ROBUST FEEDSTOCK SELECTION SYSTEM FOR THE CHEMICAL PROCESS INDUSTRIES UNDER MARKET AND OPERATIONAL UNCERTAINTY

• US 20170308831A1
• Filed: 04/20/2016
• Published: 10/26/2017
• Est. Priority Date: 04/20/2016
• Status: Active Grant

• Current Assignee: Aspentech Corporation
• Original Assignee: Aspen Technology (Telefonaktiebolaget LM Ericsson)
• Inventors: Apap, Robert M., Varvarezos, Dimitrios
• Granted Patent: US 10,755,214 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 17/11 for solving equations , e.g. nonlinear equations, general mathematical optimization problems optimization specially adapted for a specific administrative, business or logistic context G06Q10/04
• CPC Class Codes: G06F 17/18 for evaluating statistical data , e.g. average values, frequency distributions, probability functions, regression analysis forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G06Q 10/06313 Resource planning in a project environment
• CPC Class Codes: G06Q 10/06375 Prediction of business process outcome or impact based on a proposed change
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities
• CPC Class Codes: G06Q 30/0206 Price or cost determination based on market factors
• CPC Class Codes: G06Q 50/04 Manufacturing
• CPC Class Codes: G06Q 50/06 Energy or water supply
• CPC Class Codes: Y02P 90/30 Computing systems specially adapted for manufacturing

Abstract

A computer system and method optimize feedstock selection planning for an industrial process by evaluating first and second stages at separate intervals throughout the planning process. Evaluating the first stage determines a set of robust feedstocks to procure on long-term contracts. The computer system and method solve, in parallel, multiple simulation cases of a non-linear model generated with different expectation values for uncertain input parameters related to selecting feedstocks to procure on long-term contracts. Probabilistic analyses on the solutions from the simulation cases, including the application of chance-constraints, determine the set of robust feedstocks to procure on long-term contracts. Evaluating the second stage determines a set of robust feedstocks to procure in the spot market, using the information from the first stage. Specifically, the computer system and method solve each of multiple new simulation cases of the non-linear model, generated with different expectation values for uncertain input parameters related to selecting feedstocks to procure in the spot market. Each simulation case is solved to determine breakeven prices for one or more available spot feedstocks, and probabilistic analyses are performed on the breakeven prices for these spot feedstocks to determine a set of robust feedstocks to procure in the spot market.

Claims

1. A computer-implemented method for feedstock selection planning for an industrial process, the method comprising:

a) evaluating a first stage and a second stage at separate time intervals throughout a feedstock selection planning process of a subject industrial plant;

b) determining, in the evaluation of the first stage, a set of robust feedstocks to procure on long-term contracts, wherein the determining includes;

modeling feedstock procurement on long-term contracts as a first set of simulation cases of a non-linear model, wherein the first set of simulation cases includes a set of uncertain input parameters representing uncertainty related to selecting feedstocks to procure on long-term contracts;

solving each simulation case of the first set of simulation cases, in parallel, to model different feedstock selection outcomes, wherein the modeled outcomes include optimal feedstocks, feedstock volumes, and operation conditions with respect to the given simulation case; and

performing probabilistic analysis of the different modeled outcomes from the first set of simulation cases to determine the set of robust feedstocks to procure on long-term contracts; and

c) providing to the subject industrial plant the determined set of robust feedstocks, said providing being in a manner enabling programming of operations of the subject industrial plant.

2. The method of claim 1, wherein the non-linear model is a mixed-integer, non-linear optimization problem (MINLP) model, and each simulation case comprises an independent instance of the MINLP model with a different set of realization values provided for the respective set of uncertain input parameters.

3. The method of claim 1, wherein the simulation cases are generated by:

(i) fitting a multivariate distribution to the uncertain input parameter data using kernel density estimation (KDE), or capturing correlations among the uncertain input parameters using a copula-based approach, and (ii) applying Monte-Carlo sampling to the respective multivariate distribution or the captured correlations.

4. The method of claim 1, wherein the uncertain input parameters represent one or more of uncertain market conditions, uncertain operation conditions, and other uncertain industrial conditions.

