Unlocking Efficiency: Data science to improve OEE

This graph is interactive! (drag the dashed — lines)

The OEE (Overall Equipment Effectiveness) indicator is used in manufacturing industries to rate the operational efficiency of machines and production processes. It was created in Japan in the 1960s as part of the TPM (Total Productive Maintenance) methodology, developed by Seiichi Nakajima. This metric is aligned with the Lean production philosophy, which focuses on the continuous improvement of processes, enabling its presence in all industrial sectors. This improvement is directly linked to the adoption of new technologies, such as data science, which results in a more accurate calculation of OEE. According to the National Confederation of Industry (2022), the Brazilian industry is increasingly digitalized, with the automotive sector as its leader.

OEE is calculated from three key factors:

• Availability: Measures the time the machine is actually available to produce compared to the scheduled time. It takes into account planned and unplanned downtime (such as maintenance and failures). Its importance lies in the fact that the higher the availability, the more time the equipment spends producing, which increases overall efficiency.

Calculation: Availability = Actual Production Time / Planned Time;

• Performance: Evaluate the actual production speed against the machine’s rated speed. This means that having good performance is synonymous with the equipment operating close to its maximum capacity. It already includes losses due to short downtime and reduced speed operation.

Calculation: Performance = Real Production / Ideal Production X 100

• Quality: Measures the quantity of good products to the total products manufactured, considering losses and rework. More efficient equipment generates fewer defects, improving the overall quality of production.

Calculation: Quality = Good Products / Total Produced X 100;

Exploring OEE with Data Science

The interpretation of the OEE calculation may vary, but, in general, industries seek a number above 85%, considered the world standard. Levels below 60% indicate serious problems in production, and levels between 60% and 85% show significant opportunities for improvement.

To achieve this result, it is necessary to collect complete operational data of the machine or production line. Those responsible need to understand what are the factory’s setpoints, that is, the ideal parameters of time, quantity and speed, for example. After that, it is necessary to evaluate the real production metrics, so that it is possible to identify points for improvement. This includes the studying unplanned downtime, low-speed operation, and units that needed rework.

Data collection is the first step to calculate OEE. The data is collected by monitoring devices such as the ST-One® Hardware. It records this information simultaneously and integrates data from various sources such as PLCs (Programmable Logic Controllers) and SCADA (Supervisory Control and Data Acquisition) systems. Then, the collected data must be stored in a cloud database, where each event (start-up, stoppage, etc.) is recorded in a timestamp. In case of larger structures, data lakes are used to store large volumes of data, structured or unstructured ones, which can be processed later. After that, the data is processed and applied to the OEE formula. Here, aggregation algorithms are used to gather the individual data from the machines. The results are presented in dashboards that show OEE in real-time, using data visualization libraries or solutions such as Stash Platform™. This intelligent view allows for preventive analysis based on historical data and scheduling alerts if OEE falls below the threshold.

Copyright: ST-One

Features in the calculation of OEE

Some factors act as groundbreakings elements during the OEE calculation process. Among them are aggregation algorithms, which are used to combine or summarize multiple values into a single value. They can come from a variety of sources, such as different machines on a production line, or in different time periods. Their goal is to create a representative metric that synthesizes the raw data, making analysis easier and faster, and aiding in decision-making.

“Aggregation is the process of combing a set of data, transforming and programming it in a cyclical way so that you have only a single piece of data. This is widely used in OEE because it is very complex, with several variables, and it is possible to reduce them to a single indicator, which is the percentage of OEE” – Daniel Michalichyn, Head of Development at ST-One

If the OEE calculation is done for a single machine, the formula process is straightforward. Now, if the intention is to calculate the OEE of a production line or a factory, it is necessary to aggregate the individual OEEs of the machines or lines. This can be done in different ways, such as by weighted average, where the contribution of each OEE is weighted by the operating time or quantity:

Weighted average OEE=∑(OEE of each machine × Machine Weight)/∑Machine Weight​

This approach is most accurate when different machines have different levels of importance or output. The choice of aggregation method depends on the complexity of the system and on the desired accuracy. Also, for large volumes of data, the efficiency resulting from the aggregation algorithm is indispensable, because methods like this are computationally more intensive. Finally, it is necessary to highlight the time optimization and the understanding of data with ease provided by aggregation.

