WO2014123993A1 - System, method and apparatus for determining properties of product or process streams - Google Patents

System, method and apparatus for determining properties of product or process streams Download PDF

Info

Publication number
WO2014123993A1
WO2014123993A1 PCT/US2014/014857 US2014014857W WO2014123993A1 WO 2014123993 A1 WO2014123993 A1 WO 2014123993A1 US 2014014857 W US2014014857 W US 2014014857W WO 2014123993 A1 WO2014123993 A1 WO 2014123993A1
Authority
WO
WIPO (PCT)
Prior art keywords
analyzer
model
dependent variables
variables
inferential
Prior art date
Application number
PCT/US2014/014857
Other languages
French (fr)
Inventor
W. Marcus Trygstad
Original Assignee
Yokogawa Corporation Of America
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Yokogawa Corporation Of America filed Critical Yokogawa Corporation Of America
Priority to JP2015556250A priority Critical patent/JP2016507113A/en
Priority to EP14748750.8A priority patent/EP2954427A4/en
Priority to CN201480017536.2A priority patent/CN105229627B/en
Publication of WO2014123993A1 publication Critical patent/WO2014123993A1/en

Links

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/048—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators using a predictor
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation

Definitions

  • the present invention generally relates to a system, method, and apparatus of estimating properties of hydrocarbon streams.
  • the systems, methods, and apparatuses of the present invention relates to a measuring properties of hydrocarbon streams with high frequency and a high degree of accuracy.
  • Soft sensors are appealing for at least two reasons. First, many times they do not require the installation of additional sensors in the process unit because they typically rely upon measurements such as temperature, pressure, and flow rate, which may already be available. Second, with the advent of distributed control systems, the input measurements typically relied upon by soft sensors are substantially available in real-time, having discrete sampling rates of one second or even less. These advantages at least address the disadvantage of time delay associated with primary measurements by providing property predictions at intervals that are greater than those typically required by process control systems. Additionally, they obviate the need to physically obtain a sample, eliminating the issues of representative sampling and sample integrity.
  • the common practice therefore is to use periodic laboratory measurements to update the outputs from inferential analyzers to deal with a variety of variables.
  • This strategy may improve the quality of inferential predictions across (i) the full range of possible feed qualities; (ii) changes in the condition of the process system, e.g. fouling; and (iii) changes in the response of sensors whose readings are the independent variables, e.g. those resulting from simple drift or from replacement of a faulty sensor with a new one.
  • issues attach to this approach which limit the possibilities for optimizing the performance of continuous processes, which may include maximizing throughput, adhering closely to product quality targets, minimizing energy usage, extending catalyst service life, and the like.
  • An aspect of the invention includes a system for determining a property of a process stream.
  • the system includes a hard analyzer configured to measure one or more dependent variables associated with a sample of the process stream, where a dependent variable is also a property of the sample.
  • An inferential analyzer includes a model of aspects of the process, and is configured to capture one or more independent process variables associated with the process stream and to use the independent process variables to generate one or more inferred dependent variables.
  • a comparison module generates a comparison of the dependent variables measured by the hard analyzer and the dependent variables inferred by the inferential analyzer. The system uses the comparison to adjust one or more aspects of the model and/or adjust one or more inferred dependent variables and/or control the process stream.
  • Another aspect of the invention includes a method for determining a property of a process stream using the aforementioned system.
  • FIG. 1A is a functional block diagram of an embodiment of the present invention.
  • Fig. IB is a flowchart showing steps for model development and property prediction through application of a property model, that may be used with embodiments of the present invention
  • FIG. 2 is a flowchart showing steps of predicting a property using a model-based soft analyzer, and updating the model in conjunction with a hard analyzer according to embodiment of the invention.
  • FIG. 3 is a flowchart showing steps of determining a property of a process stream according to an embodiment of the invention.
  • hard analyzer refers to an analyzer, including online or offline laboratory based devices, capable of substantially directly sensing and measuring a value for a particular property associated with a process stream and/or a sample of the process stream. These measured properties are referred to as "dependent variables.”
  • dependent variables may include concentrations of components in the sample and characteristics of the sample including "hot properties" such as flash point and distillation properties, or cold properties such as cloud point, freezing point, cold filter plugging point, and pour point of diesel.
  • hot properties such as flash point and distillation properties
  • cold properties such as cloud point, freezing point, cold filter plugging point, and pour point of diesel.
  • these analyzers include, respectively, the FPA-4, DPA-4, CPA-4, FRP-4, CFPP-4.2, and the PPA-4 available from Bartec Benke GmbH, Hamburg, Germany, while the GC8000 online GC by Yokogawa Electric
  • the terms "primary measurement”, “measurement”, “property values”, and “dependent variables” refer to the output of a hard analyzer. These measurements or dependent variables may be the result of a single analysis or the average of a plurality of analyses.
  • the terms “inferential analyzer” and “soft sensor” refer to a computer-implemented model-based analyzer that relies on various independent variables captured by sensors associated with a DCS (Distributed Control System), to infer values for various dependent variables of a process fluid.
  • DCS Distributed Control System
  • inferred dependent variables are also referred to as “predicted dependent variables", “inferential” or “model-based” measurements, “properties of interest”, or simply “predictions”.
  • An example of an “inferential analyzer” that may be used in embodiments of the present invention includes the ExaRQE Robust Quality EstimatorTM from Yokogawa Electric Corporation, Tokyo, Japan.
  • a "process stream” or “process fluid” is any stream or fluid associated with a process.
  • process streams include any intermediate process stream of a process within a process plant.
  • a "product stream” or “product fluid” is a process stream that is the output from an operating unit within a production process and may be the feed to another operating unit or may substantially become the basis for a product produced by a process plant.
  • a "product stream,” which is a more limiting case of a process stream may still encompass process streams that may be subjected to further modification by other processing units or blending within a tank or a blender, for example, to provide the desired properties of the finished product.
  • the methods and systems of the invention are intended to pertain to all process streams and all product streams, which may be a more limiting class of process streams. In certain embodiments, the methods and systems of the invention may be applied to the finished product itself.
  • the terms "computer” and "end-user device” are meant to encompass a workstation, personal computer, personal digital assistant (PDA), wireless telephone, or any other suitable computing device including a processor, a computer readable medium upon which computer readable program code (including instructions and/or data) may be disposed, and a user interface.
  • the term 'real time' refers to sensing and responding to external events nearly simultaneously (e.g., within milliseconds or microseconds) with their occurrence, or sufficiently fast to enable the device to keep up with an external process.
  • the frequency of "real-time" data is defined in terms relative, not absolute.
  • DCS distributed Control System
  • Centum VPTM commercially available from Yokogawa Electric Corporation, Tokyo, Japan.
  • the process streams may be
  • the systems and methods of the invention may be used to determine a property or properties of a product stream.
  • the systems and methods, according to certain embodiments of the invention provide measurements frequently and with a high degree of accuracy.
  • the inventor has conceived of novel and unique methods and systems to integrate statistical models and statistical analysis/analyses with process models and process operations for determining a property or properties of a process stream. Certain embodiments of the invention are directed to determining a property or properties of a hydrocarbon process stream. Yet other embodiments of the invention are directed to determining a property or properties of a hydrocarbon product stream. Still yet other embodiments of the invention are directed to determining a property or properties of a finished hydrocarbon product.
  • the inventor has conceived of a system and method for combining two approaches to process operations and control by integrating one or more analyzers or methods of analysis with a process model for inferential analysis that is particularly suited to such one or more analyzers.
  • the process model may be selected specifically based upon the types of analyzer or analyzers being used, the inferential predictions may be adapted or tuned on the basis of less frequent, discrete measurement results from the one or more analyzers.
  • the inventive system and method provides real-time prediction of results with a degree of accuracy that approaches that of the analytic measurements provided by the analyzer or analyzers whose measurement cycle times represent time frames that are substantially longer than real time.
  • the primary measurement is provided by at least one hard analyzer. In a particular embodiment of the invention, the primary
  • the system and method of the invention offers benefits associated with optimization of the associated chemical, petrochemical, and/or refinery process.
  • the number of primary measurements (dependent variables) provided by an online analyzer supports a methodology for adapting or tuning (i.e., adjusting) the predicted dependent variables provided by an inferential analyzer.
  • the adaptation or tuning procedure may be statistically improved relative to the conventional approaches by the relatively frequent capture and use of relatively numerous variables, for model adaptation or tuning.
  • adaption or tuning of the model using the procedures of the systems or methods of the invention is improved over the more conventional techniques of adjusting a bias or scaling factor that is applied to the estimated value.
  • the more frequent feedback, for greater numbers of variables, for example, provided by an online analyzer or online analyzers enhance statistical certainty of the correction of the model due to the larger number of primary test results available over a period of time.
  • a multiple number of measurement values returned by the online analyzer may be compared using, for example, statistical techniques to determine whether any number of the values should be discarded. I.e., if a value falls outside of a statistically defined range, the values may be discarded preventing incorrect adaptation or tuning of the model from being performed.
  • a statistical analysis or statistical diagnostic procedure is performed on the hard analyzer value and associated inferential value to determine the validity or invalidity of any one or more of the hard analyzer measurements and the inferential measurements.
  • the model adaptation or tuning procedure may be configured to wait for the return of a valid analyzer measurement or to use one or more previous valid analyzer measurements for model adaptation or tuning.
  • model adaptation or tuning may be performed to increase the accuracy of the inferential model.
  • a valid hard analyzer measurement or even a series of valid hard analyzer measurements may be held for subsequent use in model adaptation or tuning.
  • the procedure for using previous valid hard analyzer measurement or measurements for model adaptation and/or tuning will consider the age of such valid hard analyzer measurements and/or whether the operation of the process is compatible with the use of such values in adapting or tuning the model to provide improved inferential measurements
  • a validation procedure for the inferential measurement may involve a procedure as further described herein that uses the most recent inferential measurement(s). Further pursuant to this embodiment, the plurality of inferential measurement data points may have been collected over a period of at most about 2 minutes, at most about 5 minutes, at most about 10 minutes, at most about 12 minutes, at most about 15 minutes, at most about 20 minutes, at most about 24 minutes, at most about 30 minutes, at most about 40 minutes, at most about 45 minutes, at most about 60 minutes, at most about 90 minutes, at most about 120 minutes, at most about 150 minutes, at most about 3 hours, at most about 4 hours, at most about 5 hours, at most about 6 hours at most about 7 hours, at most about 8 hours, at most about 10 hours, at most about 12 hours, at most about 18 hours, at most bout 1 day, at most about 2 days, or at most about 1 week.
  • the substantially fixed, discrete interval for collecting the inferential measurement data points for use in the validation procedures may be about 1 second, about 2 seconds, about 5 seconds, about 6 seconds, about 10 seconds, about 12 seconds, about 15 seconds, about 20 seconds, about 24 seconds, about 30 seconds, about 40 seconds, about 45 seconds, about 1 minute, about 90 seconds, about 2 minutes, about 150 seconds, about 3 minutes, about 4 minutes, about 5 minutes about 6 minutes, about 10 minutes, about 12 minutes about 15 minutes, about 20 minutes, about 24 minutes, about 30 minutes, about 45 minutes, about 1 hour, or about 2 hours.
  • any one or more of the independent variables used to calculate an inferential measurement may be evaluated to determine a validity of the inferential measurement.
  • the evaluation of these independent variables may be used instead of or in addition to the most recent inferential measurement or a plurality of inferential measurement data points as further described herein.
  • the most recent value of any of these independent variables may be used or a plurality of measurement data points of any of these independent variables may be used in determining a validity of the inferential measurement.
  • a validation procedure for the hard analyzer measurement may involve a procedure as further described herein that uses the most recent hard analyzer measurement(s). Further pursuant to this embodiment, the plurality of inferential measurement data points may have been collected over a period of at most about 5 minutes, at most about 10 minutes, at most about 12 minutes, at most about 15 minutes, at most about 20 minutes, at most about 24 minutes, at most about 30 minutes, at most about 40 minutes, at most about 45 minutes, at most about 60 minutes, at most about 90 minutes, at most about 120 minutes, at most about 150 minutes, at most about 3 hours, at most about 4 hours, at most about 5 hours, at most about 6 hours at most about 7 hours, at most about 8 hours, at most about 10 hours, at most about 12 hours, at most about 18 hours, at most bout 1 day, at most about 2 days, or at most about 1 week.
  • the primary measurement may be available at about every week, at about every day, at about every 12-hour shift, at about every 8-hour shift, at about every 4 hours, at about every 2 hours, at about every 1 hour, at about every 45 minutes, at about every 30 minutes, at about every 20 minutes, at about every 10 minutes, at about every 6 minutes, at about every 5 minutes, at about every 4 minutes, at about every 3 minutes, at about every 2 minutes, at about every 1 minute, at about every 30 seconds, at about every 15 seconds, at about every 10 seconds, at about every 5 seconds, or at about every 1 second.
  • the primary measurement may be available at most about every week, at most about every day, at most about every 12-hour shift, at most about every 8-hour shift, at most about every 4 hours, at most about every 2 hours, at most about every 1 hour, at most about every 30 minutes, at most about every 20 minutes, at most about every 10 minutes, at most about every 6 minutes, at most about every 5 minutes, and at most about every 1 minute.
  • a plurality of hard analyzer measurement data points are collected, such data points may be collected on a substantially fixed, discrete interval.
  • the substantially fixed, discrete interval for collecting the inferential measurement data points for use in the validation procedures may be about 2 seconds, about 5 seconds, about 6 seconds, about 10 seconds, about 12 seconds, about 15 seconds, about 20 seconds, about 24 seconds, about 30 seconds, about 40 seconds, about 45 seconds, about 1 minute, about 90 seconds, about 2 minutes, about 150 seconds, about 3 minutes, about 4 minutes, about 5 minutes about 6 minutes, about 10 minutes, about 12 minutes about 15 minutes, about 20 minutes, about 24 minutes, about 30 minutes, about 45 minutes, about 1 hour, about 2 hours, about 4 hours, about 5 hours, about 6 hours, about 8 hours, about 12 hours, about 18 hours, about 1 day, or about 2 days.
  • validation of the inferential measurement is performed using the most recent value of the hard analyzer measurement and comprises a comparison of the inferential measurement, without adaptation or tuning, to the hard analyzer measurement.
  • the validation procedure comprises a simple evaluation of "closeness" of the inferential measurement to the hard analyzer measurement.
  • the validation procedure involves determination of the statistical deviation between the values and whether that deviation is statistical within the expected range of deviations that occur between these variables.
  • a population of sample data for either one of or both of the inferential measurement and the analytical measurement may be used in the validation procedure.
  • statistical values may be used in the validation procedure.
  • representative of the population e.g. , without limitation, median, mean, standard deviation, and/or variance— may be used in the validation procedure for the inferential measurement against the hard analyzer measurement.
  • the validation procedure of the inferential measurement and even for the hard analyzer measurement involves a multivariate procedure. Further pursuant to these embodiments of the invention, the inferential measurement and the hard analyzer measurement are each examined and validated substantially independent of the performance of each of these variables according to the procedures further provided herein.
  • an inferential model for predicting the 20 percent boiling temperature for diesel was developed using independent process variables, including two temperature variables, Ti and T 2 , from the process; two pressure variables, Pi and P 2 , measured from the process; and a flowrate, F, measured from the process, e.g., by sensors and supplied to DCS 320 (Fig. 1A).
  • Ti, T 2 , Pi, P 2 , and F are the independent variables supplied to inferential analyzer 312 (Fig. 1A) for the inferential model prediction and are considered the multivariable space for the inferential model.
  • These variables may become part of the multivariate statistical validation procedure for the inferential measurement.
  • these variables may be used by the inferential analyzer to generate various inferred dependent variables that may then be compared to the measurement results (dependent variables) generated by hard analyzer 310 (Fig. 1A).
