTEM

1. Introduction

In this chapter, we shall consider two scenarios of pipelines carrying oil. We shall also introduce some of the basic ideas of manipulating graphs, which are discussed in more detail later in this manual.

1.1 Scenario 1

In the first scenario, we consider a single pipeline that is carrying oil from a platform to a terminal. This is the scenario of a sub-sea pipeline.

1.2 Scenario 2

In the second scenario, we consider two cross-country pipelines that are connected by a short pipe. This pipeline is a cross-country pipeline.

2. Sub-sea Pipeline Modelling

This example is based on a network devised by ZADCO in Abu Dhabi, which includes some pioneering work. The paper based on this pioneering work is reproduced as the appendix to this chapter of the training manual. In this problem, we demonstrate three aspects of modelling:

• Creation of a user defined pipe schedule.

• The effect of valve closure and closure time.

• The effect of a pipe rupture. A summary of the data is presented below, along with dialog boxes and descriptions of the results.

2.1 Units

The units that are to be used are essentially metric, but with m³/hr for flow rate. In this case, we specify that the units are user-defined. These units can be set using the PIPENET VISION menu system via either (a) Options | Units, if the Windows menu style is used, or (b) Init | Units, if the PIPENET VISION menu style is used. The menu style can be changed via the Windows Menu. In the remainder of this document, it is assumed that the Windows menu style is being used.


- 2 -


- 3 -

The actual units for the model are tabulated below.

Variable Unit

Length metres

Diameter mm

Velocity m/sec

Temperature Celcius

Density kg/m3

Viscosity cP

Time seconds

Mass kg

Mass rate kg/s

Torque Nm

Inertia kg m2

Force N

Volume m3

Surface Tension N/m

Thermal Conductivity W/(m K)

Heat Capacity J/kg K

Young’s Modulus Pa

Pressure Bar Gauge

Flow type Volumetric

Flow rate m3/hr

Input these units into the Transient Module. Note that, to make the flow-rate units visible, double-click on “Flow type”.

2.2 Simulation time

The first simulation is to begin at 0 seconds and end at 240 seconds. Note that the end time will be increased subsequently, for some of the later simulations. The time step will be user-defined, with a value of 0.05 seconds. Enter these values in the appropriate dialog box (from Options | Module options).

2.3 Pipe Data

In this problem, we shall use a user-defined pipe schedule, the data for which is as follows.


- 4 -

Pipe schedule name: Special Steel Pipe roughness: 0.0457 mm Poisson’s ratio: 0.33 Young’s modulus: 2.279 x 109 Pa

The diameters of the pipes that are to be used in the network are defined in the following table.

Nominal Bore (mm) Internal Bore (mm) External Diameter (mm) 600 581.2 609.6 700 679.5 711.2 750 730.3 762.0

800 781.1 812.8

850 831.9 863.6 900 882.7 914.4

Enter this data in the dialog box that arises from Libraries | Schedules, and then click on the OK Button.

All of the library data is stored in a separate file that has an extension of “.Lib”. PIPENET VISION automatically saves the library file once the Sunrise Data File (which has an


- 5 -

extension of “.sdf”) is saved. The root name of the library file is the same as the root name of the Sunrise Data File.

2.4 Fluid Properties

The liquid properties are shown below.

Density: 877 kg/m3 Viscosity: 6.8 Cp Bulk modulus: 1.5GPa Vapour pressure: -0.996263

Enter these fluid properties by selecting Options | Fluid, and completing the dialog box as shown below.

2.5 Pump Data

The following values are to be used to define the pump curve.


- 6 -

Flow rate (m3/hr) Pressure (bar)

0 64.877

6000 59.186

8000 56.91

10000 53.495

12000 46.666

The dialog box for the pump is obtained as follows.

The dialog box for the pump curve should be completed as shown below.


- 7 -

2.6 The Network

The schematic diagram of the network is shown below. Input this network, and ensure that the pipes are created in the same order, so that the network is consistent with the data table for the pipes. Note that two components are coloured green, as results have been selected for these components (by right-clicking on the component, and choosing “Select Results”, “Variable v Time”, and “All”).

The variation of the length of the pipes in the network is indicated in the diagram below (which can be obtained from Colouration | Simple rules).

The data for the pipes in the network is shown below.

Pipe Label Diameter (mm) Length (m) Elevation (m)

1 800 25 0

2 800 40 -10 3 800 60 -40 4 800 100 -20 5 800 10000 0 6 800 5000 10 7 800 5000 -10

8 800 2000 10 9 800 3000 -20 10 800 3000 20 11 800 2000 20 12 800 60 30 13 800 1000 10


- 8 -

Once the data for Pipe 1 has been entered, the Properties Window for this pipe should be as follows.

