Unit closed properly: what a complete record looks like
Unit closed properly grades the record rather than the vegetation. GO 95 Rule 18A(1)(b) sets a three fact standard, but it opens on completion of the corrective action, so it reaches remedial work alone. On routine cycle work the utility's own procedure is the whole standard, and the test is whether an outside reviewer can re-run the count.
A distribution work order packet closes on Friday, October 3, 2025. A status changes in a work management system, a completion date lands in a field, and 38 spans stop being work in progress. Nothing about the trees changed that afternoon. What changed is what the utility is now entitled to say about them.
Unit closed properly is the audit line that tests that claim. It is graded on the record rather than on the conductor, so it can fail on a span where every cut was correct.
The unit is whatever the closing system says it is
The unit is a creature of the program rather than of the code. Each program defines its own, and that definition decides what a closure is able to assert.
PG&E states what a distribution unit has to be before quality control will accept it. Its Quality Control Vegetation Management Business Process Document, MID Quality 0002 at Revision 02, published April 10, 2024, puts four conditions in one line:
Work Order (WO) packets with all spans 100% VMI complete, with all VMI work completed in the current Calendar Year, from the current project plan and in the HFTD (Tier 2 and 3).
Every span inspected. The inspections inside one calendar year. The packet drawn from the current plan. The location inside Tier 2 or Tier 3. A packet failing any one of those is out of scope whatever its status field says, and the separate readiness gate that 85 percent of prescribed tree work is completed and documented sits on top of them.
PacifiCorp closes a different object. Its evidence to the state is a work release, one per circuit on distribution and one per line on transmission, carrying a distance in miles rather than a count of trees. The 2025 Wildfire Mitigation Plan Independent Evaluator Annual Implementation Report for the utility was prepared by NV5 and Guidehouse for California's Office of Energy Infrastructure Safety. It records the evaluator asking for one work release per circuit across 20 circuits, and one per line across 11 transmission lines.
Neither unit converts into the other after the fact. A program that prescribes by tree and closes by mile carries both numbers or it carries an argument, because the tree count is what the crew was paid against and the mileage is what the state was promised. Defining the lot before the sample is drawn is the same decision, made early and on purpose.
The three facts California asks for attach to a correction
GO 95 Rule 18A is the only place in the California stack that states what a finished piece of work has to show. Rule 18A(1)(a) opens by naming what the rule is about:
Each company (including utilities and CIPs) is responsible for taking appropriate corrective action to remedy Safety Hazards and GO 95 nonconformances posed by its facilities.
Rule 18A(1)(b) then sets the record standard, and it begins on the same trigger:
Upon completion of the corrective action, the company's records shall show, with sufficient detail, the nature of the work, the date, and the identity of persons performing the work.
Three facts. What each regime asks a clearance record to carry is its own subject, and so is the ten year clock that hangs off that opening clause.
Read the trigger, because most closed units never reach it. A span on a routine cycle that was in conformance on the day it was inspected posed no Safety Hazard and no GO 95 nonconformance. The crew prunes it to hold the next cycle, the unit closes, and Rule 18A has not been engaged at any point. For that unit the Commission leaves the content of the record entirely to the company.
So the answer to who decides is split. On corrective work the Commission fixes a floor of three facts and the company's own auditable maintenance program builds above it. On cycle work, which is the bulk of a vegetation program by volume, the standard is entirely the contract, the utility procedure, and the wildfire mitigation plan commitment behind the cycle. An auditor grading unit closed properly on a cycle span is grading a contract term, and the finding has to cite one.
A closure picks a status, and the list of statuses is short
Closed and complete are separate states, and the vocabulary that separates them is defined. MID Quality 0002 carries the definition on page 8:
Constraint: A condition where a property cannot be accessed due to customer access or contact issues, weather, accessibility, safety concerns, other inspections are in progress, all work is not complete, or there are other issues that prevent an accurate inspection.