  • These measurement results may be based on a single sample or series of samples over a period of time, as further described herein, may be used in determining the validity of the inferential measurement.
  • the validation procedure analyzes the values of the independent variables supplied by DCS 320 to consider whether the model used by inferential analyzer 312 has been developed for that sample space. If the model has not been developed for that space, then the validation procedure for the inferential measurement may indicate that the inferential values for these measurements or series of measurements are not valid, for example.
  • hard analyzer measurements may involve a multivariate data array suitable for analysis.
  • a SimDist analysis by gas chromatography provides a gas chromatogram that itself is a multivariable data array. Such information may be further analyzed, as further provided herein, in determining the validity of the hard analyzer measurement.
  • a hard analyzer may produce more than one hard analyzer measurements. Further pursuant to these embodiments, the hard analyzer measurements themselves may be considered a multivariate data array useful for determining the validity of any of these more than one hard analyzer measurements.
  • a statistical process control procedure is used to monitor the running changes of any one of or all of the hard analyzer measurement or measurements and the inferential measurement or measurements and to determine whether certain of these values should be identified as invalid.
  • other statistical procedures may be used to determine the validity of any one of or all of the hard analyzer measurement or measurements and the inferential measurement or measurements. For example, a statistical analysis may be used to establish an expected value for each measurement and based upon a comparison of the actual variability in the measure data versus the typically variability, a deviation from the expected means test may be used to determine whether a measurement is valid or not. Indeed, any statistical procedure known in the art may be used to determine the validity of any or even all of the measurement values.
  • a goodness of fit test procedure may be used using, for example, Chi Square test statistics to determine whether the inferential measurement should be accepted or rejected.
  • the hard analyzer measurement may similarly be subjected to a test procedure to determine whether a certain value should be accepted or rejected.
  • a message/messages and/or a signal/signals is/are generated when any one or both of the inferential measurement and/or the hard analyzer measurement is/are determined to be invalid.
  • the message or messages may be sent to an operator and/or to the DCS 320 for real time control of the process 10.
  • the signal or signals may be integrated into other routines in determining the status and validity of those routines.
  • the signal or signals may be used by a control procedure to determine whether control should continue on the hard analyzer measurement and/or the inferential
  • the measurement generated by the hard analyzer may be a special sample measurement that is used in addition to or instead of a routinely taken primary measurement.
  • the measurement generated by the hard analyzer may be available on discontinuous sampling intervals.
  • certain methods and systems of the invention may apply discretization techniques to process the discontinuously sampled measurements.
  • a simple interpolation is used to discretize
  • discontinuous sample intervals In yet certain other embodiments of the invention, a first order model may be used to discretize discontinuous sample intervals. In still yet certain other embodiments of the invention, a second order model may be used to discretize discontinuous sample intervals. In still yet certain other embodiments of the invention, a high order model may be used to discretize discontinuous sample intervals.
  • a system or method of the invention may use a technique to replace certain independent variables that are found to be invalid or unavailable.
  • a primary measurement may be unavailable for any number of reasons.
  • the primary measurement may be unavailable because the hard analyzer and/or instrument or instrumentation used to provide the measurement is not functioning correctly, not functioning at all, or being serviced, for example.
  • Any of the methods described herein for discretizing measurements over discontinuous sampling intervals may be used to replace certain primary measurements that are invalid or not available at a certain sampling interval. For example, any of a simple interpolation, a first order model, a second order model, or a higher order model may be used to replace certain primary measurements that are invalid or not available at a certain sampling interval.
  • the model used to determine inferential measurements may be tuned using any one or more of a procedure for adjusting bias, a procedure for adjusting a ratio multiplier, and a statistical procedure for model adaptation or tuning.
  • the model adaptation or tuning procedure may operate on one or more model parameters.
  • the model adaptation or tuning procedure may operate on one or more predicted values.
  • the model adaptation or tuning procedure may operate on any combination of one or more model parameters and one or more predicted values.
  • statistical validation techniques are used to determine validity of the primary measurement and the inferential measurement.
  • other validation techniques may be used to determine the validity of the primary measurement and/or the inferential measurement.
  • simple high and low limit checking may be used to determine validity of the primary measurement and/or the inferential measurement.
  • measurements used in the inferential estimation procedure may be used to determine the validity of the primary measurement and/or the inferential measurement.
  • a smart transmitter provides the status of any of the primary measurement and/or any of the independent measurements used in the inferential estimation procedure.
  • An embodiment of the invention is directed to determining ASTM properties of hydrocarbon streams.
  • ASTM properties include, for example, flash point, initial boiling point, ten (10) percent boiling point, twenty (20) percent boiling point, fifty (50) percent boiling point, seventy (70) percent boiling point, ninety (90) percent boiling point, and final point.
  • chromatographic properties typically provided by chromatographic analyzers.
  • a non- limiting example of such chromatographic properties includes any typically measured distillation properties.
  • chromatograph values determined in accordance with the systems and methods of the invention include concentrations of certain compounds in the process stream.
  • Such compounds in accordance with certain embodiments of the invention, may include paraffins, isoparaffins, olefins, naphthenes, and/or aromatics.
  • concentrations of all compounds of this exemplary embodiment are provided, the analysis is collectively referred to as "PIONA.”
  • concentrations of all compounds of this exemplary embodiment, except isoparaffins, are provided, the analysis is collectively referred to as "PONA.”
  • the invention may be directed to inferential analyzers using yet other primary test methods such as, for example, cold properties, which may include pour point and cloud point.
  • the invention may be directed to inferential analyzers using yet other primary test methods such as, for example, percent sulfur, initial boiling point(s) and final boiling point(s); flash point; total aromatics (or paraffins, isoparaffins, olefins, and/or naphthenes), for example, in naphtha feed to ethylene crackers.
  • an inferential analyzer of the invention may be applied to any primary test method whether such methods are known or later invented.
  • inferential predictions of dependent variables may be made using independent variables that include any one or more of temperature, pressure, or flow rate. Certain embodiments of the invention are directed to providing inferential predictions of dependent variables that may include any or any number of density, viscosity, refractive index, conductivity, pH, and any combinations thereof.
  • any variable that can be measured, whether now or later developed, may be used as an independent variable for determining inferential predictions.
  • the methods and systems of the invention may be applied to any type of process including, but not limited to, refinery processes, hydrocarbon processes, and chemical processes.
  • the methods and systems of the invention may be applied to continuous processes.
  • the methods and systems of the invention may be applied to batch processes.
  • the methods and systems of the invention may be applied to semi -batch processes.
  • the methods and systems of the invention may be applied to any combination of continuous processes, batch processes, and semi-batch processes.
  • An aspect of the invention provides systems and methods in which a hard analyzer may measure the independent variables, such as any one or multiple combinations of pressure(s), temperature(s), and flow rate(s) that are used in an inferential model whereby the inferential model generates a prediction(s) of the composition of the process fluid/sample.
  • a hard analyzer may measure the independent variables, such as any one or multiple combinations of pressure(s), temperature(s), and flow rate(s) that are used in an inferential model whereby the inferential model generates a prediction(s) of the composition of the process fluid/sample.
  • validation of both the independent variables and the composition of the samples may be provided, similar to any of the validation techniques as further described herein for inferential measurements and hard analyzer measurements.
  • validation of the predictions may be provided, similar to any of the validation techniques as further described herein for inferential measurements and hard analyzer measurements.
  • a model provides an inferential, modeled, referenced, and secondary measurement of the laboratory measurement.
  • the model may be adapted or tuned using the procedures of the invention, as further described herein, based upon other comparative measurements, for example.
  • An embodiment of the invention includes a system having a hard analyzer configured to generate one or more corresponding dependent variables of a process sample.
  • the system also includes an inferential analyzer configured to capture and use one or more independent variables, along with a model, to generate one or more inferential dependent variables that correspond to the dependent variables generated by the hard analyzer.
  • the system is also configured to generate a comparison of the measured dependent variables to the inferred dependent variables, and to use the comparison to adapt one or more parameters of the model.
  • discretization of the corresponding measurement may be performed according to the procedures provided herein.
  • an enhanced process control system 300 is configured for enhancing the operation and performance of an inferential analyzer by means of one or more hard analyzers.
  • System 300 includes a hard analyzer 310 that performs online capture and/or analysis of a sample of process stream 10.
  • the hard analyzer measures one or more dependent variables.
  • the sample is thus a representation of the process stream which is bounded in terms both physical and temporal, e.g. it is the physical subset of the process stream corresponding to a particular point and time in the process.
  • System 300 also includes an inferential analyzer 312 configured to receive various independent process variables, e.g., from a DCS (Distributed Control System) 320, and to use those process variables to predict the values of various properties (inferred dependent variables) of the process, using a model.
  • the DCS 320 is configured to capture various independent process variables from process 10, for use in control and operation of the process 10.
  • a CQM (Cognitive Quality Manager) 316 compares the dependent variables generated by the hard analyzer 310 with the inferred dependent variables generated by inferential analyzer 312, and uses the comparison to update the model used by the inferential analyzer 312 and/or adjust values for the inferred dependent variables which are then sent to a Process Optimizer 314.
  • the CQM is thus a processor-enabled module configured to use the comparison to validate the property predictions received from inferential analyzer 312, and to then offer options for adjusting those predictions and/or updating the inferential models.
  • the Process Optimizer 314 may receive predictions from the CQM 316 for one or more streams and stream properties, and then convert the validated, adjusted predictions into processing directives, such as adjusted process control inputs, which are sent to the DCS 320 for its control of the process 10. It should be recognized that the Process Optimizer 314 may receive inputs from sources other than the CQM 316, e.g. it may also receive and operate on independent process variables received directly from DCS 320.
  • APC Advanced Process Control
  • MPC Model Predictive Control
  • ExasmocTM Multivariable Optimizing Control commercially available from Yokogawa Electric Corporation, Tokyo, Japan.
  • the model may be generated in any conventional manner, e.g., based on (a) the application of first principles, (b) on a "black box” approach that is strictly empirical e.g. based on statistical or neural network algorithms, or (c) a combination of the first-principles and empirical approaches, and is an inferential model.
  • a representative example of model generation suitable for use in embodiments hereof, is shown and described hereinbelow with respect to Fig. IB.
  • Examples of the independent variables used in system 300 may include but are not limited to temperature, pressure, the rate of addition or consumption of a catalyst or reagent, flow, the input (output) of thermal energy into (from) the process or a point in the process, refractive index, conductivity, viscosity, opacity, pH, density, intensity responses measured by photometric or spectrometric devices at particular wavelengths or frequencies, and the like.
  • the independent process variables may be organized by plotting their values (y- values) according to kind (source/origin or type) arrayed across the x-axis to form a property spectrum (also referred to as a sample spectrum or spectra) 322, in which the independent variables plotted in a given property spectrum are related to each other temporally, e.g. they were captured substantially at the same time.
  • This sample spectrum 322 may then be used to update or generate the property model, so that the model is based on a population of sample spectra 322 and their corresponding property values (dependent variables).
  • the sample spectrum 322 may be supplied directly to the CQM 316 as shown at 323, to enable the CQM to receive the property spectrum 322 in an "un-processed" or unaltered state.
  • a direct channel 323 may be used, it should be recognized that spectrum 322 may also be supplied in its unaltered state to the CQM 316 through the inferential analyzer 312, e.g., by tunneling through the analyzer 312.
  • a particular population of known sample spectra may also be used as a calibration sample set for the inferential model.
  • the inferential analyzer 312 is thus configured to optionally predict a plurality of properties for a single stream or for a plurality of streams associated with said process, the predictions being based on one or more property spectra 322.
  • Exemplary methods used by process optimizer 314 to monitor/control/optimize process 10 may include one or more of: (a) multivariate method to determine whether the sample spectrum 322 belongs to the same population of sample spectra as a calibration sample set; (b) multivariate method to classify any spectral outlier according to cause; (c) replacing in the sample spectrum 322 of the false spectral outlier; (d) determining whether the outlier condition is due to a particular sensor by comparison to a baseline; (e) correcting the outlier condition by replacement with a model value; (f) comparing the inferred dependent variables with the dependent variables; (g) wherein the comparing (f) optionally includes averaging, regression, a damping function, or a vector-based algorithm, to provide a mathematical relationship between the inferred dependent variables and the dependent variables, which mathematical relationship is then applied to subsequent inferential predictions to produce a corrected inferential value; (h) estimating the lag time between values predicted by the inferential analyzer and those measured
  • the method (a) may include determining the validity of a property prediction supplied by inferential analyzer 312, on the basis of one or more metrics which may be obtained by a multivariate method, e.g. the application to the sample spectrum of a multivariate model such as a PLS, PCR, or Principal Component Analysis (PC A) model based on the calibration sample set, which method and metrics are known to those skilled in the art of chemometric analysis, are capable of identifying if sample spectrum belongs to the same population of sample spectra in the calibration sample set, and which include but are not limited to the Mahalanobis distance, which is an expression of a sample's spatial relationship to the calibration sample set, or the spectral residual, which is a measure of how well a multivariate model accounts for all independent variables in a sample spectrum.
  • a multivariate method e.g. the application to the sample spectrum of a multivariate model such as a PLS, PCR, or Principal Component Analysis (PC A) model based on the calibration sample set,
  • Method (b) may include further applying the multivariate method to classify any spectral outlier according to cause, e.g. whether an independent variable in the sample spectrum 322 is anomalous (e.g. because a sensor that is the source of an independent process variable is faulty), or alternatively if the sample spectrum 322 represents process conditions, sample chemistry, or a combination thereof which are distinctly different from those in the calibration sample set, these outliers being referred to, respectively, as false and true spectral outliers.
  • cause e.g. whether an independent variable in the sample spectrum 322 is anomalous (e.g. because a sensor that is the source of an independent process variable is faulty), or alternatively if the sample spectrum 322 represents process conditions, sample chemistry, or a combination thereof which are distinctly different from those in the calibration sample set, these outliers being referred to, respectively, as false and true spectral outliers.
  • Method (c) may include inferring and replacing in the sample spectrum 322 of the false spectral outlier a value for the faulty independent variable by methods known to those skilled in the art of multivariate modeling methods (chemometric s), e.g. a method based on Principle Component Analysis (PCA), to permit the reliable prediction of the dependent variable (the sample property) by the inferential analyzer.
  • chemometric s e.g. a method based on Principle Component Analysis (PCA)
  • Method (d) may include determining if the outlier condition is because the response function of a sensor that is the source of an independent process variable has changed compared with the response function for the same independent process variable when property spectra in the calibration sample set were recorded, where the change may be due to aging, fouling, calibration drift, or the replacement of a faulty sensor with a different sensor whose response function is not identical to that of the sensor used to generate property spectra in the calibration sample set, and where the change may be characterized by a mathematical function known to those skilled in the art, e.g. a simple linear equation involving slope and/or offset, or a second order equation, a multivariable function based on PCA, and the like.
