ABSTRACT The Negative Pressure Wave (NPW) leak detection method has been increasing in popularity recently due to the promise of highly accurate leak location calculations compounded with response times on the order of seconds. Additionally, when compared to other leak detection methods such as Real Time Transient Models (RTTM), NPW can be implemented on many different pipelines in a relatively short period of time with only the need for pressure transmitters. However, as with all engineering applications, there are tradeoffs to consider. In this analysis, a NPW solution was evaluated on two different liquid pipelines running offline simulators: one pipeline transports crude oil and one pipeline transports diesel. It was observed that, under very specific operating and test conditions, NPW could perform adequately. The conditions which allow for accurate NPW performance are limited when considering real world operations. With this consideration, NPW can be a complimentary method to other leak detection methodologies. This analysis is intended to provide pipeline operating companies with a greater understanding of when and where an NPW solution may be appropriate, as well as the limitations and challenges associated with the NPW methodology.

Abstract For many years, the pipeline industry has worked on leak detection technologies to accurately predict and mitigate the occurrence of leaks in pipelines. However, this is not an easy task since many systems have different characteristics and the existing technologies have specific ranges of applications. Nevertheless, the technological progress has reduced the gap between accurate leak predictions and false alarms in single-phase systems. A new challenge has been growing leak detection technologies with multiphase production systems since many of the current leak detection techniques limit their applications to single-phase systems. Although, there are some technologies that claim multiphase capabilities; it is not clear yet how they overcome the challenges imposed by the hydrodynamics of a multiphase system. Therefore, the main objective of this paper is to assess a multiphase system and simulate leak events to gain a better understating of its hydrodynamic behavior during the transient leak event. The studied multiphase system operates with three phases: oil, water, and gas; thus, flow patterns vary from stratified to severe slugging for some operating conditions and locations. The 30-in. pipeline system starts from an offshore production-separation platform which is located approximately 55 miles offshore and delivers its production to a separation processing plant on-shore. A leak condition was simulated in the middle of the trunkline to determine its effect on the hydrodynamics of the system and both ends facilities. Different gas oil ratios (GOR) and pipeline pressures were evaluated to determine the effect of those conditions on the leak detectability. One key parameter to accurate determination of a leak is to precisely differentiate or filter a leak signal (pressure, flow rate, pressure wave, etc.) from normal pipeline data noise. In this case that was even more challenging since a slugging condition was observed in the normal operation of the multiphase system which generates flow and pressure fluctuations that could overlap the leak data. This paper presents the results obtained for the steady-state prior leakage and transient leakage cases. In addition, a sensitivity analysis of different gas oil ratios is presented as well to map out its effect on the leak detectability. Introduction and Background A concern for operating a multiphase production system under higher pressures and gas-oil-ratios was expressed during a capacity increase study since that would reduce its reliability and integrity while increasing the operational risk. The offshore field demonstrated a potential for producing 5-7% more crude oil and gas than that of its original design capacity and the capacity study indicated a good feasibility of operating at those conditions. Therefore, a risk assessment of the system was required to ensure it function within safety compliance and quantify the risk associated with these new operating conditions. In addition, a part of the IMP possible mechanical failures (leaks) and their effect has to be evaluated to determine the consequence of a leak and how to better control it. Consequently, a leak simulation analysis was conducted to understand how the system will behave under leak conditions, to estimate the time to possibly detect a leak with the current technology, and to get an approximation of the amount of product spill for mitigation actions.

Abstract This work focus on the simulation of gas pipeline dynamic models in view to develop a leakage detection tool. The gas dynamics in the pipes is represented by a system of nonlinear partial differential equations. The linear partial differential equations is reduced to a transfer function model. Taking advantage of an electrical analogy, a pipeline can be represented by a two port network where gas mass flows behave like electrical currents and pressures like voltages. Thence, four transfer functions quadripole models are found to describe the gas pipeline dynamics, depending on the variable of interest at the boundaries. These models are simple enough to be used in the control and management of the network. These models have been validated using operational data and used to simulate a leakage. Introduction Natural gas sustains one quarter of the global energy necessities [8] with the world gas networks representing many kilometers of pipelines, manifold branches, pumps, compressors, valves, among others; being therefore a large scale and complex system to control and maintain [1]. The control and the safe operation of these systems are crucial due to grave consequences that may result from default operation [5], mainly leaks. Model-based (MB) software methods continually measure pressure and/or mass flow signals frequently at the intake and outtake of the pipeline, so instrumentation is usually limited to the extremes of the pipelines and consequently economical to implement. The most reliable among these methods are based on nonlinear partial differential equations (PDE) to describe the gas dynamics [3]. These PDE are not modular and its resolution can lead to a non-efficient and high complexity models; nevertheless PDE can be linearized and leading to simple and accurate models [6,3]. In this paper, we propose a model focus on the modelling and simulation of the pipeline dynamics, and able to improve leakage detection/location techniques. Therefore, we describe four quadripole models which are derived using an electrical analogy. These models bestow boundary values of pressure and mass flow, as well as intermediate values, allowing for obtain mass flow and pressure calculated profiles. The analysis of the discrepancies between calculated and measured values allows investigating the effect of a leakage. The article is organized as follows: In Section 2, transfer function (TF) models are derived from a nonlinear first order hyperbolic PDE. Four different TF quadripole models are described in Section 3. In Section 4 the described models are validated using real data. An approach to determine intermediate pressure and mass flow values is presented in Section 5 and adapted to the leakage situation in Section 6. The last section provides some brief conclusions and directions for future work.

