Use analytics to improve a FMS
The performance of an automated flexible machining system (FMS) depends on the interaction between scheduling, part programming, operations, engineering design, quality control, tooling, purchasing, and more. This encourages a holistic view of the entire system instead of focusing on each component (such as part programming) in isolation. Recent advances in big data technology make it easier than ever to understand the entire manufacturing process, investigate trade-offs, and calculate metrics on monthly cost-per-piece, resource utilization, and system bottlenecks. Data analytics leads to understanding the entire system as a whole and actionable insights for improvements and future investments in automation and process improvement.
Actionable Insights
Data collection and analytics focuses on presenting actionable insights. Raw data collection and information visualization can generate many metrics and insights into a system, but the challenge is to produce clear, relevant, specific insights. For a FMS, in our experience there are two metrics which lead to actionable insights: cost-per-piece and resource utilization.
Cost-per-piece is a great actionable insight because it is easy to understand, specific to the ultimate purpose of the FMS, highly relevant to the business, and can be clearly presented to all stakeholders. For ordering, an accurate analysis of cost-per-piece allows easy quoting of work. For accounting, cost-per-piece allows justifications for future capital investments in new machines or new automation. For operational management, cost-per-piece helps understand the impact of a process change. That is, management can see if a process change has improved cost-per-piece and therefore if the process change should be kept or adjusted. The entire business benefits from visible and accurate cost-per-piece metrics, because various departments and employees can see that even if a change requires them to perform extra work, it improves the overall business value of the system. The only data required from daily operations is the monthly quantity of parts produced, and it is implemented on the flexibility analysis page.
While it is the ultimate goal, overall cost-per-piece is not a useful metric to improve the system. Sure, a cost-per-piece breakdown will show the costs of labor, machining, and tooling but how do you turn that into a plan to change something? Also, focusing on improving cost-per-piece directly risks optimizing cost by sacrificing quality. Instead, the best metric for continuous improvement is the bottlenecks and utilization of system resources, since the goal is to produce more stuff with the same machines and quality.
Utilization and bottleneck metrics are more difficult to turn into actionable insights. Many companies generate OEE (operational equipment effectiveness) data or collect machine utilization and bottleneck information, but seemingly inconsequential details in one area can have large impacts in other areas. For example, the chosen part programming technique might require complex operational part tracking on the shop floor or scheduling techniques might increase quality control overhead. When each department focuses only on their own slice of the system, they can't turn these insights into actions.
To turn the system utilization and performance metrics into actionable insights, we suggest the creation of a tactics council. The tactics council is made up of a representative from each department and component of the system, and the council's job is to decide on the overall tactics for system operation. By combining representatives from all areas of the system, the tactics council can produce actions (changes in how the system operates) based on the information and metrics on OEE and system performance. Helpful efficiency metrics and graphs are displayed in the flexibility analysis page.
Tactics Council
How can we produce actions from the various insights and information on OEE, machine utilization, labor use, and system performance? Because improving a FMS requires coordination between various stakeholders across several departments, a great management technique is a tactics council made up of a representative from part programming, operations, scheduling, tooling, quality control, accounting, and anyone else involved in the system. The council is the primary recipient of the system performance metrics, meets once every 6 weeks or once every two months, reviews the data, brainstorms improvements, and has final authority on the overall tactics for how the FMS operates.
Plan. Do. Check. Adjust. The PDCA technique is an effective strategy to improve the performance of an existing FMS and is ideal for the tactics council. During the tactics council meeting, the council will perform the plan and adjust phases of PDCA and in the 6-weeks to two months between meetings, individual departments will perform the do and check phases of PDCA. The tactics council should focus on only a single or at most two improvements at once.
For example, say that an improvement idea is to implement material tracking by adding a serial and barcode to each part during machining and then scanning the barcode at the inspection station, allowing the inspection operator to view the pallet, machine, and time for this specific part. The first step is to develop the plan, which would include the cost of purchasing a barcode printer, an analysis of the extra time at the load station applying the barcode to the part, a target for reducing the percentage of parts that require inspection, the cost savings from the reduced inspections, and other facets such as training. This plan is presented to the tactics council, the council discusses it, and say the tactics council agrees to implement this change. In the 6-weeks to two months between council meetings, operations will purchase the barcode printer, start barcoding parts, and start collecting data (the do and check steps of PDCA). The inspection stand will start using the barcodes and importantly record data on quality improvements. The next tactics council meeting can then review the data to understand the impact on overall cost-per-piece and system performance. By reviewing these metrics, the council can decide on any adjustments if necessary.
The tactics council is an important management technique because it encourages buy-in from everyone to improve overall cost-per-piece and system performance instead of focusing on individual isolated metrics such as cycle times or inspection rates. Typical improvements are similar to the previous example, where one department (such as operations) has extra work and a different department (such as quality control) sees an improvement. The tactics council and the repetitive nature of the PDCA technique keeps everyone happy, because the next improvement implemented by the council might improve operations while the inspection stand might have some extra work. Since everyone has a representative on the tactics council and the tactics council focuses on part cost-per-piece and total system performance, one department refusing to implement a change is greatly reduced.
Continuous Improvement
The utilization and bottleneck metrics which lead to actionable insights vary by system and part mix. The overall system OEE is easy to calculate by taking the total quantity of parts produced in a month, adding up their planned time, and dividing by the hours in a month. But the OEE is not an actionable insight: is the OEE lowered by a pallet traffic jam, a problem in a part-program, tooling capacity limitations, cart contention, load station inattention, or something else? This is where the repetitive nature of the PDCA technique shines. Every two months, the tactics council meets, reviews the available data, and can suggest new metrics based on what was learned from the previous reports. For example, perhaps the overview of the system identifies that pallet 5 has very irregular and erratic flow times. This could be caused by competition between pallets, lack of flexibility, traffic jams, a problem in the part-program of a part run on pallet 5, lack of tool capacity, or something else. In this case, more detailed data collection can focus on pallet 5 to help identify the problem.
Cost report
Calculating cost-per-piece requires splitting large fixed costs such as machine depreciation and labor across the parts that were produced. To do so, we suggest a monthly analysis window and dividing costs by planned use of resources. Collect data on the total quantity of parts produced during a month, along with their planned use of each resource such as machining time, load station time, tooling use, and inspection time and rates. The total cost is then divided among the parts according to weights based on their planned use of system resources. The cost-per-part can then be exported to the ERP and combined with order data from the ERP to calculate a cost-per-order.
The cost-per-piece is the primary metric and insight produced by the tactics council. It should be broadcast to the entire business as an actionable insight for quoting work, future capital investment, and general accounting. Cost-per-piece should be produced by the tactics council itself and not management because the council is in the best position to decide on cost trade-offs. For example, there is a trade-off between quality and inspections and cost, and the council is in the best position to decide on the required quality and inspections and then work to reduce costs only under these constraints without sacrificing quality.