A Pipe Network Data Model for Defect Tracking

A Pipe Network Data Model for Defect Tracking

Paul C. Marsh
Geomatics Engineering Manager and Mike Garaci, Senior GIS Analyst
The Pressure Pipe Inspection Company 4700 Dixie Road
Mississauga, Ontario L4W 2R1 Canada

Abstract:
A typical Pipe Network model is composed of nodes and links for contiguous waterlines; however, current inspection technologies collect more detail than this traditional model can present. This paper outlines a new hybrid model approach for managing the hydraulic requirements of water systems with inspection and defect tracking activities. Using UML and starting with the ArcGIS water model prepared by ESRI, the hybrid model tracks individual pipe segments with inspection and integrity information while still maintaining the hydraulic modelling capability.

New inspection techniques offer pipeline operators to grow beyond direct condition information and map placement of assets. As an example, the RFEC/TC technology for PCCP pipe offers utilities information not only on a per pipe segment basis but also at sub-regions along each pipe. Pipe failure modelling, risk to operations management, and cost analysis involve more detailed, and direct information leading to better investment decisions. The hybrid model displays data in plan and profile views. The plan view of a typical water network drawing shows the horizontal position of links and nodes and integrates with hydraulic analysis programs. An associated profile view, similar to as-constructed drawings, presents the individual pipe segments and their related defect information.

The new model lends itself to numerous other data presentation options to highlight asset tracking, data trending and analysis.

Introduction
Traditional Pipe Network Models use a node and link database to represent Water Conveyance Systems. Links are water mains that convey water from one point, or node, to another point, or second node. Thus, the atomically defined database elements are Waterlines from start to end and nodes. Nodes are typically anything that controls the flow of water or changes the properties of the waterline, e.g. a coupling between waterlines of differing materials. Changing Requirements This node link data structure was suitable for collecting water related infrastructure asset management data. Examples include fire hydrant testing results to establish C factors for waterlines. In addition, this data structure is good for water quality data from water monitoring at consumers and from within the distribution system. Water supply capacity planning is particularly well suited to this data structure as hydraulic analysis software uses the same data structure.

Recent advances in non-destructive testing of pipes have begun to collect greater amounts of and more detailed data than is suitable for this traditional data structure. These technologies include; CCTV inspection of Wastewater pipes, magnetic flux leakage to determine pitting and Remote Field Eddy Current/Transformer Coupling, (RFEC/TC). For the purposes of our discussion, we will focus on RFEC/TC inspections and the types of data produced. We will also limit our discussion to the application of RFEC/TC inspections to Pre-Stressed Concrete Cylinder Pipe or PCCP.

A RFEC/TC inspection is performed from the inside of a pipe by a remote unit that is operated by staff who ensure that the equipment is properly collecting data while it is propelled through the pipe. The equipment collects a signal later analyzed for pipe structural defect detection.

Figure 1 Typical RFEC/TC Scan Result
The signal identifies the start and end of each pipe segment. Defects are tracked according to their relationship with a specific pipe segment and related to an overall position within a waterline.

New Problems:
PPIC has inspected over 2,000 kilometres of PCCP water conveyance infrastructure. With all of this experience collecting, interpreting and analyzing data, there are some common problems addressed with the new hybridized data model.

#1 When a defective pipe is reported, it is often just one pipe or even several pipes spaced out over several thousand feet. Pipes are then replaced individually and not an entire waterline between nodes. The past practices of replacing an entire line between two nodes are too costly for transmission mains. This is one of the major advantages for performing the RFEC/TC inspection. The RFEC/TC inspection identifies a specific pipe segment for replacement or repair instead of replacing an entire line.

#2 In almost every inspection, there is a discrepancy found between what was actually inspected from what was reported on As Constructed plans. The RFEC/TC inspection is sensitive to pipe properties and when a series of pipes are out of sequence from that reported on the Construction Lay Schedule or the As Constructed Plans, this is detected in our signal. This means that an RFEC/TC inspection can report a more accurate profile of pipe data that is typically contained in most clients engineering documents of record.

#3 The use of accurate GPS survey technology to geographically fix the position of surface features is very important to the accurate reporting of pipe defect position. This would be true for subsurface features like bends and other pipe route inflection points; however, because they are buried infrastructure, fixing their exact position is not possible with a GPS survey technology.

