Figure 1
Common GIS Errors
Title of Paper: Efficient Pre-processing of Water Infrastructure GIS Data Facilitates Water System Hydraulic and Water Quality Model Construction
Abstract
Water distribution system GIS data meets a variety of needs including automated mapping, facilities management, work order processing, meter records and billing, and emergency planning. Detailed hydrants, valve, fitting, and pipe topology and attribute data typically exceeds the level of detail required for hydraulic and water quality modeling of distribution systems. This paper presents an ArcView application which process detailed water system data in a systematic manner, resulting in a data set which supports water system model construction. The process completed by this ArcVew "Generalization" application include:
Computer modeling in the water industry has come a long way. It serves as a key tool in identifying and planning capital improvements as well as a base for hydraulic analysis and now, water quality analysis. Historically, the number of pipes contained within a water distribution model have been limited by software capabilities and the time required to physically construct each node and main within the hydraulic model. In addition, the ability to accurately assign demands to nodes in the model was time consuming due to the format of metered sales data. But the evolution of Geographic Information Systems (GIS) and their ability to synthesize large amounts of demographic, metered water use, and physical facility data have resulted in models which are increasingly larger and more complex. This technological advancement has benefited utilities because it maximizes their initial investment to their GIS system, and provides more accurate hydraulic models resulting in improved service to their customers.
A Brief History of GIS
GIS was originally developed by the Canadian government to inventory their vast amount of natural resources more efficiently than could be accomplished by conventional mapping methods. The basis of GIS development was production of maps to inventory resources. Therefore, GIS was viewed primarily as a map-making tool.
As it was further developed, GIS was used for the organization and analysis of geographic information. "Organization" of data includes all aspects of acquisition, upkeep, integration, storage, and exchange of geographic information. "Analysis" of data includes retrieval, reporting, and presentation. But the core analytical functionality of GIS goes beyond retrieval and presentation. Commonly available analysis functions can be categorized as search, location analysis, terrain analysis, flow analysis and network, spatial analysis and statistics, and measurements.
The continued development of GIS products combined with the increasing power of desktop PCs has opened up numerous opportunities in terms of use and integration with other business processes. Therefore, most water utilities are actively investigating how they can fully utilize the GIS data which is available.
Typical GIS Data Utilized for Hydraulic Modeling
Typical data which would be retrieved from a GIS system for a water
distribution master plan include the coverages and associated database
information shown in Table 1.
|
GIS Information Used for Hydraulic Modeling |
|
|
|
|
| Main | From Node, To Node, Diameter, Length, Material, Age |
| Node | Node Location, Elevation, Type (Tee, Cross, Etc.) |
| Population Zones | Zone Number, Existing Population, Future Population |
| Metered Water Use Zones | Zone Number, Number of Accounts, Average Day (AD), Maximum Day (MD), and Maximum Hour (MH) Water Use |
Our experience has shown that the quality of the GIS data is generally less than adequate for use in construction of hydraulic models. Not to diminish the benefits and rewards associated with use of GIS data, but quality limitations continue to be a problem during hydraulic model construction. Two problems typically encountered when applying GIS data for model construction include compatibility of the GIS format and resolution of data gaps and errors.
By format of the GIS data, we are referring to the database design or the data capture techniques which were used in the formulation of the GIS system. Typically, the water distribution portion of the GIS is formatted for other purposes; most commonly, Automated Mapping/Facilities Management (AM/FM). Although this format is good for maintenance and management of the distribution system, it is not suitable for construction of hydraulic models. Common data format problems encountered are inclusion of hydrants, line valves, and tees or crosses. In order for the GIS data to be usable for hydraulic modeling, this excess information must be eliminated.
Use of GIS data also requires determination and correction of spatial data errors. Examples of spatial errors are illustrated on Figure 1. Point A is an example of a digitizing error commonly referred to as an under/overshoot. These digitizing errors are typically data entry errors which can be visually identified and corrected based on the utilities' water map books. Points B and C show system gaps which usually are the result of data which was not captured.
Figure 1
Common GIS Errors
Gaps in the GIS network are unacceptable for construction of a hydraulic model due to the discontinuities they create. Point B represents a gap error which would be identified when executing the hydraulic model because it creates an isolation in the distribution system. The gap error represented by Point C is much more difficult to identify because during execution of the model, it will not create an error. This gap may result in significant deficiencies in hydraulic capacity or circulation problems with respect to water quality. Resolution of these errors will require extensive model review against water system records, and will be necessary to develop an accurate hydraulic model. These errors must be addressed by the QA/QC process performed during the data capture phase of the GIS construction.
Generalization - Reduce, Skeletonize, and Trim
Generalization may be viewed as an interpretation process which leads to a higher level view of some phenomena - looking at them 'at a smaller scale'. This paradigm is always used in any generalization activity, whether spatial or statistical.