5. The method of claim 1, wherein performing probabilistic analysis of the different modeled outcomes further comprises:

selecting robust feedstocks from the optimal feedstocks of the different modeled outcomes;

generating a probabilistic feed slate distribution from the optimal feedstocks of the different modeled outcomes; and

applying a threshold probability to the probabilistic feed slate distribution to select one or more robust feedstocks from the distributed feedstocks; and

establishing an optimal feedstock volume for each selected robust feedstock by;

generating a probabilistic distribution of feedstock volumes provided from the different modeled outcomes for the given robust feedstocks, and

applying joint chance constraints derived from the probabilistic distribution of the feedstock volumes to establish optimal feedstock procurement volumes for the given robust feedstocks.

6. The method of claim 5, wherein performing probabilistic analysis of the different modeled outcomes further comprises:

generating a robust base case of the non-linear model, the robust base case being generated to include the joint chance constraints and a capacity-fulfillment constraint; and

solving the generated robust base case, the solving determining the optimal procurement volumes corresponding to the set of robust feedstocks to procure on long-term contracts.

7. The method of claim 1, wherein the determined robust feedstocks to procure on long-term contracts are validated by:

comparing expected profits from the determined set of robust feedstocks from the first stage against expected profits from a deterministic approach.

8. The method of claim 1, further comprising:

determining, in the evaluation of the second stage, a set of robust feedstocks to procure in a spot market, wherein the determining includes;

modeling feedstock procurement in the spot market as a second set of simulation cases of the non-linear model, wherein the second set of simulation cases includes a set of uncertain input parameters representing uncertainty related to selecting feedstocks to procure in the spot market,

solving each simulation case of the second set of simulation cases to model different breakeven prices for one or more spot feedstocks, and

performing probabilistic analysis on the different breakeven prices for each of the one or more spot feedstocks, wherein the performing determining which of the one or more spot feedstocks to procure in the spot market; and

determining optimal operating conditions to process the determined robust feedstocks from each stage in order to produce a set of products.

9. The method of claim 8, wherein solving each simulation case of the second set of simulation cases further comprises:

for each simulation case of the second set of simulation cases;

solving the given simulation case with the determined set of robust feedstocks from the first stage, the solving determining an optimal value of an objective function for the given simulation case; and

for each spot feedstock of a set of available spot feedstocks;

generating a feedstock case by forcing a fixed procurement volume of the given spot feedstock into the given simulation case,

solving the feedstock case to determine an optimal value of an objective function for the given feedstock case, and

calculating a breakeven price for the given spot feedstock based on;

(i) the determined optimal objective function value for the given simulation case, and (ii) the determined optimal objective function value for the given feedstock case.

10. The method of claim 9, wherein performing probabilistic analysis on the different breakeven prices for each spot feedstock of a set of available spot feedstocks, further comprises:

generating ECDFs for breakeven analysis, each ECDF representing, for each spot feedstock, a correspondence between a calculated breakeven price and a risk level; and

ranking each spot feedstock of the set of available spot feedstocks for breakeven analysis, the ranking determined by defining an incremental profit value between a market value and each calculated breakeven price for the given spot feedstock; and

based on the ECDFs and the ranking, determining the set of robust feedstocks to procure in the spot market.

21. The method of claim 1, wherein programming of operations includes programming any of a plant system application, a blending control system, a process control system, and other control systems.

11. A computer system for feedstock selection planning for an industrial process, the system comprising:

one or more processors and associated memory, the one or more processors configured to evaluate a first stage and a second stage at separate intervals throughout a feedstock selection planning process of a subject industrial plant, the one or more processors determining, in the evaluation of the first stage, a set of robust feedstocks to procure on long-term contracts, wherein the one or more processors comprise;

a modeler engine configured to model feedstock procurement on long-term contracts as a first set of simulation cases of a non-linear model, wherein the first set of simulation cases includes a set of uncertain input parameters representing uncertainty related to selecting feedstocks to procure on long-term contracts;

a solver engine configured to solve each simulation case of the first set of simulation cases, in parallel, to model different feedstock selection outcomes, wherein the modeled outcomes include optimal feedstocks, feedstock volumes, and operation conditions with respect to the given simulation case; and

a solution analyzer configured to perform probabilistic analysis of the different modeled outcomes from the first set of simulation cases to determine the set of robust feedstocks to procure on long-term contracts, the solution analyzer providing to the subject industrial plant the determined set of robust feedstocks in a manner enabling programming of operations of the subject industrial plant.