#4 The reproduction of the original survey line stationing is very difficult to establish within the GIS and often similarly difficult to establish in the field. Consequently, when representing pipe geometry as a profile the original station series is not preserved exactly as it was recorded. Thus reporting a pipe as being at station XX+YY from the GIS Profile is not usually consistent with the client’s original station series.

#5 When a pipe is removed from service it is not typically cast aside but is usually kept for analysis. Thus each pipe removed from service represents considerable value for information both for RFEC/TC and for long term asset management information. This creates a need to preserve data within the data structure rather than merely keeping a representation of current assets.

Possible Options:
ArcGIS Water Wastewater data model, (ArcFM) This data model presents some of the objects that we need but it does not specifically contain pipe segments. This data model is based on the same node link models implemented previously and by many other industry groups.

Dynamic Segmentation
Using a linear addressing scheme to create dynamic segmentation of a waterline would present an opportunity to represent pipe segments as an event table with specific geometry from the linear address scheme. Thus, a pipe would be from station XX to station YY and with a waterline between nodes 43 and 44 a specific geographic position would be recorded based on the dynamic segmentation of the waterline. There are circumstances when this data structure is quite effective for pipe property data; however, it is very dependent on the accuracy of the linear addressing scheme and this creates an unacceptable physical representation dependency. In particular, the use of interpolated positions based on vertexes of the waterline does not provide adequate pipe segment and pipe joint positional data.

Pipeline Open Data Standard, PODS
A consortium of companies working from the base of a single project for a particular client created PODS version 1.0. The use of the word standard with respect to the work undertaken is not appropriate as changes have been made by several companies including a new version of the model created using UML diagramming with ESRI extensions. This is really a competitive commercial data structure and not a true standard. While there are some interesting aspects to this data structure, its background in the oil & gas industry limits its applicability to the water industry. Specifically, there are not the data structures that we need to address the jointing aspects of PCCP pipe. Steel pipe is typically constructed and installed as a welded continuous pipe while PCCP and Ductile pipe are constructed of short, (12 ft, 16 ft & 20 ft) segments with bell and spigot ends that possess a gasket connection.

A Proposed Solution

Conceptual Approach
Our approach is based on work performed by Intergraph for its Smart Plant Software used on North Sea Oil Platforms. This conceptual approach introduces five high level objects used in the data model as shown in the following figure.

Figure 2 Data Model Approach
Figure courtesy of Bob Humphrey of Intergraph This approach provides two distinct objects that fulfill differing functions within our data model. The Plant Item object represents the process function of the plant or in this case Water Transmission facility. If we are only interested in material flow properties, then we need go no further than this object. However, process functions are fulfilled by Service Items or equipment. This establishes a relationship between a process representation and an equipment representation.

This figure of the Data Model Implementation shows the use of ArcGIS Water Data model for Plant Items or process model while the Service item is the PPIC Pipe Segmented Data Model.

Process components verses Equipment components
The process portion of the data model maintains the original node link data model approach. The use of ArcGIS water data model provides a node link data model but any of the several in the water industry would also work. The node link data model provides everything in the database that would be needed for a detailed hydraulic assessment for water capacity planning purposes. The Equipment component portion of the database repeats an instance within the database for each process component but provides greater attribute detail. In fact, there is a many to many relationship created in the database between equipment components and process components. This provides historical tracking of equipment, (pipes) for providing water conveyance within a waterline.

Plan verses Profile
Modelling pipelines of any sort presents a common modelling problem of representing a three dimensional system in a two dimensional space. Despite improvements in 3D rendering software, modelling systems in three dimensions continues to challenge pipeline data models. Our model structure is a 2½ dimensional representation of the pipeline and provides for the three dimensional real-world pipeline via separate, but associated plan and profile views. A plan view of the pipeline represents the process elements of the database model in a planimetric layout traditionally used in most pipeline or network models. Nodes represent all types of fittings such as manholes, air valves, or elbows while edges characterize the pipeline between nodes. Information about the edge pertains to the entire length of pipeline between the two nodes. At this level, traditional network and hydraulic modelling can occur. For example, one can perform water traces, isolate valves, or determine disconnected areas of the network In contrast, the profile view presents the individual pipe segments. This side view of the pipeline does not lend itself to any world coordinate system. Thus, we plot the pipe segments using the pipe lay station number for the X-axis against elevation on the Y-axis. Most evident in the profile view are elevation changes between pipe segments because our model incorporates the slope at every pipe joint. Each pipe segment also displays its inspection result and defect location (see figure 4).

