Professional Documents
Culture Documents
Accepted for presentation at European Shipbuilding and Repair Conference on November 2, 2004 in London.
SUMMARY
Recently, several major shipyards in the US have begun to implement a Lean Manufacturing strategy. During the last
two years Northrop Grumman Ship Systems (NGSS) has been transitioning their Lean Manufacturing program to a Lean
Six Sigma program in conjunction with the Lean Six Sigma in Shipbuilding project sponsored by Advanced Maritime
Technology Application Center (AMTAC) at University of New Orleans.
In this paper, we first describe the Lean Six Sigma approach used for process improvement at Avondale, Ingalls and
Gulfport shipyards of NGSS along with the training requirements for implementation. The workforce training strategy
will be described to help understand the prerequisites for shipyard implementation. Then domino effects and resulting
financial impact of defects are presented.
1985 1990 1993 1994 1995 1998 1999 2000 2001 2002 2003 2004
Combining
CombiningSixSixSigma
SigmaQuality
Qualitywith
withLean
LeanSpeed
Speed
Reducing
Reducingcomplexity
complexitythrough
throughLean
Lean
Increasing
Increasingprocess
processcapability
capabilitythrough
throughSix
SixSigma
Sigma
1/27/04
Getting the right project for the LSS training class is Companies which developed CoPQ systems can choose
essential for successful workforce education. the projects based on the bottom line impact. However,
many organizations implementing Six Sigma tend to where the defect was discovered. However, the excess
use a prioritization matrix [10]. These matrixes are costs attributed to the defect are much more extensive.
completed by utilizing customer data which, by its Upstream and downstream processes may be affected
nature, does not account for actual costs of poor by having to reschedule work. The rescheduling of
quality. work at one station may cause priorities at other
stations to change. When those stations priorities
More time needs to be spent training Champions, Black change still others will change, and so forth.
Belts, and Green Belts on the impact of CoPQ, Rescheduling costs include, expenses such as extra
including domino effects in the manufacturing setups and tear-downs, idle time, expediting costs for
process. The emphasis on expanding CoPQ coverage to transporting materials, the cost of planners and
include domino effects in training classes and schedulers needed to make the changes, and the costs
mentoring sessions can make project impact assessment of delivery delays to other products whose priorities
more realistic. were reduced.
The need to modify the existing Lean Manufacturing These problems are particularly severe in lean/Just-in-
implementation by introducing a Six Sigma toolbox time (JIT) systems due to the reduced amount of work-
was recognized in complex product manufacturing as a in-process (WIP) inventory. When a stations schedule
way of dealing with variation in a resource constrained changes and the inventory needed for the new highest
environment with an expanding workload. It was also priority job is available, the ripple effect is attenuated
discovered that CoPQ is more complex to quantify in or even terminated. However, in lean/JIT systems
this scenario. Where as domino effects are significant where a conscious effort has been made to reduce WIP
when quantifying CoPQ, a literature survey showed inventory, it is necessary to adjust the priorities at
that there is very little research done on this subject, supplying stations. Thus, the ripple continues, but the
also referred by some as snowball or ripple effects. energy in ripples can be absorbed by inventory.
The study of CoPQ and domino effects is needed to
clearly reap the benefits of integrating Lean and Six Accounting for the domino effect is difficult for
Sigma methodologies. CoPQ and domino effect many companies. By using the conventional unit-based
concepts should also be included in both Lean and Six cost systems, there is no one way to accurately
Sigma training courses. determine ripple effects [12]. A study was also done
using transition probabilities to explain how failure
As far as project selection is concerned, including costs flow and accumulate in the entire production
CoPQ assessment as a supplement to expert-opinion- system. The only problem with using transition
based project selection the matrix criteria will become probabilities is that they can be subjective in nature
more effective. An example of such a matrix is given [13].
in Figure 1, where AOR means Area of Responsibility
and BB means Black Belt [11]. As shown in Figure 1, The problem with Job cost systems is that they capture
even though Six Sigma is a data driven methodology, costs by having time and materials charged to the jobs
numerous organizations use subjective judgment and being processed. For instance, if a worker is setting up
expert-opinion-based weights as the primary method to work on Product A, their time is charged to Product
for project selection. The obstacle in using CoPQ data A. However, if a problem with Product B forces the
including domino effects for project selection is the worker to break the setup for Product A and do Product
missing link between CoPQ data and accounting chart B, the time spent tearing down and then later resetting
of accounts. If a company associates the accounting up for Product A is still charged to Product A.