12. The system of claim 11, wherein the non-linear model is a mixed-integer, non-linear optimization problem (MINLP) model, and each simulation case comprises an independent instance of the MINLP model with a different set of realization values provided for the respective set of uncertain input parameters.

13. The system of claim 11, wherein the modeler engine is further configured to generate the simulation cases by:

(i) fitting a multivariate distribution to the uncertain input parameter data using kernel density estimation (KDE), or capturing correlations among the uncertain input parameters using a copula-based approach, and (ii) applying Monte-Carlo sampling to the respective multivariate distribution or the captured correlations.

14. The system of claim 11, wherein the uncertain input parameters represent one or more of uncertain market conditions, uncertain operation conditions, and other uncertain industrial conditions.

15. The system of claim 11, wherein the solution analyzer is further configured to:

select robust feedstocks from the optimal feedstocks of the different modeled outcomes by;

generating a probabilistic feed slate distribution from the optimal feedstocks of the different modeled outcomes; and

applying a threshold probability to the probabilistic feed slate distribution to select one or more robust feedstocks from the distributed feedstocks; and

establish an optimal feedstock volume for each selected robust feedstock by;

generating a probabilistic distribution of feedstock volumes provided from the different modeled outcomes for the given robust feedstocks, and

applying joint chance constraints derived from the probabilistic distribution of the feedstock volumes to establish optimal feedstock procurement volumes for the given robust feedstocks.

16. The system of claim 15, wherein the solution analyzer is further configured to:

generate a robust base case of the non-linear model, the robust base case being generated to include the joint chance constraints and a capacity-fulfillment constraint; and

solve the generated robust base case, the solving determining the optimal procurement volumes corresponding to the set of robust feedstocks to procure on long-term contracts.

17. The system of claim 11, wherein the one or more processors are further configured to validate the determined robust feedstocks to procure on long-term contracts by:

comparing expected profits from the determined set of robust feedstocks from the first stage against expected profits from a deterministic approach.

18. The system of claim 11, further comprising:

the one or more processors configured to determine, in evaluation of the second stage, a set of robust feedstocks to procure in the spot market, wherein the one or more processors comprise;

the modeler engine configured to model feedstock procurement in a spot market as a second set of simulation cases of the non-linear model, wherein the second set of simulation cases includes a set of uncertain input parameters representing uncertainty related to selecting feedstocks to procure in the spot market,

the solver engine configured to solve each simulation case of the second set of simulation cases to model different breakeven prices for one or more spot feedstocks, and

the solution analyzer configured to perform probabilistic analysis on the different breakeven prices for each of the one or more spot feedstocks, wherein the performing determining which of the one or more spot feedstocks to procure in the spot market; and

the one or more processors further configured to determine optimal operating conditions to process the determined robust feedstocks from each stage in order to produce a set of products.

19. The system of claim 18, wherein the solver engine is further configured to:

for each simulation case of the second set of simulation cases;

solve the given simulation case with the determined set of robust feedstocks from the first stage, the solving determining an optimal value of an objective function for the given simulation case; and

for each spot feedstock of a set of available spot feedstocks;

generate a feedstock case by forcing a fixed procurement volume of the given spot feedstock into the given simulation case,

solve the feedstock case to determine an optimal value of an objective function for the given feedstock case, and

calculate a breakeven price for the given spot feedstock based on;

(i) the determined optimal objective function value for the given simulation case, and (ii) the determined optimal objective function value for the given feedstock case.

20. The system of claim 19, wherein the solution analyzer is further configured to:

generate ECDFs for breakeven analysis, each ECDF representing, for each spot feedstock, a correspondence between a calculated breakeven price and a risk level; and

rank each spot feedstock of the set of available spot feedstocks for breakeven analysis, the ranking determined by defining an incremental profit value between a market value and each calculated breakeven price for the given spot feedstock; and

based on the ECDFs and the ranking, determine the set of robust feedstocks to procure in the spot market.