US 7,257,451 B2 · Method for creating a linear programming model of an industrial process facility

Abstract

A method for creating a linear programming (LP) model of an industrial process facility is disclosed. The model may be used for interactively simulating and/or optimizing the operation of the facility to facilitate feedstock selection and/or other economic analyses. According to the inventive method, a first principles reference tool is used to create a non-linear reference model of the entire facility. Then, independent input variables and key output variables are specified, and initial values for the independent input variables are specified. Base case values for each key output variable are determined by running the non-linear reference model using the specified initial values for the independent input variables. Next, partial derivatives for each key output variable with respect to each independent input variable are determined. A matrix is constructed containing the base case values and partial derivatives. The matrix serves as the depiction of the industrial process facility in the final LP, which may also include prices, availabilities, and other external constraints.

Claims32

Dependent claims

A method for creating a linear programming (LP) model of an industrial process facility, said method comprising the steps of:

(a) using a first principles reference tool to create a non-linear reference model of said industrial process facility;
(b) specifying a plurality of independent input variables and a plurality of key output variables;
(c) specifying an initial value for each of said independent input variables;
(d) running said reference model to determine base case values for each of said key output variables based on said initial values of said independent input variables;
(e) determining partial derivatives for each of said key output variables with respect to each of said independent input variables;
(f) creating a matrix at least (n+1) columns wide by at least m rows high, where “n” is the number of independent input variables and “m” is the number of key output variables, one column of said matrix containing said base case values for each of said key output variables from step (d) and the other n columns of said matrix containing said partial derivatives for each of said key output variables with respect to each independent input variable from step (e) except, optionally, those partial derivatives having a value smaller than a pre-selected minimum value;
and
(g) constructing a LP model of said industrial process facility, said LP model containing said matrix and being adapted to calculate the value of one or more of said key output variables from said matrix and a set of known and/or assumed values for said independent input variables.

5 more elements

(a) using a first principles reference tool to create a non-linear reference model of said steam cracker;
(b) specifying a plurality of independent input variables and a plurality of key output variables;
(c) specifying an initial value for each of said independent input variables;
(d) running said reference model to determine base case values for each of said key output variables based on said initial values of said independent input variables;
(e) determining partial derivatives for each of said key output variables with respect to each of said independent input variables;
(f) creating a matrix at least (n+1) columns wide by at least m rows high, where “n” is the number of independent input variables and “m” is the number of key output variables, one column of said matrix containing said base case values for each of said key output variables from step (d) and the other n columns of said matrix containing said partial derivatives for each of said key output variables with respect to each independent input variable from step (e) except, optionally, those partial derivatives having a value smaller than a pre-selected minimum value;
and
(g) constructing a LP model of said steam cracker, said LP model containing said matrix and being adapted to calculate the value of one or more of said key output variables from said matrix and a set of known and/or assumed values for said independent input variables.

5 more elements

A method for creating a linear programming (LP) model of a steam cracker, said steam cracker adapted to convert various hydrocarbon feedstocks into other hydrocarbon products, said steam cracker including at least a furnace section and a recovery section, said furnace section having multiple possible feed types (or headers of feed types) and/or multiple furnace types, said method comprising the steps of:

