Evaluating the Company’s Situation

Preter was founded in 1986 as a one-person company, and in the beginning it had only conventional turning and milling machines at its disposal. Over the years the company has grown to twelve people and it has become a specialist for single-piece and small-batch manufacturing. Today, the company manufactures custom metal parts on around 10 CNC machines. Typical operations required by the products include drilling, milling, outside processes at subcontractors (for example, coating) and finishing processes such as assembly operations. A major issue faced by the company is that qualified workers are hard to find due to the location of the factory. Consequently, improving processes and dealing with existing capacity in an optimal way is key to Preter’s ongoing success.

Upon reviewing the company’s situation, I felt that the small-batch and custom environment meant that QRM would be well-suited as a strategy for Preter. At our first meeting, I presented the QRM strategy and its benefits for this type of business. They had never heard about QRM and some of the concepts looked odd to them. However, the concept of POLCA seemed clear and attractive to them. With POLCA, they thought that they would be able to improve their performance significantly. They understood the principle of not pushing too much material onto the shop floor and to downstream resources, and that this would help reduce work-in-process (WIP) and shorten lead times. Again, they pressured me that their goal was to get software that would solve their problems, which were primarily a lack of transparency, inadequate planning, and no follow-up control on the shop floor. However, I convinced management to invest in a few days of analysis to develop a plan for how to move forward prior to jumping into the implementation of a software package.

Our initial analysis revealed numbers that are typical of this type of business. The average lead time, as measured by the MCT metric (see Appendix A) ranged from 20 to 70 days, with about 40 days as an average. Management had a strong focus on maximizing the utilization of machines (aiming for 90–95%), avoiding setup times through batching, and making products to stock based on forecasts. As a result, the touch time on orders (more formally, the gray space, as defined in Appendix A) never exceeded 2% of the MCT—this meant that during 98% of the time, products were just sitting on the shop floor! Finally, the metric of most importance to the customer, namely the on-time delivery, was below 60%.

We also reviewed the shop floor layout (Figure 15.1) and realized that there were no organized flow patterns. Because of the custom nature of the business, each work order would just go from one machine to any other machine, based on the operations needed and which machine had the right capability as well as the availability at that moment. Hence, job routings crisscrossed all over the shop floor, creating the classic “spaghetti flow” that is discussed in the literature on job shops.

Figure 15.1Shop floor layout at the Preter factory.

Opportunities and Challenges for Implementing POLCA

Based on our review of the lead times, on-time delivery, and shop floor layout, we felt there was definitely opportunity for applying both QRM and POLCA. However, we felt that some challenges would have to be overcome for this type of company. The first challenge was the small size of the company, which meant there were few machines and seldom duplicate resources. This limited the possibilities for restructuring into cells. Nevertheless, we tried to determine some families of orders (known as FTMS in the QRM terminology, see Appendix B) around which to create cells. We found that in order to keep their customers happy, the company accepted many quick turnaround jobs, which they just added onto their existing heavy schedule! As expected, these jobs jeopardized the delivery of their regular orders. So, we thought it made sense to dedicate some capacity (specific machines) to these quick-turn jobs. Thus, we created just two FTMSs, one for the regular orders and one for the quick-turn jobs. We used this approach to create cells for the processing of these two types of jobs. Due to the small number of machines, sometimes a “cell” would only be one machine, but as explained in Chapter 2, POLCA can be used between standalone machines as well.

The second challenge we faced was that some orders had few operations, i.e., very short routings. We knew that the benefits of POLCA increase with the complexity resulting from the number of cells and the network of routings, and the management of the company wondered if POLCA would have a significant impact with all the simple routings, and this also made them a bit hesitant about going forward. Nevertheless, I was pretty sure that the move to cells—including planning on a cell level (which includes the combination of several operations), dedicating capacity to the different types of jobs, having a clear policy on spare capacity, and using POLCA to manage and control the flow—would still provide positive results. I was able to convince the company to move forward with the implementation.

Cell Formation and POLCA Launch

Our approach was to start with basic QRM training for everybody in the company, and then to brainstorm together with everyone to create the cell structure that would support the two FTMSs. Finally, we would implement POLCA using our timeaxx software.

Through the brainstorming we came up with three cells: a 5-Axis Cell consisting of machines C40-1 and C40-2; an NEX Cell with machines NEX250 and NEX250-II; and a Mill-Turn Cell consisting of machines C42 and SQT300. Ideally, we would have liked to place these sets of machines physically together, but management at Preter was reluctant to move machines since they had just moved to their new building and installed every machine with expensive foundations. So, we agreed to go for a combination of physical and virtual cells. Figure 15.2 shows the three cells: you can see the 5-Axis Cell is a physical cell, while the other two are virtual cells, which means that the machines are not collocated, but they work organizationally as one planning unit. We also dedicated an older machine (600E) to be used for spare capacity. The remaining machines were left as standalone machines (or “one-machine cells”).

