Processing, June 2018
Cover Series Process Automation IIoT Creat i n g c o n t e x t w i t h i n process data In process automation analytics effective use of data depends on creating context within the data to accelerate insights By Michael Risse Seeq Corporation very Industrial Internet of Things IIoT application starts with sensors which create the data needed for advanced analytics In process plants and facilities these sensors typically measure pressure flow temperature level and other parameters Additional process variables of interest are generated by analyzers power measurement devices motor drives and other automation system components Still more sources of data for analysis may include raw material pricing utility rates weather conditions shipping costs etc The sensor data is typically stored in a time series database usually a process historian Other relevant data may be found in relational databases but the primary source of operations production data in most process plants is a historian As its name implies a relational database such as the Microsoft SQL Server is designed for grouping data records so users can easily establish relationships for example relating a company to the orders it has open By contrast a time series database simply stores all data sequentially with time stamps with no effort made to establish relationships among the data points To address the lack of relationships among data points users create asset models and hierarchies to define the relationships showing how sensor data relates to assets lines plants etc but there is still an issue within sensor data streams which is a lack of context In other words context requires defining specific time ranges within sensor data and relating these time ranges to asset stages production modes products and so forth Because no context or relationships have been established among data points in a time series database the person performing the advanced analytics must execute this task This article shows how this can be done with results and benefits gained when using advanced analytics software to define context and accelerate insights But first it discusses why time series databases are favored in the process industries instead of relational databases which would have at least some context predefined E In other words context requires defining specific time ranges within sensor data and relating these time ranges to asset stages production modes products and so forth 16 Processing JUNE 2018 Figure 1 The volume of data generated in process and other manufacturing plants precludes the use of relational databases for storage Graphic courtesy of McKinsey Global Institute Figure 2 Creating and sharing insights into process data requires completion of these steps in a sequential fashion Graphic courtesy of Seeq 1 Discrete manufacturing constitutes 1072 petabytes Process manufacturing 740 petabytes SOURCE IDC McKinsey Global Institute analysis
You must have JavaScript enabled to view digital editions.