Alan Bryant (Occidental) provided a sneak preview to the ISA112 Scada standard. Work on ISA112 began in in 2016 on reference architectures, a common terminology and lifecycle guidance for Scada systems. ISA112 is also to clarify how other ISA standards for cyber security, alarm management and safety relate to Scada. A first draft of the new standard is now out for comments with publication set for 2022. More information from the ISA.
Arlen Nipper (Cirrus Link) described MQTT and SparkPlug as central to the digital transformation of the oilfield. Cirrus Link has added to the Sparkplug B interface to MQTT to create a ‘complete’ industrial internet environment. The company has worked with some 35 manufacturers to implement Sparkplug B natively on their devices. Cirrus Link’s Chariot can be deployed as standalone MQTT server or data can be ‘injected’ to Azure, Google, AWS, IBM and others (Rockwell, ABB) from nearly 100 Scada electronic flow meter protocols. The ‘simple and open’ MQTT spec ‘has become the dominant IoT/IIoT messaging transport’ thanks to its low bandwidth requirements, secure comms and access control. Sparkplug B adds an operations technology-centric tag namespace and payload definition that supports auto-discovery and a ‘single source of truth’ at the edge.
Brandon Davis, a regular at the Wellsite Automation event since 2016, presented his evaluation of Scada systems for Red Bluff Resources. The company previously had multiple 3rd party systems that would not talk to each other, had multiple interfaces for operators and lacked data consistency. Similar issues were found in the automation package and artificial lift systems. Davis has evaluated message-based systems and found that these technologically attractive solutions (especially to OEMs) currently suffer from poor support from device vendors, although this is changing. Another consideration is data storage – either to a real-time data historian or cloud-based. The historian offers strong ad-hoc visualization and analytics tools with both MQTT and OPC interfaces. The cloud offers more IOT functionality but currently is not designed for oil and gas and requires a significant development effort.
Red Bluff’s current solution now embeds Signal Fire’s Pressure Scout wireless telemetry solution, Redlion Graphite HMI and Crimson software that offers protocol conversion to MQTT and cloud storage to AWS and/or Azure. The Apache ActiveMQ MQTT broker also ran.
Evan Rynearson (Middle Fork Energy Partners) showed how MQTT has been deployed to decrease communications bandwidth from the wellsite and increase efficiency. Rynearson recently challenged the LinkedIn community to ‘give me some examples of real world dollar value from a digital transformation’. The post generated considerable interest*. Scada and automation folks have been building infrastructure for their entire careers. But MQTT ‘takes it up a level’ and drives huge efficiencies. Problem is that ‘Scada folks hate change’. The appropriate strategy for MQTT deployment is therefore to keep and utilize what has already been built. Scada today is complex, intertwined and interdependent. But big benefits are achievable with little investment. The secret is templating common solutions, an approach that is already well understood by construction, maintenance and other fields. Templates leverage a unified namespace and data sources now supply unambiguous tag names. As field devices still run their native formats, these need to be converted to MQTT, adding the template-derived metadata. Middle Fork uses the low cost Ignition Edge Scada server with and OPC DA Module. The whole Scada system is now published into the MQTT infrastructure and namespace. Scada data is now available to all parts of the business. Rynearson advises starting with alarms than are pushed to operators in real time. Looking further up the value chain, ‘predictive models are impossible to scale without a unified namespace and known data intervals’. Rynearson acknowledged Sigit’s help with the namespace.
* Although not too many ‘real world examples’.
Hoss Belyadi showed how Vine Oil & Gas has leveraged machine learning to develop a ‘dynamic completions optimization workflow’ for its Haynesville Shale operations. Data mining and machine learning techniques are essential to extract information and knowledge from very large raw data sets. Working with a dataset of 222 Haynesville wells Belyadi studied the Pearson correlation coefficient to detect and remove collinearity in the data. A variety of ML approaches (feature selection and ranking, grid search…) were trialled. Feature ranking with Random Forest was deemed to be a ‘very powerful’ supervised ML algorithm. Combining many decision trees into a single model is the fundamental concept behind using random forest. Support Vector Machines and neural nets were also ‘powerful’ as was a ‘one variable at a time’ sensitivity analysis of the SVM output. The results show that a reduced cluster spacing showed, inter alia, that water/proppant loading and cluster spacing produce the most significant impact on production performance across operated and non-operated wells.
More from the conference website.
* For more on workflows for building ML models read Belyadi’s book ‘Hydraulic Fracturing in Unconventional Reservoirs. ISBN: 9780128176658, Elsevier, June 2019.
© Oil IT Journal - all rights reserved.