Editor’s Note: The following article is based on “
Improving on-time delivery with sequential manufacturing,” presented at the FMA Annual Meeting, Feb.
24-26, 2016, by Keith Stout, vice president of operations, Ace Metal Cra;s, www.acemetal.com. Stout
spoke as a user representative of ECi So;ware Solutions, www.ecisolutions.com.
Imagine you’re waiting in line at Disneyland’s Space Mountain. The sign at the back of the line says “30-minute wait from this point.” That’s reasonable, so you get in the queue. And you wait 30
minutes. Then 40. Then 50—and finally, at long last,
you get on the ride. You’re fuming. You just spent
50 minutes staring at the person’s head in front of
you. What took so long? What was the holdup?
The ride operations supervisor—deep inside Space
Mountain, invisible to you—wasn’t having a good
day. First, he had to stop the ride unexpectedly for
five minutes to allow a passenger with health issues
to get o;. Second, he had to deal with a big group
of riders who all had Fast Passes, which they got in
advance (and paid more for) to guarantee a set time
to ride. These riders were given a 30-minute window,
and it turned out that nearly all Fast Pass holders for
that window arrived at nearly the same time. The result: A bottleneck that sent ripples throughout the
system, forcing everyone else without a Fast Pass to
wait 20 minutes longer than expected.
This illustrates a classic problem in custom fabrication. Like riders reading the wait-time sign at
Space Mountain, fab shop customers are given a
certain delivery date when they place an order. Like
the ride shutting down for a few minutes, a machine
breaks or setups take longer than expected. Then
the expedites happen (the Fast Pass holders arrive)
as we bend over backward to try to meet customer
demand. We can handle a few Fast Pass holders, but
if they all arrive at the same time (which, of course,
they usually do), chaos ensues. Overtime mounts,
and we deliver jobs late.
We could extend lead times, and we could even
turn some of the Fast Pass holders away. And in
many cases, “just saying no” is the right thing to
do. But we hesitate, knowing that competitors will
probably say “yes.” Moreover, many of us operate
with a “we’ll do what it takes to get it done” mentality. A;er all, it’s how we’ve grown our businesses
into what they are today.
Eventually, though, being late about 15 percent
of the time becomes the new normal. We o;en
just throw up our hands and think that late deliveries are a fact of life in custom fabrication. We can’t
avoid unpredictability in high-product-mix operations, particularly when we need to deal with such
unreasonable customer demands, so we continue
with ad-hoc scheduling methods that work about
85 percent of the time.
At Ace Metal Cra;s, a custom stainless steel fabricator near Chicago, we found that late deliveries
are indeed avoidable. We now schedule through
our ERP so;ware using a method called sequential
manufacturing. This focuses not on machine utilization, but instead on job velocity: how long a job
takes to make it through the plant. When people
stand in line—be they a regular rider or a Fast Pass
holder—they know that the posted wait time (
delivery date) is almost always met.
Our on-time delivery is better than 95 percent,
and people aren’t breaking a sweat to make it happen. How? They receive and fabricate jobs in the
Not So Simple
Waiting in line at Space Mountain is more like low-product-mix assembly than high-product-mix custom fabrication. The ride takes a set amount of time,
it follows one track, and no one gets out of line. It’s
like one or just a few products progressing on an assembly line based on a single takt time.
We as custom fabricators process thousands of
di;erent parts, each requiring di;erent processing
times, resources, and routings. One piece is cut on
the laser and then goes to deburring, then forming,
then back to deburring and weld prep, then finally
to welding. Another piece is cut with a saw, goes to
weld prep, then right to welding and assembly. Yet
another piece arrives from an outside supplier. The
challenge is for all parts to arrive in welding and assembly at the same time, in sequence.
What prevents this from happening? At Ace, it was
all the wasted setups and teardowns, what’s called
context switching. A welder would have several jobs
on his schedule. As he finished setting up the first
job, a material handler arrived with an expedited
job. So the welder had to tear down and set up
again to push the rush job on through.
The welder couldn’t see what was coming; he
had no idea that an expedite was in the pipeline.
That was a relatively easy fix. We now have monitors posted at every department, showing what
jobs are arriving when. But just having a screen and
posting the schedule doesn’t solve the root causes.
Sure, it makes things visual, which is important, but
it doesn’t magically make the right parts appear at
the welder’s workstation at the right time.
To that end, we had to rethink how jobs were released to the floor. We needed to start with the operation in which multiple parts need to arrive at the
same time. For us, that was usually welding or assembly. We then work backward from there.
To make sure parts arrived in welding when
needed, two elements needed to be in place. First,
we needed accurate processing times. Second, we
needed to schedule sequentially and resist batching jobs together for better material utilization or
Actual Processing Times
Workers report actual job cycle times continually.
This works because it’s easy and seamless, and
people are given the tools to make it happen. Usually it’s just a matter of scanning a bar code on a
part. Operators aren’t waiting in line to clock in a
job at a computer terminal. The time measurement
involves not just the machine cycle time, but also
the material handling time. Even if the laser flies
through a nest, we still need to account for the time
people spend sorting parts and delivering them to
the next operation.
Note that we’re a custom fabricator, so estimating
processing times is an imperfect science, especially
for new jobs. But the more data we have, the more
we can adjust and perfect those estimated processing times.
The secrets of
How sequential manufacturing
works in custom fabrication