Method and system for process schedule reconciliation using machine learning and algebraic model optimization
A computer-implemented method and system for process schedule reconciliation receives a scheduling model and an initial schedule for reconciliation, where the initial schedule includes projected plant data. Current plant data is imported into the system, and dynamic optimization data representing trends in process data at time-varied values for key process and operation parameters are identified. The current plant data and projected plant data is processed using mathematical modeling techniques to identify event boundaries, stream flowrates associated with tanks and process units. The system builds an optimization model applying identified event boundaries, stream flowrates, dynamic optimization data, key scheduling parameters and pre-determined constraints along a period of time that includes priority slots to reconcile the projected plant data of the initial schedule with the current plant data, and then solves the optimization model to develop a reconciled schedule.
1. A computer-implemented method of process control and schedule reconciliation, comprising:
receiving by at least one digital processor a scheduling model formed of an initial schedule for reconciliation, the initial schedule having projected plant data for a first period of time regarding at least one chemical process of a subject plant;
importing, into a memory area accessible by the at least one digital processor, measured plant data for the first period of time regarding the at least one chemical process;
identifying dynamic optimization data representing trends in plant-wide process production data at time-varied values for key process and operation parameters;
processing the imported measured plant data and the projected plant data using mathematical modeling techniques and identifying event boundaries, and stream flowrates of materials associated with tanks and process units of the subject plant during the first period of time;
building an optimization model that mathematically represents a schedule reconciliation problem of processes of the subject plant during the first period of time, the optimization model applying penalties for deviating from: (i) the identified event boundaries during the first period of time, (ii) the identified stream flowrates of materials during the first period of time, (iii) the identified dynamic optimization data, (iv) pre-determined constraints, and (v) the initial schedule for the first period of time, along a period of time that includes priority slots such that solving the optimization model results in a reconciled schedule for the first period of time regarding the at least one chemical process that aligns the projected plant data for the first period of time regarding the at least one chemical process with the measured plant data for the first period of time regarding the at least one chemical process;
solving the optimization model to develop the reconciled schedule for the first period of time regarding the at least one chemical process, the reconciled schedule having reconciled plant data for the first period of time regarding the at least one chemical process consistent with the identified event boundaries and the identified stream flowrates of materials identified from the processing of the measured plant data and the projected plant data including minimizing deviation from the projected plant data, and providing a current event status, said importing, identifying, processing, building, and solving being automatically performed by the at least one digital processor;
automatically building a future schedule for a second period of time succeeding the first period of time in computer memory, using the reconciled plant data; and
controlling the subject plant during the second period of time using the built future schedule by automatically adjusting start time, stop time, and flow rates of tank to tank transfers and of operating modes of the process units.
2. The computer-implemented method of claim 1 wherein the pre-determined constraints include operation constraints.
3. The computer-implemented method of claim 1 wherein the pre-determined constraints include scheduling constraints.
4. The computer-implemented method of claim 1 wherein the pre-determined constraints are further applied across multiple periods.
5. The computer-implemented method of claim 4 wherein reconciling the projected plant data further includes determining the timing of the priority slots within period boundaries.
6. The computer-implemented method of claim 1 wherein the dynamic optimization data is based on reconciliation data obtained once every few minutes.
7. The computer-implemented method of claim 1 wherein the scheduling model and optimization model consist of different types of resources, including transportation modes, tanks, mixers, splitters and production units, that represent corresponding resources in a refinery scheduling operations system.
8. A computer-implemented method of process control and schedule reconciliation, comprising:
receiving a scheduling model formed of an initial schedule for reconciliation, the initial schedule having projected plant data for a first period of time regarding at least one chemical process of a subject plant;
importing, into a computer memory area, measured plant data for the first period of time regarding the at least one chemical process;
processing the imported measured plant data and the projected plant data using mathematical modeling techniques and identifying event boundaries, and stream flowrates of materials associated with tanks and process units of the subject plant during the first period of time;
importing historical scheduling data into the computer memory area, the historical scheduling data including historical reconciled schedules, and corresponding initial plant data and projected plant data for the historical reconciled schedules;
processing the imported historical scheduling data to develop machine learning and statistical models to identify key scheduling parameters and decisions;
building an optimization model that mathematically represents a schedule reconciliation problem of processes of the subject plant during the first period of time, the optimization model applying penalties for deviating from: (i) the identified event boundaries during the first period of time, (ii) the identified stream flowrates of materials during the first period of time, (iii) key scheduling parameters, (iv) pre-determined constraints, and (v) the initial schedule for the first period of time, along a period of time that includes priority slots such that solving the optimization model results in a reconciled schedule for the first period of time regarding the at least one chemical process that aligns the projected plant data for the first period of time regarding the at least one chemical process with the measured plant data for the first period of time regarding the at least one chemical process;
solving the optimization model to develop the reconciled schedule for the first period of time regarding the at least one chemical process, the reconciled schedule having reconciled plant data for the first period of time regarding the at least one chemical process consistent with the identified event boundaries and the identified stream flowrates of materials identified from the processing of the measured plant data and the projected plant data including minimizing deviation from the projected plant data and providing a current event status, said importing, identifying, processing, building, and solving being automatically performed by at least one digital processor;
automatically building a future schedule for a second period of time succeeding the first period of time in computer memory, using the reconciled plant data; and
controlling the subject plant during the second period of time using the future schedule by automatically adjusting start time, stop time, and flow rates of tank to tank transfers and of operating modes of the process units.