In order to effectively use GIS digital data for water distribution models, specialized applications are required to reliably and expeditiously convert the data into a hydraulic model. The specialized applications required must have capability for data reduction, skeletonization, and trimming; hereinafter referred to as RST. RST applications process detailed GIS data to efficiently construct water system models as discussed in the following sections.
Reduce
"Reduce" applications require the capability to remove excessive pipe segmentation caused by valves, fire hydrants, and data capture processes. The execution of the application should remove (dissolve) interior nodes on pipe reaches and combine the associated pipe segments into single pipes. For example, the nodes to be "reduced" are shown on Figure 2 as open circles and nodes to remain are filled circles.
Figure 2
Reduce Application
The Reduce procedure combines segments in a pipe theme (polyline) based on attributes relating to length and diameter. The procedure coordinates changes (deletion of points) with a corresponding point theme representing line segment junctions. The procedure outputs (1) a new line theme and (2) a new point theme specified by the user. The output themes appear in the current view upon completion of the procedure.
Skeletonize
"Skeletonize" application requires the capability to remove small diameter pipes as specified. All pipes with diameters below the specified diameter should be removed. Small pipes required to complete system loops should be protected against removal during execution of the application. An example of skeletonization is shown on Figure 3. The 4-inch main enclosed by the polygon should be removed during skeletonization of the 4-inch mains.
Figure 3
Skeletonize Application
The skeletonize procedure removes segments from a pipe theme (polyline) based on (1) a users specified skeletonization diameter and (2) manual selection/deselection. The procedure coordinates changes (deletion of points) with a corresponding point theme representing line segment junctions. The procedure outputs (1) a new line theme and (2) a new point theme specified y the user. The output themes appear in the current view upon completion of the procedure.
Trim
"Trim" application should remove short pipe segments including dead ends and hydrant leads as shown on Figure 4.
Figure 4
Trim Application
The Trim procedure removes segments from a pipe theme (polyline) based on minimum pipe length. The procedure coordinates changes (deletion of points) with a corresponding point theme representing line segment junctions. The procedure outputs (1) a new line theme and (2) a new point theme specified by the user.
Final Model Pipe Segment Selection and Processing
It should be recognized that RST is an iterative process. Processing of detailed GIS pipe data is expected to require a number of passes of the Rdeuce, Skeletonize, and Trim routines to yield acceptable hydraulic model input data. The RST routines will complement each other. For example, Reduction of excessive pipe nodes and segments will require fewer pipe segments to be manually selected for protection and removal for construction of an effective hydraulic model.
Impact of GIS Integration
There are definite technical advantages associated with the use of GIS for the construction of hydraulic models and the evaluation of water demands. Nevertheless, it is important to understand the hydraulic and cost impact associated with constructing a model which includes 6-inch and 8-inch mains, as compared to the models constructed 10 years ago which typically included only 12-inch and larger mains. These impacts are evaluated in the following section.
Hydraulic Model Size
Typically, 10 years ago when a water master plan was performed, model construction involved digitizing of the system over water plat maps or electronic base maps constructed using computer-assisted drafting (CAD) packages. Models were constructed using the 12-inch and larger mains with smaller mains as required to complete loops. The person-hours required to manually construct a model which included all the 6-inch and 8-inch mains was simply cost-prohibitive. But the evolution and implementation of GIS technologies, specifically RST applications, have allowed for construction of models which include all 6-inch and 8-inch mains with minimal impacts to project cost.
To better understand the hydraulic impact of smaller mains within the
distribution system, refer to Table 2 which includes the actual summation
of all pipe diameters and their associated length for another project in
the Midwest that will be designated Utility A. From these diameters and
lengths, a volume can be calculated for each pipe size. Since the larger
diameter pipes possess greater area, the cumulative volume increases more
rapidly than the cumulative length. Also note that an increase in system
volume, which relates to transmission capacity, can be recognized by each
successive step of including smaller pipe sizes. For example, constructing
a model using only 16-inch and larger mains would require inputting only
16.4 percent of the overall system length, but would capture over 52 percent
of the system volume. Taking two more steps to include 14-inch and 12-inch
mains would produce an overall system length and volume of 34.7 percent
and 75.6 percent, respectively.
