One, if not the most important, KPI to assess production is the Overall Equipment Effectiveness (OEE). It multiplies availability, performance and quality within a designated factory to derive a statement about how effective the equipment is utilized. Through implementing this powerful KPI, companies were able to significantly improve their production processes and develop industry-wide best practice examples.
Using the OEE as the key driver for operational excellence focuses all shop floor activities on guaranteeing the uptime of equipment. In exceptionally manual work-load heavy industries or segments where human labor can not be replaced by machines (e.g. assembly) the Overall Labor Effectiveness (OLE) is used to describe productivity. Its structure follows the same pattern as the OEE. But it only measures jobs that generate value in building a physical product. OEE and OLE focus on long-term horizons and are therefore not feasible for agile production and balancing of short-term shopfloor activities.
The expenses in high-cost countries for labor costs in relation to revenue have a significant impact on companies’ cost and therefore profit, e.g. 26% in machinery and 28% in the metal processing industry in Germany. Thus, additional, more granular KPIs are needed to assess shop floor activity effectiveness.
For this reason, we at WORKERBASE and GKN PM have jointly developed the Overall Activity Effectiveness (OAE). This new KPI can be measured using the Workerbase system and has been introduced and evaluated at GKN’s digital manufacturing reference plant: Based on a fully connected factory and real-time work coordination all operational tasks involved in producing a product are dynamically managed by a load balancing system. Tasks are assigned based on skills and priority to the next free employee to guarantee optimal and prioritized execution of processes.
Each task has a timestamp and a status, that describes its progress:
The timestamps generated with the WORKERBASE system are used to determine the share of value-adding activities. To prevent abuse of personal data all data points are anonymized and aggregated on a higher stage. This way KPIs on segment and section levels can be used for data-driven decisions without interfering with individual data.
Starting from the total shift time, planned downtime (e.g. breaks) is deducted to derive at the available shift time. Same as the OEE the OAE consists of 3 factors: Availability, Performance and Agility. Multiplying these factors describes the share of value-adding activities to the available shift time as a percentage (see figure 1).
Most production tasks are scheduled in advance, such as setups, material movements, maintenance, assembly, machine interruptions and quality tasks. However, some unscheduled tasks need to be executed immediately to guarantee a smooth production. Thus, availability calculates the share of scheduled task time (including idle time) of the total available time. By increasing availability and minimizing unscheduled tasks, the planning quality can be improved. Companies can better respond to changing production demands with more flexible shift planning and optimize workforce utilization and planning.
Since most task durations fluctuate around a specific length, deviations through small issues, errands, miscommunication, etc. need to be taken into account. Analogously to the OEE the performance can be calculated in two ways:
1. Deducting deviations from scheduled task time to get to operational task time
2. Multiplying ideal task time with the occurred number of tasks.
Using the second method, ideal task times are derived from best practice examples (e.g. 10% quantile of task duration) and multiplied with the occurred scheduled tasks during the considered period. That way performance benchmarks productivity against an actual target.
Operational task time is further divided into task response time and task in progress time. Task response time characterizes the time between notifying the first available employee and acceptance of the task. The remaining task in progress time represents the actual time that is spent on value-adding processes.
If a task's response time is exceptionally high, skill distribution on the shopfloor might not reflect actual possessed skills. Tasks are therefore not executed and put on hold. A low response time thus does not necessarily indicate low individual performance but points to wrong planning and insufficient skill distribution.
Working with OAE offers production planners full transparency of the ongoing work on an aggregated and consolidated level without accessing personal information. In-depth analysis of tasks assignments and response times are possible in an anonymized way. Thus, production planners are able to assess the effectiveness of new procedures and layouts and can constantly benchmark new processes. In addition, the OAE uncovers potential skill shortages during certain shifts and thus enables competency management measures which are aligned with real needs.
The OAE KPI is not meant to offer management the ability to assess and fine-control the performance of individual employees. However, by using the WORKERBASE system, employees can self-control their individual performance. For example, operators can receive a daily report stating central KPI´s of their day e.g. actual working hours, performance benchmark vs. previous days. Through detailed insights, gamification structures are implemented and the continuous improvement initiatives are fostered on a regular basis.
To drive short term improvements, WORKERBASE and GKN PM have developed a toolbox of different levers and implementation strategies for OAE optimization. Figure 2 depicts the root causes of efficiency losses and the generic measures from a management perspective. The toolbox enables decision-makers to define and detect the biggest levers for improvement based on real time data, bringing transparency into manual and hard to measure cross-functional operations.
By using this concept, scheduled recurring tasks are constantly optimized and deviations from ideal task time are reduced. This enables management to conduct planning and forecasting more accurately, resulting in production to run more smoothly while reducing operational costs. In addition, by focusing on unscheduled tasks and the reason for their occurrence, ad-hoc events are transformed into scheduled tasks using predictive operations. With this concept, manual operations are load balanced to further smoothen peak workload and increase uptime of equipment.
To smoothen shop floor activity levels and to increase OAE, predictive operations are needed. Based on historical data, activities are forecasted and integrated into the task backlog before they actually occur. Figure 3 depicts the open tasks in logistics at one of our partners’ manufacturing sites. While the baseline displays the number of employees working in logistics, the blue line represents the number of open tasks. In this example for predictive operations, the occurrence of tasks can be predicted based on preceding processes and sensor data. The assignment of those tasks can be scheduled to smoothen activity levels. Thus, interruptions are prevented by scheduling countermeasures before peaks occur.
The prediction quality can be expressed in two ways taking into account the time a task spent in status open. First, the average response time measures how long a task on average is dwelling in status open. This indicates how fast people are able to respond to upcoming tasks for a given period.
Second, the prediction performance is calculated dividing the time at least one task is in the backlog by the available time of all employees working in a designated area.
Using these two metrics enables managers to assess the quality of the prediction by comparing before and after values. The impact of different triggers and order variations on the shop floor and their effect can directly be correlated to these KPIs. Thus dynamic adjustments to the order queue (pearl chain) are made to optimize a certain target (e.g. throughput time, load, due date, etc.), based on machine learning and historical data.
Introducing the OAE as a new metric creates the same transparency OEE brought to the machine world into the manual manufacturing world and enables data-driven manufacturing. While the OEE is often perceived as a measure used from management, OAE can easily be explained to operators, especially when broken down into its individual easy-to-understand components. Through real-time performance evaluation, operators are able to link their actions directly to the output and therefore accept OAE as the key metric for agile production. Thus, OAE has the potential to revolutionize the way work is scheduled and executed, paving the way for efficient use of manual activities through agile manufacturing in high waged countries.
Marius Maier, Digital Transformation Consultant, WORKERBASE,
Thorsten Krüger, Co-Founder, WORKERBASE,
Paul Mairl, Chief Digital Officer, GKN PM