Walkthrough

What goes in a turnover package

Last updated 3 August 2026

Every section of a complete electrical turnover package, walked through on one worked job — then the five cross-checks that send a package back, shown against that same job so you can see exactly what a reviewer sees.

The client’s specification defines the section list, not this page. Section names, order and numbering vary by owner, by spec and by jurisdiction, and some clients want things here that nobody else asks for. What follows is a common shape, not a standard. Read the turnover requirements in your contract before you build the binder, not after.

The job

Everything below runs on one example, small enough to follow end to end and large enough to touch every section.

Project 2614 — replace MCC-2, a 480 V motor control center at a water treatment plant. Fed from existing switchgear SWGR-1 through breaker 52-4. Fourteen motor feeders out of the new MCC, a ground grid extension at the pad, and one instrument loop tied back into the DCS.

Tag	What it is
`C-2101`	Feeder, `SWGR-1` breaker `52-4` to the MCC-2 main — 3/C 500 kcmil
`C-2114`	Branch feeder, MCC-2 bucket 4B to pump motor `P-201A` — 3/C #2 AWG
`P-201A`	75 hp pump motor
`L-4412`	Loop — level transmitter `LT-4412` via marshalling cabinet `MC-1` to the DCS
`GG-2`	Ground grid extension at the MCC-2 pad
`51-2`	Overcurrent element on `52-4`, settings sheet rev C
`E-2614-03`	One-line and feeder schedule drawing, currently rev D

The sections, in the order they get read

A turnover package is not filed in the order the work happened. It is filed in the order a reviewer needs to build confidence: what governs the work, what was installed, what proves it, what went wrong, who did it, and what allowed them on site. That is the order Sparky QC’s job book assembles in, and it is the order below.

Group	Sections	What the reviewer is establishing
What governs it	Documents · Electrical Procedures · Drawings · Tags	Which revision of which spec, procedure and drawing the work was actually built and tested to. Everything downstream is judged against this, so a stale revision here poisons the whole book.
What was installed	Cable Schedule · Cable Pulls · Terminations · Conduit & Raceway · Cable Tray · Grounding & Bonding · Telecom & Data · Panels & Switchgear · Panel Schedules · Transformers · Motors · Lighting · Heat Trace · Equipment · Instrumentation	A complete inventory of the physical work, keyed by tag. The cable schedule is the spine — almost every test record in the next group points back at a row in it.
What proves it works	Insulation Resistance (Megger) · Continuity · Ground Resistance · Loop Checks · Functional Tests · Relay & Protection Testing · Torque Records · Phase Rotation & Voltage	That each item in the inventory was tested, by a qualified person, with a calibrated instrument, against a stated criterion, after the work it tests.
What was found and fixed	Non-Conformance Reports · Requests for Information · Observations · Auditor Punch List · Punch List	That problems were raised, dispositioned and closed with evidence — and, crucially, that anything retested after a fix has a test record dated after the fix.
Who did it, with what	Materials · Calibrations · Electrician Qualifications · Job Safety Analysis · Safety · Equipment Certifications	That the people were qualified on the day, the instruments were in calibration on the day, and the material installed is what was specified.
What allowed the work	Permits · One-Calls · Subcontractors · Weather	Energized work permits and LOTO, 811 tickets for the duct bank and grid work, subcontractor insurance, and the conditions on the day a weather-sensitive test was run.

A section with no records is not automatically a gap — project 2614 has no heat trace and no lighting scope, and empty dividers for both are correct. A section you expected to be populated and which is empty is the one to ask about, and the time to ask is before the binder ships.

Following one cable all the way through

This is the test of whether a package is real. Take one tag and walk it across every section it touches, in order. Here is C-2114, the feeder to pump motor P-201A:

Drawings — E-2614-03 rev D shows the feeder and its breaker.
Cable Schedule — the row: C-2114, MCC-2 bucket 4B to P-201A, 3/C #2 AWG, 600 V, routed via tray T-12 and conduit C-441, taken from E-2614-03 rev D.
Cable Pulls — date pulled, reel and lot, 214 ft installed, no jacket damage.
Terminations — two records, one per end. Lug type, and the core landed on each terminal.
Torque Records — the value applied at each end, the source of that value (the bucket manufacturer’s label), and torque wrench TW-06 with its calibration date.
Continuity — each core identified at both ends, proving the conductor landed where the schedule says it did.
Insulation Resistance — six readings for a three-core cable, test voltage, duration, conductor temperature, and the acceptance criterion from the spec.
Phase Rotation & Voltage — rotation confirmed at the motor terminals before coupling.
Motors — P-201A nameplate data and its own winding insulation test.
Functional Tests — the run test, with the starter, overload and remote stop proven.

Ten sections, one tag, and every one of them carries the string C-2114. That is the whole trick. When it holds, a reviewer can satisfy themselves about that cable in about two minutes; when it breaks anywhere, they stop sampling and start auditing.