system with the CoPQ system including domino Similarly, if a problem with Product B causes the
effects data, Six Sigma project selection can be based worker to have to stay late to finish work on Product A,
on hard data. that was interrupted by Product B, the overtime will be
charged to Product A. To better demonstrate impact of
the domino effects in manufacturing an example is
4. IMPACT OF DOMINO EFFECTS IN given in the next section.
SHIPBUILDING PROCESSES
This section discusses the domino effect of poor 4.1 SHIPBUILDING EXAMPLE
quality. First, a discussion of the typical breakdown of
the costs of poor quality (CoPQ) is presented. Next, The following example will help to better illustrate the
the often neglected domino effect costs are reviewed. domino effect in shipbuilding. The panel production
line in Figure 3 consists of three sequential production
When a failure occurs somewhere in the system, the stations: plate joining, stiffener positioning, and
effects ripple throughout the system in a manner similar stiffener welding. For simplicity, each station is
to that of an earthquake. Obviously rework must be assumed to have a lead-time of 2 hours for every panel
performed and scrap may be generated at all of the and a 1 hour setup to change from one panel to the
stations between where the defect was created and
next. Travel time between stations is considered (rework and scrap) are easily captured by many
negligible. accounting systems. However, consider the domino
effects of this problem. Using the schedule in Figure
A simple Gantt chart schedule by station appears in 4 where there is a 4 hour inventory buffer, note that the
Figures 4 and 5. The schedule in Figure 4 assumes plate joining station can rework Panel A from hours 3
there is a 4 hour buffer of WIP inventory in between to 5. Thus, Panel A will arrive at the stiffener
each of the stations. That means that the schedule is set positioning station at hour 5, which is 2 hours prior to
such that panels arrive at production stations 4 hours when it is needed. Therefore, no downstream
prior to when they are scheduled to be processed. As is rescheduling is needed. Note, however, that what was
shown below, this inventory helps buffer the panel a 4 hour buffer is now down to 2 hours. There is only
production from quality problems. However, there is a room now for one more such problem before the
significant cost associated with having that extra protective buffer is fully depleted. Assuming that the
inventory. plate joining station can work some overtime to get
caught back up, the WIP buffer can then be increased
Figure 5 shows the schedules for the stations assuming back to 4 hours. However, as noted before, the job that
no WIP buffers. This is essentially the ideal JIT gets charged for the overtime will likely not be Panel
environment. Certainly, few if any manufacturing A.
facilities have achieved this ideal. It is used here
merely to help illustrate how the domino effect is Now consider the ideal lean situation depicted in Figure
exacerbated by lean systems that still have quality 5. Here there is no protective buffer of WIP inventory.
disruptions. Thus, even if a problem occurs at the plate joining
station and is detected at the plate joining station,
significant domino effects are felt throughout the
Plate Stiffener Stiffener panel production. Since the plate joining station will
Joining Positioning Welding
have to spend time reworking Panel A, the stiffener
positioning and stiffener welding stations will have
holes in their schedules later where they were supposed
Plates &
WIP WIP Panel to be processing Panel A. However, the lack of
Stiffeners
material will cause them to go idle for 2 hours. Thus,
Panel A will be late by 2 hours. Depending on the
Figure 3. Panel Manufacturing Process situation this could result in lost incentives, delay
charges, expedited transportation/delivery costs, and/or
lost customer goodwill. Furthermore, the problem is
not just limited to Panel A. All successive panels will
incur the same CoPQ until some excess capacity (i.e.
overtime) is available that allows the panel production
to catch up.
