Granted Patent B2

US 12,572,889 · App. 17/661,323 · Granted Mar 10, 2026

Inventors: Sebastian Terrazas-Moreno (Houston, TX); Dimitrios Varvarezos (Houston, TX)

Assignee: AspenTech Corporation

G06Q10/087G06Q10/04G06Q10/067

Abstract

Claims (90)

1 . A computer-implemented method for automatically controlling physical operations in an industrial supply chain, the method comprising, by a processor:

for each node in a supply chain where the supply chain is formed of multiple nodes:

obtaining, in memory associated with the processor, an input-output model for the node; and

generating an individual equation-oriented model using the obtained input-output model corresponding to the node;

integrating the generated individual equation-oriented models of the multiple nodes with a linking structure to form an optimization model of the supply chain, the optimization model of the supply chain including a plurality of variables;

for each variable of the plurality of variables included in the optimization model, assigning, in a data structure in the memory, at least one respective category to the variable selected from a group of three or more categories, wherein the at least one respective category assigned to each variable is: (i) variables connecting nodes in the supply chain, (ii) variables in an objective function, (iii) variables in linear constraints in the optimization model, or (iv) variables exclusively in nonlinear constraints;

using the at least one respective category assigned to each variable of the plurality of variables in the data structure in the memory, solving the optimization model of the supply chain to determine a value for at least one variable of the plurality, wherein solving the optimization model using the at least one respective category assigned to each variable includes:

applying a respective solving procedure, to each variable of the plurality of variables, where each respective solving procedure is dictated by the at least one respective category assigned to the variable in the data structure in the memory;

grouping nonlinear constraints in the optimization model; and

removing the grouped nonlinear constraints from the optimization model by:

for each nonlinear constraint, creating a corresponding linear constraint by substituting each variable categorized as exclusively in the nonlinear constraint, with a lower and upper bound of the variable, wherein the lower and upper bound of the variable represent a linearized projection of the nonlinear constraints; and

replacing each nonlinear constraint in the optimization model with the created corresponding linear constraint; and

outputting a signal indicating the determined value and automatically controlling a given node of the multiple nodes in the supply chain in accordance with the determined value by one or more computing devices at the given node implementing instructions responsive to the signal to control physical operations at the given node.

2 . The method of claim 1 wherein each obtained input-output model includes one or more inputs and one or more outputs, wherein the one or more inputs are configured to be manipulated to optimize the one or more outputs with respect to an objective function or performance indicator.

3 . The method of claim 1 wherein generating an individual equation-oriented model comprises at least one of:

processing a given input-output model to generate a matrix indicating logistic, economic, and operational constraints for a supply chain node corresponding to the given input-output model;

using a given input-output model and fitting parameters to a first-principles engineering model; and

processing a given input-output model using at least one of statistics, machine learning, and artificial intelligence.

4 . The method of claim 1 wherein the linking structure includes at least one of:

mathematical variables and equations connecting the generated individual equation-oriented models.

5 . The method of claim 1 wherein solving the optimization model comprises:

grouping linear constraints in the optimization model;

solving the optimization model with the removed nonlinear constraints to determine optimal values for variables categorized as connecting nodes in the supply chain or variables in an objective function;

passing the determined optimal values back to given individual equation-oriented models that contain the nonlinear constraints;

solving the given individual equation-oriented models, wherein:

if an individual equation-oriented model is feasible the solving converges; and

if an individual equation-oriented model is infeasible, the solving does not converge and, in response, (i) creating a cut constraint in terms of variables categorized as at least one of: connecting nodes in the supply chain and in an objective function and (ii) adding the cut constraint to the optimization model; and

iteratively solving the optimization model until solving each individual equation-oriented model converges.

6 . The method of claim 1 further comprising:

obtaining the outputted signal;

determining a control parameter of the given node using the determined value, from the obtained signal, in a given individual equation-oriented model indicating behavior of the given node; and

operating the given node in accordance with the determined control parameter.

7 . The method of claim 1 wherein nodes in the supply chain include: a refinery node, a petrochemical plant node, and a polymer plant node.

8 . A computer system for automatically controlling physical operations in an industrial supply chain, the system comprising:

a processor; and

a memory with computer code instructions stored thereon, the processor and the memory, with the computer code instructions, being configured to cause the system to:

for each node in a supply chain where the supply chain is formed of multiple nodes:

obtain, in the memory, an input-output model for the node; and

generate an individual equation-oriented model using the obtained input-output model corresponding to the node;

integrate the generated individual equation-oriented models of the multiple nodes with a linking structure to form an optimization model of the supply chain, the optimization model of the supply chain including a plurality of variables;

for each variable of the plurality of variables included in the optimization model, assign, in a data structure in the memory, at least one respective category to the variable selected from a group of three or more categories, wherein the at least one respective category assigned to each variable is: (i) variables connecting nodes in the supply chain, (ii) variables in an objective function, (iii) variables in linear constraints in the optimization model, or (iv) variables exclusively in nonlinear constraints;

using the at least one respective category assigned to each variable of the plurality of variables in the data structure in the memory, solve the optimization model of the supply chain to determine a value for at least one variable of the plurality, wherein solving the optimization model using the at least one respective category assigned to each variable includes:

applying a respective solving procedure, to each variable of the plurality of variables, where each respective solving procedure is dictated by the at least one respective category assigned to the variable in the data structure in the memory;

grouping nonlinear constraints in the optimization model; and

removing the grouped nonlinear constraints from the optimization model by:

for each nonlinear constraint, creating a corresponding linear constraint by substituting each variable categorized as exclusively in the nonlinear constraint, with a lower and upper bound of the variable; and

replacing each nonlinear constraint in the optimization model with the created corresponding linear constraint; and

output a signal indicating the determined value and automatically control a given node of the multiple nodes in the supply chain in accordance with the determined value by one or more computing devices at the given node implementing instructions responsive to the signal to control physical operations at the given node.