Abstract Analysing the leak detection sensitivity for a particular system, or predicting how operational transients influence it, are common challenges for leak detection professionals. An empirical approach can be followed: modeling the pipeline system, including all the inaccuracies due to the instruments and communication systems. The sensitivity can then be inferred by running specific scenarios. Designing the scenarios is not a trivial task. The system response to transients provides information about sensitivity degradation. This information is used to plot a sensitivity map, which in turn helps to understand the more favorable and less favorable sensitivities for a particular system. Colombian oil company, Ecopetrol, followed this practical procedure for a new pipeline in the commissioning phase. This allowed a pre-determined estimate of the system sensitivity and an evaluation on the operational transients involved in the leak detection tests. The present paper describes the approach which Ecopetrol followed during model construction, the design of scenarios and the results obtained. API 1149 steady state and transient predictions are also compared with the model results. A sensitivity map considering both sources of data is analyzed. Finally, the paper presents a discussion about the validity of this procedure and its generic application. Introduction Ecopetrol, the national oil and gas company in Colombia, carried out a leak detection sensitivity analysis for a pipeline prior to commissioning. The procedure included the configuration of an off-line model, the design of leak test scenarios, the addition of uncertainties to the hydraulic variables and running a leak detection system. The leak test results from the model are compared with the API 1149 steady state estimations and the sensitivity degradation due to uncertainties in system variables.

Abstract In the recent decades, new leak detection technologies for hydrocarbon pipelines are emerging constantly, and conventional technologies are evolving to achieve better performance. It gives pipeline operators a good chance to detect leaks in the early phase and protect the environment from contamination by leaked hydrocarbons. However, this also makes the selection of leak detection methods more complicated. Two major tasks are required during the selection of leak detection methods: appropriately defining the selection criteria, and evaluating the leak detection capability of candidate technologies. For internal leak detection technologies, API 1130 (API, 2012) provides guidance to industry on the selection criteria and API 1149 (API, 1993) attempts to define the evaluation technique. Applying an evaluation technique that considers all likely operating conditions in its risk assessment is advisable. The process of combining of all the potential performance scores of a leak detection technology for various operating conditions into a single fitness factor can be complicated. This paper recommends the use of a decision tree to calculate the leak detection methods' fitness factor. In this analysis, many facets of leak detection are considered. They include the physical principles of the technologies as well as their performance under different operation conditions. The paper describes a systematic approach of evaluating the various leak detection methodologies against specific design criteria. The appendix includes descriptions of the strengths and limitations that the various leak detection techniques may have given a particular set of leak characteristics, operating conditions, physical design considerations, commercial implications, construction costs and operational costs. The paper proposes that a qualitative or semi-quantitative analysis on individual pipelines over all expected operating conditions could determine the best technology(s) to apply in mitigating the risk of undetected leaks in pipelines. Introduction There are multiple real time leak detection technologies, and they all have strengths and weaknesses. Leaks can present themselves in different ways, and every pipeline has different physical design and operational parameters. Determining the best leak detection technology(s) for a specific application can be complex. Every leak detection technology can cover different portions of the risk spectrum but no single technology will mitigate all of the risk as illustrated in Figure 1.

ABSTRACT Noise is prevalent in all pipeline instrumentation and measurement system data and has been shown to affect Computational Pipeline Modeling (CPM) based leak detection (LD) systems in practice. Noise present in data can be random or repetitive, occurring continuous or in isolated bursts and can have a multitude of sources. At Enbridge, the leak detectability of CPM LD system is notably affected by the presence of noise in the pipeline data. Data smoothing of noisy pipeline data is a non-trivial process; these techniques are used not only to eliminate "noise" from the operating data, but also to extract real trends and patterns within the data, and to maintain important characteristics of the signal itself, particularly during transient time periods. The data smoothing initiative began within the Leak Detection group around August 2011 and the wavelet transformation technique was adopted as the data smoothing methodology to remove or attenuate the noise present in the pipeline data to improve the leak detectability of current LD system (MBS or Mass Balance System). Extensive offline studies and pilot tests including simulated leak tests, fluid withdrawal tests, and API 1149/API 1130 calculations have been carried out to prove this concept [1]. The offline studies have shown promising results and Enbridge has decided to pilot this technology with real time SCADA data in 2013. If successful this pilot will be the catalyst for the full implementation of this technology on all Pipelines. Online implementation is a non-trivial process as it includes not only modifications to the real time communication architecture between the SCADA system and the leak detection CPM, but also includes code integration, windows service development, and dealing with performance issues arising from the online assessment and evaluation. To properly assess performance it has also been commited to performing a full fluid withdrawal test during this Pilot to assess real time performance of Data Smoothing. This paper reviews the data smoothing methodology employed, and discusses the issues faced during the online implementation as well as the online test results. Extensive pre-implementation and online tests have been carried out to evaluate and assess this methodology. These test results indicate that a data smoothing technique has a substantial effect on the data quality which in turn has improved the detectability of current leak detection system.