Figure 6. New Schedule after Reworking Panel A (S Since 2000, authors have been involved in Six Sigma
= setup) implementation, starting the Continuous Improvement
of Drydocking Project sponsored by the Office of
Naval Research via the Gulf Coast Region Maritime
Figure 4 illustrated that without the 4 hour buffer a Technology Center at the University of New Orleans.
problem that goes undetected becomes more Pilot projects included reduction of growth work during
problematic. In this case the problem is found at hour drydocking, ship repair specification development,
5. The plate joining station is in the middle of crane availability improvement, paint cost reduction,
processing Panel B at this point. Therefore, it is ship repair contract development, error reduction in
necessary for the plate joining station to either break shipyard accounts payable process, as well as tow tank
the setup for Panel B and immediately do Panel A process improvement. The first wave of LSS
(Figure 7) or wait for Panel B to be completed and Implementations included accuracy of deflanging I-
insert Panel A (Figure 8). Based on meeting beams, and accuracy of stiffener fitting process at panel
completion times, not allowing setups to be broken line, which are led by the authors as well as numerous
(Figure 8) performs better. Figure 7 shows that Panel B pilot projects led by NGSS employees. Some of these
would be completed by the stiffener welding station on projects are summarized below.
time. However, Panel A would be late by 4 hours (hour
13 rather than hour 9), Panel C would be 3 hours late
(hour 16 rather than hour 13) and Panel D would also 5.1 ACCURACY OF DEFLANGING OF I-
be 3 hours late (hour 19 rather than 16). Thus, the BEAMS
CoPQ including the domino effect is the cost of
being late for panels A, C, and D; rework costs for A baseline study was conducted to identify the
Panel A at both plate joining and stiffener positioning performance of the current process and adherence to
stations; and idle time for the stiffener welding station. accepted hull standards. The focus of the project was
on deflanging I-beams after data collected from the
baseline study revealed excessive variation in the
process. This variation creates welding re-work and
results in added weight to the structure. The standard
procedure for the process was analyzed and found to be
inefficient in some areas. This tends to increase cycle
time and likely adds to process variation. Maintenance
practices are also studied to identify the effectiveness
of the current maintenance policy and to examine the
Figure 7. New Schedule to Rework A with Broken
availability of the machine. To reduce the variation of
Setups Allowed (S = Setup) the deflanging process and more effectively meet hull
standards, several improvement areas were identified.
A modification to the web sensor that controls the
plasma cutting torches was deemed necessary to reduce
deflanging variation. The modification significantly
improved process control limits and also reduced the
amount of material left on the structure after being cut
[15].
board displaying both neat edge alignment and January February March April May
Downtime (hrs)
150.0
90
also give management a picture of the processes 100.0
Downtime (hrs)
70
without interfering with production [15]. 50.0
0.0
50 January February March April May
30
y
PM
m
or
ist
e
al
r
es
ts
-10
ve
we
the
lle
Tim
e
tric
o
t
Ho
ak
era
Tra
gn
Bo
Tro
Po
a
Br
We
Ma
p
Ele
Sto
Ge
LSS is now used extensively to improve the
Availability metrics of equipment at NGSS. One such
project is focusing on the Availability of a critical Figure 9 Crane Downtime Chart Indicating Total
crane, and uses data generated by the Production (Inset) and Categorized Downtime [16]
organization to identify common failure categories and
to work with the Maintenance organization to perform
root cause analysis and corrective actions (RCCA). 6. VALIDATION OF SAVINGS
The data collected by Production not only reports As mentioned in the section 4 of this paper, accounting
downtime due to mechanical failures, but also due to for all CoPQ costs represents a major challenge when
non-maintenance related events causing stop time. The trying to quantify the benefits of LSS projects.
data collection system also has a category to record Financial Indicators are based on investment and
weather related downtime since coastal weather savings. For shipyard LSS projects, there are two
regularly causes this particular crane to be idle. investors: shipyard and ship-owner (customer). If
Generally, these two categories are the portion of project is successful savings accrue to shipyard or ship-
downtime that is not related to mechanical failure or owner or both shipyard and ship-owner. This situation
any other maintenance related activity. results in two indications of financial worth, one to
each investor.
