CORRECT A DROP IN OPERATING EFFICIENCY
The Challenge:
The contribution margin in our contract manufacturing operation dropped 1.5% two quarter’s ago; this quarter it’s down by another 2%.  It’s a small company, a 1-2% fluctuation is not unusual, but 2 quarters in a row indicates a potential problem.  How do we find the problem and correct it?

The Solution:
We currently track labor content at each of our 4 workstations, overall company-wide, direct material costs and customer rejections.  A machine operator runs 3 machines, and usually 3-4 different projects each shift; we run 2 shifts.  We do not have the ability to track machine time by project.  Our staff is not new, so learning curve issues over the last 6 months are eliminated.  Customer rejections have not changed significantly over this period.  Direct material usage has increased slightly, but accounts for about 1% of the 3.5% increase and that’s within our normal margin of error.  Direct labor costs have increased by about 3% over the last 6 months.  

I selected 4 projects that were started and completed over the 2-quarter period, ranging from simple to complex.  I had timecards pulled for each project and a comparison made between quoted and actual labor at every workstation.  I had the same project timecards pulled from 18-24 months ago.  It looked like the labor content in the CNC router workstation had increased – although there were different operators and a couple of the projects were run on different shifts, so it was hard to tell for sure, and there was always the question of time card accuracy because each operator is running 3 machines with different projects on each machine.  

I focused on labor at the CNC router workstation.  I asked the shop manager and machine operators if they had any ideas as to why they’d be taking longer to run parts.  One of the operators mentioned that one machine was hard to run good parts when the machines were running fast.  I asked the manager how often the calibration routines were being run…I asked when the machines were last serviced, it had been over a year.  I asked him to start scheduling maintenance, 1 machine per month (a machine is down for 24 hours for maintenance and we were running at over 95% capacity on the machines).  I had him run the calibration tests and they came out good.  We tracked the number of machined pieces per person per shift for a week; there wasn’t much difference between operators and shifts, so the problem was in the machines.  I talked to the operators again, grilled them on the machines, “Which was the best machine?”, “Why”, “How often are you changing cutters?”, “Do you think the calibration routine is accurate?”, “When is the first machine due for maintenance?”, “What speeds and feeds are you running at?” and finally…”Tell me again why you can’t speed them up?”  The answer was “Well, then we can’t get it to run good parts”.  It turns out that the night shift operator, figured out how to reduce the internal feed rates on all the machines so he could consistently make parts within tolerance.  This change over-rode the programmed feed rate, so everything in the CNC shop had been “dumbed” down to run slower.  Everyone had become used to the new speeds and didn’t even realize they were slower than in the past.  They were happy to have figured out how to consistently make good parts.  The operator could barely remember doing it…the calibration routine was good because it was running slower also… The machines needed service; the bearings were worn, creating out of tolerance parts when the machines were run at their fastest speeds.

We immediately started pulling machines out of service for maintenance, we called customers to re-schedule deliveries, we changed the permissions on the machines’ controllers to disable an operator’s ability to change internal settings, we re-implemented a routine maintenance program, and added a step to the calibration routine to note machine speed and feed rates used on the calibration part.

RE-ENGINEER 120 ENCLOSURES AND RE-PROGRAM 8000 CAM PROGRAMS IN 6 MONTHS
The Challenge:
A French company, La Tolerie Plastique (LTP), had acquired our business.  They used different CNC routers and had slightly different methods of assembly and operations in the shop.  We had an existing customer base of approximately 120 regularly re-ordered plastic enclosures, utilizing 8000 CAM programs which would need to be re-engineered for the new processes and re-programmed for the new CNC machines.  We had only 2 engineers, and they were both usually busy with new projects, or supporting the manufacturing operations.  How do we get this re-engineering and re-programming effort done without significantly affecting our deliveries and our ability to take new projects?

The Solution:
LTP had a team of 5 engineers and designers, and a 3-shift operation nearly 5 times larger than ours.  They also had a much higher capability to design and build complex enclosures.  Materials were different enough so that we could not substitute.  There was a 9 hour time difference, 1 of the LTP engineers spoke English fluently and none of us spoke French.

I called our top 40 customers to introduce and explain the new owner, our new capabilities, and to get an idea of who would be re-ordering product in the next 6 months.  We prioritized the projects by schedule, and by complexity.  The most complex projects we sent to France for re-engineering and re-programming. We also sent 3 pallets of a variety of our materials so they could manufacture the first run of parts when we received an order and fabricate a prototype for any products they engineered there.  

I contracted the writing of an algorithm to reverse-translate our CNC machine language back to a DXF format.  With some human intervention, the DXF format could be imported into the French CAM programming software.  This reverse translation process was used for the simpler projects, some were run in our shop, some were run in France, depending on our respective capacities at the time.

Finally, complex new projects were sent to France, the simpler ones were engineered and run in our shop.  The English speaking French engineer was shifted to a later work day, working from home part of the time to honor his family commitments, and to coincide with our time zone and our customer’s time zones.  He managed his design team and was our single point of contact.  For the new projects and for the re-programmed projects they were responsible for, we would receive a completed prototype, machine code and a complete set of shop floor documentation.

It took about 6 months to complete the re-programming of the first 8000 CAM programs, while introducing 20-24 new projects, many of which incorporated the new, enhanced capabilities of the French manufacturing system.