Implementing Lean Manufacturing
Analyze each step in the original process before making changes
Missouri University of Science
Lean manufacturing's principles can be applied to the solution of problems in every industry, including aerospace. In one case, at a plant operated by the Spicer Axle Division of Dana Corp., manufacturing engineers set out to use the six-sigma DMAIC (Design, Measure, Analyze, Improve, Control) methodology—in conjunction with lean manufacturing—to meet customer requirements related to the production of tubes used to manufacture axles. Those requirements were related to both quality and production.
The objective of the lean effort devoted to the tube line was to reduce quality defects to a six-sigma level by developing a more-efficient process. This new process would not only reduce product cost through reduction of quality defects, but also by a reduction in WIP and lot sizes, and increased throughput. It's important to note that after lean manufacturing principles were implemented, the tube line was also used as a pilot cell for other layout designs throughout the plant.
Based on the then-current-state production process, the performance measure used to determine the achievement of target goals and the process design's effectiveness was sigma level. Supplemental performance measures included scrap, WIP level, throughput, machine utilization, quality, labor utilization, and labor costs.
Manufacturing engineers were charged with designing a new process that could meet a laundry list of objectives, including:
- Improved quality,
- Decreased scrap,
- Delivery to the point of use,
- Smaller lot sizes,
- Implementation of a pull system,
- Better feedback,
- Increased production,
- Individual Responsibility,
- Decreased WIP, and
- Line flexibility
To understand the initial-state process, the lean-manufacturing team developed a process-flow diagram with the cross-functional six-sigma team. This diagram allowed identification of the actual flow or sequence of events, and definition of problem areas, redundancy, unnecessary steps, and areas presenting opportunities for simplification and standardization.
After mapping the process, the lean team collected data from the Material Review Board (MRB) bench to measure and analyze major types of defects using a Pareto Diagram. To better understand the process, the team also did a time study of the existing operation.
In the initial state, the tube line consisted of one operator and four operations, separated into two stations by a large table using a push system. The table acted as a separator between the second and third operation, creating a kanban of approximately 172 parts. In the first operation, an aluminum tube is machined to face and bore three IDs and an OD. Next, the tube is washed to remove dirt, turnings, and oils in a batch size of five parts. A bearing is then pressed into the bearing bore and, finally, a bushing is pressed into the bushing bore, creating a final product. This component is sold to an assembly plant as a component for an axle.
Operators continued to run parts until they could not stack them any higher. Then, the operator would move to the other side of the table and run the bushing press and bearing press. Finished parts were gaged 100% for assembly dimensions, and stacked on a cart. After approximately 30 parts were finished, operators packed the parts into a cardboard box, two at a time. Double handling occurred between each grouping of operations.
The first problem discovered was the line's unbalanced nature. The first station was used about 70% of the time. Operators at the second station were spending a considerable amount of their time waiting between cycle times. By combining stations one and two, room for improvement became evident with respect to individual responsibility, control of inventory by the operator, and immediate feedback when a problem occurred. The time study and the department layout reflect these findings.
Due to lack of floor space caused by the large table's footprint, the inefficient process flow and large lot size caused a higher probability of an increased number of defects. Process flow also allowed the possibility of missed operations. Missing the bearing press, a common defect, required workers to sort not only inventory, but also product that was shipped to customers, to detect problems. Four such sorting operations, undertaken to look for missing bearings, had been performed over an eight-month period just prior to this lean implementation project. Defects found by operators were not counted in defect data, because operators were able to rework the parts, and did not enter "corrected" defects into the defect log. After operators went through training, all defects were entered into the defect log before rework.
A second problem was recognized. Because of the then-current process flow, the production rate did not allow the production schedule to be met with two stations. Because operators lost track of machine cycles, machines were waiting for operator attention. Operators also tried to push parts through the first station—the bottleneck operation in the process—and then continued to manufacture the parts at the last two operations. Typically, long runs of WIP built up, and quality problems were not caught until a substantial number of defective pieces were produced.
A Failure Modes and Effects Analysis (FMEA) created by the cross-functional six-sigma team helped the lean team identify where problems were likely to occur by using a Risk Priority Number (RPN). Subsequently a more robust process and a control plan were developed. A control plan not only enables a team to understand how a defect is controlled, but also includes the measurement systems and frequencies of inspection. This control plan was a critical step, because by using Gage R&R we discovered inadequate measurement systems, as well as features that were not being controlled.
As mentioned earlier, results of the initial time study showed extremely off-balance time utilization between operations. The first station was used some 70% of the time, and the second station about 30% of the time. Combining stations one and two improved production significantly.
The next step was to chart process flow. Doing so identified the critical opportunities for time-saving steps, and revealed potential problems. The process-flow chart sequenced the most effective process for operating the line, to reduce the amount of machine idle time and optimize throughput. Initial layout was also traced to show operator travel.
The new process sought to meet all of the customer requirements for the part. Also, due to the number and severity of recent quality problems, the quality of components produced by the line had to surpass the initial-state process quality.
Capacity for the coming year was unknown, so a U-shaped cell design was selected that did not limit the number of operators, and reduced material handling within the process. Capacity for the facility with the initial-state line was 330 pieces per shift, or 990 pieces per day. The average actual production rate was 300 pieces per shift or 900 pieces per day. Total usage by all customers, however, was 1528 pieces per day, based on a five-day work week. To meet this customer demand, the line had to run overtime.
