DCP Midstream, a gas gathering and processing company, has over 60,000 miles of pipelines which are maintained in an enterprise GIS solution. The GIS database has over 158,000 linear pipeline segments in ten non-contiguous States. To import the pipelines into the "One Call, Call Before You Dig" software where only polygons are accepted, a query needed to run to capture only the company operated pipelines and to create a buffer. This process coulid not be handled by the GIS software itself, but FME came to the rescue. In this session, we will explore how FME was used to address this challenge.
1. Using FME to Overcome
General GIS Software
Limitations
Alicia Foose, DCP Midstream
2. DCP Midstream Overview
DCP Midstream, LLC, a 50-50 joint venture
between Spectra Energy and ConocoPhillips,
is headquartered in Denver, Colorado.
The Company leads the midstream segment
as one of the nation’s largest natural gas
gatherers and processors in the United
States.
DCP Midstream is the largest natural gas
liquids (NGLs) producers in the nation.
3. DCP Midstream Overview
The Company owns or operates 58 plants, 10
fractionating facilities, and approximately
60,000 miles of gathering and transmission
pipeline with connections to approximately
38,000 active receipt points.
Visit https://www.dcpmidstream.com for
more details.
6. GIS Environment
Oracle database
ESRI SDE
Pipeline Open Database Standard (PODS)
4.02 database model
http://pods.org/
The volume and complexity of data can create
challenges for GIS analysis.
7. A Recent Project
DCP Midstream went through an evaluation
process to find a software solution to manage
the One Call (Call before you dig) process.
One of the solutions only accepted polygon
features. Because all of the pipelines in the
PODS database are polylines features a buffer
needed to be created for the pilot.
To keep the comparisons similar we decided
to buffer the pipeline by one foot.
Only the location of the pipelines were of
interest.
9. Volume Of Data
We were only interested in the pipelines we
operated so a query was necessary.
The pipeline layer being used has 164,535
polylines in the database.
SQL> select count(*) from PODS.REGULATORY_SEGMENT;
COUNT(*)
----------
164535
Laptop processing capacity along with
memory limits can become an issue when
buffering this volume of data.
10. Creating The Buffer In FME
The buffer was created in FME because
It is easy to set up
It can run in the background
It doesn’t seem to use as many resources
It tends to run faster on my environment
It has an aggregate feature
It can filter attributes
11. Creating The Buffer In FME
The goal was to:
Query for only DCP Midstream Operated
pipelines.
Simplify the data by eliminating most of the
columns.
I chose to keep Region because there are only 10
regions (Regions have a logical geographical area)
Buffer the pipelines by 1 foot.
Aggregate the data.
Export the Polygon feature to an ESRI shape
file format.
12. Query For DCP Operated
The data was queried directly
from the SDE connection in
FME – this filters the data on
the fly.
The 164,535 rows were
reduced by 2,829 to total
161,706 records to buffer.
13. Transformers Used
The AttributeKeeper was used to reduce the
number of columns from 51 down to 6
keeping only REGION_NAME from the SDE
layer. Because only the location of a pipeline
was required, the associated attributes were
not needed.
14. Transformers Used
The Reprojector was used to project the data
from NAD 83 to a projection with a unit of
measure in feet.
US48-DUKE was chosen because the
projection was created for the continental US
and has relatively little overall distortion.
16. Transformers Used
The Aggregator was used to aggregate the
data using the REGION_NAME to group by.
Aggregating the data reduced the number of
records from 161,706 to 10.
17. Transformers Used
The Reprojector was used to project the data
back to NAD83.
Finally, the destination dataset was set to a
shape file format. A visualizer was used so
the output could be viewed right away.
A dissolver transformer was not used because
the aggregate combined all of the polygons
into Regions and the overlaps were not a
concern for the end use.
19. Final Results
Total Features Written 10
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Translation was SUCCESSFUL with 0 warning(s) (10 feature(s)/5295013
coordinate(s) output)
FME Session Duration: 7 minutes 9.2 seconds. (CPU: 141.9s user, 10.6s system)
END - ProcessID: 1480, peak process memory usage: 208120 kB, current process
memory usage: 53868 kB.
20. A Side By Side Comparison
A 1 foot buffer was run in Arc Info using the
same query on the same layer.
Dissolve by field - REGION_NAME was
selected because it is the closest option to the
FME Aggregate .
21. A Side By Side Comparison
Executing (Buffer_2): Buffer PODS.REGULATORY_SEGMENT Server
1_ft_Buffer.shp "1 Feet" FULL ROUND ALL #
Start Time: Thu Mar 11 08:29:01 2010
Dissolving...
Output feature 0 cannot be dissolved into other inputs because of memory
limitations
Output feature 1 cannot be dissolved into other inputs because of memory
limitations
…
…
Output feature 15 cannot be dissolved into other inputs because of memory
limitations
Executed (Buffer_2) successfully.
End Time: Thu Mar 11 11:12:32 2010 (Elapsed Time: 2 hours 43 minutes 31
seconds)
22. A Side By Side Comparison
The FME translation ran in 7 minutes 9.2
seconds with no memory errors.
The Arc Info Buffer wizard ran in 2 hours 43
minutes and 31 seconds with memory
limitations errors.
The results from either process were
acceptable.
23. Annual Tax Project
Benjamin Franklin once said that “In this
world nothing is certain but death and taxes”.
So lets talk about taxes, specifically property
taxes. You might be asking yourself what on
earth does FME have to do with property
taxes. Well here is your answer-
Each year companies with tangible assets pay
property taxes. Pipelines are not excluded.
24. The Challenge
Every State has unique taxing districts by
which they collect and distribute property
taxes.
Tax districts can change from year to year
although most remain the same.
Population shifts and demographics are the
most common cause of tax boundary
changes.
DCP Midstream operates primarily in 17
States so tax boundary maintenance is a
fairly large undertaking.
25. Tax Project
Each year the GIS department provides the
Tax department with a report of how many
feet of each pipeline is in what tax district by
install year, diameter and so on.
The first step, for States with electronic data,
is to download the current tax boundary files
and update the SDE layer with the changes.
The SDE Layer has to be topologically clean.
Neighboring states do not tend to use the
exact same state line. This creates gaps and
overlaps which are ugly to clean up
particularly along rivers.
28. Oklahoma Tax Districts
Who can tell me what changed?
Going once
Going twice
Going three time
How are you going to find out?
FME has a transformer named Matcher which
detects both geometry and attribute changes
from two files.
29. Lets See What Changed
The 2008 Oklahoma tax districts are added as
one source.
The 2009 Oklahoma tax districts are added as
another source.
Both are run through the Matcher as input.
The Not_Matched features are output to a
visualizer so they can be looked at.
The Not_Matched features are output to a
shape file to be used in ArcMap for updating
the SDE layer.