The chart in Figure 9 identifies common downtime
categories on a monthly basis beginning in January To better illustrate validation of savings we will show
2004. In this figure, the scale has been removed to the example stiffener welding project [17]. The actual
protect proprietary information. The data shows a five dollar amounts have been removed to protect
month period of corrective maintenance related proprietary information.
downtime is 47% whereas weather related and
operations related stop time are responsible for 22%
and 20% of downtime respectively. Figure 9 (inset) 6.1 SHIPYARD SAVINGS
shows how addressing the root causes of various
downtime categories has resulted in a significant Savings to shipyard include:
reduction of downtime for the crane for the January to
May time period. The total downtime has been reduced Reduced rejection of stiffener due to quality,
by 42%, but downtime due to only mechanical failure Increased processing speed allowing less labor,
has been reduced by 70%. Reduced cost of welding wire due to smaller weld size,
However, this LSS project is currently in the Improve Reduced cost of weld rework due to improved quality,
phase and there are several root cause categories and Reduced beam setup time (not yet determined ~ no
processes that still need to be addressed. It is evident estimate calculated), and
that this is causing some fluctuations in the cranes Reduced inspection time (not yet determined ~ no
monthly downtime. Once these additional RCCA tasks estimate calculated).
are completed, the project will progress into the Control
phase where an effective Control Plan will be It should be noted that weld-grind and grind costs were
developed to stabilize and monitor the cranes determined by timed experiment for sample panels
downtime. Based upon project objectives, the undergoing rework. Defects and defect rates with
associated cost before improvements and after amortized over the three projects. Supplies are also
improvements were computed. used on projects with a pro-rata absorption estimate.
The salary and training costs dominate. Some other
A summary of savings for shipyard results from these costs include the cost of hourly participants and
the project is shown in Table 1. material costs for the fabrication of test panels. Each of
these costs is determined in the order outlined here.
Table 1 Summary of Annual Cost Savings to
Shipyard [17] Table 2 summarized the costs experienced by the
Reduced rejection of stiffeners XXXXX* shipyard during the life of the project.
due to quality
Increased processing speed XXXXX Table 2 Summary of LSS Project Costs [17]
allowing less labor Type of Cost Amount
Reduced cost of welding wire XXXXX Black Belt Candidate XXXXX
due to smaller weld size Hourly Meeting Costs XXXXX
Reduced cost of weld rework XXXXX Fabrication Costs Labor XXXXX
due to improved quality Fabrication Costs Material XXXXX
Current Total Estimated XXXXX
Savings to Northrop Grumman The shipyard Return on Investment (ROI) for this
*Due to nondisclosure agreement the dollar values will project is calculated to be 2,409.7%. Clearly, this
not be shown. project has shown significant benefits to the shipyard.
The shipowner ROI is determined to be 486.7% as a
result of the stiffener welding LSS project. Clearly
6.2 SHIPOWNER SAVINGS booth shipyard and shipowner benefited. Ultimately, an
LSS project financial success depends on how the
Cost savings that are available to the ship-owner company culture is willing to change. In the next
include improved ownership costs, and in the case of section we will discuss impacts of culture on LSS
stiffener welding these ownership costs are principally implementations.
derived from reduced dead weight of the vessel. Using
methods developed by a ship program office to
estimate the beneficial effects of weight reduction, the 7. CULTURE CHANGE
savings to the ship-owner for that weight reduction can
be found, over the life of the ship. However, this is only LSS implementation requires culture change. This is
part of the weight reduction. Improvements to the especially important in shipbuilding were data driven
stiffener quality resulted in reducing its weight. A culture is relatively new. One of the characteristics of
formula developed by the ship program office includes successful LSS efforts is that top management thinks of
calculations of fuel savings for the ship-owner resulting LSS as a culture. It becomes a way of life where all
from the weight reduction. improvement efforts fit under the umbrella of LSS.
Although the LSS implementation is still relatively new
When the ship contract is a cost plus contract, savings in shipbuilding, the shipbuilder examples used in this
to the shipyard are also savings to the ship-owner. So, paper shows the full support for the LSS program.
the net savings to the ship-owner includes both the Expanding workforce training creates larger groups of
ownership costs, and the cost of construction. people that are able to speak in terms of LSS. All
However, the timing of these costs is different. Savings improvement activities start to fit under the LSS
from decreased ownership costs occur for every year of umbrella. Ultimately, the shipbuilder will reach the
ship operation, and savings from decreased cost of LSS sustainability goals once the LSS is no longer
construction represent one time savings, occurring referred to as a program. Furthermore, it may not even
when the ship is constructed. be referred to as LSS.