Just-in-time (JIT) with kanban control and theory of constraints (TOC) were implemented to continuously improve the tube-line process, and reduce nonvalue-adding processes. The existing process wasted labor and yielded a high number of quality defects. The findings of the time study on the existing process provided the basis for reducing cycle time, balancing the line, designing the flow using JIT kanbans, and scheduling with finite capacity using TOC to reduce material handling, improve quality, decrease lot sizes and WIP, and improve flow.
Some of the concepts used to improve the process included total employee involvement (TEI), smaller lot sizes, finite capacity, scheduling, point of use inventory, and improved layout. All employees and supervisors in the department were involved in all phases of the project. Their ideas and suggestions were incorporated in the planning and implementation process to gain wider acceptance of the changes to the process. Smaller lot sizes were introduced to minimize the number of parts produced before defects were detected. Kanbans were introduced (in the form of material handling racks) to control WIP and to implement a pull system. And the cellular layout decreased travel between operations.
This project had actually been proposed nine months earlier, but because of lack of support from key members of management, the project sat idle. By illustrating the cellular layout and walking through the production flow with each operator and the production manager, lean team members received approval to rearrange machinery. A simulation model was used to illustrate the difference in performance measures before and after the implementation. A plant-wide idea implementation system enabled operators to contribute suggestions to improve the process and eliminate waste on the line.
The cellular layout immediately increased production. It only allowed one part in each machine and one en-route at any given time. The table employed in the initial layout was removed, almost eliminating WIP. Because it was understood that WIP does not improve production, the pull system approach to increased production was accepted. The actual number of finished products was recognized as key. Kanbans in the form of racks hold only one in-process part, reducing WIP and increasing production.
With the reduction of WIP and implementation of a pull system, defects were detected immediately, allowing operators to adjust the machinery to produce in-tolerance parts, or notify maintenance that timely machine repair was needed. Consequently, the amount of scrap generated was dramatically reduced.
Operators were authorized to stop the line when problems arose. In the initial-state arrangement, the operators in some instances were still able to continue running parts when a subsequent operation was down. With kanban control, the layout eliminated the ability to store WIP, requiring the operator to shut down the entire line. The cellular layout provides excellent opportunities for improving communication between operators about problems and adjustments, to achieve better quality.
Because of the cellular layout, the number of operators is now flexible. Previously, the layout was organized so that a maximum of two people could run the line. Now, the number of operators can vary depending on demand.
Day-to-day inspection of the initial-state process revealed that operators spent a considerable amount of time either waiting for material-handling personnel, or actually performing the material handling. With the U-shaped cell, delivery to the point of use is more convenient for the operator. The operator places boxes of raw material on six moveable roller carts, where it's easily accessible. The six boxes are enough to last a 24-hr period.
To reduce setup times, tools needed for machine repair and adjustments are located in the cell. The screws are not standardized; tools are set up in order of increasing size to quickly identify the proper tool.
An evaluation of the improvements was conducted to quantify the benefits achieved. For three months the process was monitored to verify that it was in control. Comparison of time studies from the initial-state arrangement and the implemented layout demonstrated an increase in production from 300 to 514 finished products per shift. The new layout eliminated double handling between the second and third operations, as well as at the packing step. It also reduced throughput time by making it easier to cycle all four operations in a pull-system order. Customer demand was met by two shifts, which reduced the labor cost.
Benefits and costs were evaluated to determine the overall success of the new layout. The benefits achieved by a cellular layout include:
- Quality level increased,
- Decreased scrap,
- Decreased material handling,
- Better feedback,
- Increased production,
- Line flexibility, and
- Decreased WIP.
Production and scrap numbers were taken from 20 consecutive production day periods. The initial-state data were taken from the last 20 days before the change, and the current-state data were taken starting one month after implementation. This delay gave the machine operators an opportunity to become familiar with the layout.
In summary, the production group wanted to develop a process to decrease product cost while achieving improved quality level, decreased defects, increased production throughput, and improved production capacity, while providing a design that would not impose a limit on the number of operators.
The results of the redesign are as follows:
- WIP decreased by 97%,
- Production increased 72%,
- Scrap was reduced by 43%,
- Machine utilization increased by 50%,
- Labor utilization increased by 25%,
- Labor costs were reduced by 33%, and
- Sigma level increased from 2.6 to 2.8,
This project yielded reduced labor and scrap costs, and allowed the organization to do a better job of making deliveries on time, while allowing a smaller finished-goods inventory. Daily production numbers and single-part cycle time served as a benchmark for monitoring progress towards the realization of the goal. Although the sigma level increase was not immense, the 43% reduction in defects, 97% reduction in WIP, and production increase of 72% contributed to the project objective.
Elizabeth Cudney is an Assistant Professor at Missouri University of Science and Technology. She received her doctorate in Engineering Management from the University of Missouri–Rolla. Cudney was awarded the 2008 ASQ A.V. Feigenbaum Medal and the 2006 SME Outstanding Young Manufacturing Engineering Award. A senior member of ASQ and an ASQ Certified Quality Engineer, Manager of Quality/Operational Excellence, and Certified Six Sigma Black Belt she is a member of the ASEM, ASME, ASQ, IIE, and SAE. She can be reached at firstname.lastname@example.org.
This article was first published in the March 2010 edition of Manufacturing Engineering magazine.