Integrating GIS with risk assessment
Rich McGregor Great Lakes Gas Transmission. Troy, MI John Beets M. J. Harden Associates, Inc. Kansas City MO Greg Morris Kiefner and Associates, Inc. Worthington, OH Introduction Geographic Information Systems (GIS), if implemented correctly, provide a solid information management base for pipeline companies to undertake myriad applications in which geographic location is a basic component. Examples include basic query analysis, alignment sheet generation, facility maintenance, tax analysis, and risk or integrity management related functions. With the increased regulatory emphasis on pipeline integrity, operators are under pressure to find better, more efficient ways to safely operate their pipeline systems. One of the most important applications for this system-wide high-resolution data is risk assessment. This paper outlines the experiences of one operator in implementing a new software application that uses GIS information to conduct risk assessment. The GIS Driven by the typical motivations Great Lakes Gas Transmission Company (GLGT) began to develop their system-wide GIS application in 1995. The motivations for undertaking the effort were typical of other companies in the industry, which include:
GLGT had spent considerable time and resources researching the experiences of other operators in the development and implementation of GIS applications and had developed a good understanding of the ways in which the GIS could be used to assist other work functions within the company. Since this research was done before the GIS was developed, modifications were incorporated into the database and/or downstream applications at the outset of the project rather than later, when the costs of these changes could compromise the project. This is not to say that there were not hurdles to overcome as the project proceeded, but simply that through planning and awareness, the number of setbacks that were encountered was greatly reduced. The GLGT pipeline system is relatively new, with the first of its pipelines constructed in 1967. Because of its age and the availability of records, a vast amount of data about the pipe material, the construction practices, the hydrostatic test history and service performance were readily available. GLGT recognized the value of these data and sought a system which could best utilize what they had available to them. Prior to the development of the GIS, GLGT had begun to collect and store pipeline data electronically. A number of downstream applications were researched, including risk assessment, to determine the type of data often used in such applications. A suitable format was selected for each data set, consistent with the intended use or uses of these data, before existing data were migrated into or new data were added to their GIS. Built with the ISAT model database design The GLGT GIS application is based on the ISAT data model. The ISAT data model was developed under contract for the Gas Research Institute (GRI) by M.J. Harden Associates, Inc. and representatives from numerous operating companies. The ISAT data model represents an industry consensus data model that can be used as the core structure for a GIS. Since its creation in 1994, the ISAT data model has been used by a number of different operators for GIS. GLGT used the ISAT data model as the core database, but customized the database design to include additional data tables used in other downstream applications, including risk assessment. Risk assessment model The early model GLGT began development of an in-house relative risk ranking model in 1998. The primary purpose for developing the model was to use it to prioritize sections of the pipeline for maintenance. The initial model consisted of 11 variables describing primarily 3 failure modes and 3 consequence components. Data representing the pipeline attributes for each segment were collected and entered into an Excel spreadsheet. This spreadsheet contained the algorithms required to exercise the model and produce output. GLGT divided the pipeline into predefined segments for conducting risk assessment. A “segment” is defined as a portion of the pipeline system that can be treated as a single unit because all of the pipeline attributes (or at least those used in risk assessment) are the same throughout its length. In the initial model, GLGT chose to treat each valve section as a segment, a practice common in the industry. Valve sections represented a convenient breaking point and resulted in manageable segment lengths ranging from a few miles to almost twenty miles long. Thus, the pipeline attribute data in the risk spreadsheet was used to represent each valve section in the system. A redesigned model The risk spreadsheet had been in place only a short time before GLGT assembled a team of consultants to assist them in revising the model. The team consisted of pipeline specialists from Kiefner and Associates, Inc. (KAI), M.J. Harden Associates, Inc. (MJH), and CC Technologies, Inc. (CCT). CCT focused on the Corrosion Probability and the Stress Corrosion Cracking (SCC) Probability algorithms, while KAI focused on the comprehensive model, which included probability algorithms for other failure modes and consequences. MJH installed a new software program, PipeView Risk, to replace the risk spreadsheet. The PipeView Risk software was chosen because it was designed to interface directly with GIS as well as other databases for the purposes of conducting risk assessment. A number of significant benefits were realized with the implementation of the new risk assessment model and new software, including:
Data Processing The risk assessment software reads data directly from the GIS and other stand-alone databases to extract all the data required for risk assessment. All the data is referenced by a common location hierarchy, namely station series and stationing. The risk assessment software performs a true dynamic segmentation of the data (described in more detail later in this paper) and performs all of the calculations described in the risk assessment algorithms. The results are stored in a separate risk database used by the risk assessment software. The separate risk database allows the user the ability to reprocess or even temporarily change the data when performing what-if scenarios without compromising the integrity of the actual data. The risk results can in turn be accessed by a GIS viewer type application and displayed side by side with GIS data or on maps, drawings or alignment sheets. Specialized risk analysis and output capabilities Dynamic Segmentation Dynamic Segmentation is a term that is used often but has different meanings for different users. Dynamic Segmentation, as it is used here, is an automated process that uses high-resolution facilities information to define segments of like pipe, i.e., lengths of pipe where all the pipeline attributes (or at least the ones used in the risk model) are the same