Seven conditions under one word. Six are about the property or the calendar. The seventh, all work is not complete, is about the record: it describes a unit that reached the audit queue because something upstream reported it finished.
What follows is on page 2, in the distribution out of scope list:
Span is constrained due to incomplete work, or other issues preventing an accurate inspection.
The span leaves the sample. It does not enter the fail column and it does not move a pass rate, so a premature closure is invisible in the quality number and visible only in the count of constrained spans, which is a different report with a different audience. A program that reviews its pass rate monthly and its constraint counts annually finds out a year late. Where the constraint is a customer saying no, the two artifacts a refusal has to produce are a separate file again.
Which event the completion date names
A single unit generates four candidate dates:
The day the crew finished.
The day the paperwork was submitted.
The day a reviewer signed it.
The day the system posted the status.
Nothing requires them to fall on one day, and nothing requires the record to say which of them the completion field holds.
PacifiCorp's inspection year runs on its own clock. The evaluator found circuits inspected in late 2024 counted against the 2025 plan and recorded the explanation at page 68:
PacifiCorp explained that its inspection year runs from October to October, meaning that Q4 2024 inspections on circuits or lines planned for vegetation maintenance in 2025 are tracked as part of the 2025 work plan.
That convention is legitimate and it is stated. The trouble starts where the closure dates and the reported progress stop agreeing with it. At page 63 the same report says:
However, the IE notes that the progress reported in the 2025 Q4 QDR erroneously shows that the 826 line-miles were completed by the end of Q3, when two circuits were completed in Q4, as acknowledged by PacifiCorp.
The line-mile total was right. The distribution of it across quarters was wrong, because two closure dates sat outside the window the report claimed for them. Summarizing at page 76, the evaluator listed "incorrect completion dates" among the data quality issues it met in the work releases themselves.
The practical rule is to name the event in the field label rather than in a procedure that gets read once a year. Work completed, record submitted and accepted are three fields, and a program that keeps only one of them cannot answer a question about the other two. What the acceptance signature is attesting to is the third of those dates and the one with a name against it.
One line, two contractors, two signed work releases
The evaluator's test on a patrol inspection closure is three parts, and it is worth copying. At page 68, on a set of 11 transmission lines:
The IE examined each of the 10 work release PDFs to ensure that it contained the appropriate code for distribution patrol inspections ("FIT"), covered the full length of the line consistent with the tracking document, and had a signature attesting to the inspection's completion in 2025.
The activity code, the extent, and the signature with its date. Each is checkable from the document alone against a second document held outside the closing team. The sentence calls FIT a distribution patrol code inside a section about transmission patrols, which is the report's own slip and changes nothing about the test.
Two of the lines failed the second part in a way the paperwork concealed. At page 69:
Additionally, for two lines, PacifiCorp provided two signed work releases per line, each of which suggested the full distance of the line was completed and did not include any explanatory notes. The IE inquired about this, and PacifiCorp explained that for each of these lines, the work was split between two contractors.
Two crews, one line, and two closures each reading as the whole. Nobody signed anything untrue. The form had one distance field and no way to say "my half," so each contractor closed the only unit the form offered, and the arithmetic of the two documents added up to twice the line. The evaluator's conclusion is the sentence to remember, and the utility's fix is the cheap one:
The IE noted that signed work releases suggesting a contractor inspected the full length of the line, when that contractor only completed part of the inspection, generally degrade the value of signed work releases as evidence of inspection completion. PacifiCorp stated that it would add a "Notes" field to the work release form so that contractors could indicate when they only inspected a fraction of the distance on the work release.
A form with no field for a partial will record a whole. That is the first thing to look for in any closure screen, and it is a design question rather than a discipline question.
What the closure has to survive when the state asks for it
The closed record is what a count in a regulatory filing rests on, and the path from one to the other runs the same way every year:
The reported figure. A line in Table 1 of the Quarterly Data Report.
The population. A tracking workbook listing every unit behind that figure.
The sample. A random draw from the population, in units of the thing reported.