The data for all of the pipes in the network can be tabulated by selecting View | Data Window, and then choosing “Pipe” from the pull-down menu next to “Browse”.

Note that, at this stage, no results have been selected for the pipes.

2.7 Valve Data

Choose the valve to be a non-library linear valve, with a Cv value of 8000 (m³/hr, bar1/2), and indicate that all graphical results are required for the valve (by right-clicking on the valve, then selecting “Select Results”, “Variable v Time”, and “All”). The Properties Window for the valve is then as follows.


- 9 -

2.8 Time Step

As there are some short pipes in the system, the user-defined time step option has been chosen (as described earlier). The time step has been set to 0.05 seconds. As mentioned previously, if the user-defined time-step option is specified, PIPENET VISION automatically categorises pipes into long and short pipes. How exactly this division has been performed can be seen by selecting Calculation | Options… | Output, and clicking on the Timestep Button. It can be seen that three of the pipes (namely, Pipes 1, 2, and 14) are treated as short pipes.

3. Results

Consider the following five closure modes for the valve.

• Case 1: 20 seconds (linear)




• Case 5: 240 seconds (quadratic) In all of these cases, it is assumed that the valve starts to close after two seconds of the simulation. In the first three cases, the simulation time is 240 seconds; in the last two, the simulation time is 360 seconds. An additional case (Case 6) is considered, relating to pipe rupture, and this case is described later in this section.

3.1 Case 1: 20-second Linear Valve Closure

For the 20-second valve-closure time, the specification at the information node of the valve (Node 19) is for a linear profile.


- 10 -

The data that defines the profile (which can be displayed by double-clicking on the profile graph) is as follows.

The graph of the pressure just upstream of the valve and the valve position are plotted on the same graph. The result is shown below.


- 11 -

The PIPENET VISION Transient Module has powerful graphical-display capability. Use the following steps to produce the above graph.

• Use the command Output | Graphs…, or click on the Graph Viewer Button.

• When the graph viewer is open, ensure that the Time Graphs Tab is selected.

• Click on the prompt next to the directory tree.

• Click on the prompt next to “All variables of valve 1”.

• Tick the box by “Inlet pressure of valve 1”.

• Tick the box by “Setting of valve 1”. In order to copy the graph into a document, use Edit | Copy (Ctrl + C) in the graph viewer, and then paste the graph into the required document.


The data that defines the linear profile for the 60-second closure is as follows.


- 12 -

The graph for the 60-second valve-closure case is shown below. Note that this graph contains a title, “60-second Valve-closure Case”, which is added by supplying text in the title field of the graph viewer.

The label on the Y Axis for the valve information can be changed to “Valve Position” by clicking on “Setting of valve 1”, clicking on the Y Axis Tab below, and changing the text for the axis title.

The resulting graph is as follows.


- 13 -


The corresponding graph for 120-second valve-closure case is shown below. Labels and arrows have been added to this graph, via the Annotations Menu.

After a label has been created, double-click on the text to reveal a rectangular box.

The text can be rotated, by (a) left clicking on the node above the top of the box, (b) holding down the left-hand mouse button, and then (c) moving the mouse. The text can also be moved, by (a) moving the cursor inside the box, until the interior of the box becomes blue, (b) holing down the left-hand mouse button, then (c) dragging the box to the desired position.


- 14 -

To hide the rectangular box, double click in its interior. Arrows can be created and moved in a similar manner. To change the length of an arrow, hold down the left-hand mouse button on an edge and drag.

The legend can be removed by un-ticking the box called “Show Legend”. The resulting graph is as follows.


The graph for the 240-second valve-closure case is shown below. Note that the simulation time has been extended to 360 seconds (in both Options | Module options and Calculation | Options… | Output).


- 15 -

3.5 Case 5: 240-second Quadratic Valve Closure

A quadratic profile is obtained by changing the exponent of the ramp function to 2, as indicated below.


- 16 -

The graph for 240-second valve closure with a quadratic profile is shown below. Note that the simulation time is still 360 seconds.


- 17 -

3.6 Case 6: Pipe Rupture

We now consider the effect of a pipe rupture. Since the network has already been created, we need only place a valve at the position where the pipe ruptures, and then change the specifications as appropriate. The schematic is shown below. Note that results for Pipes 6 and 7 have been selected (as these pipes are coloured green).

The scenario that we wish to simulate is the following. A leak-detection system has been installed, and we assume that the leak-detection system sends a signal to the pump to stop (and to the downstream valve to close) within 5 seconds of detecting a leak. Let us further suppose that, upon receiving a signal to stop, the pump takes 120 seconds to spin down, and the valve takes 180 seconds to close. (Here, we are making a simplifying assumption that the pipe upstream of the valve is always full of oil.)

