Geography Provides Many Advantages
Over the past 30 years, technology has revolutionized the pipeline industry. We moved from total stations to GPS for survey activities, paper-and-pen field data collection migrated to mobile devices, and generation of alignment sheets is now completed by automated processing. While we have implemented these changes and many others, we continue to use the same system for defining the position of pipeline and inline assets: stationing.
Geographic information system (GIS) technology is now prevalent in the pipeline industry. The foundation of GIS is geography. If you operate a pipeline, it is essential to know each asset’s spatial location and identify geographic relationships. Without geographic information, it is virtually impossible to satisfy U.S. DOT operating requirements, define class locations, identify HCAs, manage pipeline integrity, conduct risk assessments, improve performance, or ensure public safety. Pipeline stationing is not a geographic location. It is a measure, similar to a house address. It seems better suited to be an attribute. Yet the industry insists on using stationing to locate assets even though pipeline companies operate geographically and maintain that it's too risky or costly to change.
Rob, I agree. As we move forward, we need to place a new focus on the spatial component of the data and view the station values as simply an alternate address and measurement of accurate pipeline lengths. Until we have the ability to more accurately model the pipelines in 3D, this station value will still be critical for ordering events and the measurement of pipeline length. Stations won’t go away anytime soon, but we can start placing less emphasis on them for the spatial positioning of events along the pipeline segments. Our current industry models place too much stock in station values, so we should start there by changing that thinking at the core of these systems
Rob,
Really glad to see this topic…I am also in agreement that stationing is more of an attribute with analytic value and not necessarily the foundation for pipeline spatial data management.
Networks, Topology and Spatial Relationships are truly the path forward for transitioning from a linear environment to a real world system of measure. As Jeff Allen stated "station value will still be critical for ordering events and the measurement of pipeline length." as we do not have a better method to maintain a chained distance and offer broader versatility yet. I am confident that we will see some very impressive innovations in the near future as this topic brings to light various arguments from both sides.
The question of maintaining historical stationing is one that we face frequently with our clients, and we often find ourselves in a position of justifying the relative complexity of implementation and maintenance of linear reference system (LRS) models in the industry. There is strong industry interest in continuously improving the accuracy and completeness of pipeline asset data in support of normal operations, maintenance, risk assessment, and integrity management. Maintaining historical stationing can make all these operations significantly more difficult, especially considering that much of the new data that can be leveraged for improvement is based on GPS measurements in conjunction with field surveys and inspections of all types.
GIS is commonly accepted as a best practice in the pipeline industry, but many implementations simply retrofit the GIS to historical methods of storing and managing pipeline data and often dramatically underutilize the data integration and analytic capabilities of the technology as a result. Many operators have experienced great success with a different approach, maintaining the historical (engineering, noncontinuous) stationing as a type of attribute while fully leveraging the GIS to model and analyze data that may be best represented and managed as GIS features. This approach can take advantage of superior data management capabilities of the GIS while easily calculating stationing references as needed for applications that require or benefit from linear referencing.
However, one nice derivative of linear referencing systems/stationing is the built-in spatial integrity for coincident online spatial features. For a transmission pipeline with physical components (coating, valves, tees, etc.) and abstract designations (DOT Class, Risk Scores, MAOP, etc.) that will always fall directly on that line, stationing references provide a proven method of storing and maintaining this data. A complete GIS implementation may also represent this data as persistent features, providing great flexibility for use and analysis within the GIS and by applications that are not spatially enabled.
Maintaining stationing has great value in supporting the integration of historical data that may only be referenced or located in this manner, but using it as a primary definition of the pipeline and online features is no longer the only viable option. If we begin to consider stationing as a type of attribute, we will have many new opportunities to increase the utilization and value of the data that we manage, improve data accuracy and quality, and simplify the ongoing management of the pipeline asset database.
Based on experience with pipeliners, it appears that stationing will remain a component of pipeline data and applications for some time. However, I do agree that technological advances will continue to change how operators manage their facilities and their data.
