CE 394K.3 - GIS in Water Resources
December 2000
Objectives
Procedure
Application of naUTilus
Results
Conclusions
Future Work
References
At the University of Texas at Austin, I am currently working on a research project to assess whether municipal sewers serve as a significant source of hazardous air pollutant emissions. The project, under the supervision of Dr. Richard L. Corsi, has three main phases. The first phase, currently underway by another graduate student, involves a series of experiments to track the migration and emission of several tracers in operating sewers. After these experiments are completed, the data obtained will be used to modify and calibrate an emissions model to better represent a municipal sewer network. The second phase of the project involves modeling the emissions from a large municipal sewer network. Phase three involves field monitoring to quantify emissions from the hot spots identified in phase two.
In this term project, the sewer network associated with one of the tracer experiments was modeled. This was done by using naUTilus, a model developed to predict volatile organic compound (VOC) emissions from industrial sewers. Data for the model's input was extracted from ArcView files, and the sewer network was determined using the network capabilities in ArcInfo 8. In addition, an interface, created to aid in the development of input files and the application of the model, between naUTilus and ArcView was evaluated.
The naUTilus model was developed at the University of Texas at Austin as a method for estimating VOC emissions from industrial sewer networks. For estimating gas-liquid mass transfer, the model uses fundamental mass transfer principles; to determine air exchange rates between sewer and ambient atmospheres, fluid mechanic and heat transfer principles are used. These parameters are based on previous field monitoring and pilot experiments. A comprehensive description of the naUTilus model is provided in naUTilus documentation (Olson et al., 1997).
Because it was developed for industrial sewers, the naUTilus model consists of two separate FORTRAN modules - ISBL and OSBL. The ISBL, "inside the battery limit", is limited to specific process units. The OSBL, "outside the battery limit", is the main collection system that connects the individual ISBLs. This is shown in the figure below.
An ISBL unit can consist of branches, junctions, manholes, drop structures, process drains, and hard-pipe connections. Likewise, an OSBL unit can consist of branches, junctions, manholes, drop structures, and output from ISBL units. In addition to these physical parameters, other inputs to the model include ambient conditions, information regarding Henry's constant, flow, and wastewater characteristics. The input parameters for both units are similar. An example of a typical input file for an OSBL is shown here.
One of the many reasons that naUTilus can often be difficult to execute is demonstrated by the input file. Creating the input file requires a knowledge of the file format and the naUTilus numbering system, and these difficulties are only enhanced when applied to a large sewer network.
To help remedy some of the difficulties posed by naUTilus, naUTilus and ArcView were integrated by Cindy How in 1998 through a series of Avenue scripts. In essence, the scripts allow ArcView to be a visual interface for naUTilus. The interface allows sewer information to be entered through a series of steps by clicking on the desired node or branch. ArcView introduces a spatial aspect that allows the connectivity of sewer elements to be more easily established. The interface also allows for the creation of input files, naUTilus execution, connection of the OSBL and ISBLs, and maintenance of the required file structure. Another significant benefit of interface is its ability to spatially display the model's results, allowing for the quick identification of emission hot spots.
However, previously the interface has only been applied to data that had been digitized or manually created. Until this project, it has not been applied to existing ArcView files. For more information regarding the specific application of the naUTilus-ArcView interface, please see the on-line user document.
Shapefiles of the Austin sewer system were obtained from the City of Austin ftp site (ftp://issweb.ci.austin.tx.us/pub/GIS-Data/WWW/Shapefiles/Wastewater/). The directory, coll_shp.zip, contained two shapefiles - one for the sewer system (coll_lin.shp) and a second containing junction information (coll_pnt.shp). Upon review, each of these files contained around 70,000 features. There is a large amount of information associated with each sewer reach and each manhole. Some of the information available for the sewer reaches include pipe length, diameter, upstream and downstream manholes, construction material, and current status. Some of the junction information included the type of junction, rim elevation, and depth. Although the dataset is extensive, not all information was present for every sewer reach or junction.
The shapefiles are in NAD 1983 datum, Texas State Plane Central Projection with coordinates in feet. The sewer map for the City of Austin, showing only the sewer reaches, for the City of Austin is shown below.
The following figure, showing the Austin sewer network classified by pipe diameter, indicates that the majority of the sewer pipes in Austin are less than 10 inches. Because most of the sewer system is conveyed by gravity flow, where small pipes flow into larger pipes, this figure also allows for an easy identification of the main sewer routes.
As mentioned previously, this project focuses on the stretch of sewer that is associated with one of the tracer experiments performed as part of the larger emissions study. The sewer network of interest is shown below, highlighted in green. It runs from the UT campus south, basically following much of Waller Creek. Although it is not shown in the selected sewer stretch, the wastewater flows are discharged to Govalle treatment plant, located east of Austin..
The sewer network was identified in the following steps:
Step 1 - In order to have a smaller, more manageable file to work with, the sewer line file (coll_lin.shp) was reduced by deleting the area south of downtown. Also, the northern and western portions were removed, essentially leaving only the area of interest, as shown below. Another shapefile showing some of the major roads in Austin was overlain to provide some orientation. The number of sewer stretches was reduced to approximately 4000.
