Dynamic constrained optimization of chemical manufacturing

Abstract

System and method for chemical manufacture utilizing a dynamic optimizer for a chemical process including upstream and downstream processes. The dynamic optimizer includes a maximum feed calculator, operable to receive one or more local constraints on the downstream processes and one or more model offsets, and execute steady state models for the downstream processes in accordance with the local constraints and the offsets to determine maximum feed capacities of the downstream processes; and a feed coordinator, operable to receive the maximum feed capacities, and execute steady state models for the upstream processes in accordance with the maximum feed capacities and a specified objective function, subject to global constraints, to determine upstream production parameters for the upstream processes, which are usable to control the upstream processes to provide feeds to the downstream processes in accordance with the determined maximum feeds and the objective function subject to the global constraints.

1 . A method for chemical manufacture, comprising: 
 determining maximum feed capacities of a plurality of downstream processes in a chemical plant subject to one or more local constraints on the downstream processes;    determining upstream production parameters for the plurality of upstream processes in accordance with the determined maximum feed capacities and a specified objective function, subject to one or more global constraints; and    controlling the plurality of upstream processes in the chemical plant in accordance with the determined upstream production parameters, thereby facilitating maximum production by the chemical plant in accordance with the specified objective function.    
   
   
       2 . The method of  claim 1 , wherein the specified objective function comprises an optimal product mix for the chemical plant, wherein said determining upstream production parameters is further based on the determined optimal product mix.  
   
   
       3 . The method of  claim 2 , wherein the specified objective function further comprises product prices.  
   
   
       4 . The method of  claim 1 , wherein the chemical plant comprises an olefins production plant.  
   
   
       5 . The method of  claim 4 , wherein the optimal product mix specifies one or more of: 
 ethylene production setpoint(s);    propylene production setpoint(s); and    one or more C4+ production setpoints.    
   
   
       6 . The method of  claim 1 , 
 wherein the plurality of downstream processes comprise separation and purification processes; and    wherein the plurality of upstream processes comprise a plurality of reactors.    
   
   
       7 . The method of  claim 6 , wherein said determining maximum feed capacities, said determining upstream production parameters, and said controlling the plurality of upstream processes implement overall chemical plant steady state optimization via a multivariable predictive dynamic controller adjusting operational targets in the reactors, including target feed rates, to achieve optimal feed rates and product mix in the separation and purification processes.  
   
   
       8 . The method of  claim 7 , wherein the specified objective function comprises an optimal product mix for the chemical plant, the method further comprising: 
 performing local steady state optimization of the separation and purification processes to optimize operation of the separation and purification processes at the determined maximum feed capacities and in accordance with the optimal product mix.    
   
   
       9 . The method of  claim 7 , further comprising: 
 determining a predicted trajectory of optimal feed rates for the separation and purification processes; and    providing the predicted trajectory as input to one or more multivariable predictive dynamic controllers controlling the separation and purification processes.    
   
   
       10 . The method of  claim 7 , wherein the plurality of upstream processes are controlled by one or more upstream multivariable predictive dynamic controllers, the method further comprising: 
 performing local steady state optimization in the upstream multivariable predictive dynamic controllers to optimize operation of the plurality of reactors at the target feed rate.    
   
   
       11 . The method of  claim 6 , wherein the separation and purification processes comprise one or more of: 
 distillation columns; and    refrigeration units.    
   
   
       12 . The method of  claim 6 , wherein the plurality of reactors comprise a plurality of furnaces.  
   
   
       13 . The method of  claim 1 , wherein said controlling the plurality of upstream processes comprises controlling one or more of: 
 feeds to the plurality of upstream processes; and    operating parameters for the plurality of upstream processes.    
   
   
       14 . The method of  claim 1 , wherein said determining the maximum feed capacities of the plurality of downstream processes comprises: 
 receiving one or more local constraints and one or more model offsets; and    executing one or more steady state models for the downstream processes in accordance with the one or more local constraints and the one or more model offsets to determine the maximum feed capacities of the plurality of downstream processes.    
   
   
       15 . The method of  claim 14 , wherein the plurality of downstream processes are controlled by one or more downstream multivariable predictive dynamic controllers comprising respective steady state models of the downstream processes, and wherein said executing the one or more steady state models comprises: 
 executing the steady state models of the one or more downstream multivariable predictive dynamic controllers in optimizer mode.    
   
