Building an ADMS network model extractor seems like a straight-forward task, but there are always complications. In this note, I lay out the major issues we see time and time again.
Source of record
ADMS data, contrary to what the ADMS vendors seem to expect, is not all in the GIS. It’s often in silos across the utility. Each attribute may also have multiple potential sources of record, i.e., transformer impedance. The rule may be
· If it’s present in GIS, use it, otherwise get it from SAP.
· If not there, it’s on Joe Smith’s spreadsheet on Drive D:.
· If none of those, use the value from the planning tool.
Often, part of the building the extractor will be to formalize the source of record away from desktop spreadsheets into GIS or EAM systems. This can add a lot of time and effort to the process, so this “Discovery” phase is best done earlier, as part of a data readiness study
Method of verification
Once you have the data extracted, we need a method to know if it’s correct or not. Having a visual display capability is really helpful. If you can view the data in the new ADMS, that’s great, but there may be obstacles which prevent that in the early stages. Having some way to validate that you’re actually getting the entirety of a feeder, whether that’s a visual inspection, or some other method, is a must. We typically export the data into the planning tool as another means of verification, which has additional benefits.
Defaults method/recording/measuring
When dealing with missing data, it’s easy to say we will just default it. However, the choice of values has a huge effect on the results of the advanced apps and they must be chosen with care, and in consultation with the ADMS vendor. Identifying bad data is a lot harder, although its likely the planning group already knows which data is good vs bad. A good example is determining that a legitimate conductor type is incorrect, for a specific conductor. You need to focus on prioritizing data corrections by its relative impact on the target applications. Conductor types and phasing are critical, compared to say conductor crossarm spacing. Transformer voltages are likewise critical to a basic powerflow, which forms the basis of most of the advanced apps.
Quality Checks/Continuous Improvement
The earlier problems are identified, the better. As you work through extractor issues, you will likely find chronic issues in internal processes for capturing data. These must be identified and addressed, not just on a one time cleanup effort, but to prevent them from re-occurring. We tell our customers that your first day of the system should be its worst day, meaning that as data quality continually improves, so does ADMS performance. While establishing data governance is a topic for another day, a continuous improvement process and mindset has to be established in order to achieve the desired outcomes of the ADMS project.
In summary, this is a problem we’ve faced as an industry for a long time. In fact I built my first extractor for load flow in the early 1990’s. To some extent while the problem is the same, our ecosystem of system of record is becoming increasingly complicated and latency and quality requirements are more stringent. This is a labor of attention to detail. Follow these points I’ve highlighted and you’re experience building the extractor will be shorter, of higher impact, and definitely more pleasant.