Pipeline &
Gas Journal
February issue 2000:
Next Generation Pipeline Monitoring
Shell Australia Experience
With Real-Time Pipeline Management Software
Ron Cramer, Marketing Manager,
Shell Services International (SSI),
Houston, Stephen Bailey, SSI,
Australia, and Douglas Robertson,
Applications Engineer, Simulutions,
Calgary, Alberta, Canada
Shell Australia recently completed the installation of a replacement Leak Detection System (LDS) on its two multi-product 8-inch pipelines running from the Geelong Refinery to the Newport Terminal. The project was managed by Shell Services International Inc. (SSI), which completed the installation within budget and in less than 12 weeks. The LDS has been successfully tested for leaks, and despite time restraints, has met, or exceeded all of the stated user requirements. Local knowledge and project management provided by SSI Engineering, Australia was key to the project success. SSI familiarity with the original and new instrumentation, hardware and software configuration facilitated speedy and seamless installation of the LDS system. Pipeline operations were maintained with no operational upsets and full compliance with regulatory license requirements, despite the fact that a new control system was being commissioned at the same time. Another key aspect to the success was user acceptance of the new LDS system—this was fostered by provision of training and support by local SSI staff.
The Geelong—Newport pipelines transport 16 million barrels per annum of product comprising some 20 different grades. The operation is continuous with operational changes ranging from minutes to days. The supervision and control of the pipeline is performed from the control room in the Newport Terminal. From there, the operator has control over all aspects of the pipeline except the pumps at the Geelong refinery. However, the controller is able to shut in the pipeline by remotely closing the valves at both ends. As well, the operator has access to a variety of measurements at each end of the pipeline. These include mass flow, pressure, density and temperature. Flow measurements are obtained via MicroMotion mass flow meters and density measurement is by Solatron/ Schulmberger. This information is supplemented in a variety of ways by the new LDS, including a real time hydraulic profile of the pipeline, batch inventory and position, leak detection information, and operational information. The LDS is a product known as LeakWarn© from Simulutions of Calgary, Alberta, Canada.
The SCADA system being used on the pipelines was a Wonderware Intouch system. It polls real time data and supplies it via DDE to the LeakWarn©, which performs leak detection, pipeline monitoring and batch tracking. SSI performed the system integration with technical assistance from Simulutions.
How It Works
LeakWarn© uses a real time transient model and a modified mass balance system to provide a suite of services designed to assist both in leak detection and safe, efficient pipeline operation. It avoids many of the extra complexities of other models while retaining the leak detection performance of more complex models. This results in faster configuration and installation, lower maintenance costs and greater reliability. This enabled the vendor to meet the three-month time limit and keep costs within budget requirements.
To aid in leak detection, the system maintains a full pipeline inventory and tracks it during operation. It also calculates a hydraulic gradient using the measurements available on the pipeline and the pipeline inventory. All of this information, measured and calculated, is made available to the operator through a Graphical User Interface (GUI). Through the GUI, the operator monitors all aspects of leak detection, batch tracking and pipeline monitoring. For leak detection, the operator has several displays to choose from.
One of these is Simulutions’ proprietary Signature Plot©, (Figure 1). This display provides an X-Y plot of the line pack change vs. volume exchange in a section of the pipeline. If the plot of these values remains on the 45° line, the pipeline is operating correctly. As the line moves upwards and to the right, the operator knows that the amount of inventory in the line is increasing (packing). Similarly, if the plot moves down and to the left, the inventory of the pipeline is decreasing (unpacking). If the line starts to move off the 45° line at an angle, this is indicative of abnormal operation such as leaks and instrument failures. In conjunction with the Signature Plot©, the operator can also view a histogram showing the status of each of 12 individual time windows. As the leak progresses, the bars (which show imbalance) increase in height, indicating an abnormal operating condition, (Figure 2). The leak warning and alarm thresholds are visible in both these windows. The yellow line represents the warning threshold and the red line represents the alarm threshold. The purple line visible opposite the red and yellow lines is the mass negative alarm line. When this threshold is exceeded, it indicates that there is an overage in the line. This means that the meters are showing more coming out of the line than coming into the line. This is usually due to meter failure or product error.
The operator can also view a complete hydraulic profile of the pipeline via a profile window, (Figure 3). With this display, the operator can track a variety of product properties including density, temperature, viscosity and many others. As well, a pressure profile, flow profile and mass flow profile are available to assist the operator in monitoring the operation of the pipeline. Finally, a color bar at the bottom of the profile window provides a color-coded list of batches in the pipeline and their position. Each color indicates a different grade of product. These colors are configurable and can be set to mimic a color coding system already in use.
During the installation and tuning phase of the product, the system demonstrated the power of its inventory tracking. During a batch change on the Black Oil Pipeline, the newly started batch was contaminated by an amount of product in a pump being started. The operator was not observing the LDS during the startup and missed this. Later, the operator discovered that LeakWarn© had inserted a second batch into the system based on the density variation. Upon inspection of the density profile and the density trend, it was determined that a small amount of off-grade product had entered the line and contaminated the current batch. Using the ETA provided by the new system, the operator was able to plan for the arrival of the off-grade product so it could be diverted to a different tank. Thus, avoiding contamination of a full tank of product.
