Step 2: Analyzing and Visualizing Operational Variance from the Generated Events
Once the unit template is applied to each of the refinery unit elements, the respective attributes are configured with real-time data stream tags. As shown in Figure 4.6, the EIDI human interface software (OSIsoft's PI Vision)
Refinery overview dashboard. (Courtesy of O.A. Bascur, OSIsoft LLC.)
Crude unit relative dashboard with unit template attributes and operating modes. (Courtesy of O.A. Bascur, OSIsoft LLC.)
displays the real-time information for all refinery units, the unit template operating modes, as well as production and consumable data.
To show a more detailed KPI dashboard for that specific unit, the user can simply click on the appropriate button (Figure 4.7).
Figure 4.7 shows a real-time crude unit dashboard displaying the key process unit attributes and the operational event notifications (shown in the rectangle at the top right) generated by the EIDI event frame subsystem. By clicking one of these notifications, a second display is generated, as shown in Figure 4.8.
The power of event framing is the ability to compare similar events, operating modes, and batches/lots against each other through online data visualization tools such as PI Vision, or to export these events and their corresponding data to other software systems for offline analysis. As previously stated, users can compare an ideal batch/event/lot data profile against other similar production runs to determine what exactly causes an ideal run.
Figure 4.8 shows the production and consumable losses, such as electricity and water, for selected operational time events.
Monica went on, "By analyzing the different operational states, we can see the variance between an expected and an actual result. The expected results are those specified in the budget or in current production schedules, produced by my planning and economics team. We get the production schedule targets from a LP model that incorporates the inventories,
Event frame display showing crude unit production and consumable losses. (Courtesy of O.A. Bascur, OSIsoft LLC.)
Unit analysis of operational variables with real-time notification ability. (Courtesy of O.A. Bascur, OSIsoft LLC.)
customer orders, and process unit parameters for several types of raw materials and processing strategies."
Figure 4.9 shows the detailed information for a selected process unit, with event start and end times, attribute history, and when the event was active.
Classifying Asset Behavior for Process Improvement
Peter surmised: "This frees up valuable time for engineers and operators to make quicker decisions. The EIDI PI Vision displays present the right data to solve most of our difficult operational problems."
The digital transformation team then put the following ideas into practice:
- 1. Build a digital plant template for each process unit.
- 2. Identify input data streams for each unit template.
- 3. Define EIDI event frames for each unit production mode.
- 4. Identify conditions to indicate various modes of operation including transition states.
- 5. Configure real-time asset analytics to record transitions from one mode to another.
- 6. Define and calculate the unit template output values (e.g., production rate, yield).
- 7. Configure notifications to alert people and systems of process mode changes.