  • a mathematical function known to those skilled in the art, e.g. a simple linear equation involving slope and/or offset, or a second order equation, a multivariable function based on PCA, and the like.
  • Method (e) may include correcting the changed sensor readings by determining the mathematical function on the basis of values that are inferred by means of the
  • Method (f) may include evaluating the performance of the inferential analyzer by the comparing of property values predicted by application of a property model to a sample spectrum with property values for corresponding samples measured by the hard analyzer 310, where the comparison is the application of suitable statistical methods known to those skilled in the art of statistical analysis, e.g. those based on the calculation of the difference between one predicted value and the corresponding measured value or between averages based on predicted and/or measured values obtained in a relatively short time interval. It is noted that the frequency of property predictions by the inferential analyzer is relatively high compared with that for measurements by hard analyzers, e.g. 6 min 1 to 60 min 1 versus 1 hr 1 to 20 hr 1 , respectively.
  • Method (g) may include optionally performing the comparison by relating a population of values (inferred dependent variables) predicted by the inferential analyzer 312 and a population of corresponding values (dependent variables) measured by the hard analyzer 310, where the populations correspond to predictions and measurements made in time frames ranging from that of the hard analyzer' s measurement cycle time to about a few hours or a few days, and where the relating is by performing on each of the populations a suitable calculation known to those skilled in the art such as one based on averaging, regression, a damping function, or a vector-based algorithm, and where the calculation yields separate results for the inferential and measured values, that for the former being
  • Method (h) may include dynamically estimating the lag time between values predicted by the inferential analyzer and those measured by a hard analyzer by methodologies known to those skilled in multivariate statistical methods, e.g. by minimizing variance in a plot of predicted values versus measured values by iteratively advancing the times for a population of measured values relative to the times of a population of predicted values.
  • the magnitude of the lag may vary as a function of a number of factors relating to the operation of the process, which, referring again to the example of CDU operation, may include changes in feed quality, feed rate, draw rates for the rundown streams, and the temperature profile of the CDU. Therefore, because lag time varies dynamically, a method for dynamically estimating the lag time may be beneficial.
  • Method (i) may include adjusting predictions from the inferential analyzer by the application of the mathematical relationship in enhancement (g).
  • Method (j) may include evaluating the difference between pairs of corresponding inferential and measured values, or between results from the aforementioned population- based comparison, to determine if a predicted value or a population of predicted values are valid, where the evaluation is the application of appropriate statistical techniques known to those skilled in the art of statistical analysis, e.g. Statistical Quality Control (SQC) methodologies, where valid predictions are those whose deviation from measured values or populations of measured values is within predetermined tolerance, e.g. the predictions are in statistical control, and where predictions that are not in statistical control are prediction outliers, which also are property outliers, and in some cases may be a concentration outlier.
  • SQC Statistical Quality Control
  • Method (k) may include updating the model used by inferential analyzer 312 with the addition of data that are relatively recent and therefore are representative of the current conditions, and also by elimination of older data that may be less representative of said conditions, where the model and/or the data include independent variables and/or measurements generated by the hard analyzer 310.
  • Method (1) may include updating the inferential model by creating and downloading into the inferential analyzer a model, which creation is based on a data set updated by the addition of new sample data for true spectral outliers and/or prediction outliers, and which model updating is performed to increase the dynamic range and/or robustness of predictions made the inferential analyzer by expanding the range of process conditions and/or chemistries represented in said data set used in model creation.
  • Method (m) may include optionally capturing in a sampling cylinder a process sample that has been identified as a true spectral outlier or a property outlier, where the capturing is accomplished by a sampling system interfaced to the process stream by appropriate means, e.g. a sample fast loop that continuously circulates material from the process stream, and which captured sample is then delivered to the local laboratory for the determination of the property value(s) of interest, which then are used to adjust the outputs of the inferential analyzer in that case in which desired property values are not available from hard analyzer 310.
  • a sampling system interfaced to the process stream by appropriate means, e.g. a sample fast loop that continuously circulates material from the process stream, and which captured sample is then delivered to the local laboratory for the determination of the property value(s) of interest, which then are used to adjust the outputs of the inferential analyzer in that case in which desired property values are not available from hard analyzer 310.
  • Method (n) may include performing model updates on the basis of schedule, e.g. about once every 10 days, or about once every 30 days, or about once every 60 days, or about once every 90 days, or on the basis of need, where the need is determined in consideration of factors including but not limited to the frequency and severity (magnitude) of true spectral outliers and/or prediction outliers, the changing of process operation or configuration in a way that results in a substantive change in property spectra, or the cumulative number of outlier samples since the previous model update, where the magnitude of true spectral outliers may be gauged by parameters know to those skilled in the art such as the Mahalanobis distance or spectral residuals.
  • the need is determined in consideration of factors including but not limited to the frequency and severity (magnitude) of true spectral outliers and/or prediction outliers, the changing of process operation or configuration in a way that results in a substantive change in property spectra, or the cumulative number of outlier samples since the previous model update, where the magnitude of true
  • methods (h)-(l) generally relate to enhancing the operation and performance of inferential analyzer 312, while methods (a) - (g) generally relate to the mathematical basis for those enhancements.
  • the hard analyzer 310 is an online gas chromatograph, e.g. a GC8000 manufactured by Yokogawa Electric Corporation, Tokyo, Japan, suitably configured to perform the simulated distillation (simdist) analysis of a single CDU rundown stream, e.g. diesel, in accordance with ASTM D2887, "Standard Test Method for Boiling Range Distribution of Petroleum Fractions by Gas Chromatography .”
  • the gas chromatograph is an ultrafast gas chromatograph (fast GC), e.g.
  • the Calidus micro GC Fecon Analytical, Lewisburg, WV
  • the Calidus micro GC capable of performing the simdist analysis of CDU rundown streams with a cycle time of about 3 minutes to 6 minutes per analysis in accordance with ASTM D7798, "Standard Test Method for Boiling Range Distribution of Petroleum Distillates with Final Boiling Points up to 538*C by Ultra Fast Gas Chromatography (UF GC)," where said simdist analysis obtains values for distillation yield temperatures such as T20 and T90, and where (i) the ultrafast GC performs the simdist analysis on each a plurality of CDU rundown streams, e.g.
  • simdist values for each of the streams are obtained as the result of a single chromatographic analysis, or as the average of results from replicate analyses performed in immediate succession; (iii) the values are used in the aforementioned enhancement of inferential predictions of distillation properties of the streams; (iv) the simdist values for each of the rundown streams are obtained sequentially; and (v) the enhanced inferential predictions are used for process control and optimization between updates of the same by the CQM engine.
  • the GC may be configured to implement the conventional ASTM D7798 simdist analysis and/or the UF GC analysis following ASTM D7798.
  • the analyzer may be a Fourier Transform Infrared (FTIR) analyzer, a Fourier Transform Near-Infrared (FTNIR) analyzer, a nuclear magnetic resonance (NMR) analyzer, a Raman analyzer, or any combination thereof.
  • the hard analyzer 310 may be a distillation analyzer configured to implement ASTM D86 "Standard Test Method for Distillation of Petroleum Products at Atmospheric Pressure.”
  • the dependent variables 40 may comprise any form of measurement detected by an analyzer.
  • the dependent variables may comprise an electrical signal having a particular voltage that is representative of a measurement, a digital value, spectra, and any combination thereof.
  • the gas chromatograph is configured to determine the properties, e.g. specifically, by means of a detailed hydrocarbon analysis (DHA), or generally by the obtaining and evaluation of a chromatogram of the hydrocarbon stream(s) using methods known to those skilled in the art; where the properties include those of gasoline, e.g. octane, percent total aromatics, percent benzene, percent olefins, distillation properties, Reid Vapor Pressure (RVP) and the like, or those of diesel, e.g. cetane number, distillation properties, polynuclear aromatics, polycyclic aromatics, and cold properties such as cloud point, pour point, cold filter plugging point, or those of naphtha, e.g.
  • DHA detailed hydrocarbon analysis
  • RVP Reid Vapor Pressure
  • the hard analyzer is optionally one or more online analyzer technologies capable of measuring properties including but not limited to the aforementioned properties of gasoline, diesel, or naphtha by means other than gas chromatography, where the application of the technologies in accordance with established, standardized test methods such as those promulgated by ASTM International may be considered a primary test method; and the qualification of a sample as an outlier is optionally based on a chemometric assessment of the chromatrogram, where the assessment is a multivariate analysis, e.g. one based on PLS, PCR, PCA, and the like.
  • the plurality of property predictions may be accomplished by means of a corresponding plurality of models (one model per property); or, a single model may be capable of predicting the plurality of properties (one model, many properties).
  • a single model may be capable of predicting the plurality of properties (one model, many properties).
  • distillation yield temperatures T20 and T90, i.e., the temperatures at which 20% and 90% of the components in a sample distill off the sample, is non-limiting.
  • any other distillation yield temperatures may be used, e.g. T50, Tss, T95 and T98.
  • distillation yield temperatures are but one expression of a sample's distillation properties, another being distillation yields determined for certain temperatures, e.g. the percent of a sample that distills by some certain temperature such a 200 °C or 300 °C, which sometimes are referred to as E200 and E300 in gasoline.
  • references to distillation yield temperature generally or specifically, e.g.
  • T20 and T90 are non-limiting, but are intended as examples of distillation properties that may be predicted by inferential analyzers or measured by a gas chromatograph.
  • simulated distillation does not apply only to the determination of distillation yield temperatures, but simply denotes the method for determining distillation properties of a hydrocarbon sample by gas chromatrography.
  • references to distillation properties are non-limiting, and that the present invention may be applied beneficially to enhance the inferential prediction of other properties, which may include so-called cold properties such as pour point, cloud point, and cold filter plugging point of diesel.
  • a multivariate mathematical model relates the response matrix to the desired property value that generally can be determined in the laboratory in accordance with a primary test method defined by an organization or authority recognized by the industry.
  • Multivariate statistical modeling techniques common algorithms include Partial Least Squares (PLS) and Principal Component Regression (PCR)) extract information distributed across the entire spectrum for correlation with primary test method results.
  • PLS Partial Least Squares
  • PCR Principal Component Regression
  • Fig. 2 is a flowchart showing the steps of predicting a property using a model- based soft sensor or model-based hard sensor according to embodiment of the invention.
  • Discrete samples 20 taken from the process 10 are measured by a hard analyzer 310, to yield dependent variables 160.
  • independent variables 30 are captured by sensors associated with DCS 320, from the process 10 at a given time t.
  • the inferential model 80 for the "soft sensor” calculates the predicted values (the inferred dependent variables) 180, which is one or more inferential measurements of the process 10.
  • the values for the dependent variables 160 are compared with the values for the inferred dependent variables 180.
  • the result of the comparison may be an adjusted inferential prediction and/or an adjustment to the model which is then sent to update the model 80 at 194, and/or to control the process 10 at 196, e.g., via DCS 320.
  • the comparison result may simply be an updated inferential model 80', which may sent at 194 to replace the model 80.
  • the analyzer may include a procedure to validate the one or more object variables according to the procedures further discussed herein.
  • Embodiments of the invention are also directed to a procedure to validate the one or more analyzer measurements.
  • the process model or process models may be updated based upon, for example, feedback from a primary measurement.
  • the process model(s) is automatically updated using laboratory data.
  • the process model(s) is automatically updated using measurements provided by one or more online analyzers.
  • the process model(s) is automatically updated using any combination of laboratory data and measurements provided by one or more online analyzers.
  • any one of or any combination of single factor correlation analysis and multivariate regression analysis may be used to update the process model or correction biases or ratios associated with the estimation of the inferential measurements.
  • principal factors that control the calculation of specific inferential measurements may be identified and a single factor correlation analysis may be performed on any of these one or more variables.
  • a quantitative model is established between the independent variables and inferential measurements and the model may be updated using a multivariate regression analysis.
  • a combination of single factor correlation analyses and multivariable regression analysis may be performed to adapt and/or tune the model used to determine one or more inferential measurements.
  • FIG. 3 is a flowchart showing the steps of determining a property of a process stream according to an embodiment of the invention.
  • the method for determining a property of a process stream 200 comprises calculating an inferential measurement for the property using a model 210, receiving an analyzer measurement of the property 220, evaluating the validity of the inferential measurement 230, evaluating the validity of the analyzer measurement 240, and adjusting the model and/or adjusting the inferential measurement at 250.
  • the method of determining a property of a process stream may comprise the step of adjusting the model using one or more prior valid analyzer measurements when the inferential measurement is invalid and the analyzer measurement is invalid. Further pursuant to this embodiment, the method of determining a property of a process stream may comprise determining whether to use the one or more prior valid analyzer measurements to adjust the model based upon at least one of the age of the one or more prior valid analyzer measurements and the current state of the process in comparison to the state of the process at the time the one or more prior valid analyzer measurements were measured. Moreover, in any of the various embodiments shown and described herein, when both the inferential predictions and the analyzer measurements are determined to be valid, the latter may be used to tune (adjust) the former.
  • Various embodiments may also discretize the analyzer measurement using, for example, any of the procedures further described herein.
  • any of the one or more prior valid analyzer measurements may be discarded and not used to adapt or tune the model.
  • any of the one or more prior valid analyzer measurements may likewise be discarded.
  • Table 1 analyzes the effectiveness of the systems and methods of the invention over the more conventional techniques using off-line measurements for process operation and control.
  • the standard error of prediction is a measure of the extent of uncertainty associated with the distillation temperatures for the 20 percent (T 2 o) and 90 percent (T90) recoveries.
  • the distillation temperatures are typically measured in the laboratory by the standard ASTM D-86 method or by a simdist method, e.g. the ASTM D-2887 or D-7798 methods while the latter can also be implemented online using a gas chromatograph, for example.
  • the mere increase in frequency of measurements results in a reduction in the standard error of prediction and an improvement in process operations.
  • the frequency provided by on-line analyzers is generally favored over laboratory analysis if the rate of return for the improvement in process operations justifies the installation and support of the on-line analyzer.
  • the standard error of prediction is always improved by integrating the analyzer with an inferential analyzer using a process model that is suited to the type of analysis being performed.
  • a conventional strategy for reducing disruptions associated with crude switches involves temporarily drawing down or reducing the crude feed rate during the change in feedstocks, and gradually increasing the rate back to a target rate.
  • the availability of property predictions that have both high frequency and high fidelity by Method 6 yield an estimated reduction in throughput losses of approximately $1,280,000 per year per 100,000 barrels per day crude distillation capacity.
  • the integrated analyzer system and method of the invention particularly leveraged by the use of on-line analyzers to enhance inferential predictions, allows adjustment of the latter on time scales sufficiently short to mitigate the impact of crude switchovers in terms of their duration and the disruption they cause to the operation of the CDU and therefore the refinery, and in the immediate example enables the maximizing of diesel production.
  • models not only include those based on multivariate statistical methods, e.g. PCA (principal component analysis), PLS (partial least squares), and MLR (multiple linear regression), but also include RBF (radial basis function), neural networks, models based on first principles, and combinations thereof, along with substantially any model types that may be developed in the future.
  • PCA principal component analysis
  • PLS partial least squares
  • MLR multiple linear regression
  • RBF radial basis function
  • neural networks models based on first principles, and combinations thereof, along with substantially any model types that may be developed in the future.

