Rick Gearheart presented Unico’s autonomous expert solution to oilfield production. One trial, dubbed the ‘data of Damocles’ saw a 3 terabyte dataset handed-over to a confident AI team. An additional offer to provide domain knowledge assistance was refused, ‘just watch the magic!’. Two months the AI experts concluded that the ‘fault reset event’ was the best ‘predictor’ of failure. What went wrong? The AI team was given too much scope. AI/IOT is a toolset not a project and requires expert input. Unico’s first foray into IoT in 2011 with GMC System pulling data Unico VFDs and displaying on a mobile/web endpoint. The first GUI was too complicated and the system was redesigned as PumpExpert. Now real time data is processes to provide simple message like ‘normal operation’, ‘pumped off’, ‘gas locked’, ‘sticking’, highlighting abnormal situations so they cannot be overlooked. Pump Expert also constantly adjusts pump params autonomously, responding to changing well conditions on a stroke by stoke basis.
Blake Burnette described IOT-eq’s use of the IoT to drive machine learning at the well site. Frack truck IoT systems have evolved from local, on site sensors and processing in 1980s, through satellite connectivity (2000). Around 2010, big data Map/Reduce algorithms were producing up to 2,000 alarms per day! Burnette observed that some findings were more useful than others. An instruction to ‘wash radiators’ could generate million dollar savings. The IoT needs attention to detail, ‘talk to people’, you need ‘boots on the ground’ to find and fix sensors and comms issues. IoT-eq has a technology partnership with Weir on a pump health monitor. The Weir/IQ Data Intelligence dashboard runs either in the cloud or at the edge depending on communications costs. Burnette referenced another IoT gateway, IOT-Glue, as offering a quick data path from data to app. IOT-Glue offers drag and drop configuration and has been used to auto fill RRC reporting e-forms and for tank monitoring and alerting for theft. The IoT can be used to monitor and understand machine learning results and avoid situation such as when ML-based control tuning is ‘learning bad habits’. Finally a warning to the DIY brigade, ‘avoid the pitfall of using consumer products in harsh environments’. More from IOT-eq.
Jim McKenney (NCC Group) stated that ‘despite the promise of IoT and predictive maintenance, these systems have failed to meet their objectives so far because of system complexity and cascading failures’. For predictive maintenance efforts to succeed, field devices, sensors, data models and the supply chain need to work at 99%+. This may not be realizable with regards to cost. IT risk is also an issue. McKenny recommends using tools like Microsoft Threat Modelling, the STRIDE/DREAD approach and ‘many more’ standards and approaches to address IT risk and engineering reliability. IoT tech can reduce wellsite opcosts through predictive maintenance, but a software lifespan of decades exposes systems to future bugs that are beyond the awareness of many IoT providers. Tracking and tracing faults across a complex landscape of software and engineering hardware is difficult. Reliability, availability and maintainability engineering principles need to be applied to IoT devices and pushed upstream. Software should be thought of as a ‘material’ that can be fixed with upgrades and devices ‘front loaded’ with unnecessary features may expose future risk.
Paul Brager (Baker Hughes) also spoke on the growing threat landscape of IT/OT convergence. Future analytics requirements will grow the attack surface even more. Legacy infrastructure is especially at risk. How much convergence is a hotly debated topic. There is also the question of which, IT or OT, is more trusted! The answer is ‘smart’ convergence that assesses supply chain attacks and risks. Operators need a cyber assessment of sourcing to avoid cascading disruption. A holistic view required, as is an intelligent separation of IT/OT, along with a response plan for the eventuality of an attack.
Bob Skiebe presented a range of automation options from Siemens Oil & Gas from field to enterprise IT/OT integration offering IoT based next gen well automation with artificial lift optimization to well AI at the edge to detect anomalies in ESP operations. At the high end, Siemens can supply an ‘integrated digital twin from subsurface to facilities’. For brownfield applications, simple video surveillance of existing gauges from a stand-alone device at the well site with a once-a-year battery change offers a ‘low cost digital transformation’.
Murat Ocalan (Rheidiant) spoke in praise of LPWAN - LoRa and LoRaWAN with an AES-secured payload along with the deployment of deep learning-based algorithms (PyTorch, Keras, TensorFlow) on edge devices (ARM MBED, RTOS, C++). Security is provide with ‘crypto-signed’ firmware.
Paul Solano presented MAC Engineering’s state of the art well site automation, based on IoT and edge computing. MQTT is the base protocol along with emerging integration standards like Sparkplug. Data from existing scada systems is translated from Modbus, OPC and other ‘legacy’ protocols. See for instance the Modbus/MQTT comparison from Novus Automation. This enables a ‘transitional’ architecture in a shift to Scada 2.0 with the incremental addition of a big data lake, enterprise asset management and more. The Scada 2.0 reference architecture includes microservices, containerization, ‘open’ APIs and native big data integration.
* Note SCADA 2.0 – is a rather loose term used by many (Siemens, InFusion, ...). See for instance the 2010 EON/OSIsoft presentation.
Louis Lambert (Redline Communications) recommends using your installed scada that is already ‘built for safety and compliance’. ‘Do not disconnect it!’ Meanwhile operators need to become familiar with private LTE cellular comms. When ready, this can provide data access from cell phones and tablets and enable video and analytics. Likewise, old two-way radios can be modernized with private push-to-talk. LTE provides a ‘path to 5G’ with a self-organizing network and dynamic autonomous configuration of new kit with the 3GPP mobile broadband standard.
Claude Baudoin from the Industrial internet consortium stated, ‘as everyone knows’ that one should ‘just follow the standards’. But there are too many in the digital oilfield, around ten at the data protocol level alone. The Industrial Internet Consortium’s high level IIoT roadmap (possibly the Industrial IoT Analytics Framework) lets users develop business cases and architectures. But the IIoT needs more IT/OT convergence and needs to ‘get out of the proof of concept jail’. The IIC was founded in 2014 by AT&T, IBM, Intel, GE, Cisco. In 2019 the IIC absorbed the Open FOG consortium. The IIC advocates standards based on the OMG’s DDS and OPC-UA gateway. Some eight vendors implement the protocol which has been used in projects by Ensign Energy, NOV, Canrig Robotics and TechnipFMC. Checkout the IIC’s Smart Factory/ML presentation.
At a more practical level Noel McKim presented Meyer’s Gen3 ‘Spyder’ multi-station greasing manifold that connects multiple wells to a central grease controller and lubrication flow meter. The system keeps personnel out of the ‘red zone ‘with secure access lockout and is said to reduce costs and improve safety.
The IQ-Hub Canadian Well Site Automation 2020 conference will be held on May 11-12, 2020.
© Oil IT Journal - all rights reserved.