The model is also able to maintain leak detection through operational transients where leaks can occur due to increased stress on the pipeline and associated equipment. It utilizes change in line pack and volume exchange to reduce traditional sources of error inherent in line pack calculations. This results in its ability to detect much smaller leaks in shorter time periods. Along with this enhanced line pack calculation, LeakWarn© uses a multi-window leak detection methodology that recognizes that smaller leaks take longer to impact pipeline operations, and therefore, take longer to detect. Each window accumulates imbalance over different time periods and can be adjusted individually to account for residual imbalance in the pipeline and obtain the required leak detection accuracy. Imbalance being the difference between measured volume exchange (flow in minus flow out) and calculated line pack (inventory) change. Once the imbalance in a window exceeds the set limits, a leak is presumed to exist.
The system also uses a variety of techniques to reduce or eliminate false alarms. A false alarm is defined as a leak alarm when no leak actually exists. This results in increased operator trust in the system and results in improved performance from the utilization of the additional information provided by the model through the GUI. These techniques include dynamic thresholds and line pack change filtering. As the pipeline enters an operating transient, the model can increase the thresholds slightly on the windows to allow for any increased uncertainty in measurements or product behavior. An exponential filter is also employed to adjust the calculated change in line pack due to operational change, to match the observed behavior of the pipeline.
All of these techniques allow for the adjustment of the model to allow for a finely tuned compromise between fast detection of larger leaks and accurate reliable detection of much smaller leaks.
Historical Database
The system maintains a historical database of the operation of the pipeline. This information is used to provide accurate estimates of the location of leaks as well as estimates of leak rates. By providing the operator with an estimate of the location of the leak, it minimizes the response time of the operator to the leak, as well as the deployment of leak containment teams and negative impact of any leaks that may occur.
This historical information is also logged for archiving. It can be replayed through the model or extracted and trended.
As mentioned previously, LeakWarn© maintains a profile of pipeline inventory. Its additional Batch Tracking module was included to track inventory and different product grades in the pipeline. By utilizing indications of the volume and product entering the line, the system calculates the position of all batch interfaces in the pipeline. It also provides estimated arrival times for each interface that can be used to plan the operation of the pipeline. The interface positions, batch volumes and arrival times are calculated and updated each cycle. To combat errors in batch information due to instrument failure, the system provides automatic and manual modification to batch interface positions, batch IDs and product IDs. The batch-tracking module operates independently of the real time model, so changes in the batch information do not affect leak detection.
System Leak Tests
After the installation phase of the product, testing was conducted on the system to ensure that the stated requirements had been met. Several live leak tests were conducted on the system using simulated leaks. Tests were conducted on both of the Black Oil Pipeline BOPL and White Oil Pipeline (WOPL), running from Geelong to Newport.
Two tests were conducted on the BOPL by connecting a road tanker to an off take valve approximately 22 miles (35 km) from Geelong. The first test was at a flow rate of 113 barrels per hours during a pipeline flow rate of approximately 755 barrels per hour. The LDS indicated a leak warning three minutes after the leak began and alarmed one-minute later. Upon detection of the leak, LeakWarn© calculated a leak rate of 138 barrels per hour.
The second test was with a leak rate of 19 barrels per hour, again during a pipeline flow of 755 barrels per hour. This time, the system entered a warning state after 22 minutes and alarmed two minutes later. Upon detecting the leak, the system indicated a loss rate of 19 barrels per hour. During both tests the loss rate and time to alarm were well within customer requirements. The tests also demonstrated that the system was performing within the stated user requirements. Even though the leak location component of the software had not been configured at the time of the first test, it had been configured for the second set of tests on the White Oil Pipeline.
The test on the WOPL was conducted with a “leak” rate of 75 barrels per hour with a pipeline flow rate of approximately 1,384 barrels per hour. The system alarmed after six minutes and calculated a rate of 62 barrels per hour. Concurrent with this, the software activated and calculated a leak location 29.6 miles (47.6 km) from the Geelong Refinery. The actual location of the delivery on the WOPL was approximately 30.1 miles (48.5 km) from Geelong, an error in location of only +/- 2% of overall pipeline length.
Conclusion
With the installation of this new LDS on their Geelong to Newport multi-products lines, Shell Australia has moved into the next generation of Leak Detection and Pipeline Monitoring systems. The new system not only provides highly reliable leak detection, it also provides a suite of additional tools to aid operators in monitoring the pipeline and making good operational decisions. P&GJ
The authors: Ron Cramer works on project managing and marketing systems for oil field automation and management at Shell Services International in Houston. Previously, he worked for Union Carbide and Polysar in research and process areas, and for Shell International in the areas of oil field automation and production systems. Cramer holds a BS degree in chemical engineering from Strathclyde University and an MS degree in chemical engineering from Waterloo University.
Stephen Bailey graduated in 1989 with a BE in Engineering and Electronics. He joined Shell as an instrument engineer at the Geelong refinery in Australia where he was responsible for design, contract management, construction supervision and commissioning of projects up to $4 million. In 1994, he took the position of Instrument Electrical Engineer in Marketing Engineering. In this position he provides on site support to field project engineers and is responsible for the development and maintenance of standards and procedures in the instrument and electrical engineering fields
Douglas Robertson obtained a Bachelor of Applied Science Degree with Engineering Physics/Computer Science from the University of British Columbia. In September 1998 he joined Simulutions where he has been focusing on model design and maintenance, site installation and tuning, and product development. P&GJ