Abstract

Systems, methods, and apparatuses are provided for determining properties of process streams, in particular, hydrocarbon processing streams. The systems, methods, and apparatuses frequently, for example, substantially in real-time, determine measurements for the properties of the process stream. The systems, methods, and apparatuses provide features that allow such properties of process streams to be accurately measured even as process conditions and other parameters that affect process operations change. More specifically, an analyzer having a measurement device configured to detect one or more independent variables of a process stream, a model configured to determine one or more analyzer measurements from the one or more independent variables, and a procedure to adjust the model using a corresponding primary measurement is disclosed.

Description

SYSTEM, METHOD AND APPARATUS FOR DETERMINING PROPERTIES OF PRODUCT OR PROCESS STREAMS
RELATED APPLICATION
[0001] This application claims priority to U.S. Non-Provisional Patent Application Ser. No.14/173301 entitled System, Method and Apparatus for Determining Properties of Product or Process Streams, filed on 2/05/2014, which claims the benefit of U.S. Provisional Patent Application Serial No. 61/760,879, entitled System, Method and Apparatus for Determining Properties of Product or Process Streams, filed on 2/05/2013, the contents of which are incorporated herein by reference in their entirety for all purposes.
TECHNOLOGICAL FIELD
[0002] The present invention generally relates to a system, method, and apparatus of estimating properties of hydrocarbon streams. In particular, the systems, methods, and apparatuses of the present invention relates to a measuring properties of hydrocarbon streams with high frequency and a high degree of accuracy.
BACKGROUND
[0003] The operations of process plants have been leveraged by the availability of analytical methods, for example, through the use of laboratory measurements or online analyzers. These types of results are valued by plant operations because they are typically regarded as reliable. For example, with respect to hydrocarbon and/or refining operations, primary analytical test methods provide a critical basis for custody transfer of products whose properties have been ascertained in accordance with industry standard test methods such as those developed and promulgated by ASTM International.
[0004] Notwithstanding the importance of these primary test methods, they do suffer from certain inadequacies. For example, laboratory measurements typically may be available only once or several times per day. Furthermore, several hours can elapse between the obtaining of a discrete sample and the reporting of results from tests performed on it, severely limiting the possibility to control the process on the basis of those results. Additionally, sample stability, sample contamination, issues of representative sampling, and uncertainty associated with the execution of test procedures may result in erroneous sample values being returned.
[0005] Improving the availability, integrity, and reproducibility of test data has in many cases motivated the on-line deployment of measurements. However, depending upon the type of measurement and analysis being performed, the cycle time for online analyzers may be on the order of ten or more minutes and, in some cases, up to and over one hour, which, in some cases, may still be inadequate for purposes of maximizing process efficiency or product quality.
[0006] Process industries have conventionally responded to the time delay issues and reliability of primary measurements by instituting secondary measurement techniques capable of predicting properties of certain process streams. Such secondary techniques commonly have included the use of some form of model, for example, multivariate statistical models capable of predicting certain properties of interest using process inputs, in which the properties of interest may be termed "dependent variables" and the process inputs may be termed "independent variables."
[0007] An important class of these model-based approaches is "Inferential analyzers," also referred to as "soft sensors" because they typically reside in software. Soft sensors are appealing for at least two reasons. First, many times they do not require the installation of additional sensors in the process unit because they typically rely upon measurements such as temperature, pressure, and flow rate, which may already be available. Second, with the advent of distributed control systems, the input measurements typically relied upon by soft sensors are substantially available in real-time, having discrete sampling rates of one second or even less. These advantages at least address the disadvantage of time delay associated with primary measurements by providing property predictions at intervals that are greater than those typically required by process control systems. Additionally, they obviate the need to physically obtain a sample, eliminating the issues of representative sampling and sample integrity.
[0008] While the fidelity of these models may be quite good over limited time periods ranging from a few hours to even perhaps a few days, conventional inferential analyzers tend to be insufficiently robust because in aggregate the independent variables that serve as inputs into the model typically relate to the chemistry of the process stream both indirectly and incompletely. They are indirect expressions of the chemistry to the extent that the readings of sensors on the process are functions of both process conditions and material in the process; they are incomplete insofar as the number of independent variables used in the models is fewer than the degrees of freedom in the system, which relate to both the process system and the material being processed through it. However, an exception may occur when steady-state or quasi- steady- state conditions prevail and many process and stream variables are nominally constant, e.g. when feed quality and the operation of the process system are substantially invariant. At such times, the independent variables may "determine" stream chemistry in the mathematical sense, and property predictions by an inferential analyzer may be extremely reliable. Yet, a fundamental issue is that models generally are correlative, and because correlation does not necessarily denote cause, inferential models may be largely empirical, with first principles having only distant influence. Indeed, the literature freely refers to the modeling approach that is perhaps most common as a "black box method." In summary, property predictions by inferential analyzers are labile to the extent that the effect (a predicted value) is removed from the primary cause (a stream property that ultimately is determined by sample chemistry).
[0009] The common practice therefore is to use periodic laboratory measurements to update the outputs from inferential analyzers to deal with a variety of variables. This strategy may improve the quality of inferential predictions across (i) the full range of possible feed qualities; (ii) changes in the condition of the process system, e.g. fouling; and (iii) changes in the response of sensors whose readings are the independent variables, e.g. those resulting from simple drift or from replacement of a faulty sensor with a new one. Nevertheless, issues attach to this approach which limit the possibilities for optimizing the performance of continuous processes, which may include maximizing throughput, adhering closely to product quality targets, minimizing energy usage, extending catalyst service life, and the like. A need remains in the art for improved model adaptation procedures, in particular, to accommodate variations in operating conditions resulting from, for example, changes in composition and properties of the feed; other process unit changes not within the scope of the model; sensor failure and/or discrepancy in process measurements; frequency of availability of laboratory analysis and/or measurements from analytical instruments for updating the model; and analyzer reliability.
[0010] Even small deviations in model accuracy can significantly impact the economies of production for large scale processes similar to those process units found in petroleum refineries and petrochemical plants. There remains a need in the art for improved methods of measuring properties of hydrocarbon streams, in particular, frequently, preferably on-line in substantially real-time. Furthermore, there remains a need in the art for improved methods of measuring properties of hydrocarbon streams with a high degree of accuracy. BRIEF SUMMARY OF EXEMPLARY EMBODIMENTS
[0011] An aspect of the invention includes a system for determining a property of a process stream. The system includes a hard analyzer configured to measure one or more dependent variables associated with a sample of the process stream, where a dependent variable is also a property of the sample. An inferential analyzer includes a model of aspects of the process, and is configured to capture one or more independent process variables associated with the process stream and to use the independent process variables to generate one or more inferred dependent variables. A comparison module generates a comparison of the dependent variables measured by the hard analyzer and the dependent variables inferred by the inferential analyzer. The system uses the comparison to adjust one or more aspects of the model and/or adjust one or more inferred dependent variables and/or control the process stream.
[0012] Another aspect of the invention includes a method for determining a property of a process stream using the aforementioned system.
[0013] These embodiments of the invention and other aspects and embodiments of the invention will become apparent upon review of the following description taken in conjunction with the accompanying drawings. The invention, though, is pointed out with particularity by the appended claims.
BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWING
[0014] Having thus described the invention in general terms, reference will now be made to the accompanying drawing, and wherein:
[0015] Fig. 1A is a functional block diagram of an embodiment of the present invention;
[0016] Fig. IB is a flowchart showing steps for model development and property prediction through application of a property model, that may be used with embodiments of the present invention;
[0017] Fig. 2 is a flowchart showing steps of predicting a property using a model-based soft analyzer, and updating the model in conjunction with a hard analyzer according to embodiment of the invention; and
[0018] Fig. 3 is a flowchart showing steps of determining a property of a process stream according to an embodiment of the invention. DETAILED DESCRIPTION
[0019] Some embodiments of the present invention will now be described more fully hereinafter with reference to the accompanying drawings, in which some, but not all embodiments of the invention are shown. Indeed, various embodiments of the invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will satisfy applicable legal requirements.
[0020] As used in the specification and in the appended claims, the singular forms "a", "an", and "the" include plural referents unless the context clearly indicates otherwise. For example, reference to "an analyzer" includes a plurality of such analyzers. In another example, reference to "an analysis" includes a plurality of such analyses.
[0021] Although specific terms are employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation. All terms, including technical and scientific terms, as used herein, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs unless a term has been otherwise defined. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning as commonly understood by a person having ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and the present disclosure. Such commonly used terms will not be interpreted in an idealized or overly formal sense unless the disclosure herein expressly so defines otherwise.
[0022] As used herein, the term "hard analyzer" refers to an analyzer, including online or offline laboratory based devices, capable of substantially directly sensing and measuring a value for a particular property associated with a process stream and/or a sample of the process stream. These measured properties are referred to as "dependent variables."
Examples of dependent variables may include concentrations of components in the sample and characteristics of the sample including "hot properties" such as flash point and distillation properties, or cold properties such as cloud point, freezing point, cold filter plugging point, and pour point of diesel. Examples of these analyzers include, respectively, the FPA-4, DPA-4, CPA-4, FRP-4, CFPP-4.2, and the PPA-4 available from Bartec Benke GmbH, Hamburg, Germany, while the GC8000 online GC by Yokogawa Electric
Corporation may be used to perform a simulated distillation (simdist) analysis. The terms "primary measurement", "measurement", "property values", and "dependent variables" refer to the output of a hard analyzer. These measurements or dependent variables may be the result of a single analysis or the average of a plurality of analyses. As used herein, the terms "inferential analyzer" and "soft sensor" refer to a computer-implemented model-based analyzer that relies on various independent variables captured by sensors associated with a DCS (Distributed Control System), to infer values for various dependent variables of a process fluid. These inferred characteristics are also referred to as "predicted dependent variables", "inferential" or "model-based" measurements, "properties of interest", or simply "predictions". An example of an "inferential analyzer" that may be used in embodiments of the present invention includes the ExaRQE Robust Quality Estimator™ from Yokogawa Electric Corporation, Tokyo, Japan.
[0023] As used herein, a "process stream" or "process fluid" is any stream or fluid associated with a process. In non-limiting examples, process streams include any intermediate process stream of a process within a process plant. As used herein, a "product stream" or "product fluid" is a process stream that is the output from an operating unit within a production process and may be the feed to another operating unit or may substantially become the basis for a product produced by a process plant. However, a "product stream," which is a more limiting case of a process stream may still encompass process streams that may be subjected to further modification by other processing units or blending within a tank or a blender, for example, to provide the desired properties of the finished product. The methods and systems of the invention are intended to pertain to all process streams and all product streams, which may be a more limiting class of process streams. In certain embodiments, the methods and systems of the invention may be applied to the finished product itself.
[0024] The terms "computer" and "end-user device" are meant to encompass a workstation, personal computer, personal digital assistant (PDA), wireless telephone, or any other suitable computing device including a processor, a computer readable medium upon which computer readable program code (including instructions and/or data) may be disposed, and a user interface. The term 'real time' refers to sensing and responding to external events nearly simultaneously (e.g., within milliseconds or microseconds) with their occurrence, or sufficiently fast to enable the device to keep up with an external process. In the context of process control, the frequency of "real-time" data is defined in terms relative, not absolute. Thus, for a process in which significant changes can occur in time frames on the order of a few minutes, new analytical results that are available every half minute are, from the viewpoint of process control, real-time. So is a measurement time interval of 5 minutes when significant process variations occur in time frames longer than 10-15 minutes. For the purpose of optimizing operation of a refinery CDU (Crude Distillation Unit), an analysis results every 1-2 minutes are real time.
[0025] The terms "DCS" and "Distributed Control System" refer to a computer- implemented process control system such as the Centum VP™ commercially available from Yokogawa Electric Corporation, Tokyo, Japan.
[0026] The availability of information substantially in real-time for properties of process streams, such as continuous process streams in chemical and petrochemical processing, hydrocarbon processing, and refining operations may many times be important to maximizing throughput, controlling product quality, and/or ensuring that operations of the process plant are environmentally sound and safe.
[0027] The inventor has conceived of systems and methods for determining properties of process streams. In an embodiment of the invention, the process streams may be
hydrocarbon process streams. In certain embodiments of the invention, the systems and methods of the invention may be used to determine a property or properties of a product stream. The systems and methods, according to certain embodiments of the invention, provide measurements frequently and with a high degree of accuracy.
[0028] The inventor has conceived of novel and unique methods and systems to integrate statistical models and statistical analysis/analyses with process models and process operations for determining a property or properties of a process stream. Certain embodiments of the invention are directed to determining a property or properties of a hydrocarbon process stream. Yet other embodiments of the invention are directed to determining a property or properties of a hydrocarbon product stream. Still yet other embodiments of the invention are directed to determining a property or properties of a finished hydrocarbon product.
[0029] In particular, the inventor has conceived of a system and method for combining two approaches to process operations and control by integrating one or more analyzers or methods of analysis with a process model for inferential analysis that is particularly suited to such one or more analyzers. Additionally, because, in certain embodiments of the invention, the process model may be selected specifically based upon the types of analyzer or analyzers being used, the inferential predictions may be adapted or tuned on the basis of less frequent, discrete measurement results from the one or more analyzers. Thus, the inventive system and method provides real-time prediction of results with a degree of accuracy that approaches that of the analytic measurements provided by the analyzer or analyzers whose measurement cycle times represent time frames that are substantially longer than real time. [0030] In an embodiment of the invention, the primary measurement is provided by at least one hard analyzer. In a particular embodiment of the invention, the primary
measurement is provided by at least one online hard analyzer, such as a GC. Without intending to be limiting, the system and method of the invention offers benefits associated with optimization of the associated chemical, petrochemical, and/or refinery process. For example, the number of primary measurements (dependent variables) provided by an online analyzer supports a methodology for adapting or tuning (i.e., adjusting) the predicted dependent variables provided by an inferential analyzer. According to certain embodiments of the invention, the adaptation or tuning procedure may be statistically improved relative to the conventional approaches by the relatively frequent capture and use of relatively numerous variables, for model adaptation or tuning.
