A Digital Plant Strategy to enable Dynamic Performance Monitoring

Version 1

    Accelerating Insights Discovery From Real-time Data To Drive Decision Making.

    The Digital Plant vision for increasing value with a data infrastructure

    The alignment to resolve business processes is the key for continuous improvement, innovation and increased profitability.  This suggested practice describes a digital
    transformation strategy to achieve key business objectives through use of highly contextualized operations data.

    The process industries face many pressures, such as increase environmental and safety governmental regulations, increased energy, water and operating
    costs to process ever more complex natural resources.

    These issues can be addressed through the abundant and growing quantity of real time data and process operations events collected from
    an enterprise’s process plant(s).

    A strategy is needed to enable operations/maintenance/management to quickly discover and investigate insights to improve bottom line profitability, as well as to improve safety and
    environmental responsibilities.

    A strategy to simplify the digitization of a process plant for faster discovery and investigation of insight is proposed.  The strategy uses a modular approach
    integrating the capabilities of capturing real time data and events. 

    One process unit template with the corresponding event frame template (Asset, Mode and selected production, energy, water, other variables) is used to aggregate the data using the gross operating modes (event frames) generated by the variation of the production variance in a production environment.  The variance is defined as the difference between the production target and the actual process feed rate to a production unit.  The production targets are obtained from the Linear Program (LP) Model of the production plant.  The strategy aims at linking the production planning targets and the actual production rate based on the capabilities of the real plant. 

    C3 Digital Refinery Capture.JPG

    A gross data classification algorithm assists in real time identification of the operating modes by defining “Running ok”, “Trouble”, “Idle”, “Down” and “Maintenance”
    operating modes for all plant process units.
    On line analytics and event frame generation are embedded in a process unit template to create these results. PI Event Frame engine is used to aggregate the production and consumable variables. PI Event Frame (EF) software
    time-slices and retrieves process data by user-supplied start and end triggers, so that customers can view PI data in terms of their own manufacturing events, such Gross Operational Events, People Shift, Raw and Product Material Grades.

    Gross Operation Mode data Classification Capture.JPG

    A gross data classification algorithm assists in real time identification of the operating modes by defining “Running ok”, “Trouble”, “Idle”, “Down” and “Maintenance” operating modes for all plant process units.
    On line analytics and event frame generation are embedded in a process unit template to create these results. PI Event Frame engine is used to aggregate the production and consumable variables. PI Event Frame (EF) software
    time-slices and retrieves process data by user-supplied start and end triggers, so that customers can view PI data in terms of their own manufacturing events, such Gross Operational Events, People Shift, Raw and Product Material Grades.

    The large set of data obtained from the Process Units Event Template (Asset, Mode and selected production, energy, water, other variables) are then visualized and further analyzed by all people in a working plant. This is
    most valuable aspect of this practice.  Microsoft PowerBI enables to visualize, benchmark and analyze the overall production effectiveness of the process plant for the enterprise. All process units with the aggregate data by year, month, day and by shift (People) is summarized.   The capabilities of creating multiple charts and analyze can a self-service activity by the different functions in a plant.

    C3 Power BI Refinery Dashboard Capture.JPG

    C3 Event Frames to BI Analysis Tool Capture.JPG

    The Event Frame Template is used to aggregate the Production and Consumables variables for each unit at the desired level of details using the PI Time Derived value algorithms.

    A
    “follow the money” strategy for management through a constant flow of information. By calculating production levels, and tracking the use of resources (equipment, energy and raw materials), the company can identify
    losses, opportunities to address problems and ways to make processes more efficient in the future.  Notifications can be send immediately after the event trouble is detected for example to start a workflow for resolution. A workflow example is shown in the Appendix.  A typical Enterprise workflow is also shown.

    In today economics with real time data reaching to the corporate office real time data with EVENTS, these events are the new transactions of the past. Especially, now that we can easily related events
    with production and consumable (estimates of the operating costs) and real capabilities of the process equipment to deliver the promised targets defined by the planning department.  This is a digital transformation in
    practice.
      As such the deployment, training and maintenance of system require a new twist as well.

    The goal of reducing the production variance is a common problem in operating plants. The targets are compared with the actual production rate and the operating costs from the actual data captured in real
    time.

    The resulting classified information can be reused to improve planning, reduce downtime and idle times, identify process improvements and enhance collaboration between operations, maintenance at the plant and the
    enterprise.  The unified process unit template for all areas at the enterprise enables benchmarking, insights and profitability improvements.  The strategy is called Follow the Money by some customers.

    A continuous improvement, innovation and increased profitability is a never ending loop.
    This is change in operational improvements using the latest technologies.  Before, several attempts were made with difficulties due the poor integration and communication networks. Sharing the
    data on the cloud delivering the aggregated data via devices and the detailed data using the same cookie cutter is the digital innovation today.

    The pursuit of better results, however defined, does not end, the goals just get reset.  One of the keys is the self-serve visualization.
    The new tools enables the collaboration between the traditional functions groups in a processing plant and the enterprise.  Very similar to how we used to use PI ProcessBook within the industrial network as the industrial desktop.  Now, the network is in the cloud.

    The aggregation of production data, equipment state and consumable totals and averages by operating mode enables to identify the opportunities to focus and to prioritize actionable improvements.

    The derived benefits of having the different set of data available enables to use the historical data for developing predictive analytics models for finding the best manipulated target variables and to avoid
    constraints.  (Prescriptive analytics).

    For example, the running ok data set for a period of time can be used to find the best fit between a target feature and a set of descriptive feature. The exercise of finding the model augments the knowledge
    based on the process operation data and find opportunities for additional improvement.  This is called understanding the data before the machine learning algorithm are found.

    Another example, is to use the running dataset with the downtime data set to classify the data using logistic regression tools to avoid abnormal situations.  These technology is well documented.

    The data hierarchy presents the steps in using the operational data and events at different stages and users.  The classification stage is vital prior to finding a data set for training and later for using the trained model for
    estimation of the estimated target features.

    After the local target features are determined at the right operating mode the coordination with other units becomes feasible.

    C3 Flow the Money v2 0726 Capture.JPG

    Data Classification and Modeling Capture.JPG

    PI Data from PI Event Frame to ML Capture.JPG

    Operation and Business People Integration Capture.JPG

    The proposed trategy provides a ONE Template to configure the processing plant for Dynamic Performamce Monitoring.

    The digital transformation is journey toward operational and business transformation.  The task is never finished.

    Steps towards Capture.JPG

    Analytics towards collaboration and action.

    1. 1. - Start with Business Process initiative: In this case reduce production variance while reducing Energy, Water and Reagents

    Strategy: Classify Production Variance data in real-time based on 5 states (running Ok, troubles, idle, down and maintenance).

    1. 2. - Digitize Plant using one Unit Process Templates (analysis and event generation) (Simple configuration and easy to maintain.  Less than a month).
    2. 3 .- Visualize Production Events and Consumables aggregated data using one Production Event Template using PowerBI (Share with many people) and PI Coresight.
    3. 4. - Generate Notifications based in the events to communicate results.

    The pursuits of better results, however defined, does not end, the goals just get reset (continuous improvement and execution).

    This paragraph is based on our draft manuscript a Journey Towards a Digital Transformation in the Process Industries.  Available in PI Square. OSIPress work in Progress.