Plan. Do. Check. Adjust. The PDCA technique is an effective technique to improve the performance of an existing flexible manufacturing system. First, produce a model of the existing system and use it to create a plan to make a small change to the system. Next, implement the change to the system and then monitor the execution of the system. Finally, compare the collected data to the model produced during planning to determine the effectiveness of the change. The PDCA technique works well for flexible manufacturing systems because typically there is not a single large problem with a manufacturing system but a large number of small problems, each of which reduce the utilization and throughput by a small amount.
What is bottlenecking the system? The overarching goal is to improve the performance of the system, but for the PDCA loop to succeed we must break this goal into manageable chunks. A bottleneck is any process or machine or tool or robot which limits the throughput of the system. Typically there is no single large bottleneck, but a large number of small issues. The key to the planning stage is to focus on only a single improvement at once.
First, the SeedTactic: FMS Insight software collects data and provides a monthly report highlighting the cell's operation. The FMS Insight Guide on Analytics describes techniques to brainstorm potential improvements. Once a potential improvement is identified, the simulation-based SeedTactic: Planning and SeedTactic: Designer tools are used to predict the utilization and operation of all components of the system, and shows the potential impact of a change to the system.
Once a plan for a single small change has been developed, management, engineering, supervisors, and operators must all coordinate together to implement the plan. Because the PDCA loop will repeat many times, all stakeholders should be involved or aware of each phase. This is in stark contrast to new systems, where engineers can develop a detailed proposal and submit it to management. When improving systems, there are many small problems and they might not all be initially known. In addition, bottlenecks sometimes occur because of habits that have formed by operators or supervisors.
Once the plan has been implemented, detailed data on the operation of the system must be collected. Data is important because typically the overall performance of the system suffers the death of a thousand cuts, so a single small change is unlikely to make a visible change in the overall system performance. For example, perhaps a fixture was identified as a bottleneck and some of the work on that fixture was moved to a different fixture. The performance of the entire system might then bottleneck on a tool and so no improvement is seen in the number of parts produced per week. It is important to collect data on the operation of the fixture to determine that at least the fixture is no longer a problem. By collecting and sharing data, everyone can be confident that progress is being made even if there is no visible improvement in the overall system performance yet. Our SeedTactic: FMS Insight tool extracts, analyzes, and presents data from a range of manufacturing systems.
The planning stage produced a detailed prediction of the operation of the system, and the check stage collected data from the operation of the system. During the adjust phase, the predicted operation and actual data is examined side-by-side to determine if the change had the expected impact. In addition, this comparison is great to learn more about the system; each system has idiosyncrasies and unique traits, and by learning these quirks they can be incorporated into the next plan. Our SeedTactic: FMS Insight tool can display this side-by-side comparison. It imports the plan and the plan's predictions and compares it to the collected data.
Because systems typically suffer from many small problems, the PDCA stages must be repeated several times. We suggest a 3-4 week iteration cycle. In the first week, supervisors or engineers develop a plan, in the second week the plan is implemented, in the third week data is collected and analyzed, and the fourth week is a cushion in case things are delayed. In our experience, this initial improvement loop takes about 6-8 hours a week of planning and data analysis and lasts for 5-7 months. It is typically only at the end of the process that any large improvements in throughput and performance appear, so data collection to show that improvements are being made is important to maintain engagement.
Once this initial improvement occurs, the check and adjust phases should continue indefinitely. This is because over time, customer order demand changes and bottlenecks can appear. For example, perhaps a product becomes more popular and the increased demand causes a fixture to become a bottleneck. The good news is that once the planning and adjusting tools are in place, it is typically only 30 minutes to an hour a week to periodically review the data.
The performance of existing manufacturing systems suffers because of many small problems. The best improvement technique is to focus on only a single small problem at once. First, develop a plan and predict the utilization and capacity of all components of the system. Next, overcome bad habits to implement the change. Finally, collect data on the actual utilization and efficiency of the components of the system and compare the actual data to the predicted plan. By iterating this improvement process, the overall throughput of the system can be drastically increased.