A path to using Big Data in the manufacturing industry has opened up because industrial Ethernet has become popular as a communication protocol for FA (Factory Automation). This serialization uses articles contributed by manufacturing industry experts in Europe to explore the actual conditions of using Big Data in the industry. The first part of this serialization focused on the unique characteristics and challenges of using Big Data of the manufacturing industry. This second part pays attention to benefits from and successful methods for using Big Data.
What are the benefits for manufacturing?
The manufacturing industry has an excellent track record of putting knowledge resources to good use. Inventions like steam power, assembly lines, automated production and global supply chains have made the production of goods ever more efficient. Today it is a highly evolved industry, making it difficult to find further productivity gains. Leveraging Big Data analytics for manufacturing operations may be the technology that enables another quantum leap in efficiency.
The types and characteristics of Big Data in the manufacturing industry were described in the first part of this article series. A Big Data study by Tata Consultancy Services details the concrete benefits for the manufacturing industry. These include improvements in product quality, supply planning, and the tracking of process or component defects. Beyond these obvious opportunities, the study also sees the potential to improve contract negotiations based on supplier performance data, to reduce cost through increased energy efficiency, and to simulate and test new manufacturing processes.
Production equipment often operates for a long time and its failure has a huge risk of causing changes in production plans and other troubles. Utilizing the operating situation, operating history, and other data of the equipment will optimize maintenance work. Linking these types of data with product sales and other market data will significantly contribute to the planning of product supply and other purposes. It is important to fully understand benefits from using these Big Data to utilize them for actual business (Figure 1).
In fact, Schwering & Hasse Elektrodraht GmbH, a German wire manufacturer, has established a quality control system using a high-speed big data analysis system. The conventional system was able to detect wire voltage abnormality only in 100 m increments in the production process. The established quality control system obtains 4000 times more data than the conventional system so that it can detect abnormality in 25 mm increments. This significantly improves the product quality of this German manufacturer. In addition, BAXIROCA, a Spanish heater manufacturer, has constructed a system for collecting and analyzing various types of data of from the product design and manufacturing through to the after-sales service. The system analyzes whether causes of failures and problems in devices are design flaws, defects during installation, or insufficient maintenance. BAXIROCA uses analyzed causes to improve its products, enhancing its product value and customer service and reducing failures within the warranty period to cut repair costs.
The challenges of Big Data implementation
The Tata study also asked about the greatest challenges to getting value from Big Data. Not surprisingly, the list is topped by organizational issues. That a successful implementation requires a high level of trust between the data scientists and the functional managers, that business units need to share information across organizational silos, and that managers need to learn to take decisions based on data analytics, rather than on intuition.
At an INSEAD panel discussion Neil Soderlund, Partner and Managing Director at the Boston Consulting Group talked about the gap between leading-edge operations like Google and legacy companies starting to implement Big Data: “As much as they might be doing big data, they are abysmally bad at using it for value generation. The culture of organizations and perception of leaders is that a black box cannot yield a better answer than they can draw out from their years of experience. This is going to be a big differentiator, one of the things that separate companies to a far greater degree in years to come.”
Even before any of these challenges can be addressed, top management needs to be convinced to approve the necessary investments in software, infrastructure and qualified personnel. These investments quickly reach a substantial range. Bogdan Nedelcu of the University of Economic Studies in Bucharest says that a 2012 survey showed that the average survey respondent spending on Big Data was US$ 88 Mio. and that 7% of the companies invested at least US$ 500 Mio. each. And the companies in the survey expected to invest even more on Big Data, on average 75% more by 2015. Nedelcu argues that these investments are necessary to continue achieving high levels of productivity growth.
Is it worth the investment?
However, other experts warn that for some companies Big Data might simply be ‘A Bridge Too Far’. Roy Kok, VP Sales and Marketing at Ocean Data Systems: “While a valuable goal, the expense and complexity of implementations that have a heavy focus on Big Data (or perhaps a better term is simply Advanced Analytics) may be more of a challenge to maintain and use in day to day operations and the return on your investment may not be there.”
With all the hype around Big Data this is hardly ever mentioned, but there is a real risk that companies invest in data networks, software platforms, expertise and skills - and end up none the wiser.
A study by McKinsey Global Institute points out that currently “there is no empirical evidence of a link between data intensity or capital deepening in data investments and productivity in specific sectors.” Drawing from the experience of previous waves of IT-related productivity growth, McKinsey identifies three shared characteristics of successful adoption: First, IT investment tailored to sector-specific business processes and linked to key performance levers. Second, sequentially deployed IT, building capabilities over time. Third, IT investment evolving simultaneously with managerial and technical innovation. The road to successful Big Data implementation will be neither short nor smooth.
Successful methods for using Big Data
But there are steps that can be taken to shorten the path to successful Big Data analytics. Roy Kok believes that the most important item missing from most automation environments is information distribution: “Today, information is delivered primarily in the form of portals that provide access to HMI/SCADA, historians and analysis tools. The problem with this approach is that it requires the users to be proactive in the use of those systems.” He believes that a solution is needed that delivers information to users on a regular or triggered basis. This drives the user to be reactive to new information. Delivering information regularly provides a base of knowledge for understanding the usual, so that the user will be able to recognize the unusual. By delivering information and its context on a triggered basis, this approach also allows the user to take immediate action.
In an article for InformationWeek, Vishnu Bhat, Vice President and Head of Cloud Services at Infosys Limited, outlines the requirements for a Big Data platform for the manufacturing industry.
- It should include interfaces for a variety of data formats to allow enterprises to immediately access and extract filtered data from existing internal and external data sources. Furthermore, it should provide pre-built algorithms and reporting options to arrive at insights in near real-time. Ideally the platform will also provide a test mode so that companies can begin evaluating their hypotheses and analyses before they integrate the actual data sets.
- Data analysis is required to appropriately process an increasing large amount of data in real-time. In the market, there are analysis methods, tools, and services using algorithms that allow deep insights in real-time; however, since the analytical ability is a large factor of competitiveness, developing data scientists who take charge of high-level intelligence analysis is required.
- In order to prepare for the implementation of Big Data or advanced analytics, Roy Kok advises to keep the focus on the visibility of information and the broad distribution of information. “All too often the responsibility for analysis falls on just a few and outside of those, operations fall into the status quo,” he has observed. This tends to happen when information is centralized into a tool that requires proactive use. By broadening the distribution of information, and sharing information on a regular basis, the combined knowledge base of an organization can be leveraged.
It is necessary for organizations to work together on improvement by visualizing and sharing information from the device level (sensors and devices) through to the device control, MES (Manufacturing Execution System), and ERP (Enterprise Resource Planning) levels (Figure 2).
Many difficulties are in store for using Big Data; however, if expanding the amount of information and information sharing succeed and the integrated information can be utilized in organizations, we will see the investment in Big Data start to result in increase in productivity and other benefits. We will continue to pay attention to the use of Big Data.