The closure artifact, per sampled unit. The document that proves that one unit finished.
In the 2022 Wildfire Mitigation Plan Independent Evaluator Annual Report on Compliance for PacifiCorp, the fourth step arrived as one named file. The evaluator drew 29 line-miles and the utility answered with "Trees Completed Work Release 2022.pdf showing dates, authorizing signatures, supervisory approvals, and additional clarification/comments," at page 35.
The same page carries the cost of a closure set that fails to reconcile. PacifiCorp's 2022 plan committed 354 transmission line-miles, the fourth quarter report showed 386 as both target and completion, and the workbook behind it reported 374.4. Three numbers for one year of closed work, and the one the evaluator wrote down was the one the records supported.
Status words travel the same road. At page 39 the evaluator recorded that "[s]everal of the initiatives and goals were marked as 'complete' in Table 1 of the Q4 QDR while coming in under target." Complete is a status, not a measurement, and it will be read as both by somebody who never saw the work plan. What the utility receives from its QA/QC program has to include the reconciliation, because the filing will be read against the file.
The finding the audit opens is a second closure
Closing the unit and closing the finding written against it are two events in two systems, and at PG&E they belong to two groups. MID Quality 0002 assigns the second one away from the people who wrote it:
Execution is responsible for MONITORING QC findings and ensuring final closure in the QC system of record.
Quality control records its own activity in ArcGIS Field Maps and Survey123, which the same document names as its systems of record. So the finding is created in the quality system, the corrective work happens in the vegetation system, and the closure has to come back and land in the quality system. That return trip is the step that gets dropped, and it is the step an outside reviewer looks for first.
The 2022 evaluation says what the gap looks like from outside. Reviewing PacifiCorp's 2022 VM Audit Exception Report at page 55, it confirmed that the audits were being performed and by whom, then noted: "Although likely captured elsewhere, the report does not indicate whether or when the exception was resolved."
Likely captured elsewhere is the whole problem. A resolution held in another system, in another team's spreadsheet, becomes evidence only when somebody joins the two, and the joining gets done under time pressure during an audit. Where a failed unit goes and who carries it is the process behind that field.
A closed record is a statement about one day
PG&E's SW-0003 is explicit that a closure is not provisional:
Once the survey is closed as unable to access, it will be documented as a constraint. Do not make a second attempt to submit a survey if access is obtained later.
A specialist who knocked, got no answer and closed the survey has recorded something true about that day. Access granted three weeks later does not make the earlier record wrong, and editing it would destroy the only evidence of the attempt. The later visit is a new record. Amending an entry without destroying it is the general form of the same rule.
The cheaper control sits before the closure rather than after it. In the 2025 report at page 50, PacifiCorp described how work orders close on its weather station maintenance program, in a different asset class and a different system:
The work orders are only closed when there has been a verified maintenance form submitted for that equipment.
The gate is at the point of closure. A work order that has no verified form cannot reach a closed status, so the reconciliation never becomes an audit finding, because the state that would produce the finding cannot be entered. At page 76 the evaluator asks for the same discipline on the vegetation work releases, and asks for it later in the process: that they be "returned by contractors in a timely manner and checked for complete, accurate, and clear information before filing."
What a closed unit has to name
The fields below are the closure half of the file. They assume the finding, the measurement and the prescription are recorded elsewhere and reachable.
The unit, in the identifier the closing system uses. Work order packet, circuit, line, span as an ordered pair of structures, or tree. Named in the unit the figure will later be reported in, and mapped to the identifier the prescription used if those differ.
The status, from the program's own defined list. Complete, constrained, or out of scope, with the constraint category where one applies. A free text status field guarantees that next year's count is a judgment call.
The extent actually covered by this closure. The miles, spans or trees this document is claiming, which is the field the two contractors on one line had nowhere to put.
The date, with the event it names. Work completed, record submitted, or acceptance signed, labeled as one of those rather than left as "date."
