Critical Business Rule Validations That Can’t Wait
Not every data error is created equal. A few of them will stop your Utility Network cold.
Run a full data assessment on a legacy utility geodatabase and you will get a number that looks alarming. Tens of thousands of errors. Sometimes hundreds of thousands. Some people will panic, and some will write the whole thing off as noise.
Neither reaction is right. Most of those errors are inconveniences such as a blank install date, a material code that fell out of the domain, or a manhole ten feet from where it should be. They matter, but they don’t break anything. A much smaller group of errors are violations of the business rules that a Utility Network depends on to function at all. When they exist, subnetworks don’t build, traces don’t run, and the systems downstream of your GIS start making decisions on bad information.
In our data assessment methodology, those errors sit at the top of the scale: Severity 5. Keep reading to find out what puts an error in that tier, why the four categories that live there are so damaging, and why they need to be fixed before anything else.
What makes an error Severity 5
Severity in a data assessment isn’t about how many times an error occurs or how ugly it looks on a map. It’s about consequence. We rank an error at the top of the scale when it meets any of these tests:
- It prevents a subnetwork from being created or updated. Gas pressure systems and electric circuits are the operational units of a Utility Network. If a feature stops one from building, everything that relies on that subnetwork is affected.
- It makes a trace fail or return a wrong answer. An isolation trace that stops short, or a downstream trace that jumps into a neighboring feeder, is worse than no trace at all, because someone will act on it.
- It breaks a rule that downstream systems assume is true. Outage management, FLISR, pressure modeling, and leak survey scheduling all consume network data on the assumption that certain rules hold. When the rule is violated, the assumption is wrong.
- It blocks subnetwork creation outright. Some conditions cause the Utility Network to reject the feature entirely or generate Dirty Areas that must be resolved before the subnetworks can be updated.
Four categories of error consistently meet these tests. Two are domain-specific, one for gas and one for electric. Two are universal.
Gas: pressure validation
A gas distribution network is organized around pressure. Every pipe belongs to a pressure system bounded by regulators, and every pressure system carries a maximum allowable operating pressure (MAOP) that governs what can be safely connected to it. In a Utility Network, the pressure system is a subnetwork and the regulator is its controller.
That structure only works if pressure attributes are consistent across every feature in the system. The errors that violate it look like this:
- MAOP mismatches within a system. A run of main coded at 60 psig connected, with no regulator between them, to a run coded at 25 psig. One of those values is wrong, and the subnetwork can’t resolve which.
- Pressure-system boundaries that don’t line up with regulators. A pipe crosses from one pressure zone into another with no regulator at the transition, or a regulator sits in the middle of a run where both sides carry the same pressure code.
- Services that exceed the pressure of the main they tap. A service line coded higher than its source main is physically impossible and a compliance red flag.
- Missing or invalid pressure values on features the subnetwork depends on. A null pressure attribute on a main isn’t a blank cell. It’s a hole in the subnetwork.
Why this is Severity 5: a pressure system that can’t build is a pressure system your GIS can’t reason about. Isolation planning, regulator station analysis, and MAOP reporting for regulators all depend on it. And the consequence of a wrong pressure boundary isn’t a bad report. It’s a crew isolating the wrong segment.
Electric: phase validation
The electric equivalent of pressure is phase. Every conductor, device, and transformer carries a phase designation, and the Utility Network enforces phase continuity: a single-phase B lateral cannot be fed from a conductor that only carries phase A, and a three-phase device can’t sit on a two-phase line.
Legacy Geometric Networks never enforced this. The phase field was just an attribute, and it drifted for years. When a Utility Network starts enforcing it, the drift becomes visible all at once. Typical Severity 5 phase errors:
- Phase discontinuities along a feeder. Downstream conductors carry phases that don’t exist upstream. The subnetwork either fails to propagate or propagates incorrectly, and every downstream feature inherits the problem.
- Phase mismatches at tie switches. A tie switch connecting a phase ABC feeder to a phase AB feeder creates an ambiguity the network can’t resolve. Because tie switches are also where FLISR and switching orders operate, this is the single most operationally dangerous phase error.
- Transformers and services with phases their source conductor doesn’t carry. Customers appear to be served by power that, according to the model, doesn’t reach them.
- Devices with null or invalid phase codes at subnetwork boundaries. A breaker with no phase value can’t act as a subnetwork controller, and a circuit with no controller has no subnetwork.
Why this is Severity 5: phase errors don’t stay local. Because phase propagates through the subnetwork, one bad tie switch or one mis-phased riser can corrupt the phase designation of every feature downstream. Outage prediction, load balancing, and unbalanced-load studies all inherit that corruption.
Invalid geometry
Geometry errors are the least glamorous category on this list and the most likely to be a hard stop during migration. A Geometric Network tolerated a surprising amount of geometric garbage. A Utility Network does not.
- Null or empty geometry. A record exists, but the shape doesn’t. The feature can’t participate in the topology and will be flagged the moment the network is validated.
- Zero-length lines. Usually artifacts of snapping or an interrupted edit session. They break connectivity at the point where they sit and often mask a real gap.
- Self-intersecting or looped lines. A line that crosses itself has no defined direction, which matters when the network needs to determine flow.