|
Table 2 Utility A Water Main Length/Volume Statistics |
|||||||||
|
(in) |
|
||||||||
|
|
(ft) |
(mi) |
Total Length |
Length (%) |
(cu ft) |
Volume |
Volume (%) |
||
| 42 | 18 | 4,729 | 0.9 | 0.2% | 0.2% | 45,475 | 2.9% | 2.9% | |
| 36 | 5 | 510 | 0.1 | 0.0% | 0.2% | 3,603 | 0.2% | 3.1% | |
| 30 | 34 | 8,448 | 1.6 | 0.3% | 0.5% | 41,448 | 2.6% | 5.7% | |
| 24 | 114 | 30,499 | 5.8 | 1.2% | 1.7% | 95,767 | 6.0% | 11.7% | |
| 20 | 689 | 153,496 | 29.1 | 6.0% | 7.7% | 334,707 | 21.0% | 32.7% | |
| 16 | 1,214 | 223,520 | 42.3 | 8.7% | 16.3% | 311,935 | 19.6% | 52.2% | |
| 14 | 45 | 6,059 | 1.1 | 0.2% | 16.6% | 6,474 | 0.4% | 52.6% | |
| 12 | 4,466 | 466,307 | 88.3 | 18.1% | 34.7% | 366,051 | 23.0% | 75.6% | |
| 10 | 793 | 61,701 | 11.7 | 2.4% | 37.1% | 33,636 | 2.1% | 77.7% | |
| 8 | 4,928 | 457,748 | 86.7 | 17.8% | 54.8% | 159,703 | 10.0% | 87.7% | |
| 6 | 8,562 | 927,123 | 175.6 | 36.0% | 90.8% | 181,948 | 11.4% | 99.1% | |
| 4 | 1,662 | 124,465 | 23.6 | 4.8% | 95.6% | 10,856 | 0.7% | 99.8% | |
| 3 | 427 | 21,972 | 4.2 | 0.9% | 96.5% | 1,078 | 0.1% | 99.9% | |
| 2.5 | 3 | 193 | 0.0 | 0.0% | 96.5% | 7 | 0.0% | 99.9% | |
| 2.25 | 17 | 1,516 | 0.3 | 0.1% | 96.6% | 42 | 0.0% | 99.9% | |
| 2 | 1,881 | 76,320 | 14.5 | 3.0% | 99.5% | 1,664 | 0.1% | 100.0% | |
| 1.75 | 2 | 349 | 0.1 | 0.0% | 99.5% | 6 | 0.0% | 100.0% | |
| 1.5 | 47 | 4,816 | 0.9 | 0.2% | 99.7% | 59 | 0.0% | 100.0% | |
| 1.25 | 59 | 5,515 | 1.0 | 0.2% | 99.9% | 47 | 0.0% | 100.0% | |
| 1 | 17 | 1,175 | 0.2 | 0.0% | 100.0% | 6 | 0.0% | 100.0% | |
| 0.75 | 7 | 466 | 0.1 | 0.0% | 100.0% | 1 | 0.0% | 100.0% | |
| Total | 24,990 | 2,576,927 | 488 | 100.0% | 1,594,512 | 100.0% | |||
Integrating GIS into development of the hydraulic model allows for inclusion of 6-inch, 8-inch, and 10-inch mains. The impact of each step is shown on Figure 6. Initially, inclusion of each pipe step results in a steeper increase along the volume line and a more gradual increase along the length line. At a diameter of 12 inches, the slope of volume line turns gradual and length line becomes steeper. This trend continues until 6 inches, at which point both lines flatten and eventually merge.
This data would support the approach of using 12-inch and larger mains
for manual model construction because the difference between system
volume obtained and overall length digitized would be at a maximum. In
other words, the greatest amount of volume would be obtained relative to
the input required. Please also note the difference between conventional
methods and GIS methods with respect to percent of the system captured.
GIS now provides the capability to cost-effectively capture an additional
23.5 percent of the volume and 56.1 percent of the length.
Figure 6
Cumulative Length and Volume vs. Diameter
Conclusions
Today's hydraulic and water quality models have grown increasingly large and complex. Development of complex water system models requires manipulation of large volumes of data sets. Accurate determination and detailed procedures to pre-process model creation are necessary. The evolution of GIS systems has led to the availability of large data sets which can be utilized during a water master plan. In order to efficiently and effectively use this data, the capability of processing GIS information to create hydraulic models is required.
Additionally, large models can also now be justified by the ability to perform accurate demand allocation utilizing extremely detailed population and water use information. The results of these efforts, as detailed herein, are models which contain significantly greater amounts of the distribution system volume, have significantly lower cost per pipe, and show no adverse impact for overall cost to the customer. Realization of these benefits will require that consultants utilize GIS in their modeling efforts to be competitive in the marketplace.
References
Edward J. Koval
Project Manager
Black & Veatch
8400 Ward Parkway
Kansas City, MO 64114
Tel: 913.458.3513
Fax: 913.458.3730
Jospeh Q. Pindell
PDA, Inc.
7701 College Blvd.
Overland Park, KS 66210
Tel: 913.469.8700