Remote - Winter 2013 - (Page 8)

Feature Article Enhanced SCADA Access and Big Data Lead to New Analytics & Optimization Capabilities Mike Brulé, Big Data Solutions, IBM The previous article discussed how lightweight, event-driven data access and real-time message-oriented middleware (MOM) brokers enable access and acquisition of more of the telemetry data in remote field locations. Message Queuing Telemetry Transport (MQTT) and MOM also provide the means to transfer data to control loops and to other systems. An additional set of new information technologies called "Big Data and Analytics" has also emerged. In this article, we will explore how instrumentation and SCADA data are integrated with other data sources to analyze more data across more subject areas than ever before. These technologies help engineers and operations personnel enhance workflows and improve operations. Through integration and collaboration with the larger company, they provide even broader business opportunities. Big Data and Analytics Big Data technologies originally come from the companies with webcentric businesses, e.g., Google, Yahoo and Facebook. Is Big Data useful to the process industries? Yes, but not in the same ways as for Web companies. One popular way to portray Big Data and Analytics is in terms of the four V's of data: volume, velocity, variety and veracity. The IT infrastructure for managing and analyzing the Big Data "V's" can be provided by massively parallel processing (MPP) data warehousing for structured relational data (e.g., financial data), streaming real-time analytical processing for highvelocity (e.g., time-series) data, and Hadoop Map/Reduce for unstructured data of volume and variety (e.g., engineering and geoscience). In upstream exploration and production (E&P), midstream pipelines and transportation, and downstream refining and petrochemicals, the emphasis on data-in-motion is unique, owing to the many sensors that continually stream real-time data from the field, e.g., SCADA data.1,2 The oil and gas industry has developed over many decades powerful physics-based models that can predict the behavior of complex oilfield processes. Such modeling and simulation based on first principles will be the mainstay of predictive methods for years to come. However, opportunity exists to augment the established physical modeling methods with new data-driven analytics, including statistical analytics, text analytics, AI and machine-learning. These additional methods of Big Data and predictive analytics are especially useful in achieving the goals of Integrated Operations.4,9 OT + IT The E&P industry version of the "Internet of Things" relates to the many sensors, control and automation systems in oilfield operations. Today's isolated IT and OT (information technology and operations technology) systems do not connect oilfield operations with corporate business well (see Figure 1). The transfer of data from the oilfield is limited by two key factors that leave much valuable data stranded. The first is low bandwidth (discussed in the first article), and the second is a lack of flexible, real-time management systems for Big Data. To achieve a wider range of businessimprovement opportunities through analysis and optimization, field OT and corporate IT systems must be integrated and the resulting analytics and optimization processes converged. Figure 1 - Silo'd OT and IT systems provide little support for integrated operations 8 www.RemoteMagazine.com Supporting safe, efficient and profitable operations demands accurate, coherent, and timely information from all digital technologies, including improved access and transfer of real-time and fine-grained OT and IT data, which implies that all, not just part, of the data generated in the field OT must be sharable with corporate IT and vice versa. When IT and OT systems are converged, a new level of deep and reactive analytics can be supported across many different subject areas and functions, e.g., supply chain, procurement, inventory and materials management, reliability and maintainability, turnaround scheduling and resource optimization. New Big Data & Analytics technologies come to play at the dotted line between OT and IT, and in addition to MQTT and MOM technologies, provide the means for all data to be shared between field operations and the corporate offices. Stream Computing and Real-Time Data-Flow Architecture The oil and gas industry, with its many instruments and control systems, is often associated with real-time manufacturing operations. The flow of telemetry data parallel the flows of the hydrocarbon streams being produced, processed and transported. To realize all the goals of Integrated Operations in real time, a very-low-latency architecture is required. Stream computing, MPP data warehousing5, and Hadoop Map/Reduce support real-time analytics far better than traditional transactional relational databases, so that the resulting system is more responsive, resilient and flexible for integrated operations. Big Data and Analytics are most effective when data-in-motion are combined with data-at-rest in a real-time data-flow architecture with adaptive analytics.1 Stream computing is a new way of Figure 2 - Stream computing applies to high-volume and high-velocity analyzing highdata, whether structured, semistructured or unstructured frequency data for real-time complexevent-processing (CEP) and for scoring data against a physics-based or empirical model for predictive analytics, without having to store the data (Figure 3). It's also well-suited to improving the data quality of noisy and sparse data. Hadoop Map/Reduce and other NoSQL approaches are a new way of analyzing massive volumes of engineering and other petrotechnical data, whether semistructured or unstructured. This new Big Data infrastructure can be used to support the E&P industry's many physics-based methods in modeling and simulation, over a wide range of disciplines from geology and geophysics to reservoir, production and facilities engineering. Highvolume data of many different types can be landed on a real-time message bus, routed through a streaming real-time analytical processing engine, to a Big Data repository (Hadoop), and analyzed to provide many insights for advanced, real/right-time decision-making. CEP capabilities are valuable in oilfield-event exception management, to issue alarms and alert operators so that they can avert a problem before it happens. Stream computing is not just a rules-based CEP method. However, stream computing goes beyond CEP by providing a full modeling and simulation development and runtime environment that can be used to more finely control and optimize a process, taking the industry a major step closer to automation. Stream computing can model a complex physics-based process, for example, to set chokes, in gas or steam-injection processes, which are outside the capabilities of today's CEP engines. The real-time data-flow architecture in Figure 4 tips the balance more toward lower latency by supporting modeling and analysis while "data is in motion." The dated, but well-known n-tier abstraction of user interface, busi- http://www.RemoteMagazine.com

Table of Contents for the Digital Edition of Remote - Winter 2013

Editor’s Choice
Message Oriented Middleware - The Future of SCADA
Enhanced SCADA Access and Big Data Lead to New Analytics & Optimization Capabilities
Approaches to Powering Telecom Sites
Satellite Communications for Water Metering and Other Water Applications
Remote Magazine Launches Internet of Things North America
Geospatially Integrated Surveillance Systems
Tier 1 Operator Case Study: Intelligent Site Management
SCADA - The Brain of the Smart Grid
SCADA
Networking
Security
Onsite Power
Industry News

Remote - Winter 2013