ABSTRACT: Leak detection systems come in many "flavors", each with distinct advantages, disadvantages and sometimes unique features and capabilities. Needless to say, performance and cost vary greatly as well. It can be a daunting task to decide which technology to invest in for a specific pipeline and a few wrong turns in the decision making process can end up being very costly. It is easy to say " I want a leak detection system capable of detecting a ten gallon spill in less than a minute, locating it within an accuracy of 10 feet ". It might even be possible to install a system that can achieve this, but what about the cost? If this system costs ten times as much as another system with slightly lower performance, is it really worth spending the extra money? What is the difference in the total cost of a specific incident when you factor in the various cost components? This paper puts some of the leak detection metrics and selection criteria into perspective and illustrates the importance of various factors under different conditions. This is accomplished by using some hypothetical, yet realistic, examples of various leak incidents. BACKGROUND American Petroleum Institute's publication "API RP 1130" provides an excellent overview of most of the leak detection methods available for liquid pipelines - and many of these apply to gas pipelines as well. API RP 1130 also provides a definition of the performance metrics and other selection criteria that should be considered during the selection phase. Sometimes the selection criteria are not carefully considered and prioritized, which can lead to a poorly written Request for Quotation (RFQ) issued to various system vendors. To make matters worse, this is often done using a format that originally was part of API 1155 – now an appendix to API RP 1130.

ABSTRACT: When a long gas pipeline is severed, gas escapes through a rarefaction wave that leads away from the break at sonic velocity. Immediately after rupture, the classical inviscid solution of rarefaction holds in which change proceeds evenly across the wave and the discharge is choked. After traveling a relatively short distance (a few hundred diameters) however, friction takes hold and reshapes the inviscid profiles so that the principal changes occur near the outlet. The practical consequences of friction are continual reduction of discharge rate and smoothing of the wavefront. Until the rarefaction reaches a distant boundary, depressurization is closely approximated by the similarity solution of a diffusion equation for density supplemented by evanescent boundary layers at the outlet and leading edge. The similarity solution yields a simple formula for the declining discharge rate with time and reveals that Excess Flow Valves, designed to isolate the line automatically, will not close if placed too far apart. INTRODUCTION Accidents in which a pipeline is severed are consequential. Large quantities of toxic or flammable material are rapidly discharged to atmosphere, and prompt action must be taken to protect the neighboring population. Understanding discharge dynamics in these cases is essential to assess potential hazards and develop appropriate risk mitigation plans. Dynamics at the leading edge of the depressurization wave that sweeps through the gas within the pipe also matters. Leak detection systems that rely on prompt identification of the rarefaction wavefront at critical locations may siginificantly lower risk posed by pipeline rupture. Local leak detection is often tied to Excess Flow Valves that shut automatically when the flow exceeds a threshold value. These nonlinear equations extend classical inviscid gas dynamics with algebraic friction and heat transfer terms that account for interaction of the gas with pipe walls.

ABSTRACT This paper describes in detail the key experiences of a well-established Oil & Gas company in Latin America related to the implementation and evolution of a Pipeline Advanced Application project; how these applications have helped the organization to improve the knowledge of its own networks, and how they have provided a solid foundation to manage the critical problem of product theft. Ecopetrol S.A. (Ecopetrol) is responsible to build and operate the whole Colombian transport infrastructure and distribution of oil & gas and refined products by means of its 8,500 kilometer network of pipelines which run from the production centers to the refineries and ports in the Atlantic and Pacific oceans. Ecopetrol has the responsibility to guarantee supplies of oil and biofuels production throughout Colombia, whether its own production or through domestic or foreign third party providers (Figure 1). Ecopetrol started implementing real-time online application technology in 1998, mainly motivated by the desire to eliminate hydrocarbon leaks, thefts and losses occurred in the provision of the transport service. The paper is supported by field data gathered by Ecopetrol that describes the evolution and performance of the application as well as the overall performance achieved. COMPANY DESCRIPTION The pipeline network of Ecopetrol consists of 50 transportation systems, including more than 52 supply, delivery, storage and pump stations. The transportation network is grouped into two main systems: multiproduct pipelines and crude pipelines. Due to the complex topography of Colombia, including the Andes mountain chain located at the central and west side of the country, almost all of these pipelines have elevation profiles with dramatic variations from 0 meters to 3,000 meters (0 to 10,000 feet) and temperature changes of up to 30° C (54° F) within a few miles Technology Background Ecopetrol controls and operates all its multiproduct pipelines through a SCADA system located at the Main Control Center in Bogota. On top of the SCADA system Ecopetrol implemented a set of advanced applications. These applications are based on a real time transient model, which consists of a hydraulic and thermodynamic representation of the product's physical behavior while traveling through the pipelines. The model takes into consideration heat exchange with the surrounding environment, the physical properties of the pipeline and the physical properties of the transported products. The model thoroughly details the pipeline's present state under any operating scenario. Using the real time model as a platform for Ecopetrol's multiproduct pipelines the following applications have been configured: Leak Detection & Location, Shut-In Leak Detection, Batch Tracking, Scraper Tracking, and an Operator Trainer tool. Currently Ecopetrol has all of its 14 multiproduct pipelines configured and running with applications for Leak Detection, Batch Tracking and Scraper Tracking.