   
       16 . The method of  claim 15 , wherein the one or more local constraints and the one or more model offsets are provided by the one or more downstream multivariable predictive dynamic controllers.  
   
   
       17 . The method of  claim 16 , wherein the one or more local constraints comprise one or more of: 
 one or more product targets for the plurality of downstream processes;    one or more limits for the plurality of downstream processes.    
   
   
       18 . The method of  claim 15 , wherein said executing the one or more steady state models comprises: 
 executing the steady state models of the one or more downstream multivariable predictive dynamic controllers in combination under combinatorial constraints in accordance with connectivity among the plurality of downstream processes.    
   
   
       19 . The method of  claim 14 , wherein said executing the one or more steady state models in accordance with the one or more model offsets comprises adjusting outputs of the steady state models in accordance with the model offsets to match outputs of the chemical plant.  
   
   
       20 . The method of  claim 14 , further comprising: 
 determining the one or more model offsets by comparing outputs of the one or more steady state models and outputs of the chemical plant.    
   
   
       21 . The method of  claim 14 , wherein said determining the maximum feed capacities of the plurality of downstream processes comprises: 
 executing combined steady state models of the one or more downstream multivariable dynamic controllers in an optimizer.    
   
   
       22 . The method of  claim 14 , wherein each of the steady state models comprises one or more of: 
 a neural network; and    a support vector machine.    
   
   
       23 . The method of  claim 14 , wherein the steady state models comprise neural networks, wherein the neural networks are based on physical models of the downstream processes.  
   
   
       24 . The method of  claim 14 , wherein the steady state models comprise neural networks, wherein the neural networks are based on empirical models of the downstream processes.  
   
   
       25 . The method of  claim 1 , wherein said determining the upstream production parameters for the plurality of upstream processes comprises one or more of: 
 determining target feed rates for multiple types of feedstocks for the plurality of upstream processes;    determining cracking severity for each of the upstream processes.    
   
   
       26 . The method of  claim 25 , wherein said determining target feed rates comprises determining the target feed rates based on one or more of: 
 a specified priority; and    economic value of resulting products.    
   
   
       27 . The method of  claim 1 , wherein said controlling the plurality of upstream processes in the chemical plant in accordance with the determined upstream production parameters comprises: 
 each of a plurality of upstream multivariable predictive dynamic controllers controlling a respective reactor.    
   
   
       28 . The method of  claim 1 , wherein the one or more global constraints comprise one or more of: 
 total feedstock flow of each of a plurality of feedstocks;    ratio of each of the plurality of feedstocks to the total feed; and    maximum difference between individual upstream feeds and the average of the upstream feeds.    
   
   
       29 . The method of  claim 1 , further comprising: 
 performing said determining maximum feed capacities, said determining upstream production parameters, and said controlling the plurality of upstream processes in an iterative manner to dynamically optimize operation of the chemical plant.    
   
   
       30 . The method of  claim 1 , wherein the chemical plant comprises a polymer production plant.  
   
   
       31 . A dynamic optimizer for a chemical process comprising a plurality of upstream processes and a plurality of downstream processes, comprising: 
 a maximum feed calculator, operable to: 
 receive one or more local constraints and one or more model offsets;  
 execute one or more steady state models for the downstream processes in accordance with the one or more local constraints on the plurality of downstream processes and the one or more model offsets to determine maximum feed capacities of the plurality of downstream processes; and  
   a feed coordinator, operable to: 
 receive the determined maximum feed capacities of the plurality of downstream processes; and  
 execute one or more steady state models for the upstream processes in accordance with the determined maximum feed capacities and a specified objective function, subject to one or more global constraints, to determine upstream production parameters for the plurality of upstream processes;  
   wherein the determined upstream production parameters are usable to control the plurality of upstream processes to provide feeds to the plurality of downstream processes in accordance with the determined maximum feeds and the specified objective function subject to the one or more global constraints.    
   
   
       32 . A memory medium that stores program instructions for controlling a chemical process, wherein the program instructions are computer executable to perform: 
 determining maximum feed capacities of a plurality of downstream processes in a chemical plant subject to one or more local constraints on the downstream processes;    determining upstream production parameters for the plurality of upstream processes in accordance with the determined maximum feed capacities and a specified objective function, subject to one or more global constraints; and    controlling the plurality of upstream processes in the chemical plant in accordance with the determined upstream production parameters, thereby facilitating maximum production by the chemical plant in accordance with the specified objective function.