Process Devised For Monitoring Leak Threats Using GIS

Distribution Integrity Management Rule
By James Stout, UGI Utilities, Inc., and Tony Sileo, Opvantek, Inc. | February 2009 Vol. 236 No. 2
  • Filter repairs by cause and material to focus on specific threat categories.
  • Cover at least one normal leak survey cycle (two is better) across the entire system.

By at least capturing the leaks discovered during the most recent leak survey on every active pipe in the network, a risk-assessment system will initially focus attention on areas with recent corrosion, joint leaks, or other time-correlated threats.

When attempting to associate a leak repair with the correct facility in a geographic information system (GIS), there is a series of challenges that must be addressed. These include poor address quality, no zip code or city, multiple candidate address locations, facilities drawn relative to inaccurate (legacy) land-base maps, and retired facilities that are no longer available in the geographic information system (GIS).

Geocoding Process

To address all of these challenges, it is best to implement a multi-stage approach to leak geocoding and pipe association:

  1. Construct Candidate Addresses. For each record, construct a series of possible addresses using each of the possible street aliases and each overlapping zip code.
  2. Geocode. Obtain all candidate locations for each candidate address.
  3. Find Candidate Facilities. For each candidate location, search (in the GIS) for candidate facilities based on attributes of the repair record (main pipe, service pipe, valve, regulator and material type.). Use a maximum search radius that is based on the relative accuracy of the land-base. Widening the search radius will lead to more false positive matches and will also increase the overall processing time.
  4. Compute Confidence Score. For each candidate facility, use a combination of available attributes in the repair record and the facility record to compute a score indicating the confidence that it is the correct facility. Use the actual proximity as part of the confidence score (preference for facilities that are closer to the address location when all other attributes match).
  5. Select Best Location. Select the candidate site and nearby facility with the highest confidence score. Place the leak on the facility, at the point closest to the best address location.

It is also worthwhile to retain the following information as part of the process:

  • Coordinates of the best address location.
  • Confidence score.
  • Direct relationship of the leak to the best matched facility.

Finally, note that the same basic approach can be applied to match pipe inspection reports or service card locations to the best location in the GIS.

What About The Rest?

Once all leaks have been processed and matched to the best candidate location and facility, there will be a (hopefully small) set of leaks with either no candidate location, or with confidence score below some established threshold. It may be worth manual intervention on some or all of these remaining leaks to attempt to determine where they belong. When considering this remaining manual effort, it is also feasible to filter the list to include only recent leaks or only leaks of certain types. This determination might be based on an overall threat analysis (e.g., manually review all remaining cast iron breaks, but only look at corrosion leaks from the last five years).

Reusable Tools