Location of leak


- 18 -

Fortunately, seawater has a higher density than oil. Consequently, the external pressure at the point of rupture is higher than the pressure inside the pipe, and so the flow will eventually stop. The external pressure at the point of rupture is calculated using the fact that the rupture occurs 60 metres below sea level. The density may be assumed to be 1025 kg/m3, and so the pressure is therefore

(1025 x 9.81 x 60) Pa = 603315 Pa = 6.03315 bar. We assume that the pipe rupture starts two seconds into the simulation, and finishes three seconds later. The properties of the new pipe (at the location of the leak) are as follows.

The valve properties are as follows.

The operation of the new valve (at its information node) is defined as follows. Note that the simulation time is 360 seconds.


- 19 -

The variation of the flow rate (at the inlet of Pipes 6 and 7) with time is plotted below. The flow rate through the pump momentarily increases, because the resistance to flow decreases when the pipe ruptures. However, fairly soon afterwards, the pump spins down, and so the flow rate decreases.


- 20 -

This example is based on pioneering work done by ZADCO. A publication based on their work is presented in the following appendix.


- 21 -

APPENDIX 1. TECHNICAL PAPER PRESENTED BY ZADCO

ARAB OIL & GAS SHOW, ABU DHABI – SEPTEMBER 2001

THE USE OF PIPENET IN MODELLING PIPELINES AND LEAKS

SUBSEA AND ONSHORE PIPELINES

Eur. Ing. Dr. Waheed Al-Rafai, ZADCO, United Arab Emirates

Dr. Dev Sebastian, Sunrise Systems Ltd, United Kingdom

In this paper we present results based on the pioneering work done by ZADCO in pipeline

design. We believe that this represents a major step achieved by ZADCO in developing

techniques for pipeline design and sets a new worldwide standard. The project concerned

with the integrity modelling of the arterial oil pipeline, which is a major asset of ZADCO.

The water content of the main oil line is low at present but it is expected to increase in the

future. This brings with it the danger of significantly increased pressure surges due to

increased water cut, even though the valve closure time may remain constant. The use of

state-of-the-art techniques developed by ZADCO is invaluable in optimising and planning

costly subsea rehabilitation activities, and in quantifying and justifying the benefit of

installing a leak detection system in support of improved pipeline operation.

The paper also gives an introduction to the role played by the PIPENET software in this

application.


- 22 -

SUBSEA PIPELINE MODELLING In this example we consider modelling a pipeline which carries oil from an offshore platform to onshore reception facilities.

• The effect of valve closure and closure time

• The effect of a pipe rupture

The objective in the first case is to determine the relationship between the valve closure time and the maximum pressure with the view of determining the optimum valve closure time. This calculation is particularly important where the integrity demand on the pipeline progressively increases due to weakening by corrosion, the need to transfer greater quantities of oil and an increase in the amount of produced water. By selecting an optimum valve closure time, which is inevitably a compromise between the emergency shutdown requirement and pipeline integrity constraints due to corrosion, the inspection frequency as calculated by Risk Based Inspection (RBI) and the time to repair the line can also be optimised. The objective in the second case is to minimise the environmental effect and the waste caused by the occurrence of a pipe rupture under the sea. During a leak every second counts and quick response by a leak detection system is critical for improved safety especially lines containing H2S. For the purpose of comparison, it was assumed that it took 15 minutes to detect a leak manually and a further 1 minute to shutdown the pump. On the other hand, with a leak detection system installed, it took 4 minutes to detect a leak a further 1 minute to shutdown the pump. The estimated oil which is drained into the sea is an important consideration in this. For the valve closure surge analysis the network in the schematic form is shown below.

The pipeline is approximately 35 km of 200 mm pipe following the profile of the seabed. The lowest point of the pipeline is 80 m below the level of the platform. Oil is pumped by a booster/transfer pump and there is an isolation valve at the end of the pipeline. Consider the following four valve closure cases.

60 sec 120 sec 240 sec 600 sec (quadratic valve closure)

In the first scenario, the valve is set to close in 60 sec. The wave speed is 1159 m/sec.


- 23 -

The period for the pressure wave to return to the valve after traversing the length of the pipe is approximately 60.4 sec. As this time, which is sometimes referred to as the critical time, is longer than the valve closure time, this scenario is likely to generate the maximum surge pressure.

As expected the maximum pressure occurs at the lowest point in the system and reaches a value of 95.3 bar. In the second scenario the valve closure time is increased to 120 sec. One would expect the pressure surge to decrease a little but not very significantly. This is because in a system of this type, the pressure surge can be expected to decrease significantly only after the valve closure time is many times the critical time. As described in the previous paragraph the critical time is the time it takes for the pressure wave emanating from the valve to travel the length of the system and return.