(a) using a first principles reference tool to create a non-linear reference model of said steam cracker;
(b) creating a LP model of said furnace section by:
(1) selecting a feed type and a furnace type,
(2) specifying feed composition, feed rate, and furnace operating conditions for the selected feed type and furnace type,
(3) establishing a base case value for furnace yields and furnace energy consumption using a furnace yields tool,
(4) determining partial derivatives of the furnace yields and furnace energy consumption with respect to the feedstock flow rates and feedstock compositions (or, alternatively, flow rates for each feedstock component) and each furnace operating condition,
(5) creating a matrix for said feed type and furnace type combination, said matrix containing said base case values for each of said furnace yields and furnace energy consumption from step (b)(3) and said partial derivatives of the furnace yields and furnace energy consumption from step (b)(4) except, optionally, those partial derivatives having a value smaller than a pre-selected minimum value,
(6) repeating steps (b)(2) through (b)(5) for other desired feed type and furnace type combination in said furnace section, and
(7) constructing a LP model of said furnace section, said LP model containing said matrix for each feed type and furnace type combination in said furnace section and being adapted to calculate the products from said furnace section based on a set of known and/or assumed inputs to said furnace section;
(c) creating a LP model of said recovery section by:
(1) specifying independent input variables and key output variables for said recovery section,
(2) specifying an initial value for each of said recovery section independent input variables, said recovery section independent input variables including flow rates of products from said furnace section,
(3) establishing a base case value for each of said recovery section key output variables by running said reference model using said initial values for said recovery section independent input variables,
(4) determining partial derivatives for each of said recovery section key output variables with respect to each of said recovery section independent input variables,
(5) creating a recovery section matrix containing said base case values for each recovery section key output variable from step (c)(3) and said partial derivatives for each of said recovery section key output variables with respect to each recovery section independent input variable from step (c)(4) except, optionally, those partial derivatives having a value smaller than a pre-selected minimum value, and
(6) constructing a LP model of said recovery section, said LP model containing said recovery section matrix and being adapted to calculate the value of one or more of said recovery section key output variables from said recovery section matrix and a set of known and/or assumed values for said recovery section independent input variables;
and
(d) using said LP models of said furnace section and said recovery section as said LP model of said steam cracker.

15 more elements

A method for creating a linear programming (LP) model of an industrial process facility, said industrial process facility including at least two separate sections that can be operated either jointly or separately, said method comprising the steps of:

(a) using a first principles reference tool to create a non-linear reference model of said industrial process facility;
(b) creating a LP model for each of said separate sections of said industrial process facility by:
(1) selecting a first section of said industrial process facility,
(2) specifying independent input variables and key output variables for said section,
(3) specifying an initial value for each of said independent input variables,
(4) establishing a base case value for each of said key output variables by running said reference model using said initial values for said independent input variables,
(5) determining partial derivatives for each of said key output variables with respect to each of said independent input variables,
(6) creating a matrix for said section of said industrial process facility, said matrix containing said base case values for each of said output variables from step (b)(4) and said partial derivatives of each of said output variables with respect to each input variable from step (b)(5) except, optionally, those partial derivatives having a value smaller than a pre-selected minimum value,
(7) constructing a LP model of said first section of said industrial process facility, said LP model containing said matrix and being adapted to calculate the value of one or more of said output variables from said matrix and a set of known and/or assumed values for said input variables, and
(8) repeating steps (b)(2) through (b)(7) for each other section of said industrial process facility;
and
(c) combining said LP models of each section from step (b) to form said LP model of said industrial process facility.

9 more elements

A method for operating an industrial process facility, said method comprising the steps of:

(a) creating a linear programming (LP) model of said industrial process facility by:
(1) using a first principles reference tool to create a non-linear reference model of said industrial process facility,
(2) specifying a plurality of independent input variables and a plurality of key output variables,
(3) specifying an initial value for each of said independent input variables,
(4) running said reference model to determine base case values for each of said key output variables based on said initial values of said independent input variables,
(5) determining partial derivatives for each of said key output variables with respect to each of said independent input variables,
(6) creating a matrix at least (n+1) columns wide by at least m rows high, where “n” is the number of independent input variables and “m” is the number of key output variables, one column of said matrix containing said base case values for each of said key output variables from step (a)(4) and the other n columns of said matrix containing said partial derivatives for each of said key output variables with respect to each independent input variable from step (a)(5) except, optionally, those partial derivatives having a value smaller than a pre-selected minimum value, and
(7) constructing a LP model of said industrial process facility, said LP model containing said matrix and being adapted to calculate the value of one or more of said key output variables from said matrix and a set of known and/or assumed values for said independent input variables;
(b) using said LP model to optimize the operation of said petroleum and/or petrochemical facility;
and
(c) adjusting at least one input condition or operating parameter of said industrial process facility based on the results of said optimization.

8 more elements

Dependencies are inferred from the claim text.

Method for creating a linear programming model of an industrial process facility

Life of this patent

Method for creating a linear programming model of an industrial process facility

Life of this patent

Abstract

Claims32