9. The computer-implemented method of claim 8 wherein the pre-determined constraints include operation constraints.
10. The computer-implemented method of claim 8 wherein the pre-determined constraints include scheduling constraints.
11. The computer-implemented method of claim 8 wherein the pre-determined constraints are further applied across multiple periods of priority slots.
12. The computer-implemented method of claim 11 wherein reconciling the projected plant data further includes determining the timing of the priority slots within period boundaries.
13. The computer-implemented method of claim 8 wherein the scheduling model and optimization model consist of different types of resources, including transportation modes, tanks, mixers, splitters and production units, that represent corresponding resources in a refinery scheduling operations system.
14. A computer-implemented method of process control and schedule reconciliation, comprising:
receiving by at least one digital processor a scheduling model formed of an initial schedule for reconciliation, the initial schedule having projected plant data for a first period of time regarding at least one chemical process of a subject plant;
importing, into a memory area accessible by the at least one digital processor, measured plant data for the first period of time regarding the at least one chemical process;
identifying dynamic optimization data representing trends in plant-wide process production data at time varied values for key process parameters;
processing the imported measured plant data and the projected plant data using mathematical modeling techniques and identifying event boundaries, and stream flowrates of materials associated with tanks and process units of the subject plant during the first period of time;
importing historical scheduling data into the memory area; the historical scheduling data having historical reconciled schedules and their corresponding initial state data and projected state data for the historical reconciled schedules;
processing the imported historical scheduling data to develop machine learning and statistical models to identify key scheduling parameters;
building an optimization model that mathematically represents a schedule reconciliation problem of processes of the subject plant during the first period of time, the optimization model applying penalties for deviating from: (i) the identified event boundaries during the first period of time, (ii) the identified stream flowrates of materials during the first period of time, (iii) the identified dynamic optimization data, (iv) key scheduling parameters and pre-determined constraints, and (v) the initial schedule for the first period of time, along a period of time that includes priority slots such that solving the optimization model results in a reconciled schedule for the first period of time regarding the at least one chemical process that aligns the projected plant data for the first period of time regarding the at least one chemical process with the measured plant data for the first period of time regarding the at least one chemical process;
solving the optimization model to develop the reconciled schedule for the first period of time regarding the at least one chemical process, the reconciled schedule having reconciled plant data for the first period of time regarding the at least one chemical process consistent with the identified event boundaries and the identified stream flowrates of materials identified from the processing of the measured plant data and the projected plant data including minimizing deviation from the projected plant data, and providing a current event status, said importing, identifying, processing, building, and solving being automatically performed by the at least one digital processor;
automatically building a future schedule for a second period of time succeeding the first period of time, using the reconciled plant data; and
controlling the subject plant during the second period of time using the future schedule by automatically adjusting start time, stop time, and flow rates of tank to tank transfers and of operating modes of the process units.
15. A computer-based process modeling and simulation system, comprising:
a modeling subsystem performing schedule reconciliation on a chemical process of a subject industrial plant, the modeling subsystem (i) receiving by at least one digital processor a scheduling model formed of an initial schedule for reconciliation, the initial schedule having projected plant data for a first period of time regarding the chemical process of the subject plant, (ii) importing, into a memory area accessible by the at least one digital processor, measured plant data for the first period of time regarding the chemical process, (iii) identifying dynamic optimization data representing trends in plant-wide process production data at time varied values for key process parameters; (iv) processing the imported measured plant data and projected plant data using mathematical modeling techniques and identifying event boundaries, and stream flowrates properties associated with tanks and process units of the subject plant during the first period of time, (v) importing historical scheduling data into the memory area; the historical scheduling data having historical reconciled schedules and their corresponding initial state data and projected state data, (vi) processing the imported historical scheduling data to develop machine learning and statistical models to identify key scheduling parameters; (vii) building an optimization model that mathematically represents a schedule reconciliation problem of processes of the subject plant during the first period of time, the optimization model applying penalties for deviating from: (a) the identified event boundaries during the first period of time, (b) the identified stream flowrates of materials during the first period of time, (c) the identified dynamic optimization data, (d) key scheduling parameters and pre-determined constraints, (e) the initial schedule for the first period of time, along a period of time that includes priority slots such that solving the optimization model results in a reconciled schedule for the first period of time regarding the at least one chemical process that aligns the projected plant data for the first period of time regarding the chemical process with the measured plant data for the first period of time regarding the chemical process, and (viii) solving the optimization model to develop the reconciled schedule for the first period of time regarding the at least one chemical process having reconciled plant data for the first period of time regarding the chemical process, the reconciled plant data consistent with the identified event boundaries and the identified stream flowrates of materials identified from the processing of the measured plant data and the projected plant data including minimizing deviation from the projected plant data, and providing a current event status, said importing, identifying, processing, building, and solving being automatically performed by the at least one digital processor and;
an interface coupled to the modeling subsystem in a manner that enables improvements in performance of the processes of the subject industrial plant by automatically adjusting start time, stop time, and flow rates of tank to tank transfers and of operating modes of the process units based on a future schedule for a second period of time succeeding the first period of time, automatically built using the reconciled plant data, to control the processes of the subject plant during the second period of time.