The maximum pressure again occurs at lowest point in the system and reaches 92.5 bar. As expected this is a little less than the maximum pressure with 60 sec valve closure time but not greatly.


- 24 -

The pressure peak occurs at the lowest point and has a value of 88.9 bar. In the next case we consider a valve closure time of 600 sec with a quadratic pattern. The advantage of quadratic valve closure is the following. Generally, the pressure surge is created during the final stages of valve closure. With quadratic valve closure the valve closes quickly to begin with and slowly during the final moments. So, within a given valve closure time, the effective rate of closure during the critical period is slow.

The maximum pressure at the lowest point of the system is 69.1 bar. It is difficult to reduce this significantly for the following reason. The closed head of the pump is 57 bar. The additional pressure due to static head is approximately 7 bar. The pressure at the lowest point would therefore be 64 bar even without any pressure surge. The next scenario we consider is the case in which the pipe ruptures on the seabed. This is potentially a serious hazard from two points of view. In an area like the Gulf leakage of oil into the sea could be a major disaster. Furthermore, the sheer waste is something the operator has to contend with. One major issue in a matter like this is the analysis of the economics of the system. Is it cost effective to install a leak detection system? It would therefore be of interest to consider two cases.

• The case in which a leak detection system has been installed.


- 25 -

• The case in which a leak detection system has not been installed. In both cases we assume that the leak takes 30 sec to fully develop. The leak itself occurs approximately 15 km downstream of the pump. In the first case the leak is detected 240 sec after it begins and a signal is sent to the pump to stop and the valve to close. After receiving the signal to stop, the pump takes 60 sec to wind down. The valve closes in 180 sec after receiving the signal to close. The system schematic and the graphical results are shown below.

In the second case we assume that the pump continues to operate and the valve remains open even after the leak starts. The operators manually detect that there has been a leak and the system is shutdown 15 minutes after the leak starts.

As expected, the case without the installation of the leak detection system results in a considerably greater environmental impact. In addition what PIPENET can do is to estimate the amount of oil which has leaked into the sea in the above cases. Furthermore, PIPENET can be used to assess the impact of parameters such as the response time of the leak detection system, the spindown time of the pump, the valve closure time and other parameters.


- 26 -

Amount of oil leaked with leak detection system - 600 m

3

Amount of oil leaked without leak detection system - 2070 m3

ONSHORE PIPELINE MODELLING The second case we consider is an onshore cross-country pipeline system. The system imports oil from three tanks in a tank farm and delivers to two delivery points using two parallel pipelines. The oil is pumped by one pumping station consisting of four pumps, connected in the form of two parallel sets. The parallel pipes have an interconnecting pipe approximately half way along.

We model the case in which both pipes rupture approximately in the same location. The leak fully develops in 10 secs. The following scenarios are considered.

• In the first scenario we assume that a leak detection system has been installed which sends a signal to shut down the pumps, within 5 sec of the leak beginning to develop. The pumps themselves take 60 sec to spin down. (Graph 2.1.)

• In the second scenario we consider the case where a leak detection system has not been installed. The pumps continue to operate normally even after the leak occurs. (graph 2.2.)

In both the scenarios there is a rush of oil when the leak occurs. However, in the case where a leak detection system has been installed, the flow rapidly goes down to almost zero. There is a small remaining flowrate because of the static head caused by the oil level in the tanks. In the scenario without the leak detection system, the flowrate through


- 27 -

the leak continues at a substantial level. PIPENET can be used to estimate important factors such as the volume of leakage and the impact of parameters which are under the control of the design engineer. CONCLUSION ZADCO has achieved pioneering leadership in the field of developing pipeline design techniques. This has been achieved by using the PIPENET software. In this paper we have shown how to achieve practical benefits to support pipeline integrity risk management activities. This is an important issue in the Arabian Gulf. THE AUTHORS Dr Waheed Al-Rafai obtained his PhD in Fluid Mechanics in 1990. He worked for Brown & Root Energy Services in the Arabian Gulf, USA and the UK. He now works for ZADCO in the UAE, with responsibility for developing a Pipeline Integrity Risk Management System for an extensive subsea pipeline network. He is a Fellow of the Institution of Mechanical Engineers in London and has a Master of Business Administration degree. He is the author of a number of papers on pipelines and related technologies. Dr Dev Sebastian obtained his PhD in mathematics from Imperial College, London. He also has an MSc in Chemical Engineering. After working for BOC and CAD Centre, he is now the Marketing Manager at Sunrise Systems in Cambridge, United Kingdom. He is the author of a number of papers on numerical methods. The authors would like to thank ZADCO for giving permission to present this paper.

trans chapter3

Documents

pipenet vision menu

options units

pipenet vision menu

actual units

pipe data

flowrate units visible

user defined pipe schedule

userdefined pipe schedule