ABSTRACT In parallel with the design and construction of the Keystone pipeline, TransCanada and Telvent developed a fully integrated simulation environment including a dynamic hydraulic model, production SCADA system, simulated programmable logic controllers (PLCs), online leak detection, and batch tracking. This system served to test and develop control logic, the SCADA database, SCADA applications, online leak detection, and operational procedures prior to putting the pipeline in to production. Not only did this system allow Keystone to meet its regulatory training requirements, but it was able to predict interactions between field and control center pressure control systems, identify vulnerable areas of the pipeline during scenarios, verify over pressure protection systems, and allowed for initial tuning of the leak detection system prior to line fill. INTRODUCTION Typically the development of SCADA systems, online applications, control systems, and dynamic pipeline simulations are done as separate or sequential exercises. For existing pipelines this is often a product of technological development over the life of the pipeline. For smaller pipelines or new branches on larger existing lines, it is often impractical to invest in line-specific simulation or training systems. The Keystone Pipeline provided the unique opportunity of developing a full scope training, analysis and testing system for a large-scale green field project including all of these dynamic online systems. Through the course of the project a number of unexpected advantages were gained through what has been termed "full scope simulation." The Keystone Pipeline is a 1,854 mile (2,984-kilometre) crude pipeline beginning in Hardisty, Alberta and delivering to markets at Wood River and Patoka, Illinois and in the future to Cushing, Oklahoma (See Figure 1). It utilizes approximately 537 miles (864 kilometers) of existing gas pipeline and is designed to initially carry 435,000 barrels per day, which will later be increased to a nominal capacity of 590,000 barrels per day. The original purpose of this full scope simulator was to train operators on the specific dynamics of the Keystone pipeline; however, the core advantage of the system lies in the ability to dynamically test "live systems" in an offline environment that is nearly indistinguishable from the real world. Great strides were made in pre-tuning and testing of online leak detection, batch tracking, pressure control, batch cutting, swing logic and terminal control. In addition, engineering and support staff gained hands on training and experience with "live" systems while assisting in the development of the training system. The end result is a higher preparedness for operators, shorter commissioning times for online systems, and an efficient start up process for the pipeline. In the end this translates to an operation that is safer, more efficient, and more cost effective.

ABSTRACT Leak detection is critical to the safety of the environment and of increasing importance to those responsible and accountable for pipeline operations. The ability to detect and locate leaks in tight-line operation has been understood and advanced for quite some time. Slack-line leak detection, although acknowledged to be a more difficult task, is now more practical. Through the pipeline case studies presented herein, it will be demonstrated that it is quite practical to provide slack-line leak detection with reasonable thresholds and averaging times. The result is added leak detection protection for pipelines that must operate in slack or go through slack conditions during pipeline operation. As modeling performance improves and requirements become more stringent, the boundaries of what is possible and what is practical, regarding accuracy and resolution, are continually closing. This paper will explore the leak thresholds that are possible in slack conditions. To date, this area of leak detection has proven to be very difficult and, thus, ignored in most leak detection systems. Two pipelines, both located in the United States, are used as the basis upon which this analysis is performed. The first, ESMD, runs from East Sparta, Ohio to Midland, Pennsylvania and operates in slack line during start up and shut down. The second, BYST, runs from Byron, Wyoming to Silver Tip, Montana and operates in slack line throughout its steady state operation. The following topics will be addressed: Topic A sensitivity study for the parameters that becomes important during slack line conditions. Additional parameters become important when a pipeline transitions into slack line operation. These parameters have a distinct impact on slack line leak detection performance, being more significant than their impact to tight line conditions. Topic The determination of thresholds and associated averaging periods that are possible to achieve a balance between minimum thresholds and minimal false detections. INTRODUCTION Through empirical work, it has become apparent that the parameters that have been documented and accepted as forming the basis of tight-line leak detection are actually insufficient in leak detection operation in slack-line conditions. The analysis of two real-world case studies will clarify the additional parameters differentiating slack line leak detection. The following list attempts to identify these additional parameters and how they contribute to accurate slack-line leak detection. Elevation Effects Pressure Taps The elevation of the measurement of the pressure is important to leak detection in slack line conditions. The measured field pressure value is used to update the model pressure. The exact elevation of the pipe along with the pressure measurement is important because everyone (1) foot of error in elevation results in approximately 0.5 psi of error in pressure determination. In tight line leak detection this is not very critical because it results only in a small error in the pressure drop calculations, which can be easily compensated by tuning. However, in slack conditions, this elevation error directly affects two critical items: the determination of the onset of slack and

