Short, true accounts from migration engineering. Identifiers withheld, quantities altered, failure modes exactly as found.
VALD ENGINEERING · FIELD NOTES · No. 1
The Panel That Was Never Commissioned
How a dozen live analog points came to be documented on dead steel - and how the record was corrected before a crew walked to it.Download PDF ↓
THE SETUP
During migration engineering for a large municipal wastewater treatment plant, the issued wiring schedule for one control unit listed a dozen in-service analog points terminated at a remote I/O panel nearby. On paper, everything agreed: the control database bound those points to the panel, and the schedule repeated the database.
THE CONTRADICTION
Reconstructing the installed plant from primary evidence told a different story. The remote I/O panel had been staged years earlier for a tie-over that never happened. It was never commissioned. The bindings were software configuration overlaid on the old I/O - the field conductors never moved. Every dump of the control database ever taken repeated the same fiction, because a database can only tell you what it was told.
THE RESOLUTION
Cross-vintage reconciliation - an older database export diffed against a recent point census - recovered the actual legacy terminations channel by channel, including landings no surviving record had ever serialized, rebuilt from channel-allocation arithmetic. The corrected schedule was issued before the package reached a cutover plan. Without the catch, a termination crew works dead steel on outage day, on the clock, while the live conductors sit two cabinets away.
THE LESSON
A clean conversion of the wrong plant is still wrong. Documentation is a claim; the installed system is the fact. Migration engineering that does not reconcile the two before the outage window is scheduling its surprises for the one day they cost the most.
VALD ENGINEERING · FIELD NOTES · No. 2
Off By One Slot, Three Panels Out of Four
The I/O lists were right. The drawings were right. Joining them without checking would have landed every point one slot wrong.Download PDF ↓
THE SETUP
A migration design package for a set of sibling control panels arrived in two halves: I/O list spreadsheets defining every point, and panel drawings defining every terminal. Each half was internally consistent. The natural next step - the step every schedule builder takes - is to join them: spreadsheet row to drawing terminal, by slot and point number.
THE CONTRADICTION
Reconstructing both halves into one governed model surfaced the trap: on most of the sibling panels, the spreadsheets numbered their slots offset by one relative to the drawings. Only one panel of the set was drawing-aligned. Nothing in either document said so. Every panel looked like every other panel, and both halves of the package were "correct" - they just disagreed about what slot one meant.
THE RESOLUTION
Exact-match cross-referencing of field device tags between the two halves detected and calibrated the offset panel by panel - the alignment that produced matched tags was decisive on every panel, and the assembled schedule carries a validation column so any row that did not reconcile says so on its face. A crew working a hand-joined schedule would have terminated every point one slot off on most of the panels, and the errors would have surfaced one loop check at a time, during the outage.
THE LESSON
Two internally consistent documents can still contradict each other, and the contradiction lives in exactly the join nobody documents. Migration engineering that treats document assembly as clerical work schedules that discovery for cutover day. Reconcile the join, prove it with the plants own tags, and make the validation visible in the deliverable.
