The DIST dataset originates from a methanol distillation process within a chemical plant, encapsulating a month's worth of distributed control system (DCS) data. It offers a granular snapshot of operational dynamics through a variety of process and control parameters, essential for understanding the interactions within control loops. The dataset's value is further enhanced by its rich portrayal of real-time variations in system behavior across different conditions, serving as a crucial asset for crafting sophisticated process control, fault detection, and optimization solutions.

Covering July 1 to August 1, 2023, the DIST dataset consists of 133,920 data points at a 20-second sampling interval, ensuring high-resolution data collection. Data is collected from 37 sensors at each interval, encompassing 9 process variables (PV), 9 operational variables (OP), and 19 monitoring variables.

The data architecture of the dataset is detailed as follows:

1. data_array:
   • The data_array serves as the foundational element of the dataset, encapsulating the sensor-generated data.
   • It adopts a bidimensional array structure, with each row distinctly representing a temporal sampling instance and each column corresponding to a specific sensor variable.
   • The dataset element data_array[i, j] holds the value recorded by the j-th sensor at the i-th chronological sampling point.
2. time_stamp_array:
   • This array contains the time stamps correlated with each data sampling instance in data_array.
   • The element time_stamp_array[i] aligns with and timestamps the sensor readings in the corresponding data_array[i, :].
   • Reflecting a high-resolution data capture approach, the dataset maintains a consistent 20-second interval between each successive temporal sampling point, offering a detailed chronological progression of the distillation process.
3. vars_index_dict:
   • This dictionary provides a mapping schema, linking sensor variable names to their respective indices within data_array.
   • An entry in vars_index_dict, such as 'LIC02.PV': 3, indicates that data pertaining to LIC02.PV is located in the third column of data_array.

The following diagram presents a simplified flowchart of the methanol distillation process. A detailed process flow diagram with specific sensor locations will be published upon obtaining permission from the data source.

• The ID consists of two parts, separated by a period. The first part represents the circuit and function to which the sensor belongs, and the second part represents the category of the sensor. In the first part, the initial letter indicates the type of controlled variable: F for Flow, P for Pressure, T for Temperature, L for Level. The subsequent letters indicate the function of the sensor, where I stands for Indication, C for Control. The numeric part represents the circuit number. In the second part, PV stands for Process Variable, OP for Operational Variable. For example, FIC05.PV represents a Flow sensor with Indication and Control functions, numbered 05, and PV indicates that the data set is the value of its process variable.
• Type DV stands for "Disturbance Variable." These are variables that impact the process but are not directly controlled within the process control loop. In this context, DV signifies important process variables that are external to the control loop, such as temperature, pressure, and gas flow in various parts of the tower.