The maintenance of station value(s) on pipeline features enables so much of the required linear analysis. Station values, both as attribution and measures, serve to benefit, not replace, the increasingly beneficial use of the geodatabase with spatial features which replace the traditional pure event driven linear referenced data storage data models.
Chaining and station references will likely be used less in the collection and re-location of features and activities in the field. GPS will continue to provide field staff easier access to more precise geographic locations. Mobile and desktop GIS software will translate our field determined geographic locations to online features with their own station value(s) and back again to coordinates for use in the field, again aided by GPS. High resolution terrain models produced with LIDAR permit quite good estimates of slope distances reducing the need to drag chain over miles of pipeline routes. Technological advances serve to propel the evolution of pipeline data processes, improve practices and make processes more cost effective.
In the end, workflow processes can only change at rates tolerated by the operator’s staff. The degree to which an operator's legacy data must be maintained alongside or integrated with the GIS will also influence change in workflows. Pipeliners trust their slack chain stations, paper alignment sheets and spreadsheets of stationed ordered facility inventories. They are often wary of a pure GPS/GIS solution. That in itself ought to prove that it is too risky to stop using stationing, at least overnight.
Rob, the fact is that stationing should be optional for pipeline operators. The GIS should support stationed and nonstationed pipe. We currently have two ways of modeling pipelines in GIS—one where stationing isn’t supported at all (or very well), and the other where stationing is fully supported and is basically required. Looking ahead, one GIS system should be able to fully support linear referencing but not require it everywhere. This requires a more flexible approach to modeling pipeline data.
I think it is important to differentiate stationing from linear referencing—they are different. Linear referencing is simply assigning a linear distance (measure) along the pipeline at each x, y, z vertex. It doesn’t use series or station equations. Linear referencing allows features to be located geographically and by linear distance. Linear referencing is a valuable tool for reporting, analysis, and data integration. Looking ahead, I can see an increased need for linear referencing in pipeline GIS.
Support for stations can be implemented by using linear referencing or not. Most transmission pipe today uses stationing with linear referencing. Your suggestion is to treat stationing simply as an attribute. For operators that really use stationing, that doesn’t solve the root problem. Stationing is not going away anytime soon for most transmission operators, so until then, it needs to be supported.
This topic is more pertinent for distribution and gathering pipeline operators since it is more common to find hybrid stationed and nonstationed pipe here. In these systems, the GIS should ideally support linear referencing and allow the operator to use stationing when they need it, but not require it for nonstationed assets. This would allow stationed and nonstationed assets to be managed in the same GIS, provide the benefits of linear referencing for pipeline integrity management, and allow operators to transition away from stationing if and when they choose to.
For the purposes of systems already functioning, assets, facilities and other such sites are based on linear referencing and some with spatial attributes. Sometimes linear referencing produces variances in measurements such as when there is a requirement for replacing pipe and stationing equations to accommodate this. Simply dropping stationing off pipelines would cause a lot of expense on the side of the operator, because whatever system (I personally would like to further develop the 3D modeling method) replaces traditional stationing, the amount of re-training employees, re-aligning massive amounts of data and new methods of collecting and maintaining data will require new equipment. In our current economic climate I do not believe this is going to be a viable alternative at the moment. Therefore I do not believe stationing will disappear, the most reasonable alternative would be to maintain two data-sets - each tied to each other - one maintaining stationing series, and second a x,y,z or spatial method. In addition one must consider the fact that some data is from as-the pipe lies area (meaning captured from a Geo-pig) or as-the-surface lies, meaning a surficial surface. I hope this makes some sort of sense! I appreciate any feedback.
Rob’s comments with respect to GPS are right on the button. With the proliferation GPS-based field data collection, locating features based on stationing is becoming increasingly anachronistic. Additionally, gathering and distribution operators have, for the most part, never relied on stationing. As regulatory pressure increases on the gathering and distribution communities, it is clear that ESRI and its business partners must find ways to support pipeline users who have no desire or direct use for stationing.