Figure 4 Typical Plan & Profile GIS View
A unique PipeID uniquely identifies pipe segments, which are joined to GIS features in the profile and plan views. Likewise, a SysID uniquely identifies each node and waterline in the plan view. The pipe segments in the profile view and the process components of the plan view connect in the data model via these two identifiers. By giving each pipe segment the SysID of its process counterpart, we create a relationship between the process and equipment representations in the data model. If a pipe segment is replaced on a particular waterline, its SysID value is set to null but the PipeID is maintained and so are all of the inspection results that are linked to that pipe.

Survey Stations and GPS
In order for the database to be the basis of field operations and investigations, accurate spatial information is a necessity. Much of the PCCP water infrastructure across North America is over 30 years old. The age of the system coupled with outdated as-built lay drawings leaves many pipeline managers speculating about the exact location and composition of their system. While an RFEC/TC inspection provides the precise number, type, and relative position of pipe segments along a transmission line, the inspection cannot give real world coordinates to these features.

A GPS (Global Positioning System) survey provides centimetre level accuracies for surface features (e.g. manholes, air valves, blow offs, marked point intersections). After an RFEC/TC inspection has accurately corrected the pipelines lay schedule, interpolating the buried pipe segment’s real world position requires at least two GPS points (up and downstream from the pipe segment) in addition to pipe length, slope, and pipeline bearing characteristics. With a GPS survey, if a pipeline between two known points contains numerous changes in horizontal bearing, the accuracy of a pipe segment location in such an area will decrease. The error is estimated to be a maximum of 75% of a standard pipe length; however, there are circumstances where this could be greater depending on length between nodes and number of bearing changes.

In most pipeline Construction documents there are inconsistencies with the original lay schedules station numbering. Station numbering is the accumulated distance travelled by the pipeline from an arbitrary starting point (usually the beginning of the pipeline). During construction, a survey creates this numbering system, however; when contractors build pipelines using small individual sections that are later joined, the station numbering survey tends to have errors that are corrected by using a mathematical equation to adjust the Station numbers. Ultimately, the original station numbering is not a true representation of the distance travelled by the pipeline but is a relative positioning technique.

Our data model corrects the station survey because the model accounts for pipe length, horizontal, and vertical changes in bearing. Thus, pipeline operators can determine a pipe segment location using interpolating GPS coordinates or the recalculated station number line. Additionally, to avoid confusion and permit an agreeable transition between the old and new station number systems, the data model references the original station number for each pipe segment while preserving the specific pipe geometry.

Managing Infrastructure Assets
The pipes and other system features themselves require some analytical and historical tracking. Conducting a risk management assessment for a pipeline on a joint-by-joint basis requires the capabilities of a fully functioning GIS environment and thorough data model because of the large amounts of data and computational time required.

Even a simple comparison report of changes in pre-stressing wire breaks between inspections is a relatively simple procedure that would take a long time to prepare without the use of a GIS given the number of pipe joints per mile of pipeline. Calculating which pipes will fail is a more complicated study because of the many different variables involved other than the number of pipe breaks. Figure 5 demonstrates one method of determining those pipes to replace by looking for individual pipes with a certain number of wire breaks, in a specific area of the pipe, bar rating, and wire pitch.

Figure 5 Typical Break Tracking View
Accurate failure assessment requires not only knowledge of a pipe segments condition, but the position of defects on the pipe. In the data model, each defect is located circumferentially and/or axially in the pipe segment and given a unique identifier. Storing this information in the data model allows users to track individual defect growth and relates them to individual pipe segments. For example, one inspection may observe two mortar cracks in a pipe segment at 1.5 metres and 2.5 metres from the downstream end. A subsequent inspection reveals one long mortar crack centred at 2 metres from the downstream end. With the data model in place, a new inspector or analyst quickly knows that the two defects have become one and not the sudden appearance of a delaminating mortar coating.

Overall System Integration Opportunities
A prime motivating factor in implementing this new data model is the need to perform new analysis and new types of tracking of pipe segments. The ArcWater data model does not implement Pipe Segments and the use of the Pipe segmented data model (the equipment portion of this data model approach) provides a unique integration opportunity to a variety of Maintenance & Operations programs.

A primary integration point is the use of Maintenance Management System, (MMS) software to track operation activities and preventative maintenance procedures. By creating an explicit pipe segment object in the data model, investigations are tracked via an MMS. These investigations would include pipe inspections for leakage, inspections including RFEC/TC, failure investigations, Pipe Maintenance records including Rehabilitation and Repair works. Since the PipeID is preserved and never deleted, a pipe continues to maintain a history even after it is no longer in service.

