Where to start reducing machine downtime: a worked month from one downtime log

Step 2: total lost time and availability

Planned production time is the shift minus the break: 450 minutes a shift, 9,000 minutes over 20 shifts. The 37 stops add up to 746 minutes, 8.3% of planned production time.

Availability counts every stop long enough to need a reason, planned or not: run time is planned production time minus all stop time, changeovers included. The jams and sensor trips under 5 minutes are small stops, which the six big losses count against performance, so they are in the lost time but not in availability. Write your rule down once and keep it, or next month will not compare with this one. The six big losses worksheet puts each kind of loss under its OEE factor.

Step 4: split the many short stops from the few long ones

The same minutes can come from a few long stops or from many short ones, and the fix is different. Split the log into three groups.

Long stops are a repair time and root cause problem. The capper's three cylinder failures took 64 minutes each on average (mean time to repair) and came once every 2,770 minutes of run time, about one every six shifts (mean time between failures). Ask why the part fails and whether it gives a warning you could check for: a breakdown analysis report on the failures, then the maintenance strategy selector for the failure mode. The MTBF and MTTR calculator does the arithmetic for any machine.

Short stops are a frequency problem. Each jam took 3 minutes on average, but one came every 554 minutes of run time, a little more than a shift apart. No repair fixes them, because nothing is broken. They come from the condition of the machine and how it is run: dirt, worn guides, settings that drift, material loaded differently each time. That is the work of autonomous maintenance, where operators clean, inspect and set their own machine to a standard, and of standard work.

When the stops happen says more. 12 of the 15 jams started in the first hour after the 06:00 start and 2 more in the hour after the break, while the long stops are spread across the day. A pattern at start-up points at something about starting: glue not up to temperature, cases brought in cold from the warehouse, a setting changed overnight. It is a lead to check on the floor, not a cause yet.

Step 6: a 30, 60 and 90 day plan

1. 1

   Days 1 to 30: the capper cylinder

   Maintenance writes one breakdown analysis report covering all three failures: the same cylinder or not, air supply quality, mounting and alignment, how long each seal lasted. Run the failure mode through the maintenance strategy selector. If a slower stroke or an air leak gives enough warning, a regular check can catch the next one before it stops the line; if not, plan the replacement or change the design. Target, written as a count: B3 stops a month from 3 to 0.

2. 2

   Days 1 to 30: the start-up jams

   The team leader and the case packer operators watch the first hour on three mornings and write down what is different at start-up. Turn what they find into a start-up standard, add the cleaning and setting points to the autonomous maintenance checklist, and teach the new step to every shift with a one point lesson. Target: M1 jams a month from 15 to 5 or fewer.

3. 3

   Days 31 to 60: read month two, then start on changeovers

   Read the second month's log exactly the same way (step 7). Then run a SMED workshop on the filler changeover: film one, sort the steps into work that needs the line stopped and work that can be done while it runs with the SMED changeover worksheet, and borrow ideas from the SMED examples. The average changeover this month was 41 minutes; set the target after the first study, not before.

4. 4

   Days 61 to 90: hold the gains and pick the next loss

   Keep the new check in the maintenance schedule and the start-up step in the standard, read month three and redraw the Pareto. The next target comes from the new top five, not from this month's list. Review the codes too: if the capper torque alarm (Z1) comes back, give it a code of its own.

Definitions and sources

• Six big losses: from Seiichi Nakajima's TPM work (Introduction to TPM, Productivity Press, 1988), read here through OEE.com (Vorne), "Six Big Losses", and Moradizadeh and Mayorga, FCTA 2014 conference paper, section 2. Breakdowns and setup and adjustment are availability losses; small stops and reduced speed are performance losses; start-up rejects and production rejects are quality losses.
• Availability is run time ÷ planned production time, where planned production time is the shift minus breaks and run time is planned production time minus all stop time, planned and unplanned (OEE.com, "Calculating OEE").
• Small stops are short stops the operator clears without maintenance. Vorne's TPM guide puts them under five minutes; OEE.com says where to draw the line is a site rule, for example giving every stop over two minutes a reason. The 5-minute limit here is the default in the downtime reason codes template.
• Vorne, "Reduce and Avoid Machine Downtime": capture a reason for every downtime event, watch the constraint and the machines that starve or block it, and pick one loss in the top five that the team is confident it can improve.
• MTBF is operating time ÷ failures and MTTR is repair time ÷ repairs, as in the MTBF and MTTR calculator. Cost of a stop is hours × cost per hour plus the restart loss, as in the downtime cost calculator.