[0031] Additionally, adaption or tuning of the model using the procedures of the systems or methods of the invention is improved over the more conventional techniques of adjusting a bias or scaling factor that is applied to the estimated value. The more frequent feedback, for greater numbers of variables, for example, provided by an online analyzer or online analyzers enhance statistical certainty of the correction of the model due to the larger number of primary test results available over a period of time.
[0032] In addition, a multiple number of measurement values returned by the online analyzer may be compared using, for example, statistical techniques to determine whether any number of the values should be discarded. I.e., if a value falls outside of a statistically defined range, the values may be discarded preventing incorrect adaptation or tuning of the model from being performed.
[0033] In an embodiment of the invention, a statistical analysis or statistical diagnostic procedure is performed on the hard analyzer value and associated inferential value to determine the validity or invalidity of any one or more of the hard analyzer measurements and the inferential measurements. In some embodiments of the invention, when the hard analyzer measurement is determined to be invalid, then the model for the inferential measurement will not be updated using the hard analyzer measurement. In certain embodiments of the invention, the model adaptation or tuning procedure may be configured to wait for the return of a valid analyzer measurement or to use one or more previous valid analyzer measurements for model adaptation or tuning.
[0034] With respect to the inferential measurement, it may be more appropriate to say that evaluating the validity of the inferential measurement is a validation of such inferential measurement. Under normal circumstances, according to an embodiment of the invention, model adaptation or tuning may be performed to increase the accuracy of the inferential model.
[0035] In certain embodiments of the invention, a valid hard analyzer measurement or even a series of valid hard analyzer measurements may be held for subsequent use in model adaptation or tuning. In certain embodiments of the invention, the procedure for using previous valid hard analyzer measurement or measurements for model adaptation and/or tuning will consider the age of such valid hard analyzer measurements and/or whether the operation of the process is compatible with the use of such values in adapting or tuning the model to provide improved inferential measurements
[0036] In an embodiment of the invention, a validation procedure for the inferential measurement may involve a procedure as further described herein that uses the most recent inferential measurement(s). Further pursuant to this embodiment, the plurality of inferential measurement data points may have been collected over a period of at most about 2 minutes, at most about 5 minutes, at most about 10 minutes, at most about 12 minutes, at most about 15 minutes, at most about 20 minutes, at most about 24 minutes, at most about 30 minutes, at most about 40 minutes, at most about 45 minutes, at most about 60 minutes, at most about 90 minutes, at most about 120 minutes, at most about 150 minutes, at most about 3 hours, at most about 4 hours, at most about 5 hours, at most about 6 hours at most about 7 hours, at most about 8 hours, at most about 10 hours, at most about 12 hours, at most about 18 hours, at most bout 1 day, at most about 2 days, or at most about 1 week.
[0037] Even further pursuant to the embodiment where a plurality of inferential measurement data points are collected, such data points may be collected on a substantially fixed, discrete interval. In certain embodiments of the invention, the substantially fixed, discrete interval for collecting the inferential measurement data points for use in the validation procedures may be about 1 second, about 2 seconds, about 5 seconds, about 6 seconds, about 10 seconds, about 12 seconds, about 15 seconds, about 20 seconds, about 24 seconds, about 30 seconds, about 40 seconds, about 45 seconds, about 1 minute, about 90 seconds, about 2 minutes, about 150 seconds, about 3 minutes, about 4 minutes, about 5 minutes about 6 minutes, about 10 minutes, about 12 minutes about 15 minutes, about 20 minutes, about 24 minutes, about 30 minutes, about 45 minutes, about 1 hour, or about 2 hours.
[0038] In certain embodiments of the invention, any one or more of the independent variables used to calculate an inferential measurement may be evaluated to determine a validity of the inferential measurement. For example, the evaluation of these independent variables may be used instead of or in addition to the most recent inferential measurement or a plurality of inferential measurement data points as further described herein. Similarly the most recent value of any of these independent variables may be used or a plurality of measurement data points of any of these independent variables may be used in determining a validity of the inferential measurement. Pursuant to the embodiments of the invention involving a plurality of measurement data points of any of these one or more independent variables such plurality of measurement data points may be collected over a period and a substantially fixed, discrete time interval similar to the range of periods and fixed discrete time intervals further described herein for the inferential measurement.
[0039] In an embodiment of the invention, a validation procedure for the hard analyzer measurement may involve a procedure as further described herein that uses the most recent hard analyzer measurement(s). Further pursuant to this embodiment, the plurality of inferential measurement data points may have been collected over a period of at most about 5 minutes, at most about 10 minutes, at most about 12 minutes, at most about 15 minutes, at most about 20 minutes, at most about 24 minutes, at most about 30 minutes, at most about 40 minutes, at most about 45 minutes, at most about 60 minutes, at most about 90 minutes, at most about 120 minutes, at most about 150 minutes, at most about 3 hours, at most about 4 hours, at most about 5 hours, at most about 6 hours at most about 7 hours, at most about 8 hours, at most about 10 hours, at most about 12 hours, at most about 18 hours, at most bout 1 day, at most about 2 days, or at most about 1 week.
[0040] In an embodiment of the invention, the primary measurement may be available at about every week, at about every day, at about every 12-hour shift, at about every 8-hour shift, at about every 4 hours, at about every 2 hours, at about every 1 hour, at about every 45 minutes, at about every 30 minutes, at about every 20 minutes, at about every 10 minutes, at about every 6 minutes, at about every 5 minutes, at about every 4 minutes, at about every 3 minutes, at about every 2 minutes, at about every 1 minute, at about every 30 seconds, at about every 15 seconds, at about every 10 seconds, at about every 5 seconds, or at about every 1 second.
[0041] In an embodiment of the invention, the primary measurement may be available at most about every week, at most about every day, at most about every 12-hour shift, at most about every 8-hour shift, at most about every 4 hours, at most about every 2 hours, at most about every 1 hour, at most about every 30 minutes, at most about every 20 minutes, at most about every 10 minutes, at most about every 6 minutes, at most about every 5 minutes, and at most about every 1 minute. [0042] Even further pursuant to the embodiment wherein a plurality of hard analyzer measurement data points are collected, such data points may be collected on a substantially fixed, discrete interval. In certain embodiments of the invention, the substantially fixed, discrete interval for collecting the inferential measurement data points for use in the validation procedures may be about 2 seconds, about 5 seconds, about 6 seconds, about 10 seconds, about 12 seconds, about 15 seconds, about 20 seconds, about 24 seconds, about 30 seconds, about 40 seconds, about 45 seconds, about 1 minute, about 90 seconds, about 2 minutes, about 150 seconds, about 3 minutes, about 4 minutes, about 5 minutes about 6 minutes, about 10 minutes, about 12 minutes about 15 minutes, about 20 minutes, about 24 minutes, about 30 minutes, about 45 minutes, about 1 hour, about 2 hours, about 4 hours, about 5 hours, about 6 hours, about 8 hours, about 12 hours, about 18 hours, about 1 day, or about 2 days.
[0043] In one embodiment of the invention, validation of the inferential measurement is performed using the most recent value of the hard analyzer measurement and comprises a comparison of the inferential measurement, without adaptation or tuning, to the hard analyzer measurement. In an embodiment of the invention, the validation procedure comprises a simple evaluation of "closeness" of the inferential measurement to the hard analyzer measurement. In certain embodiments of the invention, the validation procedure involves determination of the statistical deviation between the values and whether that deviation is statistical within the expected range of deviations that occur between these variables.
[0044] In another embodiment of the invention, a population of sample data for either one of or both of the inferential measurement and the analytical measurement may be used in the validation procedure. In certain embodiments of the invention, statistical values
representative of the population— e.g. , without limitation, median, mean, standard deviation, and/or variance— may be used in the validation procedure for the inferential measurement against the hard analyzer measurement.
[0045] In other embodiments of the invention, the validation procedure of the inferential measurement and even for the hard analyzer measurement involves a multivariate procedure. Further pursuant to these embodiments of the invention, the inferential measurement and the hard analyzer measurement are each examined and validated substantially independent of the performance of each of these variables according to the procedures further provided herein.
[0046] In an exemplary embodiment, provided merely for purposes of illustration without any intention of being limiting, an inferential model for predicting the 20 percent boiling temperature for diesel was developed using independent process variables, including two temperature variables, Ti and T2, from the process; two pressure variables, Pi and P2, measured from the process; and a flowrate, F, measured from the process, e.g., by sensors and supplied to DCS 320 (Fig. 1A). Thus, Ti, T2, Pi, P2, and F are the independent variables supplied to inferential analyzer 312 (Fig. 1A) for the inferential model prediction and are considered the multivariable space for the inferential model. These variables may become part of the multivariate statistical validation procedure for the inferential measurement. For example, these variables may be used by the inferential analyzer to generate various inferred dependent variables that may then be compared to the measurement results (dependent variables) generated by hard analyzer 310 (Fig. 1A). These measurement results, which may be based on a single sample or series of samples over a period of time, as further described herein, may be used in determining the validity of the inferential measurement. In more specific embodiments of the invention, the validation procedure analyzes the values of the independent variables supplied by DCS 320 to consider whether the model used by inferential analyzer 312 has been developed for that sample space. If the model has not been developed for that space, then the validation procedure for the inferential measurement may indicate that the inferential values for these measurements or series of measurements are not valid, for example.
[0047] It should be noted that hard analyzer measurements may involve a multivariate data array suitable for analysis. In a non-limiting example, a SimDist analysis by gas chromatography provides a gas chromatogram that itself is a multivariable data array. Such information may be further analyzed, as further provided herein, in determining the validity of the hard analyzer measurement. In yet other embodiments of the invention, a hard analyzer may produce more than one hard analyzer measurements. Further pursuant to these embodiments, the hard analyzer measurements themselves may be considered a multivariate data array useful for determining the validity of any of these more than one hard analyzer measurements.
[0048] In certain embodiments of the invention, a statistical process control procedure is used to monitor the running changes of any one of or all of the hard analyzer measurement or measurements and the inferential measurement or measurements and to determine whether certain of these values should be identified as invalid.
[0049] In certain other embodiments of the invention, other statistical procedures may be used to determine the validity of any one of or all of the hard analyzer measurement or measurements and the inferential measurement or measurements. For example, a statistical analysis may be used to establish an expected value for each measurement and based upon a comparison of the actual variability in the measure data versus the typically variability, a deviation from the expected means test may be used to determine whether a measurement is valid or not. Indeed, any statistical procedure known in the art may be used to determine the validity of any or even all of the measurement values.
[0050] In yet other embodiments of the invention, a goodness of fit test procedure may be used using, for example, Chi Square test statistics to determine whether the inferential measurement should be accepted or rejected. The hard analyzer measurement may similarly be subjected to a test procedure to determine whether a certain value should be accepted or rejected.
[0051] In an embodiment of the invention, a message/messages and/or a signal/signals is/are generated when any one or both of the inferential measurement and/or the hard analyzer measurement is/are determined to be invalid. For example the message or messages may be sent to an operator and/or to the DCS 320 for real time control of the process 10. In certain embodiments of the invention, the signal or signals may be integrated into other routines in determining the status and validity of those routines. In certain embodiments of the invention, the signal or signals may be used by a control procedure to determine whether control should continue on the hard analyzer measurement and/or the inferential
measurement.
[0052] Traditionally, the art has not deployed a measurement from a hard analyzer when a comparable measurement is available from an inferential analyzer on a more frequent interval. When such an interval is about less than 30 minutes, the art has not recognized the benefits of deployment of the use of a primary measurement from a hard analyzer when a comparable inferential measurement is available.
[0053] In certain embodiments of the invention, the measurement generated by the hard analyzer may be a special sample measurement that is used in addition to or instead of a routinely taken primary measurement. In yet other embodiments of the invention, the measurement generated by the hard analyzer may be available on discontinuous sampling intervals. Pursuant to these embodiments, certain methods and systems of the invention may apply discretization techniques to process the discontinuously sampled measurements. In certain embodiments of the invention, a simple interpolation is used to discretize
discontinuous sample intervals. In yet certain other embodiments of the invention, a first order model may be used to discretize discontinuous sample intervals. In still yet certain other embodiments of the invention, a second order model may be used to discretize discontinuous sample intervals. In still yet certain other embodiments of the invention, a high order model may be used to discretize discontinuous sample intervals.
[0054] In an embodiment, a system or method of the invention may use a technique to replace certain independent variables that are found to be invalid or unavailable. A primary measurement may be unavailable for any number of reasons. In a non-limiting example, the primary measurement may be unavailable because the hard analyzer and/or instrument or instrumentation used to provide the measurement is not functioning correctly, not functioning at all, or being serviced, for example. Any of the methods described herein for discretizing measurements over discontinuous sampling intervals may be used to replace certain primary measurements that are invalid or not available at a certain sampling interval. For example, any of a simple interpolation, a first order model, a second order model, or a higher order model may be used to replace certain primary measurements that are invalid or not available at a certain sampling interval.
[0055] In an embodiment of the invention, the model used to determine inferential measurements may be tuned using any one or more of a procedure for adjusting bias, a procedure for adjusting a ratio multiplier, and a statistical procedure for model adaptation or tuning. In an embodiment of the invention, the model adaptation or tuning procedure may operate on one or more model parameters. In another embodiment of the invention, the model adaptation or tuning procedure may operate on one or more predicted values. In yet another embodiment of the invention, the model adaptation or tuning procedure may operate on any combination of one or more model parameters and one or more predicted values.
[0056] In certain embodiments of the invention, statistical validation techniques are used to determine validity of the primary measurement and the inferential measurement. In certain other embodiments of the invention, other validation techniques may be used to determine the validity of the primary measurement and/or the inferential measurement. In certain non- limiting examples, simple high and low limit checking may be used to determine validity of the primary measurement and/or the inferential measurement. In another non- limiting example, status of a primary measurement or any one or more of the independent
measurements used in the inferential estimation procedure may be used to determine the validity of the primary measurement and/or the inferential measurement. In an exemplary embodiment, a smart transmitter provides the status of any of the primary measurement and/or any of the independent measurements used in the inferential estimation procedure.
[0057] An embodiment of the invention is directed to determining ASTM properties of hydrocarbon streams. Non-limiting examples of such ASTM properties include, for example, flash point, initial boiling point, ten (10) percent boiling point, twenty (20) percent boiling point, fifty (50) percent boiling point, seventy (70) percent boiling point, ninety (90) percent boiling point, and final point.