Step 2 - The next step was to eliminate sewers that were not presently in service. Abandoned sewers and proposed sewers, identified as "AB" and "PR", respectively, in the status column of the attribute table, were removed. Once this was completed, approximately 3140 sewer stretches remained in the file (area_l.shp).
Step 3 - The third step was to use ArcCatalog to create a geometric network. First, a personal geodatabase was created and the feature dataset was added. Next two shapefiles were imported into the geodatabase. The first contained the sewer stretches in the area of interest (area_1.shp); the second contained a sink that was located at a pump station downstream. From the personal geodatabase, a geometric network was created using the network wizard.
Step 4 - Using the network created, the sewer of interest was identified using ArcMap. First, the ancillary role of the sink was changed to a value of 2, then the network was applied. A junction flag (shown as a green square) was added at a point on the main sewer, downstream of the last sampling point. Edge barriers (red Xs) were placed adjacent to the junction flag to eliminate the trace from following these tributary or downstream sewers.
Additional edge barriers were added upstream of the tracer injection point, in order to limit the area of interest. Then a trace upstream analysis was preformed, as shown in the figure below. The selected sewers were exported to a separate file (experiment1.shp).
Additionally, the main sewer stretch was identified. This was accomplished by using the junction flag previously placed along with a second junction flag located upstream of the injection point. Then the trace path function was used. The highlighted sewer stretch, shown below, was also saved separately (mainline.shp).
Step 5 - In addition to requiring sewer data, naUTilus also requires some of the manhole information contained in the coll_pnt file. In order to obtain only the information that was needed, the select by theme tool in ArcView was used. This tool allows all the manholes (from coll_pnt.shp) to be selected that are within a small distance of the features of the selected sewer network (in this case, experiment1.shp). Through trial and error an acceptable distance was 1 foot . This was small enough to only choose the manholes directly on the sewer but large to account for any small discrepancies in alignment.
As shown below, only the manholes that are associated with the selected sewer network are highlighted. The manholes slightly off the network are not selected. This selection was then converted to a shapefile (manholes.shp). A similar procedure was performed for only the main stretch of sewer (mainline.shp) and those nodes were also converted to a separate file (mainnode.shp).
Despite my best efforts, the data contained in the City's files was not compatible with the interface, and naUTilus could not be run from within ArcView. Even as I used the interface with small example files, the model was often unstable. However, ArcView was used to extract information for the model's input.
The focus of the model is the main sewer stretch. Because municipal sewers consist mainly of long stretches, this was modeled as an OSBL unit. The sewer stretches that are tributary to main sewer stretch were considered as inflows to the sewer. Sewer characteristics, such as length and diameter, were taken directly from the database file. The contaminant modeled was highly volatile (cis 1,3-dichloropropene (C3H4Cl2)) with a Henry's constant of 0.63 m3 liquid/m3 gas. The only source of this contaminant was assumed to be from the tracer experiment. Because many of the parameters to the modeled are specific to the sewer network, a number of estimates were included in the model and may need to be adjusted as new information becomes available.
The text-based results of the naUTilus model were added to the manhole information contained in the shapefile. These manholes were then colored as shown below to represent various emissions rates. As seen, the emissions are greatest near the injection point. The other area where emissions are increased is following a drop structure. These results generally agree with what is expected. Increased stripping of the VOCs is expected in areas of low flow rates and in areas of increased mixing.
In conclusion, ArcView and ArcInfo are useful tools in the creation of naUTilus input files and the display of the model results. The network capabilities in ArcInfo allow for the easy identification of a desired sewer stretch and its tributaries. Once the network is identified, data extraction for the creation of the naUTilus input file is more straightforward when the sewer network can be visually displayed. Also as shown, the visual display created in ArcView allows easy identification of emissions hot spots predicted using naUTilus.
The interface between ArcView and naUTilus could not be applied to the data supplied by the City of Austin. Although I could not determine the exact problems, some of it could be attributed to the format of the data.
Since this project is just a small portion of the work associated with my master's thesis, there are many items that I hope to address in future work. Some of these are listed below.
Edit the ArcView scripts from the interface to better handle existing data
Adjust naUTilus to more effectively model a municipal sewer
Use experimental data to calibrate the improved naUTilus model
Apply the model to a large municipal sewer network and identify hot spots
How, Cindy Fee Ha. 1998. Fugitive Emissions of VOCs from Industrial Sewer Networks: Integration of naUTilus and ArcView, Master's Thesis.
How, Cindy. 1998. Using ArcView as a User Interface for naUTilus, on-line user document (http://www.ce.utexas.edu/prof/maidment/grad/how/avnaut.html)
Olson, D.A., S. Varma, and R.L. Corsi. 1997. naUTilus: A Model for Predicting Chemical HAP Emissions from Industrial Sewers; I. User Documentation, II. Technical Documentation, report to the United States Environmental Protection Agency.
Back to the GIS Class Page