Modern automation networks allow manufacturers to collect a vast amount of data about all aspects of the manufacturing process. The challenge now is: What do you do with all this data?
Big Data is not merely a trend in the IT industry: its use is progressing at high speed in the field of general consumer marketing. Will the use of Big Data spread to the manufacturing Industial Ethernet Book, searches for challenges and vision in the expanded use of Big Data.
A road opens for data collection in the manufacturing industry with Big Data
With Big Data having been a buzzword in the IT industry for several years, today more and more industrial automation companies start to talk about it, too. The enabling technology behind it is the rise of Industrial Ethernet as the common communication protocol in industrial automation. Before that it was difficult to access data from manufacturing environments. Until way into the second half of the 20th century, collecting data from production processes meant handwritten records of meter readings that were stored in filing cabinets.
The rise of industrial Fieldbuses improved the situation, as they enabled data collection in a more timely manner. However, this data came in a huge variety of different formats and was typically used only on a machine or control level. Industrial Ethernet finally provided the bandwidth, speed, and common data structure to exchange data all the way from the sensor and actuator level to the enterprise level. Information taken in every second by temperature, pressure, power, and other sensors yields several terabytes of information per week. However, in many cases this data is not used very efficiently. Dr. Olaf Sauer of the Fraunhofer Institute for Optronics*, System Technologies, and Image Exploitation IOSB estimates that “today’s operators use only about seven percent of this data for maintenance or protection from breakdowns.”
*Fraunhofer IOSB (Fraunhofer Institute of Optronics, System Technologies and Image Exploitation): The largest applied research institute developing applications in Europe.
What is Big Data?
The term is commonly used for a collection of data sets that are too large and complex to process with traditional relational database management systems and desktop statistics and visualization applications. What is considered “too big” of course depends on the capabilities of an organization, on what size of data sets they have typically been working with. For example, scientists working in the fields of meteorology, genomics, or complex physics simulations are used to processing data in the terabyte range and over the years have developed the necessary computer platforms and software applications.
But big data is not only about sheer size. Doug Laney, a Research VP for Gartner Research, defined the challenge as being three-dimensional, and introduced the 3V model, with increasing volume (amount of data), velocity (speed of data in and out), and variety (range of data types and sources). This 3V model is today commonly used in the industry to describe Big Data challenges and opportunities. The situation that all of these three Vs are sharply expanding can be said to be Big Data itself. Meanwhile, some have added a fourth dimension, veracity and opportunities, the truthfulness or reliability of the data collected.
This 3V model is frequently and widely used today to explain Big Data. In the manufacturing industry, in particular, the velocity (frequency of data occurrence) should be noted.
In the field of general consumer marketing, Big Data is being used to offer services that respond to consumer needs and for product planning. For example, cameras are installed on vending machines to collect data on purchasers, and the date and time purchases are made, and sensors are used to collect environmental information such as weather and temperature. Such data, together with Big Data, will be used for processing such information to be used for marketing. In this case, several hours to several days would be sufficient for data collection and analysis (frequency of data occurrence).
Meanwhile, the manufacturing industry controls production facilities and machinery in production lines in "real time." Therefore, Big Data (Figure 2) consisting of further automated production lines' log data, various information from sensors, and environmental information within and outside of the plant needs to be collected and analyzed at the frequency of several to several tens of seconds, similar to "real time." This type of data is not expected to expand as sharply as Big Data used in the field of general consumer marketing, yet this high velocity (frequency of data occurrence) is feared to result in an explosive increase of data with an addition of several sensors. I would like to point out that this is an important issue in understanding Big Data to be used in the manufacturing industry.
Challenges for the use of Big Data in the manufacturing industry
Many tools and applications have already been developed for the use of Big Data in the field of general consumer marketing.
However, some industry experts warn that the tools and software used in the consumer market cannot simply be adapted to industrial Big Data requirements. While it is common practice for Google, eBay, or Yahoo to run Big Data analytics on a cluster of independent servers, most industrial automation system architectures are configured around a centralized database. A GE Intelligent Platforms white paper entitled The Rise of Industrial Big Data points out that the complexity involved and the specialized skillsets needed to create a Hadoop environment are often beyond the capabilities of industrial businesses.
The difficulties in data analysis for manufacturing industry are to create data in the form of analytics that are derived over longer term comparisons from batch to batch, month to month, shift to shift, or order to order. Analyzing these trends can give a deeper insight into cause and effect and help to optimize industrial processes. For example, in plastics manufacturing it may be important to understand the significance of annual temperature variation on quality. A production line supervisor could analyze years of past data for anomalies and variations, and to see whether they were followed by subsequent outages to thereby enable predictive maintenance.
In the part 2 of this article, an attempt will be made to examine the merit to be gained from the use of Big Data by the manufacturing industry and issues that may arise within companies because of its use. Will use of Big Data move forward in the manufacturing industry in spite of the cost for introduction and its complexity?