Practical examples of data aggregation in the context of OEE

One way to use aggregation algorithms to calculate OEE is through time-based aggregation. For example, if the intention is to determine the OEE of a production line over a day, it is possible to aggregate each shift or hour. In this approach, each time period is assigned a weight based on its duration or the quantity produced.

It is also possible to perform aggregation by equipment. In a factory with multiple production lines, the overall OEE can be a weighted average of the OEEs from different lines. In this context, the weight is based on the production volume of each line.

The following is an example of how this would be applied directly to a calculation:

• Aggregation of availability data

Equipment A: Operational Time = 400 min, Scheduled Time = 480 min

Equipment B: Operating Time = 350 min, Scheduled Time = 480 min

Total Availability = (400+350) / (480+480) = 750 / 960 ≈ 78.1%

• Performance Data Aggregation

Equipment A: Performance = 90%

Equipment B: Performance = 85%

Average Performance = 90+85/2 =87.5%

• Quality data aggregation

Production Line A: Quality = 98%

Production Line B: Quality = 95%

Average Quality = 98+95/2 = 96.5%

• OEE aggregation

Equipment A: OEE = 80%

Equipment B: OEE = 75%

Average OEE = 80+75​/2 = 77.5%

• Trend analysis to predict improvement or regression in equipment efficiency

Monthly Availability: January = 85%, February = 87%, March = 90%

Availability Tendency= 85+87+90/3 ≈ 87.3%

Data aggregation in the context of OEE is a powerful tool for production management, enabling a complete and detailed view of operational performance. By using time aggregation, it is possible to identify variations in efficiency throughout the day, making it easier to identify specific bottlenecks across different shifts. Equipment aggregation allows performance to be compared between different production lines, highlighting which equipment or processes need improvement.

In addition, combining availability, performance, and quality data into a single OEE indicator provides a consolidated metric that reflects overall production efficiency. This integrated approach not only facilitates strategic decision-making, but also fosters a culture of continuous improvement, fostering optimization and operational excellence.

Copyright: ST-One

Challenges in the OEE implementation in industry

As previously discussed, the implementation of OEE in the industry involves several steps. To execute this process without problems, the industry must pay attention to the following points:

• Data collection: Implementing a system that collects data accurately and consistently can be challenging, especially in operations that involve manual steps;
• Culture Commitment: A methodology focused on continuous improvement requires buy-in at all levels of the organization;
• Interpretation and action: To fully utilized monitored OEE, it is crucial to interpret the results and act on them to improve the outcome;
• Continuous review: Regularly reviewing data and strategies ensures that the plant is operating efficiently and productively;

In addition to those bullet points, it is necessary to relate OEE to other metrics. One such metric is TEEP (Total Effective Equipment Performance), that measures efficiency by considering all available time, including periods outside of scheduled hours. It is calculated by multiplying OEE by utilization:

TEEP = OEE X Utilization

For example, if a factory has an OEE of 75% and utilization of available time is 80%, the TEEP would be:

TEEP = 0.75 X 0.80 = 0.60 or 60%

When talking about productivity metrics, it is important to establish the differences between OEE and overall productivity. Although related, OEE specifically focuses on equipment effectiveness, while productivity can include factors like labor and process efficiency.

In addition to the previous challenges, the implementation of OEE in the industry can be enhanced with technologies such as the Internet of Things (IoT) and data analytics. These allow for more accurate and real-time data collection, making it easier to identify patterns and anomalies. The integration of OEE with management systems (ERP) also provides a holistic view of the operation, aligning the efficiency of the teams with the company’s strategic objectives.

Finally, comparing OEE to industry benchmarks can help identify opportunities for improvement and set realistic and challenging goals. With these complementary approaches, the industry can not only overcome the challenges of OEE implementation, but also achieve higher levels of efficiency and competitiveness. Learn more about us