[0058] Other embodiments of the invention are directed to determining properties typically provided by chromatographic analyzers. A non- limiting example of such chromatographic properties includes any typically measured distillation properties. In certain embodiments of the invention, chromatograph values determined in accordance with the systems and methods of the invention include concentrations of certain compounds in the process stream. Such compounds, in accordance with certain embodiments of the invention, may include paraffins, isoparaffins, olefins, naphthenes, and/or aromatics. When concentrations of all compounds of this exemplary embodiment are provided, the analysis is collectively referred to as "PIONA." When concentrations of all compounds of this exemplary embodiment, except isoparaffins, are provided, the analysis is collectively referred to as "PONA."
[0059] In certain embodiments, the invention may be directed to inferential analyzers using yet other primary test methods such as, for example, cold properties, which may include pour point and cloud point. In yet other embodiments, the invention may be directed to inferential analyzers using yet other primary test methods such as, for example, percent sulfur, initial boiling point(s) and final boiling point(s); flash point; total aromatics (or paraffins, isoparaffins, olefins, and/or naphthenes), for example, in naphtha feed to ethylene crackers. Indeed an inferential analyzer of the invention may be applied to any primary test method whether such methods are known or later invented.
[0060] In certain embodiments of the invention, inferential predictions of dependent variables may be made using independent variables that include any one or more of temperature, pressure, or flow rate. Certain embodiments of the invention are directed to providing inferential predictions of dependent variables that may include any or any number of density, viscosity, refractive index, conductivity, pH, and any combinations thereof.
Indeed, any variable that can be measured, whether now or later developed, may be used as an independent variable for determining inferential predictions.
[0061] The methods and systems of the invention may be applied to any type of process including, but not limited to, refinery processes, hydrocarbon processes, and chemical processes. In certain embodiments, the methods and systems of the invention may be applied to continuous processes. In certain other embodiments, the methods and systems of the invention may be applied to batch processes. In yet certain other embodiments, the methods and systems of the invention may be applied to semi -batch processes. In still certain other embodiments, the methods and systems of the invention may be applied to any combination of continuous processes, batch processes, and semi-batch processes.
[0062] An aspect of the invention provides systems and methods in which a hard analyzer may measure the independent variables, such as any one or multiple combinations of pressure(s), temperature(s), and flow rate(s) that are used in an inferential model whereby the inferential model generates a prediction(s) of the composition of the process fluid/sample. Pursuant to this aspect of the invention, according to an embodiment of the invention, any or all of the procedure of the methods and systems of the invention as described herein may be used.
[0063] According certain embodiments of the invention, validation of both the independent variables and the composition of the samples may be provided, similar to any of the validation techniques as further described herein for inferential measurements and hard analyzer measurements. According some embodiments of the invention, validation of the predictions may be provided, similar to any of the validation techniques as further described herein for inferential measurements and hard analyzer measurements. Furthermore, in certain embodiments of the invention, a model provides an inferential, modeled, referenced, and secondary measurement of the laboratory measurement. In further embodiments of the invention, the model may be adapted or tuned using the procedures of the invention, as further described herein, based upon other comparative measurements, for example.
[0064] An embodiment of the invention includes a system having a hard analyzer configured to generate one or more corresponding dependent variables of a process sample. The system also includes an inferential analyzer configured to capture and use one or more independent variables, along with a model, to generate one or more inferential dependent variables that correspond to the dependent variables generated by the hard analyzer. The system is also configured to generate a comparison of the measured dependent variables to the inferred dependent variables, and to use the comparison to adapt one or more parameters of the model. In certain embodiments of the invention, discretization of the corresponding measurement may be performed according to the procedures provided herein.
[0065] Turning now to Fig. 1A, an exemplary embodiment of the present invention is shown and described. As shown, an enhanced process control system 300 is configured for enhancing the operation and performance of an inferential analyzer by means of one or more hard analyzers. System 300 includes a hard analyzer 310 that performs online capture and/or analysis of a sample of process stream 10. Thus, as discussed hereinabove, the hard analyzer measures one or more dependent variables. The sample is thus a representation of the process stream which is bounded in terms both physical and temporal, e.g. it is the physical subset of the process stream corresponding to a particular point and time in the process. System 300 also includes an inferential analyzer 312 configured to receive various independent process variables, e.g., from a DCS (Distributed Control System) 320, and to use those process variables to predict the values of various properties (inferred dependent variables) of the process, using a model. The DCS 320 is configured to capture various independent process variables from process 10, for use in control and operation of the process 10. A CQM (Cognitive Quality Manager) 316 compares the dependent variables generated by the hard analyzer 310 with the inferred dependent variables generated by inferential analyzer 312, and uses the comparison to update the model used by the inferential analyzer 312 and/or adjust values for the inferred dependent variables which are then sent to a Process Optimizer 314. The CQM is thus a processor-enabled module configured to use the comparison to validate the property predictions received from inferential analyzer 312, and to then offer options for adjusting those predictions and/or updating the inferential models. The Process Optimizer 314 may receive predictions from the CQM 316 for one or more streams and stream properties, and then convert the validated, adjusted predictions into processing directives, such as adjusted process control inputs, which are sent to the DCS 320 for its control of the process 10. It should be recognized that the Process Optimizer 314 may receive inputs from sources other than the CQM 316, e.g. it may also receive and operate on independent process variables received directly from DCS 320. Examples of a Process Optimizer that may be used in embodiments of the present invention includes APC (Advanced Process Control) generally, which also may include MPC (Model Predictive Control) systems such as Exasmoc™ Multivariable Optimizing Control commercially available from Yokogawa Electric Corporation, Tokyo, Japan. The model may be generated in any conventional manner, e.g., based on (a) the application of first principles, (b) on a "black box" approach that is strictly empirical e.g. based on statistical or neural network algorithms, or (c) a combination of the first-principles and empirical approaches, and is an inferential model. A representative example of model generation suitable for use in embodiments hereof, is shown and described hereinbelow with respect to Fig. IB.
[0066] Examples of the independent variables used in system 300, including both independent process variables and independent sample variables, may include but are not limited to temperature, pressure, the rate of addition or consumption of a catalyst or reagent, flow, the input (output) of thermal energy into (from) the process or a point in the process, refractive index, conductivity, viscosity, opacity, pH, density, intensity responses measured by photometric or spectrometric devices at particular wavelengths or frequencies, and the like.
[0067] Optionally, e.g., in order to facilitate the operation of the inferential analyzer 312 and/or CQM 316, the independent process variables may be organized by plotting their values (y- values) according to kind (source/origin or type) arrayed across the x-axis to form a property spectrum (also referred to as a sample spectrum or spectra) 322, in which the independent variables plotted in a given property spectrum are related to each other temporally, e.g. they were captured substantially at the same time. This sample spectrum 322 may then be used to update or generate the property model, so that the model is based on a population of sample spectra 322 and their corresponding property values (dependent variables).
[0068] Optionally, the sample spectrum 322 may be supplied directly to the CQM 316 as shown at 323, to enable the CQM to receive the property spectrum 322 in an "un-processed" or unaltered state. And while a direct channel 323 may be used, it should be recognized that spectrum 322 may also be supplied in its unaltered state to the CQM 316 through the inferential analyzer 312, e.g., by tunneling through the analyzer 312.
[0069] A particular population of known sample spectra may also be used as a calibration sample set for the inferential model. The inferential analyzer 312 is thus configured to optionally predict a plurality of properties for a single stream or for a plurality of streams associated with said process, the predictions being based on one or more property spectra 322.
[0070] Exemplary methods used by process optimizer 314 to monitor/control/optimize process 10 may include one or more of: (a) multivariate method to determine whether the sample spectrum 322 belongs to the same population of sample spectra as a calibration sample set; (b) multivariate method to classify any spectral outlier according to cause; (c) replacing in the sample spectrum 322 of the false spectral outlier; (d) determining whether the outlier condition is due to a particular sensor by comparison to a baseline; (e) correcting the outlier condition by replacement with a model value; (f) comparing the inferred dependent variables with the dependent variables; (g) wherein the comparing (f) optionally includes averaging, regression, a damping function, or a vector-based algorithm, to provide a mathematical relationship between the inferred dependent variables and the dependent variables, which mathematical relationship is then applied to subsequent inferential predictions to produce a corrected inferential value; (h) estimating the lag time between values predicted by the inferential analyzer and those measured by a hard analyzer; (i) adjusting predictions from the inferential analyzer; (j) evaluating the difference between corresponding inferential and measured values using statistical analysis; (k) updating the model with the addition of recent data and the elimination of older data; (1) updating the model by adding sample data for spectral outliers and/or prediction outliers; (m) optionally capturing a sample that has been identified as a true spectral outlier or a property outlier, and using its property value(s) to adjust the outputs of the inferential analyzer; and (n) performing model updates on the basis of schedule, and/or on the basis of need as determined by factors including frequency and severity of spectral outliers and/or prediction outliers, changing of process operation or configuration. Each of the exemplary methods (a)-(n) will be described in greater detail as follows.
[0071] The method (a) may include determining the validity of a property prediction supplied by inferential analyzer 312, on the basis of one or more metrics which may be obtained by a multivariate method, e.g. the application to the sample spectrum of a multivariate model such as a PLS, PCR, or Principal Component Analysis (PC A) model based on the calibration sample set, which method and metrics are known to those skilled in the art of chemometric analysis, are capable of identifying if sample spectrum belongs to the same population of sample spectra in the calibration sample set, and which include but are not limited to the Mahalanobis distance, which is an expression of a sample's spatial relationship to the calibration sample set, or the spectral residual, which is a measure of how well a multivariate model accounts for all independent variables in a sample spectrum.
[0072] Method (b) may include further applying the multivariate method to classify any spectral outlier according to cause, e.g. whether an independent variable in the sample spectrum 322 is anomalous (e.g. because a sensor that is the source of an independent process variable is faulty), or alternatively if the sample spectrum 322 represents process conditions, sample chemistry, or a combination thereof which are distinctly different from those in the calibration sample set, these outliers being referred to, respectively, as false and true spectral outliers.
[0073] Method (c) may include inferring and replacing in the sample spectrum 322 of the false spectral outlier a value for the faulty independent variable by methods known to those skilled in the art of multivariate modeling methods (chemometric s), e.g. a method based on Principle Component Analysis (PCA), to permit the reliable prediction of the dependent variable (the sample property) by the inferential analyzer. [0074] Method (d) may include determining if the outlier condition is because the response function of a sensor that is the source of an independent process variable has changed compared with the response function for the same independent process variable when property spectra in the calibration sample set were recorded, where the change may be due to aging, fouling, calibration drift, or the replacement of a faulty sensor with a different sensor whose response function is not identical to that of the sensor used to generate property spectra in the calibration sample set, and where the change may be characterized by a mathematical function known to those skilled in the art, e.g. a simple linear equation involving slope and/or offset, or a second order equation, a multivariable function based on PCA, and the like.
[0075] Method (e) may include correcting the changed sensor readings by determining the mathematical function on the basis of values that are inferred by means of the
aforementioned multivariate modeling methods and also responses of the sensor that are independent process variables in outlier sample spectra.
[0076] Method (f) may include evaluating the performance of the inferential analyzer by the comparing of property values predicted by application of a property model to a sample spectrum with property values for corresponding samples measured by the hard analyzer 310, where the comparison is the application of suitable statistical methods known to those skilled in the art of statistical analysis, e.g. those based on the calculation of the difference between one predicted value and the corresponding measured value or between averages based on predicted and/or measured values obtained in a relatively short time interval. It is noted that the frequency of property predictions by the inferential analyzer is relatively high compared with that for measurements by hard analyzers, e.g. 6 min 1 to 60 min 1 versus 1 hr1 to 20 hr1, respectively.
[0077] Method (g) may include optionally performing the comparison by relating a population of values (inferred dependent variables) predicted by the inferential analyzer 312 and a population of corresponding values (dependent variables) measured by the hard analyzer 310, where the populations correspond to predictions and measurements made in time frames ranging from that of the hard analyzer' s measurement cycle time to about a few hours or a few days, and where the relating is by performing on each of the populations a suitable calculation known to those skilled in the art such as one based on averaging, regression, a damping function, or a vector-based algorithm, and where the calculation yields separate results for the inferential and measured values, that for the former being
mathematically related to that for the latter, e.g. as a bias, a first order linear equation or a second order polynomial, and the like, which mathematical relationship is then applied to subsequent inferential predictions, and which application produces a corrected inferential value.
[0078] Method (h) may include dynamically estimating the lag time between values predicted by the inferential analyzer and those measured by a hard analyzer by methodologies known to those skilled in multivariate statistical methods, e.g. by minimizing variance in a plot of predicted values versus measured values by iteratively advancing the times for a population of measured values relative to the times of a population of predicted values.
For example, in the case of diesel rundown from a CDU, changes in the independent variables P/T/flow, substantially describe in real time the changes occurring at the points in the process where those properties are being measured, whereas the manifestation of those changes in the diesel rundown stream, e.g. in values for the dependent variables T20 and T90, may lag for a relatively long periods of time ranging from perhaps 15 minutes to 30 minutes. T20 and T90 are the temperatures at which 20% and 90% of the components in a sample distill off the sample, respectively. Therefore, a temporal reconciliation is necessary to ensure that property values measured by a hard analyzer are properly paired with the corresponding property spectrum before performing any calculations described hereinabove to enhance the operation and performance of an inferential analyzer by means of measurements by a hard analyzer. Given that the magnitude of the lag may vary as a function of a number of factors relating to the operation of the process, which, referring again to the example of CDU operation, may include changes in feed quality, feed rate, draw rates for the rundown streams, and the temperature profile of the CDU. Therefore, because lag time varies dynamically, a method for dynamically estimating the lag time may be beneficial.
[0079] Method (i) may include adjusting predictions from the inferential analyzer by the application of the mathematical relationship in enhancement (g).
[0080] Method (j) may include evaluating the difference between pairs of corresponding inferential and measured values, or between results from the aforementioned population- based comparison, to determine if a predicted value or a population of predicted values are valid, where the evaluation is the application of appropriate statistical techniques known to those skilled in the art of statistical analysis, e.g. Statistical Quality Control (SQC) methodologies, where valid predictions are those whose deviation from measured values or populations of measured values is within predetermined tolerance, e.g. the predictions are in statistical control, and where predictions that are not in statistical control are prediction outliers, which also are property outliers, and in some cases may be a concentration outlier. [0081] Method (k) may include updating the model used by inferential analyzer 312 with the addition of data that are relatively recent and therefore are representative of the current conditions, and also by elimination of older data that may be less representative of said conditions, where the model and/or the data include independent variables and/or measurements generated by the hard analyzer 310.
