The manufacturing industry is known for the large data sets it produces. These data sets include both structured and unstructured data from production processes and for a long time they went uncollected until the advent of Industry 4.0.
Earlier advancements such as lean manufacturing methodologies leveraged data. To implement lean processes, the manufacturer had to capture machine-related data and inventory usage to reduce waste. The collected data sets were used to calculate machine utilization metrics such as overall equipment effectiveness (OEE) and to determine the number of available resources.
Applying data to implement lean manufacturing drastically cut down waste from individual machines and operations but it didn’t provide comprehensive facility-wide insight into the workings of the factory. The concept of Industry 4.0 intends to take the optimization of individual equipment to the next level by optimizing the interrelated operations that define manufacturing. Accomplishing this requires collating datasets from every aspect of a production cycle to optimize productivity.
Capturing Big Data on the Factory Floor
Leveraging big data to gain insight into manufacturing processes starts with capturing data. First, collecting data from machines equipped with modern communication technologies such as Wi-Fi is a different ball game to collecting data from legacy equipment with analog technology. The modern equipment can be plugged into networks to transfer data to the cloud or a centralized data aggregation platform while with legacy systems data must be extracted and transferred to the centralized platform.
Today, smart devices built for the industrial sector can be plugged onto the analog I/O or ports of legacy systems to capture data. Challenges with capturing data from legacy equipment are not the only issues manufacturers have faced with capturing facility-wide data. Traditionally, data concerning the shop floor’s environment such as temperature, facility layout, and data concerning material handling systems have been categorized as unstructured data and difficult to capture.
IoT now empowers manufacturers with the ability to capture unstructured data from the shop floor. IoT also enables the capture and flow of data in real-time across the factory floor to provide digital transformation technologies with the data they need to analyze shop floor operations.
Use Cases for Big Data in Manufacturing
Taking advantage of captured data sets is the next rung on the ladder once a manufacturer has successfully implemented a process to collect shop floor data. Use cases for leveraging big data include:
1. Predictive Maintenance
Currently, the most popular use case for historical data sets is optimizing maintenance strategies and reducing downtime through predictive planning. Predictive maintenance involves capturing historical operational data of shop floor assets to determine breakdown patterns of the asset and its components.
Successful predictive maintenance strategies reduce unplanned downtime caused by defective equipment by 75%. One such example is the predictive maintenance strategy of BASF, the largest chemical manufacturing company in the world. To eliminate its challenges with unplanned equipment downtime, the company implemented a data capturing strategy using IIoT solutions from Schneider Electric to capture machine data.
Leveraging big data, the company was able to capture 100 condition variables that relate to the health of its equipment across 63 of its shop floor assets. Analyzing the captured data-enabled BASF to drastically reduce its downtime and increase the life-cycle of its machines.
2. Condition Monitoring
While predictive maintenance actively monitors machine performance through data collection tools, condition monitoring attempts to discover anomalies across plant operations in real-time. Deploying IIoT and smart devices on the shop floor enable manufacturers to capture the data required to drive condition monitoring applications. Digital transformation tools such as the digital twin leverage collected data sets to build virtual representations of physical factory operations. The digital twin is then used to monitor operational processes within the manufacturing floor.
The real-time monitoring of wind turbines to ensure optimal performance and to gain insight into turbine operations is an example of the application of condition monitoring. The example of Brüel and Kjær Vibro, a German-Danish condition monitoring company highlights the importance of real-time asset management. The company continuously monitored turbine operations using 100s of sensors across for wind turbine installations. Utilizing condition monitoring, the company was able to forestall damages, pinpoint potential failure points, and gain insight into turbine operations to make informed decisions.
3. Production Forecasting
Getting correct answers to ‘what-if’ questions is the best way to determine the number of resources a production cycle will require to meet fluctuating demand. ‘What-if’ evaluations also assist manufacturers with deciding how available resources should be assigned to meet demand deadlines and improve customer satisfaction. Capturing demand data and production-related data can assist manufacturers with accurate production forecasts.
Data from historical demand cycles provide the foundation for demand forecasting while shop floor data enables the ability to evaluate production processes to meet increased demand.
An example is BAE system, a defense contractor’s, reliance on simulation technology to analyze its production data. Expecting increased demand, the contractor needed to develop an optimized schedule and properly allocate resources to meet production deadlines. To achieve this, a simulation model of its factory operations was created using historical data. The simulation model assisted BAE answer questions relating to its production capacity and resource allocation. BAE also developed a risk-based schedule to ensure it meets customer demand with quality throughput.
4. Improving Throughput
Gaining insight into the combination of factors that assisted a manufacturer to achieve optimized productivity is the surest way to recreate optimized processes. Effectively improving throughput starts with capturing supply chain data, inventory data, machine utilization data, and attaching optimal working processes to these statistics. The optimized data become benchmark data which can then be recreated over and over again.
Fastenal, an original equipment manufacturer, leveraged big data sets to develop benchmark data for its operational processes. Utilizing the analyzed benchmarks, the OEM was able to save approximately 100 hours wasted on unnecessary operations each month. Taking advantage of benchmark data enabled the OEM to improve its productivity and ability to efficiently meet its demand requirements.
5. Implementing Industry 4.0 Business Models
The goal of the 4th industrial revolution is the smart factory where interexchange of data is possible and analysis happens in real-time to ensure assets can make accurate decisions without human intervention. Capturing big data sets from the factory floor to discover patterns that simplify the decision-making process for machines is required to achieve Industry 4.0. Leveraging big data and machine learning, equipment within the factory floor are provided with the historical context needed to take specific actions.
Conclusion
Realizing the smart factory of the future requires improved data capturing capabilities and leveraging tools such as simulation modeling, the digital twin, and forecasting technologies to gain insight. The use cases highlighted here are just a subset of some of the ways data can be utilized to improve manufacturing operations. Manufacturers are expected to continuously push the application boundaries by developing more innovative ways to use big data.