The Survey That Does Not Add Up

Reportable as-is: the readings taken with verified interruption, a checked half-cell, and a healthy meter.

Must not be mixed with anything else: the 300 m of current-on readings. Those values include the IR drop across the soil and across coating holidays, so they read more protective than the pipe actually is. They cannot be plotted on the same curve as instant-off data or used to evaluate the -0.850 V instant-off criterion. Mark them as a data gap and re-survey that segment.

Must be flagged as suspect:

• Everything taken with the low-battery indicator on. The instrument may have been perfectly accurate, but nobody can demonstrate that after the fact. Without a check, the honest status is 'unverified', and unverified readings should be re-taken, not quietly included.
• Everything after the 1.8 km half-cell swap, if the trend shows a step at that point. A swap with no comparison check introduces an unknown offset. Look at the data around 1.8 km: if there is a discontinuity in the trend, the readings either side are not on the same basis.

Transcription: verify the spreadsheet against the paper originals before anything is interpreted. Two weeks and a different person is exactly where transposition errors hide - and a mis-keyed anomaly sends a crew to dig the wrong spot. Nothing gets deleted; every issue becomes an annotation.

From the survey alone: conclude that there is an anomaly worth investigating. Not that there is a coating fault.

A 20 m dip that recovers is consistent with several things, and they lead to different actions:

• a genuine coating holiday or mechanical damage at the road crossing;
• a casing or insulator that changes the local environment;
• interference from the foreign pipeline crossing 60 m away - its rectifier could be discharging onto this line in a localised cell;
• a local change in backfill or moisture that alters the measured potential without any change in protection;
• a survey artefact - half-cell placement or contact at that point.

Next steps, cheapest first:

1. Re-read that segment with a second half-cell and verified contact; confirm the dip reproduces.
2. Interrupt the foreign rectifier on a synchronised cycle and re-survey; if the dip disappears or moves, it is interference, not coating.
3. If the dip is real and steady-state, use DCVG or a Pearson survey to size and locate the feature.
4. Only then schedule an excavation.

Reporting language: 'anomaly, unclassified, under investigation', with the evidence. Calling it a coating fault on the strength of one pass of a CIS is how unnecessary digs get budgeted and real interference goes unfixed.

Traceability from raw data to reported number:

• original field notes, the spreadsheet, and the script or method used to produce the plotted values;
• who took the data, who transcribed it, who reviewed it, and on what dates.

Instrumentation:

• meter make/model/serial and the date of its last calibration;
• half-cell IDs, and the result and date of each comparison check - including the 1.8 km swap, which currently has none;
• lead length and condition.

Method:

• half-cell placement (over the pipe, offset distance, surface condition);
• interruption cycle, timing source, and the fact that GPS sync was lost for 300 m;
• rectifier state during each block of readings.

Conditions: weather, soil moisture, temperature.

Deviations and corrective actions: an explicit section listing the lost sync, the low battery, the unchecked swap, and what was done about each - re-survey scheduled, data excluded, trend checked.

Why this matters: the engineer is spending money on digs based on this document. If a dig misses, the record is what shows whether the data were wrong, the interpretation was wrong, or the pipe changed. A record that only contains the good numbers cannot answer that question.