Five ways this package comes back

All five are real classes of finding, written here against project 2614. None of them is about whether the electrical work was any good.

1. An instrument out of calibration

Nine insulation resistance reports, one megohmmeter, calibration expired a week before the first of them.

The Calibrations section holds the certificate for megohmmeter S/N 4471102, calibration due 14 June. The insulation resistance reports for C-2110 through C-2118 are dated 21 to 24 June.

Nine test reports were produced by an instrument that, on paper, could not be trusted on the day. It does not matter that the readings look sensible. The certificate is in the package, the dates are in the package, and the comparison takes a reviewer under a minute. Every one of those tests gets redone — after the crew has demobilized, which is what makes this the expensive one.

2. A test with no acceptance criterion

A number and the word PASS, with nothing recorded to compare the number against.

The megger sheet for C-2117 reads 1,400 MΩ — PASS. There is no test voltage, no duration, no temperature, no phase combination identified, and no criterion.

Whoever wrote it knew it was fine. Nobody else can. A three-core cable owes six readings and this is one number; without a test voltage the value is meaningless; without a temperature it cannot be corrected to whatever base the criterion assumes; and with no criterion recorded, “PASS” is an opinion. This is the cheapest finding to prevent and the most common one there is — the field notes go into why the temperature in particular matters.

3. A loop signed off at two points

Two signatures that between them cover the whole loop, and no single check that ever spanned it after the rework.

Loop L-4412 carries two sign-offs:

12 June — transmitter LT-4412 to marshalling cabinet MC-1, injected and read, signed.
19 June — MC-1 to the DCS input, injected and read, signed.

In between, on 16 June, the terminals in MC-1 were remade to close NCR-014.

Read the dates in order. The first half was proven before the rework. The second half was proven after it. Nothing has ever been injected at the transmitter and read at the DCS since the joint was disturbed. The package contains two signatures and zero end-to-end loop checks of the loop as it now stands, and the reviewer sees it the moment they lay the three dates next to each other.

This is the pattern to watch for generally: a loop proven in halves is not a loop proven, and splitting the check is exactly what happens when a rework lands between two site visits.

4. A count that does not reconcile

Fourteen feeders on the drawing, thirteen rows on the schedule.

E-2614-03 went to rev D to add a feeder for a new sump pump. The cable schedule was built from rev C and never rebased.

Nobody has to read a single test result to find this, which is why it is often the first thing checked. It also means one installed feeder has no schedule row, and therefore no pull record, no termination record and no megger result — a whole missing chain rather than a clerical slip.

5. A relay tested to a superseded settings revision

As-left values that are correct for a settings sheet withdrawn two weeks before the test.

Element 51-2 was tested on 9 June against settings rev B. The engineer issued rev C on 27 May.

The as-found and as-left values are internally consistent and completely fine — against the wrong sheet. This one is nobody’s fault in the field, and that is the point: the test tech mobilized with the revision that was current when they were scheduled, and no mechanism existed to tell them it had moved.

The pass to run before you hand it over

All five above are findable from inside your own records, weeks before a reviewer sees them. A morning spent this way is the highest-return work in the whole handover:

Trace five tags end to end, picked genuinely at random rather than the five you know are tidy. If a trace takes longer than a couple of minutes, that is the finding.
Lay every calibration certificate against the reports that instrument produced and check the dates enclose them. Mechanical, fast, and it is finding number one.
Sort every test record by date and read it against the rework. Any test dated before an NCR closure that touched the same equipment is suspect. That is findings one, three and five.
Reconcile counts against the current drawing revision, not the one your schedule was built from. That is finding four.
Read ten test sheets as a stranger. Does each one state what was tested, at what voltage, for how long, at what temperature, and against what criterion? That is finding two.
Check every empty section is empty on purpose, and note in the transmittal why.

Where Sparky QC stops. The Audit tab finds what is objectively true from the records — an expired calibration or qualification, an open NCR or RFI, a punch item not signed off, a row whose document slots are all empty. That covers finding 1 and part of the routine housekeeping.

It does not catch findings 2 through 5. It does not judge whether a criterion is present or sensible, it does not order test dates against rework, it does not reconcile a count against a drawing, and it does not know that a loop proven in halves was never proven whole. Those are still a person tracing a sample. How Sparky QC works sets out exactly which checks run and which do not.

What this page is not

It is not your client’s section list. Read the turnover requirements in the contract; they win every time.
It does not tell you what to test or to what limit. That comes from the spec and the engineer of record — the spec quick reference covers which standard asks for which record.
Project 2614 is a worked example, not a real job. The tags and dates are made up to make the failures legible; the failure patterns are not.
It is not exhaustive. Medium-voltage cable, protection coordination, fire alarm, hazardous locations and commissioning of packaged equipment each carry turnover requirements this walkthrough does not cover.

Corrections and disagreements are welcome and get published. Write to info@sparkyqc.com.

See also the blank form pack — the sheets most of these records go on, free and ungated.

What goes in a turnover package — sparkyqc.com/turnover-package