Shift hand-offs, logs and documentation

Twice a day, at 07:00 and 19:00, responsibility for the whole campus passes from one crew to another. For a few minutes, everything the outgoing crew knows about the plant, every open job, every alarm being watched, every valve in an unusual position, every vendor on site, must pass to people who were not there. If something is lost in that moment, the incoming crew begins its shift with a wrong picture of the facility, and nothing in the plant will tell them so until something goes wrong. Investigations of accidents in many industries, from hospitals to chemical plants, find hand-offs among the commonest points where critical information is lost. This lesson is about doing them well, and about the logs and records that make them possible.

Why hand-offs fail

Hand-offs fail in predictable ways. The outgoing crew is tired and wants to go home; the incoming crew is fresh but has no context. Things that seemed obvious to the people who lived through them are not mentioned, because they seem too obvious to say. Items that were fixed during the shift are mentioned, while the temporary arrangements that fixed them, a valve left half open, an alarm inhibited, a unit in manual, are forgotten. Unwritten agreements, such as a tenant's request to call before any work near their cage, vanish. And when the hand-off is a casual chat, what gets said depends on who is talking.

Stonebridge's answer, like that of most well-run facilities, is a structured hand-off: the same items in the same order every time, spoken face to face and written down, at the same place, with the incoming shift lead signing that they have accepted the facility.

The structure of a hand-off

The hand-off happens at the redundancy whiteboard in the operations center, with the outgoing and incoming shift leads and as many technicians from both crews as can attend. It follows four headings, which every crew knows by the word SAFE.

Two parts of the hand-off deserve special attention. The first is anything not in its normal state, because the plant's normal state is what every procedure assumes. A valve left closed, a breaker left open, a unit in manual or maintenance mode, a jumper or bypass fitted to a controller, a door propped open for a delivery: each changes how the plant will behave if something else goes wrong. The second is inhibited alarms. An alarm inhibited for planned work, so that floor tile lifting does not set off the leak detection, is a hole in the facility's protection. Every inhibited alarm is on the board with the time, the reason and the person who inhibited it, and the hand-off confirms whether it is still needed.

Face to face, then in writing

A hand-off is spoken and written, and both matter. Speaking allows questions: the incoming lead can ask what the trend on CRAH-A09 looked like, or whether the tenant was happy with the explanation. Writing makes sure nothing depends on memory and leaves a record. At Stonebridge, the outgoing shift lead writes a hand-off summary in the electronic shift log before the hand-off begins, the leads talk through it at the board, the incoming lead adds anything learned in the conversation, and both sign. If the incoming lead is not satisfied with the state of anything, the hand-off waits until it is explained, even if that keeps the outgoing crew late. Dmitri says that the outgoing crew does not leave until the incoming crew owns the building.

Technicians hand off too. A technician in the middle of a job at shift change briefs the technician taking over at the equipment, not in the corridor, showing the state of the work, the step reached in the method of procedure and any lockouts in place. Lockouts are never transferred by word alone: the incoming technician applies their own lock before the outgoing technician removes theirs, following the site's lockout procedure from DCT-120.

The shift log

The shift log is the continuous written record of the facility, entry by entry, through every shift of every day. At Stonebridge it is electronic, linked to the monitoring and ticketing systems, with a printed backup kept in the operations center for times when the systems are not available. Every entry has a time, the author, the equipment tag, and what was seen or done, in plain words. Good entries are factual and specific: 14:12 RPP-A2-08 main breaker found tripped; B side carrying cage A-26; Dmitri informed. Poor entries are vague: power problem in Hall A, sorted.

A few rules keep the log useful.

• Write it as it happens, or as soon as possible. Entries written at the end of the shift from memory are less accurate and miss times.
• Use equipment tags and plain words. Use only abbreviations that everyone on the campus uses.
• Record facts and actions, not opinions about people.
• Never delete or overwrite. Corrections are made by adding a new entry that refers to the old one, so the record of what was believed at the time survives.
• Record readings and times, not only conclusions: the strainer pressure was 7.0 psi, not the strainer looked high.

The log is a legal and contractual record as well as an operational one. Incident reports, tenant disputes about service credits, insurance claims and regulatory inquiries all rely on it. An honest, careful log protects the technicians who write it, because it shows what they knew and did, when.

# Parse a shift log and build the hand-off summary: what is not normal, inhibited alarms and open items
log = [
    "07:12 | RK | CRAH-A11 | out of service for strainer cleaning | NOT_NORMAL",
    "08:10 | JP | LEAK-A4 | leak detection zone inhibited for floor work | INHIBIT",
    "09:30 | DV | GEN-D2 | in manual for starter motor repair, vendor on site | NOT_NORMAL",
    "11:45 | RK | CRAH-A11 | returned to service, strainer cleaned, dP 21 | RESTORED",
    "13:20 | JP | PUMP-CW2 | locked out by mechanical contractor for seal change | LOCKOUT",
    "16:05 | DV | GEN-D2 | repair complete, test start passed, left in manual for second test | NOT_NORMAL",
    "17:30 | JP | LEAK-A4 | floor work complete | INFO",
]
state = {}
for line in log:
    t, who, tag, what, kind = [p.strip() for p in line.split("|")]
    if kind == "RESTORED":
        state.pop(tag, None)
    elif kind != "INFO":
        state[tag] = (t, who, what, kind)
print("Hand-off summary: items not in their normal state")
for tag, (t, who, what, kind) in sorted(state.items(), key=lambda x: x[1][3]):
    print(f"  [{kind:10}] {tag:9} since {t} ({who}): {what}")
inhibits = [tag for tag, v in state.items() if v[3] == "INHIBIT"]
if inhibits:
    print("Check before signing: inhibited alarms still in place ->", ", ".join(inhibits))

Hand-off summary: items not in their normal state
  [INHIBIT   ] LEAK-A4   since 08:10 (JP): leak detection zone inhibited for floor work
  [LOCKOUT   ] PUMP-CW2  since 13:20 (JP): locked out by mechanical contractor for seal change
  [NOT_NORMAL] GEN-D2    since 16:05 (DV): repair complete, test start passed, left in manual for second test
Check before signing: inhibited alarms still in place -> LEAK-A4

The summary lists three items for the incoming crew: the generator still in manual pending its second test start, the pump under lockout and the leak detection zone still inhibited although the floor work finished at 17:30. It is the example from earlier in the lesson, found here by the log itself. The tags at the end of each entry, which the log's form adds from a short list, are what make this kind of summary possible.

Other records

The shift log is the center of a set of records, each with its own purpose.

Drawings and documents deserve a final word, because Lesson 6's out-of-date placard showed what happens when they drift. Every change to the plant, a new valve, a replaced panel, a re-fed circuit, must reach the drawings, the valve charts, the asset register, the monitoring system's labels and the procedures that mention it. At Stonebridge the change management process of DCT-330 includes a documentation step that cannot be closed until the operations engineer confirms all of these are updated. Technicians are often the first to notice when the drawings do not match the floor, and reporting it is part of the job.

Documentation for the next person

Roz has a test for every log entry, hand-off and record: could someone who was not here understand what happened and what to do next? The next person might be the incoming shift lead at 19:00, a vendor arriving tomorrow, the analyst reviewing an incident next month or a new technician reading the history of a unit in two years. Writing for that person takes a little more time on shift and saves a great deal of it later. It is also, as Lesson 12 shows, the thread that holds an incident response together, from the first alarm in the log to the last corrective action in the tracker.