Figure 15.2Cell structure and standalone machines.

Finally, we installed the timeaxx software, which began operation in March 2015. The POLCA loops, cards, and rules are implemented through the software. Because the company makes so many different parts and there are constantly new parts, there is a POLCA loop from almost every cell to every other cell; these are not shown on the figure as it would become very crowded. We use the Release-and-Flow version (RF-POLCA) at Preter, and the POLCA decisions are communicated to the cells by means of simple “traffic lights” on the computer screens. For example, for the initial release of jobs, timeaxx goes through the standard Decision Time Flowchart (Chapter 2) for each job and sets the traffic light to green if that job can be released to the first cell. The Authorization Date for this first step is calculated by timeaxx using sophisticated finite-capacity scheduling starting from the promised delivery date while also considering the real-time status of each job in the company. At subsequent cells, the RF-POLCA rules (Chapter 5) are used to set the traffic light to red (don’t launch) or green (okay to launch) for each job.

Figure 15.3 shows some of these features of timeaxx. The traffic lights are indicated in the left-most column. Jobs that can be started show this column in green, while those that cannot be started have this column in red. Jobs already launched in the cell are in yellow. (There is one additional color used—orange indicates a job that is in process but temporarily suspended, for example, if an operator has gone on break.) Also, in timeaxx, since we are using the computer to manage the POLCA cards, we decided on a quantum of one minute, which allows us to manage capacity at a highly refined level. However, this means that in some loops we might have thousands of POLCA cards, so we decided that instead of indicating the absolute number of cards available, we would indicate the percentage available (as a percentage of the total cards in the loop). This gives the workers and management a quick intuitive overview of the downstream capacity available in a given loop. You can see these percentages in Figure 15.3. Note that Item 5 has no cards available, so the traffic light is red. On the other hand, Item 4 does have cards available, but the traffic light is still red because the Authorization Date has not yet been reached. Finally, the small vertical bars under the percentages indicate the status of downstream cells. The idea is that if multiple jobs have the same Authorization Date and available POLCA cards, then these bars give the operator some visibility to decide which job would be better to start based on the status of downstream cells.

Figure 15.3Example of timeaxx screen for jobs at a cell.

Results from the Implementation

From other projects in much larger organizations we have seen that the more complex the structure of the shop floor, the larger the potential effect of implementing POLCA. Therefore, the project in a small organization such as Preter was challenging because of the limited room for reorganization. Nevertheless, this project has been very successful. Within six months of starting the use of timeaxx and POLCA, the company saw a 30% reduction in the MCT metric for the shop floor. This outcome continues to be achieved at the time of writing this case study. The POLCA implementation has also led to a significant reduction in customer complaints and hence higher customer satisfaction, resulting in increased customer loyalty. Therefore, we have delivered practical proof that POLCA can bring positive effects even in smaller organizations.

Besides the hard numbers, the company has benefitted from several people-oriented effects. The POLCA system along with the timeaxx features has dramatically increased transparency for both the planners and the shop floor workers. What you see can be controlled much better than using mere numbers.

Planners at Preter now feel confident about the actual situation within the company. Compared to the uncertainty in past decisions, planners now see whether a job should be started or not with the help of the simple traffic lights as described above. This is reinforced by the fact that the planners have found the starting dates from timeaxx to be very reliable for achieving the promised delivery dates, thus increasing their trust in the traffic light signals. Furthermore, planners are able to keep track of the operations being executed and the status of work orders, so they can quickly respond to queries from customers.

Even the most experienced shop floor employees have accepted the new way of working. On their tablet devices running timeaxx, they can see in real-time which order needs to be processed using the same traffic-light methodology mentioned above.

The buy-in from the people in the company can be seen in the words of one of the planners: “The daily struggle of scheduling, rescheduling and frequently expediting jobs has finally come to an end. Even if the schedule is pretty tight, the system now provides the signals and the transparency to help our people understand what needs to be done and to adapt their operations and activities in order to meet expectations.”

A final indicator for the success of our implementation is one that should not be forgotten: top management at Preter is very satisfied with what has been achieved.

About the Author

Markus Menner holds a degree in business administration in addition to his education as a specialist in computer science. He began his career as a software architect, development manager, and consultant in the area of ERP systems. He has worked as an R&D Manager at the Laboratory for OLYMPUS, as well as for Beckman Coulter, both global players in the medical device industry. In 2013, he founded axxelia with the focus on Quick Response Manufacturing (QRM) and its rollout in German-speaking markets. At the same time, he designed and developed the QRM and POLCA-based software system timeaxx, which is now in use by many customers. These customers include large, medium-sized, and small companies that use a variety of ERP-Systems including SAP, Microsoft Dynamics, SAGE, and Infor, together with timeaxx.