Step 3: Employing Offline Visualization Tools Using Unit Process Template Data

The business value of the data dramatically increases each time the data is reused for other business and operational use cases. We will now discuss how EIDI can supply contextualized, high-fidelity data for visual analysis, software modeling, and advanced analytics tools. We will also introduce the value of integrating non-EIDI analytics with real-time data. First, we examine how the team analyzes EIDI data through advanced visualization tools, such as Microsoft Power BI or Tableau. Our ability to extract EIDI data in a contextualized format allows for offline analysis. The data can be used for data cleansing or predictive analytics to apply machine learning (ML) using tools like Microsoft ML Studio or Python and R.

Advanced Visual Analytics

An effective real-time data infrastructure enables automatic, configurable import of event-framed operational data for the entire refinery. Open database connectivity (ODBC) or published data sets provide a basis for these analyses.

Figure 4.10 shows the Microsoft Power BI desktop application. Selecting the "Get Data" ribbon control (left circle) updates the EIDI event frame data. Using Microsoft Power BI and the Azure Cloud, the team can schedule an automatic update and publish the information. The Proclndustries team decides to use an EIDI ODBC query for a quick start.

Analytics tools allow multidimensional visualization of the extracted production and operations data (Ferrari and Russo 2016). The team can generate alternate views of the data displays, trends, and reports.


Microsoft Power BI desktop ingesting contextualized event-based data. (Courtesy of O.A. Bascur, OSIsoft LLC.)

Figure 4.11 shows a multidimensional data cube containing time interval data for the units, material type, operating modes, and aggregated data. Additional data for operating modes, shifts, crews, and raw material feed types allows for additional analyses. The results are cleansed before use in process modeling to highlight hidden losses by operating mode.

Tools such as Microsoft Power BI allow the user to publish dashboard reports using Azure Power BI (see Figure 4.10, "Publish" shown in the circle on the top right of the figure).

Figure 4.12 shows a Microsoft Power BI dashboard displaying water consumption for all process units for all operational states. Monica pointed out, "This lets us visualize the hidden losses of production, water, utilities, and any consumption variables chosen in the unit template and event frame template." These self-service tools allow everyone to see how well the refinery is operating. Viewing the data in many forms enables people to take corrective actions for continuous improvements and for developing predictive analytic models.

"Power BI dashboards allow us to integrate the EIDI Asset Framework calculations with event frame data. We can utilize data pivots for specific times and events to produce detailed performance analyses." Peter explained, "By making these operational events visible for all process units, we can automatically calculate an overall production effectiveness index using Microsoft Power BI." The dashboard provides insights regarding operational modes in relation to time.

This strategy used to estimate the overall production effectiveness (OPE) is called "follow the money" (Plourde 2016, 2019; Plourde et al. 2017). It assists in improving the coordination of the supply chain in large industrial


Event frame cube of data and event pivots for business intelligence analysis. (Courtesy of O.A. Bascur, OSIsoft LLC.)


Microsoft Power BI dashboard showing overall production effectiveness. (Courtesy of O.A. Bascur, OSIsoft LLC.) complexes. Bill Roberts, vice president of operations, has often expressed that the ability to detect and quantify these types of losses are enough to justify the implementation of the EIDI system.

The ability to combine data from different data sets using Power BI allows the team to access information from multiple data sets. For example, the dashboard can have tiles from refinery inventory and their terminals. The team learned that they should be creative in the design of the EIDI's asset object model and in event frame generation. That is, they should look to include any EIDI or non-EIDI data that will help them with their analysis.

Peter and some of the team members configured the EIDI Asset Framework in one day. By applying a straightforward unit data model approach and a standardized nomenclature to define assets in the refinery, they were able to evaluate production versus planning targets as requested by the refinery manager. Refinery personnel started creating ad hoc reports using Power BI.

Figure 4.13 shows a cell phone screenshot of the event-framed operational data analysis. "Once the operational event results are regularly updated


Microsoft Azure Power BI using Cortana artificial intelligence. (Courtesy of O.A. Bascur, OSIsoft LLC.) using Power BI, you can use Cortana on your devices to provide suggestions for operational data analysis," said Peter.

The smart unit template with an operational variance algorithm assists companies by

  • • Characterizing the OPE;
  • • Reporting production and consumables losses when the plant is operating in trouble (invisible losses), as well as identifying down times, idle times, and losses; and
  • • Building predictive analytic models using running OK data sets.

"Now, that we have achieved a successful pilot, we need to convince others at Proclndustries to move forward with our process improvements," said Peter.

During the team dinner afterward, they congratulated themselves on their achievement toward demonstrating the value of the EIDI, as the Proclndustries CEO and Bill Roberts had asked. They also discussed how to prioritize the many opportunities detected by this strategy.

Using Microsoft Excel Analytics Tools

Proclndustries personnel needed to extract EIDI data into Microsoft Excel. Fortunately, an add-in was available for importing current, historical, statistical, and event frame information. The objective is to allow users to create ad hoc analyses on a variety of data types.

Scheduled reports, such as a daily production report at midnight, are now available. Data export is also possible for non-EIDI analytic tools, such as MATLAB®. Figure 4.14 shows results when the equipment was running along with the predicted values and regression analysis for crude unit feed rate and electricity and water consumption.

< Prev   CONTENTS   Source   Next >