- Multipart features. A single record with two disconnected pieces is one feature to the database and two to the topology. The Utility Network won’t accept it.
- Duplicate or stacked features. Two identical points on top of each other, or a line drawn twice, generate Stacked Point Dirty Areas and double-count assets in every downstream system.
Why this is Severity 5: Invalid geometry isn’t an error the network can work around. It’s a feature the network can’t load. Every one of these must be found and fixed and finding them by hand across a million features isn’t a realistic plan.
Logical connectivity
Geometry tells you where features are. Connectivity tells you whether they’re actually joined. The two frequently disagree, and the disagreement is invisible on a map.
- Features that touch but aren’t connected. A service line whose endpoint sits a few thousandths of a foot from the main. It looks connected at every zoom level. The trace stops there.
- Junctions with the wrong valence. A valve with three edges connected where the model says two. A riser sitting mid-line instead of at the end of a service. Either one confuses subnetwork controllers or creates mid-span terminal dirty areas..
- Overlapping edges from different systems. Two mains from different pressure zones, or two conductors from different feeders, sharing vertices at the same coordinates. The network sees a connection that doesn’t exist in the field, and two subnetworks bleed into one.
- Orphaned features. Devices and lines with no connected path back to any subnetwork controller. They’re invisible to every trace and every operational system.
Why this is Severity 5: connectivity is the network. A pressure system or a circuit is nothing more than a set of features that the topology says are connected to a controller. If that connectivity is wrong, the subnetwork is wrong, and every consumer of the subnetwork is wrong with it.
What this looks like in practice
Consider an illustrative mid-size combined gas and electric utility running an assessment: roughly 400,000 gas features and 900,000 electric features. The full assessment returns something on the order of 85,000 errors across all severity levels. That number is intimidating until it’s sorted by severity, at which point the picture changes:
- Gas pressure errors: around 1,100. Mostly MAOP inconsistencies concentrated in three older service territories where data was converted from paper records, plus about 90 regulator placements that didn’t match pressure-zone boundaries.
- Electric phase errors: around 2,400. Roughly 40 of them at tie switches. Those 40 were responsible for phase corruption on 14 feeders, which accounted for most of the remaining count.
- Invalid geometry: around 650. Dominated by zero-length service lines and a few hundred duplicate poles from an old field-collection merge.
- Logical connectivity: around 1,900. Mostly near-miss service connections and valves with unexpected valence.
That’s about 6,000 Severity 5 errors out of 85,000, or roughly 7 percent. Fixing those 6,000 is what stands between this utility and a Utility Network that builds subnetworks. The other 79,000 can be worked through methodically after go-live. The 6,000 cannot.
Two things stand out in that breakdown. First, Severity 5 errors cluster. They come from specific conversion projects, specific territories, specific field-collection efforts. That makes them tractable once identified. Second, a small number of root-cause errors (the 40 tie-switch mismatches) generate a large number of symptomatic ones. Fix the root, and the count collapses.
Why they can’t wait
They’re the errors that block everything else. You can’t validate a pressure system’s attributes until the pressure system builds. You can’t test phase-dependent business rules until phase propagates. Every lower-severity fix on a broken subnetwork is work you may have to redo.
They get more expensive the later you find them. A pressure mismatch found in a pre-migration assessment is a data edit. The same mismatch found after cutover is a Dirty Area in a production network, blocking edits in that area, with an operations team waiting on the fix.
They don’t stop appearing at go-live. Field crews will continue to snap service lines a fraction short. Contractors will continue to code phases from memory. A Severity 5 error introduced into a live Utility Network has the same consequences as one that was there from the start. This is why validation for these categories has to be continuous, not a one-time gate.
How GeoData Sentry approaches it
GeoData Sentry was built for exactly this problem: running the business-rule validations that a Utility Network depends on, automatically, across the entire geodatabase, and reporting results in a form a GIS team can act on.
- Tests mapped to severity. Pressure consistency, phase continuity, geometry validity, junction-to-edge valence, vertex overlap, and logical connectivity each run as discrete, configurable tests, so the Severity 5 categories can be isolated and prioritized.
- Attribute-filtered detection. Overlap and uniqueness tests can be filtered by FEEDERID, pressure zone, or any attribute, which is how a crossover between two subnetworks gets separated from a legitimate shared vertex.
- Scheduled runs against an enterprise databases. Validation doesn’t stop at migration. The same tests run nightly or weekly against the live geodatabase, catching new Severity 5 errors before they reach a Dirty Area.
- Comparison reporting. Run-over-run comparisons show whether the Severity 5 count is falling, where new errors are being introduced, and which fixes are actually holding.
The goal isn’t zero errors across the board. That’s not a realistic target for any utility geodatabase. The goal is zero errors in the categories that stop the network from working, verified continuously, so the rest of the cleanup can happen on a schedule that makes sense.
Start with the errors that matter
If your utility is fighting a persistent backlog of Dirty Areas, the first question to answer is simple: how many Severity 5 errors do you have, and where are they?
Laurel Hill GIS can run a data assessment against your geodatabase and give you that answer. Contact us to schedule an assessment or to see a demonstration of GeoData Sentry’s Utility Network validation tests.