Environmental monitoring software is another excellent integration opportunity with this pipe data model. Soil Resistivity and Pipe to Soil potential measurements are equally well suited for analysis with the pipe segment database. This data collection activity is one that requires the use of a Linear Addressing Scheme as most of this environmental data was collected using the Station Series from the original Construction Documents. There is software required to transform from the client station series to the relative position on the ArcWater or process data model then using a profile station series for the equipment or Pipe Segment Data Model, a new Linear Address Position is determined that reflects a coordinate on the pipe profile in the GIS. In the future, this data will most likely be collected using GPS equipment and its representation in relation to specific pipe segments can be plotted and analyzed with greater accuracy and reliability.

Solution Assessment

Costs:
The cost to implement this data model is based on the assumption that an agency has not yet begun to digitize their infrastructure. A traditional data conversion project would digitize existing paper As Constructed paper drawings into a GIS for field verification over an extended period. The Hybrid Data Model approach provides a better representation of actual infrastructure with the additional expense of a condition survey of the infrastructure. The product is a data base of pipeline assets that have been field verified and are complimented by physical assessment data leading to a comprehensive Life Cycle program.

The following table outlines the typical stages of implementation and associated cost implications.

Not including the cost of data verification and conversion to digital format, the data model presents additional costs to manage data. To begin, there is the cost of training staff to use the data model and its related software applications. This must also include time needed to acclimatize staff to a new way of operating. The data model is not solely a new tool but represents a change in business philosophy whereby the pipeline operator now continually updates and references the database information. Excluding the cost of the physical assessment inspection, this data model represents an increase in data compilation costs of about 10% over a traditional data conversion project with field validation.

Benefits:
Though the data model means an increase in effort for those using it, the benefits outweigh the costs. Most apparent to those who use the data model will be the ease of retrieving, analysing, editing, and updating data. Positional accuracy and current data for field staff easily accounts for the additional costs in using the data model by making information access quick and simple. Having all the data in a central store permits analysis functions and asset tracking not possible with a paper based system. Finally, the data model satisfies the recent GASB 34 legislation that requires government agencies maintain detailed information about their assets including depreciation or the modified approach by tracking the cost to maintain that asset.

Outline of major benefits of Hybrid Pipeline Data Model:

Increased detail of representation of pipeline assets, pipe segments (bell and spigot pipes) including integration opportunities with MMS and Environmental Monitoring programs.
Provides increased granularity for planning capital programs. Target specific pipes for replacement instead of whole waterlines.
Historical pipe data maintained for perpetuity in database.
Defects are uniquely tracked over several inspections. This provides defect trend analysis and projecting future defect values.
Maintains ability to export for hydraulic analysis and present hydraulic analysis results in the GIS for calibration.
The use of RFEC/TC and this data model delivers a high value data set that is more accurate at representing existing water infrastructure than what available to a client using their own Engineering documents.

Technology Limitations:

The UML diagram of the data model uses the ESRI template and Visio 2000. This creates a data base using the shape file specification for geographic feature geometry, (points, polylines and polygons).
This approach has been validated using RFEC/TC and PCCP, it is an appropriate model for any water conveyance system that can benefit from pipe segment specific capital programs.
The ability to move upstream and downstream in the model requires network topology. Network topology implements either ESRI’s proprietary methods or PPIC topology methods.

Summary of Hybrid Pipe Data Model
This Hybrid Data model fulfilled our primary goal of performing Life Cycle Asset Management of PCCP infrastructure. It provides hydraulic analysis capabilities through its continued maintenance of the ArcGIS Water & Wastewater data model while integrating detailed Pipe Segment data objects for tracking Asset Performance of Pre-Stressed Concrete Cylinder Pipe.

Future Data Model Additions
One of the key aspects of the Intergraph approach in the North Sea was the use of standard equipment database specifications. Major equipment suppliers agreed to use these standards and delivered equipment with digital data to ensure that the database of process equipment is up to date. This digital version of an equipment change log provides detail that traditional practices did not and this lead to major savings in uptime and productivity when tied with a maintenance management system.

The standardization of data specifications for pipe suppliers is a goal of this data model; however, the authors believe that this will be a long-term goal requiring considerable effort to accomplish. The addition of document management for tracking historical and traditional paper documents will ensure a complete understanding of a pipeline infrastructure is contained within a single database instance.