throughout its length. GLGT recognized the limitations of using valve section segments, namely that they are too large and encompass too many different pipeline conditions (pipe types, soil environments, or operating parameters) to be accurately characterized by one set of pipeline attributes. Many of the valve sections would be more accurately modeled as several smaller sections, due to changes in pipeline attributes within the valve section, such as wall thickness, class location, terrain, soil conditions, or other factors considered in the risk model. The difficulty in using large, predefined segments is how to describe specific pipeline attributes accurately. For example, how do you classify the soil type and the effectiveness of the cathodic protection if 3 miles within a 12-mile-long valve section are in a wetland where conditions are good, but the balance of the segment is in rocky soil where conditions are not as good. The dynamic segmentation process would solve this problem by making at least three segments out of the valve section: one segment representing the wetland, and two other segments representing dryland portions upstream and downstream of the wetlands. This is an illustration of just one of the many types of issues that must be considered when choosing a segmenting strategy. Dynamic segmentation is performed by systematically processing data according to the stationing values, denoting points at which pipeline attributes change. Stated in simplified terms, the software starts at the beginning of the pipeline and reads the initial set of pipeline attributes (variables). It then ‘moves’ down the pipeline checking all variables until any one of them changes. Upon detecting a change, the program defines the endpoint of the first segment and starts a new segment. The process is repeated along the entire length of the pipeline, or more accurately stated, the process is conducted through the entire database. Dynamic segmentation can be performed on databases that contain pipeline attribute data of different resolutions, i.e., both valve section resolution and higher resolution. Dynamic segmentation performed on a database of valve section resolution data will return only the valve section segments. However, when at least one pipeline attribute (or better yet more than one) is defined at a resolution higher than the valve section level, dynamic segmentation will return smaller, more accurately defined segments that will enhance the risk assessment. Operators usually have data at different resolution levels, typically the valve section level and higher levels, such as start to end survey station locations. In the GLGT system, dynamic segmentation of the high resolution GIS facilities information resulted in a number of segments, allowing them to more easily isolate areas of higher risk. POE analysis In-line inspection results can be ranked and prioritized using a statistical or probabilistic approach as opposed to simply ranking the anomalies reported by the vendor by either depth or predicted failure pressure. The approach, known as Probability of Exceedance (POE) analysis, has been adopted by several operators and its use is increasing over time. The POE analysis methods evaluate the probability that an anomaly of any indicated size could be large enough to threaten the integrity of the pipeline. It also accounts for the sizing inaccuracy or bias in the tool. The POE approach can provide the operator a number of advantages over the typical schemes for prioritizing anomalies, not the least of which is a method to evaluate the benefits of further excavations versus reinspecting the pipeline. POE also provides a means for summarizing in-line inspection (ILI) results over a segment of pipeline. By applying a conservative corrosion growth rate to the ILI results, future maintenance and inspection activities can be evaluated and planned. In order to conduct a system-wide POE analysis in conjunction with dynamic segmentation, location information had to be assigned to each anomaly in the database. The POE results were used to calculate a cumulative POE value for the segment, a single value that is based upon the POE values of each anomaly located within the segment. The cumulative POE calculation had to be performed after dynamic segmentation, which defines the endpoints of the segments. The risk software automatically performed the dynamic segmentation and extracted the POE values of each anomaly located within the segment to compute the cumulative POE value. The actual ILI data, including depth, length, location, run date, and other information, are stored within the program and can be used along with POE results in what-if scenarios. What-If Scenarios Aside from calculating risk index scores for segments, the risk assessment software is also used to evaluate the effectiveness of various maintenance alternatives. Maintenance alternatives are simulated by changing the appropriate pipeline attributes, or variables, and recalculating the risk index scores for the segment or segments involved. These recalculated results, along with some fundamental cost information, can be used as the basis of a cost-benefit analysis, allowing the operator to find the most effective mitigation alternative for the lowest cost. The what-if analysis can range from relatively simple to complex. An example of the more complex analyses performed in the risk assessment software are the POE scenarios which involve modeling the excavation and repair of specific anomalies and the impact of these activities over time. Other relatively simple scenarios involve simulating a hydrostatic test or recoating project. Future goals Data integration The pipeline attribute data is stored locationally in the GIS by station value, and the risk assessment software that reads and processes this data maintains the location information during the risk assessment processing so that the segment locations and risk assessment results can be used with the GIS display tools. M.J. Harden is developing a GIS viewer application which will allow GLGT to view the risk assessment results graphically, overlaid on a map of the system. The display tool will allow the user to turn on certain features of interest, such as rectifier locations, class location, structures, or any other data set used in risk assessment or contained in the GIS. Currently, several operators, all using the ISAT data model, are working together to standardize a format for storing pipeline integrity-related data. One of the long-term goals of this effort is to be able to develop software applications that seamlessly interface with the integrity data. Applications will include risk assessment, integrity-related alignment sheet generation, and a GIS viewer, among others. GLGT is working with this group to set the standard for data integration techniques that will optimize the use of GIS and integrity-related information to improve safety and reliability. | ||
|
|