[0082] Method (1) may include updating the inferential model by creating and downloading into the inferential analyzer a model, which creation is based on a data set updated by the addition of new sample data for true spectral outliers and/or prediction outliers, and which model updating is performed to increase the dynamic range and/or robustness of predictions made the inferential analyzer by expanding the range of process conditions and/or chemistries represented in said data set used in model creation.
[0083] Method (m) may include optionally capturing in a sampling cylinder a process sample that has been identified as a true spectral outlier or a property outlier, where the capturing is accomplished by a sampling system interfaced to the process stream by appropriate means, e.g. a sample fast loop that continuously circulates material from the process stream, and which captured sample is then delivered to the local laboratory for the determination of the property value(s) of interest, which then are used to adjust the outputs of the inferential analyzer in that case in which desired property values are not available from hard analyzer 310.
[0084] Method (n) may include performing model updates on the basis of schedule, e.g. about once every 10 days, or about once every 30 days, or about once every 60 days, or about once every 90 days, or on the basis of need, where the need is determined in consideration of factors including but not limited to the frequency and severity (magnitude) of true spectral outliers and/or prediction outliers, the changing of process operation or configuration in a way that results in a substantive change in property spectra, or the cumulative number of outlier samples since the previous model update, where the magnitude of true spectral outliers may be gauged by parameters know to those skilled in the art such as the Mahalanobis distance or spectral residuals.
[0085] It should be recognized that methods (h)-(l) generally relate to enhancing the operation and performance of inferential analyzer 312, while methods (a) - (g) generally relate to the mathematical basis for those enhancements.
[0086] In a particular embodiment of the instant invention, the hard analyzer 310 is an online gas chromatograph, e.g. a GC8000 manufactured by Yokogawa Electric Corporation, Tokyo, Japan, suitably configured to perform the simulated distillation (simdist) analysis of a single CDU rundown stream, e.g. diesel, in accordance with ASTM D2887, "Standard Test Method for Boiling Range Distribution of Petroleum Fractions by Gas Chromatography ." In another embodiment, the gas chromatograph is an ultrafast gas chromatograph (fast GC), e.g. the Calidus micro GC (Falcon Analytical, Lewisburg, WV) capable of performing the simdist analysis of CDU rundown streams with a cycle time of about 3 minutes to 6 minutes per analysis in accordance with ASTM D7798, "Standard Test Method for Boiling Range Distribution of Petroleum Distillates with Final Boiling Points up to 538*C by Ultra Fast Gas Chromatography (UF GC)," where said simdist analysis obtains values for distillation yield temperatures such as T20 and T90, and where (i) the ultrafast GC performs the simdist analysis on each a plurality of CDU rundown streams, e.g. naphtha, kerosene/jet fuel, diesel, light gasoil, heavy gasoil, and the like; (ii) simdist values for each of the streams are obtained as the result of a single chromatographic analysis, or as the average of results from replicate analyses performed in immediate succession; (iii) the values are used in the aforementioned enhancement of inferential predictions of distillation properties of the streams; (iv) the simdist values for each of the rundown streams are obtained sequentially; and (v) the enhanced inferential predictions are used for process control and optimization between updates of the same by the CQM engine. For example, if simdist properties for each of four CDU rundown streams are determined as the average of three analyses by fast GC, and the total cycle time for each analysis is 5 minutes, then hard analyzer results are available every 4 x 3 x 5 = 60 minutes for the enhancing of inferential predictions on each of the four streams. In particular embodiments, the GC may be configured to implement the conventional ASTM D7798 simdist analysis and/or the UF GC analysis following ASTM D7798. In other embodiments of the invention, the analyzer may be a Fourier Transform Infrared (FTIR) analyzer, a Fourier Transform Near-Infrared (FTNIR) analyzer, a nuclear magnetic resonance (NMR) analyzer, a Raman analyzer, or any combination thereof. Still further, the hard analyzer 310 may be a distillation analyzer configured to implement ASTM D86 "Standard Test Method for Distillation of Petroleum Products at Atmospheric Pressure." However, the illustrative embodiment of Fig. 1 A is not intended to be limiting and the dependent variables 40 may comprise any form of measurement detected by an analyzer. For example, the dependent variables may comprise an electrical signal having a particular voltage that is representative of a measurement, a digital value, spectra, and any combination thereof.
[0087] In particular embodiments, the gas chromatograph is configured to determine the properties, e.g. specifically, by means of a detailed hydrocarbon analysis (DHA), or generally by the obtaining and evaluation of a chromatogram of the hydrocarbon stream(s) using methods known to those skilled in the art; where the properties include those of gasoline, e.g. octane, percent total aromatics, percent benzene, percent olefins, distillation properties, Reid Vapor Pressure (RVP) and the like, or those of diesel, e.g. cetane number, distillation properties, polynuclear aromatics, polycyclic aromatics, and cold properties such as cloud point, pour point, cold filter plugging point, or those of naphtha, e.g. PIONA and distillation properties; the hard analyzer is optionally one or more online analyzer technologies capable of measuring properties including but not limited to the aforementioned properties of gasoline, diesel, or naphtha by means other than gas chromatography, where the application of the technologies in accordance with established, standardized test methods such as those promulgated by ASTM International may be considered a primary test method; and the qualification of a sample as an outlier is optionally based on a chemometric assessment of the chromatrogram, where the assessment is a multivariate analysis, e.g. one based on PLS, PCR, PCA, and the like.
[0088] It should be understood that the plurality of property predictions may be accomplished by means of a corresponding plurality of models (one model per property); or, a single model may be capable of predicting the plurality of properties (one model, many properties). In addition, it should be recognized that reference to distillation yield
temperatures, T20 and T90, i.e., the temperatures at which 20% and 90% of the components in a sample distill off the sample, is non-limiting. Substantially any other distillation yield temperatures may be used, e.g. T50, Tss, T95 and T98. Additionally, distillation yield temperatures are but one expression of a sample's distillation properties, another being distillation yields determined for certain temperatures, e.g. the percent of a sample that distills by some certain temperature such a 200 °C or 300 °C, which sometimes are referred to as E200 and E300 in gasoline. Thus, it should be understood that references to distillation yield temperature, generally or specifically, e.g. T20 and T90, are non-limiting, but are intended as examples of distillation properties that may be predicted by inferential analyzers or measured by a gas chromatograph. Similarly, the term simulated distillation (simdist) does not apply only to the determination of distillation yield temperatures, but simply denotes the method for determining distillation properties of a hydrocarbon sample by gas chromatrography. It should be also understood that references to distillation properties are non-limiting, and that the present invention may be applied beneficially to enhance the inferential prediction of other properties, which may include so-called cold properties such as pour point, cloud point, and cold filter plugging point of diesel. [0089] Turning now to Fig. IB, a multivariate mathematical model relates the response matrix to the desired property value that generally can be determined in the laboratory in accordance with a primary test method defined by an organization or authority recognized by the industry. Multivariate statistical modeling techniques (common algorithms include Partial Least Squares (PLS) and Principal Component Regression (PCR)) extract information distributed across the entire spectrum for correlation with primary test method results.
Application of the resultant calibration models to the measured spectrum of a sample yields inferential values for those properties as shown.
[0090] Fig. 2 is a flowchart showing the steps of predicting a property using a model- based soft sensor or model-based hard sensor according to embodiment of the invention. Discrete samples 20 taken from the process 10 are measured by a hard analyzer 310, to yield dependent variables 160. Additionally, independent variables 30 are captured by sensors associated with DCS 320, from the process 10 at a given time t. Using the independent variables 30, the inferential model 80 for the "soft sensor" calculates the predicted values (the inferred dependent variables) 180, which is one or more inferential measurements of the process 10. At 190, the values for the dependent variables 160 are compared with the values for the inferred dependent variables 180. The result of the comparison may be an adjusted inferential prediction and/or an adjustment to the model which is then sent to update the model 80 at 194, and/or to control the process 10 at 196, e.g., via DCS 320. In an alternative embodiment, the comparison result may simply be an updated inferential model 80', which may sent at 194 to replace the model 80.
[0091] In certain embodiments of the invention, the analyzer may include a procedure to validate the one or more object variables according to the procedures further discussed herein. Embodiments of the invention are also directed to a procedure to validate the one or more analyzer measurements.
[0092] In an embodiment of the invention, the process model or process models may be updated based upon, for example, feedback from a primary measurement. In certain embodiments of the invention, the process model(s) is automatically updated using laboratory data. In certain other embodiments of the invention, the process model(s) is automatically updated using measurements provided by one or more online analyzers. In yet certain other embodiments of the invention, the process model(s) is automatically updated using any combination of laboratory data and measurements provided by one or more online analyzers.
[0093] In an embodiment of the invention, any one of or any combination of single factor correlation analysis and multivariate regression analysis may be used to update the process model or correction biases or ratios associated with the estimation of the inferential measurements. According to certain embodiments of the invention, principal factors that control the calculation of specific inferential measurements may be identified and a single factor correlation analysis may be performed on any of these one or more variables. In certain other embodiments of the invention, a quantitative model is established between the independent variables and inferential measurements and the model may be updated using a multivariate regression analysis. In yet other embodiments of the invention a combination of single factor correlation analyses and multivariable regression analysis may be performed to adapt and/or tune the model used to determine one or more inferential measurements.
[0094] FIG. 3 is a flowchart showing the steps of determining a property of a process stream according to an embodiment of the invention. The method for determining a property of a process stream 200 comprises calculating an inferential measurement for the property using a model 210, receiving an analyzer measurement of the property 220, evaluating the validity of the inferential measurement 230, evaluating the validity of the analyzer measurement 240, and adjusting the model and/or adjusting the inferential measurement at 250.
[0095] In certain embodiments, the method of determining a property of a process stream may comprise the step of adjusting the model using one or more prior valid analyzer measurements when the inferential measurement is invalid and the analyzer measurement is invalid. Further pursuant to this embodiment, the method of determining a property of a process stream may comprise determining whether to use the one or more prior valid analyzer measurements to adjust the model based upon at least one of the age of the one or more prior valid analyzer measurements and the current state of the process in comparison to the state of the process at the time the one or more prior valid analyzer measurements were measured. Moreover, in any of the various embodiments shown and described herein, when both the inferential predictions and the analyzer measurements are determined to be valid, the latter may be used to tune (adjust) the former.
[0096] Various embodiments may also discretize the analyzer measurement using, for example, any of the procedures further described herein.
[0097] For example, if the age of the one or more prior valid analyzer measurements is too old, then any of the one or more prior valid analyzer measurements may be discarded and not used to adapt or tune the model. According to another example, if the state of the current process is different than the state of the process at the time the one or more prior valid analyzer measurements were measured is different, then any of the one or more prior valid analyzer measurements may likewise be discarded.
EXAMPLES
[0098] Table 1 analyzes the effectiveness of the systems and methods of the invention over the more conventional techniques using off-line measurements for process operation and control. The standard error of prediction (SEP) is a measure of the extent of uncertainty associated with the distillation temperatures for the 20 percent (T2o) and 90 percent (T90) recoveries. For example, the distillation temperatures are typically measured in the laboratory by the standard ASTM D-86 method or by a simdist method, e.g. the ASTM D-2887 or D-7798 methods while the latter can also be implemented online using a gas chromatograph, for example.
Figure imgf000029_0001
[0099] As shown in Table 1 , the mere increase in frequency of measurements results in a reduction in the standard error of prediction and an improvement in process operations. Thus, the frequency provided by on-line analyzers is generally favored over laboratory analysis if the rate of return for the improvement in process operations justifies the installation and support of the on-line analyzer. Notwithstanding the frequency of the analyzer measurement, the standard error of prediction is always improved by integrating the analyzer with an inferential analyzer using a process model that is suited to the type of analysis being performed.
[00100] The economic benefit was estimated assuming the value of diesel is $6 greater per barrel than the adjacent naphtha cut and $3 greater per barrel than the adjacent gasoil cut. Thus, the refiner would be motivated to maximize production of diesel while still maintaining the specifications on the T20 and T90 distillation temperatures. [00101] Upon initial consideration of the economic benefits presented in Table 1, the benefit provided by the 6th method over methods 3 and 4 may not appear to be very compelling. Indeed, the small economic benefit may not offset the cost to install and operate an online analyzer. Given that the performance comparison in Table 1 applies to a CDU operating under nominally steady-state conditions, e.g. when the composition of the crude feed is relatively stable, this result in fact serves to make the point that soft sensors are very effective under such conditions. However, significant profit improvement may be realized by minimizing the duration and severity of disruptions to the crude distillation unit operation resulting from crude switches-i.e., changes between crude feedstocks with different properties. A conventional strategy for reducing disruptions associated with crude switches involves temporarily drawing down or reducing the crude feed rate during the change in feedstocks, and gradually increasing the rate back to a target rate. The availability of property predictions that have both high frequency and high fidelity by Method 6 yield an estimated reduction in throughput losses of approximately $1,280,000 per year per 100,000 barrels per day crude distillation capacity. Compared with conventional techniques, the integrated analyzer system and method of the invention, particularly leveraged by the use of on-line analyzers to enhance inferential predictions, allows adjustment of the latter on time scales sufficiently short to mitigate the impact of crude switchovers in terms of their duration and the disruption they cause to the operation of the CDU and therefore the refinery, and in the immediate example enables the maximizing of diesel production.
[00102] Many modifications and other embodiments of the inventions set forth herein will come to mind to one skilled in the art to which these inventions pertain having the benefit of the teachings presented in the foregoing descriptions and the associated drawings. For example, it should be recognized that the "models" shown and described herein not only include those based on multivariate statistical methods, e.g. PCA (principal component analysis), PLS (partial least squares), and MLR (multiple linear regression), but also include RBF (radial basis function), neural networks, models based on first principles, and combinations thereof, along with substantially any model types that may be developed in the future. Similarly, although various exemplary analyzer technologies have been shown and described with respect to the various embodiments herein, it should be recognized that other technologies, such as those based on TDL (tunable diode laser), QCL (quantum cascade laser), and CRD (cavity ring-down) spectroscopy, and combinations thereof, along with any future developed analyzer technology, may be used without departing from the scope of the invention. Therefore, it is to be understood that the inventions are not to be limited to the specific embodiments disclosed and that modifications and other embodiments are intended to be included within the scope of the appended claims. Moreover, although the foregoing descriptions and the associated drawings describe exemplary embodiments in the context of certain exemplary combinations of elements and/or functions, it should be appreciated that different combinations of elements and/or functions may be provided by alternative embodiments without departing from the scope of the appended claims. In this regard, for example, different combinations of elements and/or functions than those explicitly described above are also contemplated as may be set forth in some of the appended claims. Although specific terms are employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation.

