Runtime modeling approach to updating refinery planning models

A computer-implemented method of increasing efficiency in a processing plant, the method comprising, by a processor:

• creating a planning model comprising a set of model parameters that models a configuration of a plurality of reactors in the processing plant, each reactor having respective rigorous reactor model comprising a set of pre-defined independent variables and a set of pre-defined dependent variables;
• receiving, from a database, at least one of a plurality of runtime models of reduced order corresponding to at least one of the reactors, different reduced order runtime models in the database corresponding to different reactors, wherein for each received reduced order runtime model corresponding to the one reactor, the reduced order runtime model is based on a set of relationships of the rigorous reactor model between the set of pre-defined independent variables and the set of pre-defined dependent variables that correspond to real-world behavior of the one reactor, and a set of tuning factors, wherein:
• the set of relationships of the rigorous reactor model extrapolate input-output behavior of the rigorous reactor model of the one reactor, the set of relationships capture real-world behavior of the one reactor in a simplified form, and
• the tuning factors enable adjusting the runtime model to closely align the runtime model with the real-world behavior of the one reactor;
• running the received reduced order runtime models to update the set of model parameters such that the set of model parameters correspond to real-world behavior of the plurality of reactors in the processing plant, thereby creating an updated planning model;
• and
• forcing a change of the real-world behavior and output of the plurality of reactors in the processing plant by applying the updated planning model, the forced change of the real-world behavior increasing efficiency in the processing plant.

A computer-implemented method of increasing efficiency in a processing plant, the method comprising, by a processor:

• creating a planning model that employs a set of model parameters and models a configuration of a plurality of reactors in a processing plant;
• receiving, from a database, at least one of a plurality of runtime models of reduced order corresponding to at least one of the plurality of reactors, different reduced order runtime models corresponding to different reactors, the different reduced order runtime models correspond to different rigorous reactor models of the reactors, wherein for each reduced order runtime model in the database, the reduced order runtime model comprises a respective set of relationships of a rigorous reactor model between a set of pre-defined independent variables and a set of pre-defined dependent variables that correspond to real-world behavior, and a respective set of tuning factors, the received reduced order runtime models corresponding to one reactor being created by:
• from a source of templates, identifying a template corresponding to a configuration of the one reactor, wherein each template comprises the respective set of pre-defined independent variables and the respective set of pre-defined dependent variables;
• creating a rigorous reactor model of the one reactor, said created rigorous reactor model comprising the set of pre-defined independent variables and the set of pre-defined dependent variables in the identified template, the identified set of relationships of the rigorous reactor model extrapolating input-output behavior of the rigorous reactor model of the one reactor, the identified set of relationships capturing real-world behavior of the one reactor in a simplified form;
• using a simulator, running the created rigorous reactor model with possible candidate values for the set of pre-defined independent variables and the set of pre-defined dependent variables to identify a set of relationships of the rigorous reactor model between the set of pre-defined independent variables and the set of pre-defined dependent variables;
• building the reduced order runtime model based upon the identified set of relationships of the rigorous reactor model between the set of pre-defined independent variables and the set of pre-defined dependent variables;
• and
• storing the reduced order runtime model on the database,
• wherein the stored reduced order runtime model further comprises a set of tuning factors, wherein the tuning factors enable adjusting the runtime model to closely align the runtime model with the real-world behavior of the one reactor;
• running the received runtime models and updating the planning model by updating the set of model parameters such that the updated set of model parameters correspond to real-world behavior of the plurality of reactors in the processing plant;
• and
• forcing a change of the real-world behavior and output of the plurality of reactors in the processing plant by applying the updated planning model, the forced change of the real-world behavior increasing efficiency in the processing plant.

An apparatus for increasing efficiency in a processing plant, the apparatus comprising:

• computer memory;
• and
• a computer means coupled to the computer memory and executed by a processor, the computer means configured to:
• create a planning model comprising a set of model parameters that models a configuration of a plurality of reactors in processing plant, each reactor having a respective rigorous reactor model comprising a set of pre-defined independent variables and a set of pre-defined dependent variables;
• receive, from a database, at least one of a plurality of runtime models of reduced order corresponding to at least one of the reactors, different reduced order runtime models in the database corresponding to different reactors, wherein for each received reduced order runtime model corresponding to the one reactor, the runtime model is based on a set of relationships of the rigorous reactor model between the set of pre-defined independent variables and the set of pre-defined dependent variables that correspond to real-world behavior of the one reactor, and a set of tuning factors, wherein:
• the set of relationships of the rigorous reactor model extrapolate input-output behavior of the rigorous reactor model of the reactor model, the set of relationships capture real-world behavior of the one reactor in a simplified form, and
• the tuning factors enable adjusting the runtime model to closely align the runtime model with the real-world behavior of the one reactor;
• run the received reduced order runtime models to update the set of model parameters such that the set of model parameters correspond to real-world behavior of the plurality of reactors in the processing plant, thereby creating an updated planning model;
• and
• force a change of the real-world behavior and output of the plurality of reactors in the processing plant by applying the updated planning model, the forced change of the real-world behavior increasing efficiency in the processing plant.

A computer program product for increasing efficiency in a processing plant, comprising:

• a compute readable medium embodying program instructions, the program instructions, when executed by a digital processor, cause the digital processor to:
• create a planning model comprising a set of model parameters that models a configuration of a plurality of reactors in the processing plant, each reactor having a respective rigorous reactor model comprising a set of pre-defined independent variables and a set of pre-defined dependent variables;
• receive, from a database, at least one of a plurality of runtime models of reduced order corresponding to at least one of the reactors, different reduced order runtime models in the database corresponding to different reactors, wherein for each received reduced order runtime model corresponding to the one reactor, the runtime model is based on a set of relationships of the rigorous reactor model between the set of pre-defined independent variables and the set of pre-defined dependent variables that correspond to real-world behavior of the one reactor, and a set of tuning factors, wherein:
• the set of relationships of the rigorous reactor model extrapolate input-output behavior of the rigorous reactor model, the set of relationships capture real-world behavior of the one reactor in a simplified form, and
• the tuning factors enable adjusting the runtime model to closely align the runtime model with the real-world behavior of the one reactor;
• run the received reduced order runtime models to update the set of model parameters such that the set of model parameters correspond to real-world behavior of the plurality of reactors in the processing plant, thereby creating an updated planning model;
• and
• force a change of the real-world behavior and output of the plurality of reactors in the processing plant by applying the updated planning model, the forced change of the real-world behavior increasing efficiency in the processing plant.