Nearly 20 years ago I had the opportunity to learn a chilling lesson from a seasoned GIS manager who worked for a large gas utility. He was nearing retirement and had a lot of experiences to share. It has stuck with me for a long time and I have wanted to share it with others with the intent of raising the issues of the importance of managing spatial information related to utilities and infrastructure data.
Here is how it goes: This GIS manager had a team of mapping technicians that were adding gas service data to their GIS database. They were moving along, adding service, after service, connected to the gas mains. When they got to an intersection of another main they kept adding the service data. The map wound up looking like figure 1, below.
Figure 1:(Please note that these diagrams are for illustration only and have been highly simplified; they do not reflect the actual configuration of mains, valves, pressure regulators and other appurtenances that would normally show up on a detailed GIS distribution map.)
When it really should have looked like what you see in Figure 2:
It was really quite a subtle change. The service to the structure on the corner was fed from the gas main that is running North-South, rather than the actual main that was running East-West. No one noticed it, for a while, anyway.
Sometime later it was necessary to upgrade the East-West main. It was set up to run at a higher pressure and the buildings that were serviced by that main received new orifices to accommodate the change. As you might imagine, the town home on the corner did not get upgraded because the maps showed that it was serviced from a different main.
That town home does not exist anymore. It was completely destroyed in a gas explosion caused by a mapping mistake. It was fortunate that the building was unoccupied at the time so there were no injuries, but it was still a tragedy that resulted in significant property damage. As GIS professionals and engineers we have an obligation to do everything we can to promote the best practices in managing spatial information. It’s fair to ask the questions, as to where things break down under these circumstances, and how can we change things in the future to avoid these kinds of issues.
There are always good lessons to be learned from tough situations so here are a few ideas I would like to share with you. This is intended to supplement those necessary quality assurance / quality control practices that should already exist in any well-managed mapping organization.
Train your people.
Yes, it’s trite. Most training for GIS technicians is about how to lay down lines and points and how to do it efficiently, especially in the context of data conversion processes. That’s good, but training needs to also address the nature of what is being done, specifically what is being done at a business level, and the implications of things not being done in accordance with required business practices. Gas can explode, electricity can electrocute. These facts need to be understood at every working level of the business.
The time context of this particular issue, having occurred in the mid-1990s, is also important. This is work that was done at a gas utility. It was common practice back then to do much of the data maintenance work in house with trained technicians in the office. Today, much of this work is done with external contractors, mostly on an off-shore basis, which adds large distances, time zones and different languages to the challenges of proper business communications. The implications of these practices must be managed accordingly.
Be very cautious of automated processes that generate spatial information. I’m not referring to electronic translation of data, but rather auto-generation of data. For example, some organizations do not have detailed information on services so they are auto-generated, often based on customer address information. The algorithms generate data from the main to the centroid of the lot, or some other approach. This may convey important information (e.g. there is a service associated with this lot) that needs to be on a map but there can be smarter (and safer) ways to convey this information. For example, information that is auto-generated, or of more suspect quality, can be shown more subtly on the map, in grey-tones so that it will not be confused with accurately placed, field verified information.
Replicate with Caution
GIS data gets replicated. While organizations have been striving to have one common source of truth in their spatial information for several decades, the real truth is that there are copies of this information all around the organization. Some is paper, some is digital. It exists on servers, laptops, tablets and a variety of other devices. The increased availability of spatial information is a good thing, but not putting in the effort to manage changes back to the master database can create major problems.
How to properly replicate should be the subject of a separate post, but here are some things to think about. They include controlled replication schemes, proper metadata management and well-managed processes to support delivery of a common operational data.
Use Automated Validation Routines
With today’s systems, unlike 20 years ago, there is much more computing power to support improved validation of data. Different checks can be performed to ensure data connectivity and enforce business rules. It is also much more straightforward improve visual checks for those less-common issues that may come up.
That said, the problem described in this article was a tricky one that would fail most validation tests. Having the proper human intervention and expertise involved needs to always be in place as a final check.
Enhance Your Culture
Beyond training and lots of extra tools and algorithms, consider enhancing the culture of your mapping organization to align it to the mission of providing safe, reliable utility services to your organization, or the organization you serve. This can give the GIS technicians a great sense of purpose. Many of them are highly educated and want to contribute by doing more than putting down points and lines on a map.
Twenty years ago, when this incident happened, utility organizations were under the impression that data would be converted and the project would be done. We know today that GIS data conversion doesn’t really end. More information is always necessary to better manage utilities and other infrastructure-intensive organizations.
I’d like to Hear from You…
As fellow professionals in the industry, I welcome your positive contributions to how we as GIS professionals, engineers, computer scientists and information technologists can increase the quality of the products delivered by our profession. I look forward to hearing from you.
The Next Post:
While this post has focused on the technical aspects of how these problems can occur, the next post will focus on management culture and how it can impact quality of spatial information for utilities. We will use publicly available information on the 2010 San Bruno gas pipeline explosion as a case study.