This is not to say that linear referencing technology has no utility. As others have pointed out, linear referencing technology shines for pipeline analytical purposes. Linear processing using SQL that makes use of stationing attribution is in most cases considerably more efficient and reliable than spatial processing. This is easily demonstrable by comparing the results and performance of a linear Intersect operation with that of an Overlay Route Events operation. Others have also correctly observed that for many transmission operators, abandoning stationing would be prohibitively expensive in terms of staff retraining, business process realignment, and data migration. So from both an analytical and business standpoint, stationing remains relevant.
For the foreseeable future it seems we must be able to have our cake, and eat it, too. From a practical standpoint, this means retaining linear referencing, but making it completely transparent (especially for gathering and distribution users). To facilitate this, it is necessary to make M values easier to generate and access. The Linear Referencing Tools toolbox is a good first step, but ESRI could help things along with several targeted enhancements:
1) Enhance the Create Routes tool to:
a) Use 3D LENGTH as the Measure Source for PolyLineZ input features that have Z values. This allows users to easily define the linear referencing system on the basis of 3D length. This is most desirable from a pipeline user’s standpoint.
b) Add DIGITIZATION_DIRECTION as a Coordinate Priority parameter. When using shape length to define measures, this would be far more useful than the existing UPPER_LEFT, … LOWER_RIGHT parameters. Currently digitization direction is only used when specifying both From-Measure and To-Measure fields. This is a Catch-22; see item 2).
2) Enhance the Calculate Geometry… tool to extract M and Z data in addition to X and Y data from M- and Z-aware shapes. Having created routes using shape length, it would be nice to be able to easily calculate begin and end M values.
3) Enhance the Locate Features Along Routes tool to optionally add the Output Event Table fields directly to the input feature class and populate these columns with data for the closest route location.
4) Enhance the Add Route Event… tool to utilize feature classes as input data sources in addition to tables.
Items 3) and 4) allow us to easily calculate M values for near-line, offline features, and then create online representations of such features as route event layers. Such functionality would be invaluable because, as most of us have discovered, field-collected GPS points for online features almost never fall directly on the centerline stored in the GIS.
Finally, as Ron has pointed out, stationing and linear referencing are not quite the same thing. Real-world pipeline stationing is more complicated than linear referencing as implemented by ESRI. Linear referencing technology in ArcMap does not allow linear events that span across spatially connected routes. However, as every pipeliner knows, it is quite common to have linear events that span station equations (e.g. MAOP ranges, HCA’s, etc.).
The designs of both the Pipeline Open Data Standard (PODS) data model and the ArcGIS Pipeline Data Model (APDM) incorporate elaborate workarounds for the events-spanning-routes limitation. For instance, PODS explicitly implements two systems of measurement, one based on Route and Measure (for linear referencing), and one based on Series and Station (to support traditional stationing). These workarounds are cumbersome to maintain and confusing to the average user. The APDM 5.0 (and PODS ESRI Spatial) attempt to simplify the situation via extensive data model denormalization, but these efforts remain workarounds.
Interestingly, Oracle Spatial 11g has introduced new “subpath” functionality that effectively allows linear events to span routes. See: Oracle Spatial Topology and Network Data Models Developer's Guide 11g, Section 5.3.1 Subpaths. This new functionality is based on a marriage of Oracle Spatial linear referencing and geometric network technology. It seems that ESRI, as the world leader in GIS technology, might want to follow suit. It would be extremely useful to be able to define route events that span complex edges in a geometric network.
I can not imagine a pipeline operation without stationing. However, another truth is that I have encountered so many positional uncertainties and plain errors when pipeline features were located based solely on stationing that I am convinced that whenever possible the GPS coordinates should be used as a primary method of defining spatial locations, with reference to stationing.
To use or not to use stationing – does it need to be regulated or otherwise imposed on pipeline operators? If stationing outlives its usefulness, who would insist on using it?