Abstract
Dallas Water Utilities services about 1.9 million people and covers about 699 square miles. In the last two years, the Utility Automation Integration Program has successfully completed the data migration from CAD DGN format to an ArcSDE geodatabase using the ArcFM model for 4,566 miles of water mains; 4,021 miles of wastewater mains; and numerous features such as hydrants and fittings. GIS not only serves as an efficient data integrator but facilitates inter-departmental communications, supports real-time database management, and allows updating by field personnel. The spatial characteristics of GIS provide unique perspectives for engineering, design, and administrative uses.
Background
Dallas Water Utilities has historically maintained permanent paper maps of the water distribution and wastewater collection systems. There are 1755 Water System maps, and 1189 Wastewater System maps. Most of the permanent maps were digitized in MicroStation. The maps had been grouped into map tiles comprised of sixteen maps each. There are 150 map tiles each of the water distribution system, and for the wastewater system. This allowed the manipulation of the maps more efficiently. These map tiles, which included a limited amount of feature information, were migrated to an ArcSDE geodatabase.
First Step
The biggest challenge we faced was the database design phase. This process began around 4 years ago and consisted of interviewing many groups within the utility to determine their needs. These interviews provided us the opportunity to document various processes and develop a list of features and attributes. We used the basic Visio ArcFM Water Distribution Model and the ArcFM Wastewater Model provided by Esri to develop Dallas’ model. Changes were made to these Visio models in order to capture the features and attributes that had been identified through the interview process. The Visio models were then used to build geodatabases inside of SDE. Once this was completed we were able to focus on training staff and building the water and sewer networks in SDE.
The CAD Technicians have had extensive drafting experience and knowledge and understanding of the work associated with maintaining paper maps. They had been trained to use MicroStation and were familiar with the use of this software. We provided formal in-house training by Esri and informal classes by our GIS Analysts, but for the most part GIS training been comprised of on-the-job training.
The Conversion Process
A key decision was how to best establish the networks in SDE. Several years ago, management decided to digitize all the maps into the GIS. After years of having very little success and realizing that this process was time and staff intensive, focus was shifted to converting the existing MicroStation DGN files with the assistance of Esri.
As with any conversion process, challenges exist in data integrity and compatibility. Our experience with the MicroStation files was no different. Through the conversion process we were aware that the MicroStation data on different features was stored at different levels. Additionally, data was inconsistent within the individual levels. However, each MicroStation map tile had a spatial component, identifying each maps’ relationship to the other. Unfortunately, the spatial component was not referenced to the earth. With the assistance of Esri staff, we used an AML to convert the map tiles to coverages, bringing in the limited amount of attribute data from various levels. We developed two coverages: one for the Water Distribution System, and the other for the Wastewater Collection System. Another AML was then used to perform a bulk move of all the coverages to real world coordinates. The data resulting from these two processes was not always spatially accurate. This was primarily due to the original drawings not being drawn to real world coordinates (some pipe lengths were drawn to fit within a cell for visual clarity).
The next step was to use a transform procedure to get the pipe networks to consistently fit the geography. The transform procedure was done one map tile at a time. After all the maps were transformed, we loaded them into SDE. It took about one year from the beginning of the conversion process until the transform process was completed.
Ultimately, the conversion process has provided a relatively accurate geodatabase as a beginning point. Even though many hours have been spent converting and cleaning the data files, the approach has proved to be the most efficient for a system the size of Dallas.
Working with SDE
Once the data was moved into SDE we were able to join the map tiles through edge matching. We found many small breaks in the pipe lines due to the method used in the development of the original CAD drawings. In addition, there were points left behind when lines were moved, as well as duplicated features. Our editing process uses the Trace and Find Disconnected tools for building connectivity and the true pipe networks. This is a manual time-intensive and tedious operation that is still ongoing.
For SDE to work effectively and efficiently, it requires the development of a well thought-out process. Considerations include coordinating the work of each staff member, and establishing the schedules to perform reconcile, post, and compress procedures. The functionality of the reconcile and post process allows a user to “check out” a version of the parent geodatabase to make the edits or changes. Once an editor is finished making the edits, the version can be posted back to the parent geodatabase where the changes will be made permanent. To minimize the conflicts during reconcile and post, we recommend assigning specific areas of the city to each staff member. We also tried several schedules for reconcile and post in order to allow each staff member to see changes made by others to the geodatabase. The schedule and time required depends on the number of versions created, the number of changes made to each version, and of course the hardware being used. Posting and compressing procedures cannot be performed while editors are working on their versions. We also recommend that the geodatabase be compressed frequently to reduce its size and improve functionality, as a compressed geodatabase reconciles and posts more efficiently. Currently we perform this process every other day. From our experience, it is evident that SDE provides the tools to manage the geodatabases on a real-time basis.
