❝ A Cleveland contract manufacturer lost a spindle and nine working days because a night setter's warning about a worn drill reached the day shift as eight words on a whiteboard. The setting is a 118-person CNC machining and stamping plant on three shifts, with 18 handovers a day. Modelled across a year, 61 of its 214 scrap, downtime and late-shipment events over $500 trace back to something the previous shift knew, worth $192,900. The design below takes a voice note from whoever knows, turns it into a numbered brief for the next shift, asks the incoming lead to confirm each item, and escalates anything handed over three times without being closed. It costs about $85 a month to run.
The whiteboard said: "C3: drill getting tired. New ones in AM."
At 04:50 Marco, the night setter on cell 3, noticed the 12 mm drill on one of the horizontal machining centres was leaving a burr and pulling more spindle load than it had at midnight. The tool crib had no spare; a box of six was on the morning delivery. He told Dariusz, the night lead, it would "do another hundred, maybe". At 05:55 Dariusz wrote the line on the board, spent two minutes at the clock with the day lead talking about a job that was running behind, and went home. The night shift ran another forty parts on it before 06:00.
Day shift read the board and started the run. At 07:41, 63 parts in, the drill snapped and left its tip in the bore. The next tool in the program was a boring bar, and it found the stub. The spindle repair quote came to $24,800, and the machine was down for nine working days.
Everything the day shift needed was known at 04:50. It reached them as eight words on a whiteboard, and none of the eight said stop.
Below is the agent that would have carried it: how it works, how to build it, what it costs to run (about $85 a month), and a handover card you can print today, with or without it.
The plant, Marco and Dariusz are a composite, built from the patterns we see in contract manufacturers of this size, and the figures attached to them are design projections. The investigations, studies, regulations and published figures are real, and linked to their sources.
The business
A 118-person contract manufacturer in Cleveland, Ohio, doing CNC machining and metal stamping for automotive, HVAC and agricultural equipment customers. Four machining cells hold about 35 machines between them; two press lines hold ten presses. It runs three shifts, 06:00 to 14:00, 14:00 to 22:00 and 22:00 to 06:00, five days a week, plus Saturday days. That is six handover points at every shift change and 18 a day. It is not unusual: across the US, 24.9% of people in production jobs usually work something other than a day shift.
Each cell lead runs the handover. Machining uses a whiteboard by the cell; the press lines keep a paper log. The ERP knows which jobs are open and how many parts are booked against them. It has no field for "the die on press 4 started clicking at three o'clock".
The publication's Knowledge Capture Blueprint dealt with decades of expertise leaving at retirement. This one deals with what one shift knows and the next does not, every day at 06:00.
What is actually going wrong
Marco did his job. Dariusz did his. The day lead read the board. The failure sits in the format of the handover, and it is an old, well-documented failure.
The board records what happened, and the next shift needs what to do. When BP studied shift logs at its Grangemouth site, the findings, summarised in an appendix of the US Chemical Safety Board's Texas City report, were blunt: "The logbooks were mostly historical records and little information was collected on what should be done in the future" (CSB final report, page 287). "Drill getting tired" is a history. "Change the drill before you start the 4471s" is an instruction. Only one of them stops a run.
The person who knows is often not the person who hands over. Marco saw the drill. Dariusz wrote the board. On the morning of 23 March 2005 at BP's Texas City refinery, the night lead operator left about an hour early, and the log entry for the unit, as the CSB records it on page 51, read "ISOM: Brought in some raff to unit, to pack raff with." The day board operator misread it, and the day supervisor arrived more than an hour late and "did not conduct shift turnover with any night shift personnel" (CSB final report, page 51). Among its key findings (page 23), the CSB wrote that "supervisors and operators poorly communicated critical information regarding the startup during the shift turnover; BP did not have a shift turnover communication requirement for its operations staff." The explosion that followed killed 15 workers and injured 180. The handover was one of several causes the CSB identified. The mechanism is the same one on a Cleveland shop floor: a short, jargon-heavy line, handed over by someone who had not done the work, to someone who had not been there.
Nothing confirms what was received. The UK's Health and Safety Executive describes good handover as two-way, with "cross-checking of information by in-coming personnel as they assume responsibility for the task" (HSE, shift handover). A paper on the HSE's website describes 15 real handovers at a UK refinery, lasting between 1 and 11 minutes. Nearly all the incoming staff took part, and all read back through the logs, but only one took notes, and none of the 15 handovers showed all of the good practices (Lardner, 1999). In Cleveland, the two minutes at the clock went on the late job, because the late job was the one with a customer on the phone.
Writing alone loses the most. In a NASA-backed experiment, people handing over a half-finished maintenance task by paper log alone left an average of 6.2 of 8 bolts done wrong at the next shift. With an audio recording added to the paper log, the figure fell to 4.0; with a face-to-face briefing and the log, 3.5 (Parke, Hobbs and Kanki, 2010). It was a lab task with 50 participants; face-to-face scored slightly better than audio and was preferred, though the gap was not statistically significant. The direction is still clear: a voice note added to the log beat the log alone.
What it costs
The crash is the expensive morning everybody remembers. The rest of the cost comes in smaller events.
Take a year of the plant's scrap tickets, downtime codes and late shipments, 214 events over $500 in all, and ask one question of each: did the previous shift know something the next shift did not? In this design we model 61 of the 214 as yes. Run the same question over one month of your own tickets: and build the case on that number.
48 scrap and rework runs, averaging $2,900 each, $139,200. A tool, a die or a coil the night shift had doubts about, run by days as if it were fine.
11 late shipments, averaging $1,600 each in expedited freight and customer chargebacks, $17,600. A job known to be behind at 06:00 and discovered at 09:00.
2 machine crashes, $24,800 and $11,300, $36,100.
That is $192,900 in a modelled year, before counting the day supervisor's half hour each morning on the phone to people who have just gone to bed. At the Bureau of Labor Statistics' average total compensation for manufacturing, $48.62 an hour, that half hour is about $6,300 a year on its own.
The crash figure is in line with the trade estimates. A machine tool laboratory at RWTH Aachen put the average cost of exchanging the damaged spindle components after a crash at about $25,000, in a 2017 trade article. In the same article, an engineer at a crash-protection maker said that when a motor spindle must be replaced, "costs can easily rise over $45,000".
The design is next. The design, the build guide, the running costs and the take-away are free to subscribers. Enter your email below and the rest of this Blueprint opens. New ones arrive every Tuesday and Friday.
