To collect and store data and monitor systems, most manufacturers are using technologies at least 30 years old. With the competition in today's global market, most industrial companies hesitate to take advantage of new opportunities promised by the Industrial Internet of Things (IIoT), with concerns related to difficulty and cost. A recent LNS Research survey of more than 400 manufacturing executives showed the vast majority of companies do not have plans to invest in IIoT technology in the near future.
It is understandable why industrial companies are reluctant to invest in new technology when considering the expense of existing systems.
Old technologies that have been tweaked to try to take advantage of IIoT opportunities may require removing current systems. Affordable technologies developed for the Internet Age can work with existing systems to help manufacturers gain deep insight into process behavior that translates into fast return on investment (ROI).
Valuable SCADA information
Supervisory control and data acquisition (SCADA) systems were originally designed to collect data and monitor processes. Since SCADA systems generate enormous amounts of data, historians were added to store this data. Initially, historians were used to fulfill regulatory requirements, such as generating reports for government agencies. Leading industrial companies recognized the data hidden inside historians could provide valuable information on plant processes and production, but accessing and using data could be difficult. Historians weren't designed for "read" purposes or a two-way information transfer.
Manufacturing execution systems (MES) were introduced in the early 1990s in an attempt to bridge the gap between plant-floor SCADA systems and enterprise ERP software. They also promised to provide analytics, such as key performance indicator (KPI) data, to improve plant-floor operations. MES can provide more advanced capabilities than SCADA systems, but are expensive and often require extensive engineering for implementation. MES were developed for a business era in which systems were still largely siloed, and Internet optimization was largely an afterthought.
Avoid a locked-in upgrade cycle
With the amount of time and money industrial companies have spent for traditional software, we can understand the reluctance of some manufacturers to enhance existing systems. They fear that a new solution would be expensive, require extensive engineering and training for employees, and lock the company into a cycle of difficult and expensive upgrades, patches, and limited scalability. Next generation software offers ease of use and affordability.
Search engine for industry
As mentioned, accessing historian data and turning it into actionable information to improve operations has been time consuming and difficult. Data modeling applications required extensive engineering and data scientists to perform. As a result, only mission critical applications were targeted, leaving vast areas of improvement opportunities hidden.
In 2008, engineers from Covestro (then known as Bayer MaterialScience) leveraged time-series data by examining different analytics models and identifying limitations for scaling-up beyond pilot projects. Using deep knowledge of process operations the engineers created "pattern search-based discovery and predictive-style process analytics" for the average user. Unique multi-dimensional search capabilities of this platform enable users to find precise information quickly and easily, without expensive modeling projects and data scientists.
A simple example of how this works is the song title recognition application Shazam (by Shazam Entertainment Ltd.). While the technology Shazam uses is different, the concept is similar. Instead of trying to map every note in a song to its vast database of songs, Shazam uses pattern recognition software that seeks "high energy content" or the most unique features of a song then matches it to similar patterns in its database. This is a simple explanation of a complex process, but the point is that it enables users to quickly find a song title with a high rate of accuracy.
Industry demands more sophisticated algorithms beyond search software by connecting to existing historian databases then implementing a column store database layer for an index. This software makes it easy to find, filter, overlay, and compare interesting time periods to search through batches or continuous processes.
Moreover, this next generation solution enables users to search for particular operating regimes, process drifts, operator actions, process instabilities or oscillations. By combining these advanced search patterns users unlock information they need. For example, an operator compares multiple data layers or time periods to discover which sensors are more or less deviating from the baseline then make adjustments to improve production efficiency.
Contextualization, prediction
In addition to easier search, attention to process data contextualization and predictive analytics capabilities is needed. Engineers and operators can provide annotation to provide greater insight.
Predictive analytics capabilities enable an early warning detection of abnormal and undesirable process events by comparing saved historical patterns with live process data. Calculating possible trajectories of the process can predict process variables and behavior before it happens. This gives operators the ability to see if recent process changes match the expected process behavior and proactively adjust settings when it does not.