Remote Site & Equipment Management 2016 - (Page 12)

Feature Article Creating Value from IIoT Data Michael Risse, Vice President Seeq Access to local and remote assets and sensors has improved tremendously over the past two decades, and this trend is sure to accelerate as the cost of generating, collecting and storing data keeps dropping. Whether this trend is called the Industrial Internet of Things (IIoT) or something else, sensor deployments and wireless technologies (cellular, satellite, Wi-Fi, Zigbee, SigFox, WirelessHART, ISA 100) are on track to gather ever more data. This flood of data has increased the challenges to manufacturing organizations already "data rich but information poor" (Figure 1), who will need to accelerate their search for better approaches to enable timely insight and improve production outcomes. The challenge of too much data collected and too little insight is not a small question for companies, nor a new issue. According to the analyst firm McKinsey & Company (www.mckinsey.com), the upstream oil and gas industry alone represents a $50 billion opportunity for cost savings and revenue increases by implementing remote access and improved data analysis (Figure 2). Figure 1. DRIP - Spreadsheets and other tools are no match for the flood of date produced by modern local and remote monitoring systems. And that one industry segment is just a fraction of the value across all industry segments which will exceed $1.2 trillion a year by 2025 per another McKinsey report (McKinsey & Company: Internet of Things, Mapping Value beyond Hype). This goes along with dismal rates of production data used for analytics as documented by Gartner, which reports that 70 percent of factory data goes unused for improving manufacturing execution. But that 70 percent usage rate is far better than a McKinsey report which found just 1 percent of oil rig data was analyzed, and 99 percent of the data was lost before it reached operational decision makers (Figure 3). While these issues of opportunity available in existing data and waste from not using it are not new, what has changed is the modern generation of software offerings proposing to enable "actionable insight," the industry's marketing promise of the last decade. Two of these new offerings are machine learning and data analytics software, and in this article we will define and compare these approaches for enabling insight into process manufacturing data. Figure 2. In just the upstream oil and gas industry only, $50 billion in savings and increased revenue await those companies able to fully exploit the capabilities of existing remotely accessed data. Machine Learning Defined From a programming standpoint, machine learning is software using automated and iterative algorithms to learn patterns in data. Users of machine learning don't need to program the end-point solution at the outset because the algorithm adjusts itself from one Figure 3. Largely due to the difficulty of using spreadsheets and other cumbersome data analytic tools, almost all of the data acquired via remote access and other means is wasted. data point to the next to learn how to solve a particular problem. This is done using either a supervised (training set) or unsupervised starting point. 12 www.RemoteMagazine.com http://www.mckinsey.com http://www.RemoteMagazine.com

Table of Contents for the Digital Edition of Remote Site & Equipment Management 2016

Editor's Choice
Integrating Local HMI with the Cloud
Managing Distributed Energy Resources with IoT and Cloud Technologies
Three Considerations for M2M/IoT Connectivity
Creating Value from IIoT Data
Wireless Sensor Networks - Applications in Oil & Gas
Applying Remote Monitoring & Predictive Analytics to Satisfy Customers While Keeping Costs Down
M2M & IoT Products
Industry News

Remote Site & Equipment Management 2016