ABSTRACT The theft control is crucial for pipeline operators in many parts of world. The paper discusses the properties typical for theft-related leaks from liquid pipelines. Those are usually small, short-term leaks, with controlled flow. The technology and measures supporting the effective detection of such leaks are suggested. The paper documents real cases of product theft which appeared on 10"-12" pipelines, its reflection in pressure and flow trends, and the response of the leak detection system. The paper analyses the requirements put on all involved elements of technology from instruments, to telemetry, SCADA and finally leak detection methods; and touches also the processes involved in effective protection. The generally known software-based leak detection methods are analyzed in view of requirements put above. The combination of methods, with the emphasis on pressure wave analysis, is suggested, although the method limitations are acknowledged. The integration with batch tracking is suggested for exact determination of product properties in a batched pipelines. As a conclusion, the paper suggests the standard requirements to be put on a ‘theft-aware’ leak detection system, including speed of response, accuracy of localization, integration with batch tracking system, positioning of leak in a geographic information system, evidence to be generated and archived for further investigation, etc. Testing methodology proven in practice is suggested. The author is member of the pipeline modeling team in ttc, a Slovakia-based company focused in pipeline management systems. His background is in numerical mathematics. The company ttc is member of CSE Global Group. CHARACTERISTICS OF LEAKS RELATED TO THEFT The product theft from pipelines, in most cases known to author, appears as a product spill from an illegal tap, ½" - 2", with a valve to control the flow. The tap points may be in a block valve housing, on a bridge crossing or other non-buried section, or in proximity of valve stations, where there is easier for intruders to trace the pipeline in ground. Typical flow rate of leak is 2-10 m3/hr, unless it goes out of control, as when their equipment does not stand high pressure. Total size of one ‘leak incident’ is given by capacity of prepared storage, usually couple of thousand of liters. If it is a sophisticated operation, the leak is repeatedly closed and re-opened. TYPICAL LEAK INCIDENTS Type 1: ‘Offensive’ theft The violators try to reach their goal with a fast action, not taking much care for disguise. They don't expect to use their tap point repeatedly. Typical case reported from a pipeline operator is the intrusion into section valve housing through a hole sawn in the door, to not open the door limit switch, which would attract the attention of SCADA operator. The typical tapping point in the section valve housing is the degassing or sediment dumping valve.

ABSTRACT Present-day systems for modeling and control of gas transport and distribution commonly offer the Leak Detection and Leak Localization as one of their features. The function is usually based on running comparison of calculated and measured values. Sensitivity of Leak Detection, accuracy of Leak Localization, and speed of setting the above information up depend on the number and accuracy of the meters, their distribution over the pipeline system, and the current mode of transport. General specification of the sensitivity, accuracy, and speed in dependence on the above relevant parameters is not viable. Nevertheless, from experience is known that the state of the art in Leak Detection and Localization does not satisfy the requirements. Therefore, a new method has been designed that employs the effect of State Finder. For the purpose of Leak Detection and Localization, the State Finder was modified in a special way. The use of State Finder suppresses the negative influence of the inevitable errors of meters and thus remarkably sensitizes the Leak Detection, significantly enhances the accuracy of Leak Localization and considerably shortens the time for setting the alert information up. The modified State Finder is triggered periodically in frame of the cyclic simulation. Effect of the new Leak Detection and Localization Method is analyzed and demonstrated by a series of experiments on factual trunk line where no compressor station has flow rate measurement. BACKGROUND Leak detection systems for gas networks can be divided into externally based system (acoustic, fiber optic sensing cables…) and internally based system (balancing method, pressure/flow monitoring…). Below described method, based on the State Finder properties, belongs to internally based methods but does not require flow rate measurements in compressor stations. The state of network is described by a complete set of values for all process variables, namely pressures and flow rates, for the entire network. Naturally, because of the perpetual transients in the network, the state changes continually. State Finder (State Reconstruction) is a tool known from Control Theory. Employing the measured values and the mathematical model of dynamic processes running within the technical system, the State Finder computes an estimate of the actual state of the system. The ability to find the state of given dynamic system, on the basis of a set of values measured over a time interval, is called observability . Consider all non by-passed compressor stations, control valves, and closed valves being removed from the network. This divides the network in separated sub-networks. If all supplies and off-takes are measured, set points of nonpipe elements are known, and at minimum one pressure measurement exists in each sub-network, the entire system is observable. The sub-network in which the pressure measurement is for a longer time not available is not observable; the errors in flow rate measurements result in uncertainty of the in-take/off-take balance of the sub-network. Without the information about pressure the State Finder is unable to compensate the drift of system balance.