Quality Assurance/Quality Control
Procedures for Quality Assurance/Quality Control (QA/QC) need to be developed and routinely used to ensure the accuracy of the GIS. We recommend that QA/QC procedures focus on specific data that is being entered. For example, if you have an attribute on a feature with a fixed character length, you can create queries based upon the length. Results obtained from that type of query indicate a data entry error. Queries can also be developed to spot check data. In the case of a feature with many attributes, we recommend creating a QA/QC worksheet looking at all the data entered. While QA/QC procedures are time intensive, developing and using them will provide you with feedback of your staff’s abilities and the quality and accuracy of your GIS.
Attribute Population
Because we are still in the development stages of GIS, and due to our limited staff numbers, we are planning to contract out a major portion of the attribute data entry. Data that we plan to include is information currently available on the permanent maps. Some examples would be main sizes, types of valves, etc.
Communication through GIS
One of our goals is to get the word out about GIS. A beginning step in communicating the benefits of GIS was to offer training to non-GIS staff who would be using the geodatabase for information gathering purposes. We held various informal training sessions on the use of Arc Explorer which were well attended. We have created a shared drive on one of our servers where we store GIS data. This data is accessible to anyone on the network.
Communication also includes providing services to other areas of the utility. We also provide mapping and data analysis to the extent the data is available. Our work has provided printed maps showing locations of water accounts and locations of industries that discharge into the sewer system, etc. An important deliverable will be the development and deployment of GIS using ArcIMS. We are planning to begin this effort within a year.
Where are we and where are we going???
At this time we are in the process of connecting the water distribution and wastewater collection pipe networks to all the associated features. The next step is to rectify the pipe networks to the orthos. We will be using the Rubber Sheeting feature in ArcGIS 8.2 to perform this task. We are also working on entering detailed information on water structures that are not on the permanent maps. To obtain some of this information requires extensive research. An example is information regarding pump stations, elevated water tanks, water storage tanks, etc.
Even though we have a working geodatabase, we continuously review it to make sure that it meets the needs of the water utilities. We adopted a policy that when a change is needed to the geodatabase, the change is also made on the Visio Model. This allows us to have a framework from which to work. We have found that through the initial interviews, emphasis was placed on gathering real time data for a number of features that in reality are not consistent with the capabilities of a GIS, but are more appropriate for a work order management system. Emphasis was also placed on data that is not readily available, but information staff thought it would be nice to have. As we fine-tune the geodatabase we are able to prioritize the data entry needs based on input from staff to ensure that the model is valid and that it reflects actual operational standards. Due to recent input from staff we have made changes to the Visio models and in the near future we will be making those changes to the geodatabases.
Long range goals include deploying GIS to the field personnel. Initially we plan to provide field personnel with laptops or wireless handheld devices containing GIS data regarding the system. They will be trained to perform operations like valve isolation and pipe line location. This will also provide field personnel an avenue to give feedback regarding the accuracy of the model. Depending on the capabilities and training of field personnel, it is perceivable that a mechanism could be developed to allow them to make edits to the geodatabase during the course of their workday.
Summary
These past two years have been challenging. Working with ArcSDE has provided another dimension to the maintenance process of the GIS. As we continue to develop and populate the geodatabase, we foresee a growth in the interest and the use of the geodatabase by water utilities staff as a management tool on a daily basis. Deployment of GIS to the field will increase the functionality of the system, providing a more efficient and effective workforce.
Author Information
Betty Antebi-Taylor, PE, RS
Program Manager
Utility Automation & Integration
City of Dallas Water Utilities Department
2121 Main Street, Suite 400
Dallas, TX 75201
(214) 948-4227
(214) 670-5859 fax
bantebi@mail.ci.dallas.tx.us
Chris Way
GIS Analyst
Utility Automation & Integration
City of Dallas Water Utilities Department
2121 Main Street, Suite 400
Dallas, TX 75201
(214) 948-4610
(214) 670-5859 fax
cway@dwu.ci.dallas.tx.us