Data Acquisition, Validation, and Classification
"The data hierarchy shows how data is processed, transformed, and used at many levels of decision-making based on the context of time and other attributes associated with it," asserted Peter. "Once you have the right time context and operating conditions, you can apply a process model." Peter explained further, "You can calculate the efficiency of a process or predict a sensor reading."
The first step is to automatically collect data from all possible sources. The second step checks the connectivity of the subsystems and the validity of the data based on the basic limits of the sensors. The next step is data classification: Here an online analytics application checks the state of the process equipment based on basic performance rules. The state is stored as an event marker in time. This internal record of time is used to
- • Aggregate all the data associated with a corresponding asset or physical entity;
- • Establish a state that can be used to generate a notification to inform people about important updates; and
- • Use the established state to aggregate more data.
Take, for example, a storage tank. A large tank will have a very long residence time depending on its size and the velocity of the pump associated with it. The pump's status is checked every one or two minutes. These status checks create a stream of data collected in the EIDI for analysis of what is happening and when with equipment and processes in the refinery.
"This step represents the Are we on target?' assessment we discussed earlier," acknowledged Peter. "Using this type of data classification enables us to sort the data so we can use it for additional value-added calculations. It augments the engineering knowledge available from the process engineers. For example, if it is within the reasonable operational limits, we can use the data in several ways: calculating performance, estimating the fouling factor of a heat exchanger, or evaluating the property of a stream using chemical engineering principles." This idea emanates from Bill Roberts' vision of automating to achieve continuous improvement and innovation, and notifying people immediately. The vision becomes a reality by automating the system monitoring "Are we on target?" loop and generating events and notifications (Bascur and Soudek 2019).