ABSTRACT Operation of the Huldra field commenced on November 2001. It is a gas condensate field, which has been developed with a not normally manned wellhead platform remotely controlled from the existing, manned Veslefrikk B platform. A first stage separation process is installed at the platform to separate gas and liquid. Two pipelines emanate from the Huldra field - a 93.2-mile [150 Km] 20" inner diameter pipeline transporting wet gas from Huldra to the Heimdal platform and an 8" inner diameter 9.9-mile [16 Km] pipeline transporting unstabilised condensate, including water, to the Veslefrikk platform. Due to the hydrocarbon mixture, the flow through these pipelines exhibits multiphase behaviour and therefore requires a multiphase real-time simulation model for surveillance and optimised control. Statoil was responsible for development of the Huldra field and is the field operator. The transient multiphase code OLGA has been an important tool in all phases of the project from the decision to develop the field and into the production phase. The design of the Huldra wet gas pipeline was a significant improvement in multiphase flow transport technology due to the long large diameter pipeline which should be tied-in to the existing Heimdal platform with restricted liquid processing capacity. For the condensate line efficient and good environmental hydrate and wax control methods are selected. This paper describes the requirements of the simulation engine and the pipeline modelling system. This leads to a clear split between the functionality that is necessarily encapsulated in the simulation engine and the functionality that can be incorporated within the wider On-Line System environment developed by Energy Solution International for gas pipeline modelling. The aim is to provide a common level of functionality throughout the system. Functions for multiphase flow pipelines, such as: model tuning, leak detection, liquid accumulation, and composition tracking need to be considered as well. A brief description of the approach to these issues is given. The strategy for operating the pipeline and its impact on extending the functionality of the existing OLS system is presented. This paper gives a brief description of the initial operational experiences with the realtime model. Page 2 of 30 1 Introduction In November 2001, the Huldra field commenced production. Design and start-up of the transport system for wet gas and condensate from Huldra is an extension of current experience. Statoil, as the operator of the Huldra field, awarded Gassco contracts for operational support for the condensate and wet gas pipelines. Valuable experience related to operational aspects and real-time modelling of large-scale multiphase flow systems has been gained after start-up of production. Huldra is a gas/condensate field located in the North Sea at a water depth of 410 ft [125 meters]. After a first stage separation the rich gas containing water and MEG is routed 93.2 miles [150 km] to the Heimdal platform for further processing.

The Caspian Pipeline Consortium (CPC) combines the cooperative efforts of the governments of the Russian Federation, the Republic of Kazakhstan, the Sultanate of Oman and a group of international oil companies including Chevron, Mobil and Lukarko. It represents one of Russia's largest-ever industrial development ventures, spanning over 1,600 km from the rich Tengiz oil fields in Western Kazakhstan, to those along the Caspian Sea and on to the Black Sea coast in Russia, with an eventual capacity of 1.34 million barrels per day. The pipeline runs from Tengiz (Kazakhstan) to Novorossyisk (Russia) with 5 pump stations and over 100 mainline block valves. The primary mission for pipeline design and control was to provide the basis for an efficient, safe and environmentally sound operation. Environmental protection being a key concern for this project, construction would adhere to the most stringent standards of the European Union, the World Bank and other international organizations. To meet these standards, the simulation model that would eventually be used to control and operate the pipeline was also used in the design phase for engineering analysis and optimization. The system needed to include comprehensive SCADA with fully integrated pipeline simulation, leak detection and the ability to perform predictive and look-ahead modeling of the pipeline. Applications also include Threshold Setting, Maximum Throughput Calculations Pipeline Inventory Management, Scraper Tracking, Accounting, Quality Bank, Off-line Planning and Forecasting and Operator Training and Certification. Industrial-quality reliability and redundancy was required for all critical communications, leak detection, control locations and servers. INTRODUCTION On July 11, 2002, the Caspian Pipeline Consortium (CPC) reached the target of six million tons of crude oil delivered to world markets through its pipeline system from the Tengiz oil field in Kazakhstan to the Marine Terminal near Novorossiisk. The six millionth ton of crude oil was loaded on the 11th onto the tanker "Katja" - the 60th tanker since loading of the first tanker, the "Minerva Alexandra", started the comprehensive testing phase of CPC-Russia's Marine Terminal on October 13, 2001. The "Katja" was being loaded simultaneously with another tanker at total loading velocity 20,500 tons per hours, indicating that the Marine Terminal had reached its full project capacity. ChevronTexaco chartered both tankers - the shipper is Tengizchevroil. In other developments, in late June CPC-Russia obtained its operating license from the Russian Federation Ministry of Transportation to conduct all loading-unloading activities at CPC's Marine Terminal. The license, renewable every five years, was granted without any limitations. Earlier, on June 14, 2002, the State Acceptance Commission signed the Act of Acceptance of the CPC Marine Terminal, which verified that all requirements for construction, safety and environmental protection had been met successfully.