The person, and the authority they closed under. The crew or contractor who performed the work and the reviewer who accepted it, as two entries.
The prescribed against performed reconciliation. Trees prescribed, trees worked, trees remaining, and the reason for any remainder. A closure with a remainder and no reason is the case the audit is looking for.
The pointer to the finding this closure answers, where there is one. A finding identifier carried onto the closure record, so the return trip into the quality system can be demonstrated rather than asserted.
The document behind any status that is not complete. The refusal file, the permit, the access constraint note, or the exemption, named by identifier.
Assembled, one unit reads as a line an outside reviewer can test on its own:
Circuit 1104, work order packet WO-2025-4471, 38 spans, 12.4 line-miles. Status complete. Work type routine distribution trim. Prescribed 214 trees, worked 211, 3 constrained as customer refusal under refusal file R-2025-0188. Work completed September 29, 2025, record submitted October 1, 2025, accepted October 3, 2025 by [forester]. Closes QC finding QC-2025-0912, returned to the QC system of record October 6, 2025.
Every element in that line is checkable against a tracking workbook, a named procedure, or another record in the same file. The one that takes the most work to build is the last, and it is the one an evaluator asks for first.
Where the closure record stops
A closure record is evidence about work and it cannot improve the work. A span left short is short whatever the packet says, and a complete file describing an incomplete trim is worse than a thin one, because a reviewer can test every line of it and will.
The record also stops at the edge of what a status word can carry. Complete means the program's own definition of complete on that day, under that contract, in that unit. It does not mean the target was met, as page 39 of the 2022 evaluation records. It does not mean the number is right, as the 386 against 374.4 records. It means a named person, on a named day, wrote down that a defined unit was finished, and left behind enough for somebody who was not there to agree.
Further reading
What a QA/QC audit actually covers in utility vegetation work: The unit under audit, the evidence file, and the deliverables the closure record feeds.
Who signs off on a span, and what that signature means: The attestation that sits on top of the closure, and what it is attesting to.
Rework: what happens to a failed unit, and who carries it: Where a unit goes after the audit reopens it, and who owns the second closure.
Frequently asked questions
Does any California rule say what a completed vegetation work record has to contain?
GO 95 Rule 18A(1)(b) does, for corrective work. It requires that upon completion of the corrective action the company's records shall show, with sufficient detail, the nature of the work, the date, and the identity of persons performing the work. Rule 18A(1)(a) sets what that trigger covers, which is corrective action to remedy Safety Hazards and GO 95 nonconformances. A routine cycle prune on a span that was in conformance when it was inspected is neither, so no Commission content standard reaches its closure. The contract and the utility's own auditable maintenance program do.
Is a unit that was closed before the work finished a failed unit?
Usually not, and that is the problem. PG&E's MID Quality 0002 defines a Constraint to include the case where all work is not complete, and its distribution out of scope list removes a span constrained due to incomplete work from the review. The span leaves the sample rather than entering the fail column, so a premature closure does not move the QC pass rate. It shows up in the count of constrained spans, which is a different report.
Which date belongs in the completion field?
The utility's procedure decides, and the record should say which event the date names. A single unit generates the day the crew finished, the day the paperwork was submitted, the day a reviewer signed, and the day the system posted the status. California's Independent Evaluator listed incorrect completion dates among the data quality issues that degraded the value of PacifiCorp's signed work releases as evidence in its 2025 report, at page 76.
Who closes the finding a QC audit opens?
At PG&E, execution rather than quality control. MID Quality 0002 states that execution is responsible for monitoring QC findings and ensuring final closure in the QC system of record. That splits the two closures across two groups and two systems: the unit closes in the vegetation system of record, and the finding closes in the quality system of record.
Can a closed unit be reopened when the situation changes?
Not in every program. PG&E's SW-0003 tells a specialist who has closed a survey as unable to access to document it as a constraint and not to submit a second survey if access is obtained later. The closure is a statement about what was true on a date, so a later visit produces a new record rather than an edit of the old one.
