Smart, connected real-time data that is synchronized throughout the plant and global enterprise is the future for process automation, the digital transformation, and smart manufacturing. Whether working with advanced analytics, machine learning, robotics, or chatbots, having connected industrial time stamped process data enables companies to make better data based decisions that can optimize production. Manufacturers need to integrate, manage, and obtain intelligence from large volumes of disparate data from sensors, machines, cameras, energy, financial, weather, and other data sources, to make more intelligent decisions faster.
The first case study presented was by an onshore oil & gas producer who is applying advanced analytics to maximize oil and gas production by detecting abnormal operating conditions in gas lift compression. The company uses a SCADA system to manage 29000 devices logging 52,000,000 records a day.
For the analytics, the objectives were to provide tools for everyone, serve analytics to untapped customers, and provide a first line of insight into the production process. The analytics platform they applied was Seeq Workbench. The tool includes pattern search, prediction/regression and big data collection. A journal captures the process of building a model to be able to show others how the model was built and the rationale behind it.
The specific use case described was gas lift compression. There were 120 gas lift compressors, which each unit cable of moving 4-6MMCF per day. Compression is key for the artificial gas lift to stimulate production in this producing area.
The challenge they were trying to solve was to predict abnormal compressor operating conditions. The analytics help to correlate the signal data and define conditions of interest to identify abnormal events. From headquarters, they were able to identify these conditions and notify to field operations staff in advance of a problem.
In one instance, they identified a bad valve in the head of the compressor which early identification saved the failure of a cylinder, piston and turbo. This not only saved the cost of replacement of these parts, but also the avoidance of lost production.
They are currently creating dashboard views of the compressors on a report-by-exception basis as a predictive tool for the field maintenance personnel.
Results to date include expedited ad hoc investigations and enables subject matter experts to self-serve their analysis efforts.
The second case study was by an electrical power producer who is using a Chatbot for operations and maintenance on wind turbines to help substation technicians understand and analyze historical operations data. One application described was an augmented reality with GPS positioning to help maintenance personnel located assets such as 50KVA transformers. The system was integrated with their Maximo database of production assets. This application helped technicians be more efficient in finding and fixing issues on the grid.
They performed an artificial intelligence pilot for a substation transformer maintenance application. Data was collected from the control system historians, maintenance data bases and other sources. A chat bot interface allows technicians to voice query the system for facts such as alarm history, procedural steps to troubleshoot, help with specific abnormalities, etc.
The pilot proved that voice interaction with the AI engine is possible and the quality is improving at a fast rate. The AI engine can perform the data analytics on a specific set of data. Finally, even when questions are asked in different ways, the AI engine can give the same answer. The goal is to be able to provide Amazon Alexa-like functionality for the technicians.
The next phase of the project is to include all the operating manuals and specifications into the system for the AI engine to digest and have available to assist the technicians.