ABSTRACT Energia Mayakan (Mayakan) is Mexico's first privately owned gas pipeline venture. This 432 mile 24" diameter pipeline transports natural gas from a PEMEX plant to several cyclic-load power plants in the Yucatán peninsula of Mexico. On-line leak detection, off-line predicting, and off-line training models were commissioned in the summer of 2000 and have proven successful in assisting control room operators to safely and efficiently operate this pipeline. Specific real-world experiences will be presented that demonstrate how the various simulation tools have benefited the operation. Mayakan's operators use real-time and predictor simulation models to plan compressor schedules, budget fuel usage, detect leaks, and recognize maintenance needs. The training model has been used to train and to certify all control room operators. This case history also presents lessons learned during the first year of operation, and uses of the model that had not been anticipated. INTRODUCTION Mayakan initiated commercial operation in September 1999 as the first privately owned pipeline to deliver gas to third parties in Mexico. A SCADA system was installed to monitor and control the pipeline. Computer modeling tools were configured to supplement SCADA and to provide leak detection, predictive modeling, and operator training. These tools are intended to provide support to the solo control room operator in making operational decisions and to promote safety, reliability and efficiency. This paper intends to present a description of the modeling system configuration and the benefits of using these tools in Mayakan during the first year of operation. Description of the models: Mayakan currently operates without any in-house engineering staff. Solo operators in the control room use the on-line model for linepack analysis, leak detection, SCADA instrument troubleshooting, survival time analysis, and predictive modeling analysis. They use a predictor model to make decisions regarding compressor schedules, fuel budgeting, and whether they will approve or deny customer's requests to increase their gas rates over the rates that have been previously allocated. The training model is used to educate and to certify Mayakan's staff. All three of the models are described in more detail in the sections that follow. Model Communication with SCADA All of the various models communicate with SCADA using bi-directional interfaces. SCADA data feeds the on-line leak detection model, whereas the predictor and training models can begin from either a current or saved on-line model state, or manually set conditions. All models can display their results using graphical display screens and also communicate modeling results back to SCADA. The on-line and predictor models send very few data items back to SCADA, whereas the training model, which simulates all pipeline hydraulics, sends hundreds of values to a copy of the SCADA system every few seconds. On-line and Leak Detection Model The on-line model has become an integral part of gas control. Continuous monitoring of pipeline hydraulics provides leak detection alarms, leak location calculations, and line pack calculations. The on-line model is also used to provide an evergreen starting hydraulic pipeline state for predictive modeling simulations.

Introduction The project being described in this paper is one of the first of its' kind in the world and is made necessary for the future operations of the pipeline based on continuing deregulation. The effects of deregulation in countries around the world are driving pipeline companies on a global basis to re-think the way they operate and do business. The emphasis of this project is on the integration of the business and operational worlds to come into a more realistic perspective of how the physical world can impact the commercial world. The implementation of pipeline modeling systems and business systems is a part of a project to provide a comprehensive real time pipeline monitoring, leak detection, and Nomination, Allocation and Billing System for the SembGas Gas receiving and distribution pipeline system in Singapore. The system that has been provided on this project is based on Energy Solution's, gas on-line modeling and leak detection system called LICLDS and the Nomination, Allocation and Billing System called NABS. Both LICLDS and NABS are implemented on a single PC server based on an Intel processor running Microsoft Windows NT 4.0. LICLDS is an enhancement to the functions provided by the SCADA system. LICLDS includes graphical user interface, which provides access to all major modeling functions. SCADA makes field data available to the operator at designated consoles. The LICLDS system provides additional information about real time and/or future pipeline conditions, thereby enabling safe and efficient operation of the pipeline. Certain data at various measurement locations along the pipeline is passed from the SCADA system to the LICLDS model via the use of an OPC server/client interface. The real time model uses this data to generate detailed information on the pressures, flows and other values along the pipeline system. The application software detects and alarms for leaks and uncommon operating conditions. The Predictive Model provides a projection of current pipeline operation to forewarn the operator of possible minimum pressure violations. Results from the different modules can be viewed through the Graphical User Interface (GUI) including certain trends and graphical profiles of selected pipeline values. NABS is a system for managing the business transactions between the Gas Supplier and End Users connected to the pipeline system. The End Users will submit nomination requests via a NABS client installed at each End User site to the NABS host system. The NABS host system consolidates these requests and submits nomination requests to the Gas Supplier. NABS passes the nomination forecast to the LICLDS Predictive Model for further analysis as described above. The Buyer's Facility begins at the seller's delivery point. Once in the buyer's facility the gas passes through three sections: Filters Heaters, and Metering section. The gas is normally delivered dry and free of liquids and particulates at 210C. and between 550 and 600psig (37.93 - 41.38 Barg). Once in the facility the gas splits into two 24" trains that pass through a filter or heater or flow meter.