9 . The system of claim 8 wherein each obtained input-output model includes one or more inputs and one or more outputs, wherein the one or more inputs are configured to be manipulated to optimize the one or more outputs with respect to an objective function or performance indicator.

10 . The system of claim 8 wherein, in generating an individual equation-oriented model, the processor and memory, with the computer code instructions are further configured to cause the system to perform at least one of:

processing a given input-output model to generate a matrix indicating logistic, economic, and operational constraints for a supply chain node corresponding to the given input-output model;

using a given input-output model and fitting parameters to a first-principles engineering model; and

processing a given input-output model using at least one of statistics, machine learning, and artificial intelligence.

11 . The system of claim 8 wherein the linking structure includes at least one of:

mathematical variables and equations connecting the generated individual equation-oriented models.

12 . The system of claim 8 wherein, in solving the optimization model, the processor and the memory, with the computer code instructions, are configured to cause the system to:

group linear constraints in the optimization model;

solve the optimization model with the removed nonlinear constraints to determine optimal values for variables categorized as connecting nodes in the supply chain or variables in an objective function;

pass the determined optimal values back to given individual equation-oriented models that contain the nonlinear constraints;

solve the given individual equation-oriented models, wherein:

if an individual equation-oriented model is feasible the solving converges; and

if an individual equation-oriented model is infeasible, the solving does not converge and, in response, (i) creating a cut constraint in terms of variables categorized as at least one of: connecting nodes in the supply chain and in an objective function and (ii) adding the cut constraint to the optimization model; and

iteratively solve the optimization model until solving each individual equation-oriented model converges.

13 . The system of claim 8 wherein, the processor and the memory, with the computer code instructions, are configured to cause the system to:

obtain the outputted signal;

determine a control parameter of the given node using the determined value, from the obtained signal, in a given individual equation-oriented model indicating behavior of the given node; and

operate the given node in accordance with the determined control parameter.

14 . The system of claim 8 wherein nodes in the supply chain include: a refinery node, a petrochemical plant node, and a polymer plant node.

15 . A computer program product for automatically controlling physical operations in an industrial supply chain, the computer program product comprising:

one or more non-transitory computer-readable storage devices and program instructions stored on at least one of the one or more storage devices, the program instructions, when loaded and executed by a processor, cause an apparatus associated with the processor to:

for each node in a supply chain where the supply chain is formed of multiple nodes:

obtain, in memory associated with the processor, an input-output model for the node; and

generate an individual equation-oriented model using the obtained input-output model corresponding to the node;

integrate the generated individual equation-oriented models of the multiple nodes with a linking structure to form an optimization model of the supply chain, the optimization model of the supply chain including a plurality of variables;

for each variable of the plurality of variables included in the optimization model, assign, in a data structure in the memory, at least one respective category to the variable selected from a group of three or more categories, wherein the at least one respective category assigned to each variable is: (i) variables connecting nodes in the supply chain, (ii) variables in an objective function, (iii) variables in linear constraints in the optimization model, or (iv) variables exclusively in nonlinear constraints;

using the at least one respective category assigned to each variable of the plurality of variables in the data structure in the memory, solve the optimization model of the supply chain to determine a value for at least one variable of the plurality, wherein solving the optimization model using the at least one respective category assigned to each variable includes:

applying a respective solving procedure, to each variable of the plurality of variables, where each respective solving procedure is dictated by the at least one respective category assigned to the variable in the data structure in the memory;

grouping nonlinear constraints in the optimization model; and

removing the grouped nonlinear constraints from the optimization model by:

for each nonlinear constraint, creating a corresponding linear constraint by substituting each variable categorized as exclusively in the nonlinear constraint, with a lower and upper bound of the variable; and

replacing each nonlinear constraint in the optimization model with the created corresponding linear constraint; and

output a signal indicating the determined value and automatically control a given node of the multiple nodes in the supply chain in accordance with the determined value by one or more computing devices at the given node implementing instructions responsive to the signal to control physical operations at the given node.

16 . The computer program product of claim 15 wherein, in solving the optimization model, the program instructions, when loaded and executed by the processor, cause the apparatus to:

group linear constraints in the optimization model;

solve the optimization model with the removed nonlinear constraints to determine optimal values for variables categorized as connecting nodes in the supply chain or variables in an objective function;

pass the determined optimal values back to given individual equation-oriented models that contain the nonlinear constraints;

solve the given individual equation-oriented models, wherein:

if an individual equation-oriented model is feasible the solving converges; and

if an individual equation-oriented model is infeasible, the solving does not converge and, in response, (i) creating a cut constraint in terms of variables categorized as at least one of: connecting nodes in the supply chain and in an objective function and (ii) adding the cut constraint to the optimization model; and

iteratively solve the optimization model until solving each individual equation-oriented model converges.

Assignments (4)

Continuity (2)

References Cited (12)