1.0 INTRODUCTION This paper describes a Sour Gas Leak Detection System implemented by Novacorp International Consulting Inc.'s Pipeline Software Group. The emphasis of the paper is on the design considerations made in supplying a leak detection system for a gas pipeline system. The paper also presents the estimates of performance obtained from preliminary studies. The project's main goal was to provide reliable and accurate leak detection at low cost. That meant providing better accuracy than a simple volume balance at a cost considerably less than a transient model based leak detection system. A Compensated Volume Balance approach was selected since it is cost effective and sufficiently accurate when the pipeline is well instrumented. The pipeline's SCADA system design was not yet finalized so that additional pressure and temperature instruments could still be added to accommodate the leak detection system. 2.0 PIPELINE DETAILS 2.1 GEOGRAPHY The pipeline is located in a sparsely populated area of Central Alberta, Canada. It is close to a small town and a provincial campsite and crosses a river at one point. The effect of a prolonged leak of sour gas in this environmentally sensitive area would be disastrous. Therefore, the steps taken by the client to protect the public and the environment are commendable. 2.2 PIPELINE CHARACTERISTICS The pipeline is 44 miles long and 12 inches in diameter with a 0.35 inch wall thickness. Elevation of the pipeline ranges from 3250 to 4500 feet. Depth of burial is approximately three feet. The pipeline is extremely well instrumented with pressure measurements spaced from one to six miles apart. Flow, pressure and temperature are measured at the inlet and outlet ends of the pipeline with pressure and temperature measured at four intermediate block valves and pressure only measured at another eight intermediate block valves. The following instrumentations accuracy is assumed: Measurement Accuracy (%) Range pressure 0.5 0-2200 Psig NOTE: Accuracy is given as a percentage of full range. The SCADA system receives data fiom each of the RTU's along the pipeline every two minutes . 3.0 LEAK DETECTION METHODOLOGY 3.1 SIMPLE VOLUME BALANCE On the low end of the performance scale is the Simple Volume Balance leak detection method. In its simplest form leaks are detected by comparing the volume into and out of the pipeline over time. This method might be considered for an incompressible fluid such as crude oil but is highly insensitive for gas pipeline application especially in short time fiames. 3.2 TRANSIENT MODEL BASED LEAK DETECTION On the other end of the scale of leak detection algorithms is the Transient Leak Detection method. This method can provide a highly accurate means of detecting leaks in a liquid or gas pipeline. This method of leak detection comes in various forms but the basic principles are the same. The modelled state of the pipeline is used to determine the change in line pack between flow measurements.

ABSTRACT The Statpipe gas transmission system comprises totally 880 km offshore pipelines and an onshore NGL fractionation-plant. Gas from several fields located in the North Sea are transported in the system to buyers on the European Continent. Part of the gas is transported in dense phase to the NGL-plant for processing to sales gas and NGL products specifications. The gas transport through the system is monitored and controlled from a Pipeline Control Centre (PCC) located in Norway. There is implemented a real time simulator in the PCC SCADA-computer. SCADA data are used as boundary conditions for solving the transient flow equations of the real time simulator. In addition, several application programs are available for offline simulations. The present paper focus on operational use and experience with the real time simulator and applications programs. The Statpipe System The Statpipe system is a gas gathering and transmission system for transportation and treatment of gas from the Statfjord, Gullfaks and Heimdal fields in the North Sea. It includes 880 kilometers of pipeline, two offshore riser platforms and an onshore terminal for gas treatment and NGL extraction. The system was put into operation in 1985. Figure 1 shows the main dimensions and lengths. Associated rich gas is transported from the Gullfaks and Statfjord oil fields to the Kårstø NGL plant. The rich gas is transported in dense phase, i.e. at pressures higher than the cricondenbar (see figure 2). Typical operating pressures are in the range 120 to 167 barg. Design pressure of the pipeline is 172 barg. At Kårstø the heavier components of the rich gas is separated and fractionated into propane, normal-butane, iso-butane and natural gasoline. These products are stored in tanks and shipped from the gas terminal. The dry gas, mainly a mixture of methane and ethane, is sent through a pipeline to the riser platform 16/11S. At this platform the dry gas pipeline from the Heimdal field ties in, and the mixed gas is moving further to the Ekofisk Centre via riser platform 2/4S. From there on, the gas is transported in the Norpipe system to Emden in Germany, and distributed to consumers in Germany, The Netherlands, Belgium, France and Austria. The operational pressure range for the dry gas system is 60 to 116 barg. By installing additional compression facilities, the transport capacities can be increased. The design pressure for the Kårstø to 16/11S leg is 172 barg, and the Heimdal to 16/11S leg and 16/11S to 2/4S leg have design pressures of 156.8 barg. Statpipe Coordination The day to day operation of the Statpipe system is monitored and coordinated from a Pipeline Control Centre (PCC) at Karmøy, close to the Kårstø terminal, see figure 3. Information to/from the productions platforms, the gas terminal, the riser platforms and the Norpipe system operators are transferred via a telecommunication system to the control centre. This includes data, telex and voice traffic. The main goal of the operation is to meet the daily nominations of gas given by the buyers on the European Continent.