Claims

WHAT IS CLAIMED IS:
1. A system for determining a property of a process stream, the system comprising: a hard analyzer configured to generate one or more dependent variables associated with a sample of the process stream; an inferential analyzer including a model of aspects of the process, the inferential analyzer being configured to capture a plurality independent process variables associated with the process stream and to use the independent process variables to generate one or more inferred dependent variables; a comparison module configured to generate a comparison of the dependent variables and the inferred dependent variables; and the system being configured to use the comparison to adjust the model and/or to control the process stream.
2. The system of claim 1, wherein the system is configured to control the process stream by adjusting values for the inferred dependent variables.
3. The system of claim 1, further comprising a cognitive quality manager (CQM) configured to use the comparison to adjust the model.
4. The system of claim 1, further comprising a process optimizer configured to use the comparison to control the process stream.
5. The system of claim 4, further comprising a process control system operatively engaged with the process optimizer, the process control system configured to adjust one or more parameters of the process stream in response to receipt of instructions from the process optimizer.
6. The system of claim 1, wherein the inferential analyzer is configured to capture the plurality of independent process variables over a discrete interval.
7. The system of claim 6, wherein the discrete interval is at most about 240 minutes.
8. The system of claim 7, wherein the discrete interval is at most about 180 minutes.
9. The system of claim 8, wherein the discrete interval is at most about 120 minutes.
10. The system of claim 9, wherein the discrete interval is at most about 90 minutes.
11. The system of claim 10, wherein the discrete interval is at most about 60 minutes.
12. The system of claim 1, wherein the model comprises at least one of a statistical model and a statistical multivariate model.
13. The system of claim 1, wherein at least one of the independent process variables, the dependent variables, and the inferred dependent variables is discretized.
14. The system of claim 1, wherein the independent process variables include one or more of temperature, pressure, rate of addition or consumption of a catalyst or reagent, flow rate, the input or output of thermal energy to the process, refractive index, conductivity, viscosity, opacity, pH, density, intensity responses measured by photometric or spectrometric devices at particular wavelengths or frequencies, and combinations thereof.
15. The system of claim 1, being further configured to aggregate the independent process variables into a property spectrum.
16. The system of claim 15, wherein the system is configured to use the property spectrum to update the model.
17. The system of claim 1, wherein the comparison includes a validation of at least one of the independent process variables, the dependent variables, and the inferred dependent variables.
18. The system of claim 17, being configured to adjust the model and/or control the process stream once the dependent variables are validated.
19. The system of claim 18, being configured to determine whether to use the validated dependent variables to adjust the model based upon at least one of an age of the validated dependent variables and a current state of the process as compared to a state of the process at the time the validated dependent variable was measured.
20. The system of claim 17, wherein the comparison module is configured to generate the comparison using a statistical multivariate procedure.
21. The system of claim 17, wherein the comparison module is configured to generate the comparison using a procedure selected from the group consisting of:
(a) a multivariate method to determine whether the sample spectrum belongs to a population of sample spectra which includes a calibration sample set;
(b) multivariate method to classify any spectral outlier according to cause;
(c) replacing any false independent variable within the outlier sample spectrum;
(d) determining whether any outlier condition is due to a particular sensor by comparing sensor output to a baseline;
(e) correcting the outlier condition by replacement of a false independent variable with a model value;
(f) using averaging, regression, a damping function, or a vector-based algorithm, to provide a mathematical relationship between the inferred dependent variables and the dependent variables, which mathematical relationship is then applied to subsequent inferential predictions to produce a corrected inferential value;
(g) estimating a lag time of the dependent variables relative to the inferred dependent variables;
(h) applying a statistical analysis to the dependent variables and the inferred dependent variables;
(i) updating the model with the addition of recent data to and the elimination of older data from the data set used in creation of the model;
(j) updating the model by adding sample data for spectral outliers and/or prediction outliers to the data set used in creation of the model;
(k) capturing a sample that has been identified as a spectral outlier or a property outlier, and using it to adjust outputs of the inferential analyzer; (1) performing model updates on the basis of schedule;
(m) performing model updates on the basis of need as determined by factors including frequency, severity of spectral outliers and/or prediction outliers, and/or changing of process operation or configuration;
(n) applying a goodness of fit test procedure; and combinations thereof.
22. The system of claim 1, wherein the hard analyzer comprises a gas chromatograph configured to generate the dependent variables by capturing and evaluating a chromatogram of a sample of the process stream.
23. The system of claim 1, wherein the hard analyzer is at least one of a gas
chromatograph (GC), a fast GC, an ultrafast GC, a Fourier Transform Infrared (FTIR) analyzer, a Fourier Transform Near-Infrared (FTNIR) analyzer, a nuclear magnetic resonance (NMR) analyzer, a Raman analyzer, a flash point analyzer, a distillation property analyzer, a cloud point analyzer, a freezing point analyzer, a cold filter plugging point analyzer, a pour point analyzer, and a combination thereof.
24. The system of claim 1, wherein the dependent variables include one or more of octane, percent total aromatics, percent benzene, percent olefins, distillation properties, Reid Vapor Pressure (RVP), cetane number, distillation properties, polynuclear aromatics, polycyclic aromatics, cloud point, pour point, cold filter plugging point, PIONA, PONA, distillation properties, and combinations thereof.
25. The system of claim 1, wherein the inferential analyzer is configured to generate the one or more inferred dependent variables substantially in real time, and the hard analyzer is an online analyzer that generates dependent variables with a frequency that is lower than the frequency with which inferred dependent variables are generated, and wherein the comparison is used to adjust the model and/or to control the process stream substantially in real time.
26. A method for determining a property of a process stream, the method comprising: generating, with a hard analyzer, one or more dependent variables associated with a sample of the process stream; capturing, with an inferential analyzer including a model of aspects of the process, one or more independent process variables associated with the process stream and using the independent process variables to generate one or more inferred dependent variables; generating, with a comparison module, a comparison of the dependent variables and the inferred dependent variables; and using the comparison to adjust the model and/or to control the process stream.
27. The method of claim 26, wherein said using the comparison further comprises using the comparison to control the process stream by adjusting values for the inferred dependent variables.
28. The method of claim 26, further comprising using a cognitive quality manager (CQM) to adjust the model.
29. The method of claim 26, further comprising using a process optimizer to control the process stream.
30. The method of claim 29, further comprising using a process control system operatively engaged with the process optimizer, to adjust one or more parameters of the process stream in response to receipt of instructions from the process optimizer.
31. The method of claim 26, further comprising using the inferential analyzer to capture the one or more independent process variables over a discrete interval.
32. The method of claim 31, wherein the discrete interval is at most about 240 minutes.
33. The method of claim 32, wherein the discrete interval is at most about 180 minutes.
34. The method of claim 33, wherein the discrete interval is at most about 120 minutes.
35. The method of claim 34, wherein the discrete interval is at most about 90 minutes.
36. The method of claim 35, wherein the discrete interval is at most about 60 minutes.
37. The method of claim 26, further comprising the inferential analyzer using a model in the form of at least one of a statistical model and a statistical multivariate model.
38. The method of claim 26, further comprising discretizing at least one of the independent process variables, the dependent variables, and the inferred dependent variables.
39. The method of claim 26, wherein the independent process variables include one or more of temperature, pressure, rate of addition or consumption of a catalyst or reagent, flow rate, the input or output of thermal energy to the process, refractive index, conductivity, viscosity, opacity, pH, density, intensity responses measured by photometric or spectrometric devices at particular wavelengths or frequencies, and combinations thereof.
40. The method of claim 26, further comprising aggregating the independent process variables into a property spectrum.
41. The method of claim 40, further comprising using the property spectrum to update the model.
42. The method of claim 26, wherein said generating a comparison comprises validating at least one of the independent process variables, the dependent variables, and the inferred dependent variables.
43. The method of claim 42, comprising adjusting the model and/or controlling the process stream once the dependent variables are validated.
44. The method of claim 43, comprising determining whether to use the validated dependent variables to adjust the model based upon at least one of an age of the validated dependent variables and a current state of the process as compared to a state of the process at the time the validated dependent variable was measured.
45. The method of claim 42, comprising generating the comparison using a statistical multivariate procedure.
46. The method of claim 42, comprising generating the comparison using a procedure selected from the group consisting of:
(a) a multivariate method to determine whether the sample spectrum belongs to a population of sample spectra which includes a calibration sample set; (b) multivariate method to classify any spectral outlier according to cause;
(c) replacing any false independent variable within in the sample spectrum;
(d) determining whether any outlier condition is due to a particular sensor by comparing sensor output to a baseline;
(e) correcting the outlier condition by replacement with a model value;
(f) using averaging, regression, a damping function, or a vector-based algorithm, to provide a mathematical relationship between the inferred dependent variables and the dependent variables, which mathematical relationship is then applied to subsequent inferential predictions to produce a corrected inferential value;
(g) estimating a lag time of the dependent variables relative to the inferred dependent variables;
(h) applying a statistical analysis to the dependent variables and the inferred dependent variables;
(i) updating the model with the addition of recent data to and the elimination of older data from the data set used in creation of the model;
(j) updating the model by adding sample data for spectral outliers and/or prediction outliers to the data set used in creation of the model;
(k) capturing a sample that has been identified as a spectral outlier or a property outlier, and using it to adjust outputs of the inferential analyzer;
(1) performing model updates on the basis of schedule;
(m) performing model updates on the basis of need as determined by factors including frequency, severity of spectral outliers and/or prediction outliers, and/or changing of process operation or configuration;
(n) applying a goodness of fit test procedure; and combinations thereof.
47. The method of claim 26, wherein the hard analyzer comprises a gas chromatograph configured to generate the dependent variables by capturing and evaluating a chromatogram of a sample of the process stream.
48. The method of claim 26, wherein the hard analyzer is at least one of a: gas chromatograph (GC), a fast GC, an ultrafast GC, a Fourier Transform Infrared (FTIR) analyzer, a Fourier Transform Near-Infrared (FTNIR) analyzer, a nuclear magnetic resonance (NMR) analyzer, a Raman analyzer, a flash point analyzer, a distillation property analyzer, a cloud point analyzer, a freezing point analyzer, a cold filter plugging point analyzer, a pour point analyzer, and a combination thereof.
49. The method of claim 26, wherein the dependent variables include one or more of octane, percent total aromatics, percent benzene, percent olefins, distillation properties, Reid Vapor Pressure (RVP), cetane number, distillation properties, polynuclear aromatics, polycyclic aromatics, cloud point, pour point, cold filter plugging point, PIONA, PONA, distillation properties, and combinations thereof.
50. The method of claim 26, wherein said capturing and using is effected using an on-line hard analyzer substantially in real time, and the comparison is used to adjust the model and/or to control the process stream substantially in real time.
PCT/US2014/014857 2013-02-05 2014-02-05 System, method and apparatus for determining properties of product or process streams WO2014123993A1 (en)

Priority Applications (3)

Application Number Priority Date Filing Date Title
JP2015556250A JP2016507113A (en) 2013-02-05 2014-02-05 System, method, and apparatus for determining the characteristics of a product or process stream
EP14748750.8A EP2954427A4 (en) 2013-02-05 2014-02-05 System, method and apparatus for determining properties of product or process streams
CN201480017536.2A CN105229627B (en) 2013-02-05 2014-02-05 For determining system, the method and apparatus of the attribute of product stream or processing stream

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US201361760879P 2013-02-05 2013-02-05
US61/760,879 2013-02-05

Publications (1)

Publication Number Publication Date
WO2014123993A1 true WO2014123993A1 (en) 2014-08-14

Family

ID=51300105

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2014/014857 WO2014123993A1 (en) 2013-02-05 2014-02-05 System, method and apparatus for determining properties of product or process streams

Country Status (5)

Country Link
US (1) US10175661B2 (en)
EP (1) EP2954427A4 (en)
JP (1) JP2016507113A (en)
CN (1) CN105229627B (en)
WO (1) WO2014123993A1 (en)

Cited By (31)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017062416A1 (en) * 2015-10-05 2017-04-13 Honeywell International Inc. Hybrid analyzer for fluid processing processes
CN107430398A (en) * 2015-03-30 2017-12-01 环球油品公司 System and method for tuning process modeling
US10534329B2 (en) 2015-03-30 2020-01-14 Uop Llc System and method for improving performance of a plant with a furnace
US10663238B2 (en) 2017-03-28 2020-05-26 Uop Llc Detecting and correcting maldistribution in heat exchangers in a petrochemical plant or refinery
US10670353B2 (en) 2017-03-28 2020-06-02 Uop Llc Detecting and correcting cross-leakage in heat exchangers in a petrochemical plant or refinery
US10670027B2 (en) 2017-03-28 2020-06-02 Uop Llc Determining quality of gas for rotating equipment in a petrochemical plant or refinery
US10678272B2 (en) 2017-03-27 2020-06-09 Uop Llc Early prediction and detection of slide valve sticking in petrochemical plants or refineries
US10695711B2 (en) 2017-04-28 2020-06-30 Uop Llc Remote monitoring of adsorber process units
US10734098B2 (en) 2018-03-30 2020-08-04 Uop Llc Catalytic dehydrogenation catalyst health index
US10739798B2 (en) 2017-06-20 2020-08-11 Uop Llc Incipient temperature excursion mitigation and control
US10752844B2 (en) 2017-03-28 2020-08-25 Uop Llc Rotating equipment in a petrochemical plant or refinery
US10754359B2 (en) 2017-03-27 2020-08-25 Uop Llc Operating slide valves in petrochemical plants or refineries
US10752845B2 (en) 2017-03-28 2020-08-25 Uop Llc Using molecular weight and invariant mapping to determine performance of rotating equipment in a petrochemical plant or refinery
US10794644B2 (en) 2017-03-28 2020-10-06 Uop Llc Detecting and correcting thermal stresses in heat exchangers in a petrochemical plant or refinery
US10839115B2 (en) 2015-03-30 2020-11-17 Uop Llc Cleansing system for a feed composition based on environmental factors
US10844290B2 (en) 2017-03-28 2020-11-24 Uop Llc Rotating equipment in a petrochemical plant or refinery
US10901403B2 (en) 2018-02-20 2021-01-26 Uop Llc Developing linear process models using reactor kinetic equations
US10913905B2 (en) 2017-06-19 2021-02-09 Uop Llc Catalyst cycle length prediction using eigen analysis
US10953377B2 (en) 2018-12-10 2021-03-23 Uop Llc Delta temperature control of catalytic dehydrogenation process reactors
US10962302B2 (en) 2017-03-28 2021-03-30 Uop Llc Heat exchangers in a petrochemical plant or refinery
US10994240B2 (en) 2017-09-18 2021-05-04 Uop Llc Remote monitoring of pressure swing adsorption units
US11022963B2 (en) 2016-09-16 2021-06-01 Uop Llc Interactive petrochemical plant diagnostic system and method for chemical process model analysis
US11037376B2 (en) 2017-03-28 2021-06-15 Uop Llc Sensor location for rotating equipment in a petrochemical plant or refinery
US11105787B2 (en) 2017-10-20 2021-08-31 Honeywell International Inc. System and method to optimize crude oil distillation or other processing by inline analysis of crude oil properties
US11130111B2 (en) 2017-03-28 2021-09-28 Uop Llc Air-cooled heat exchangers
US11130692B2 (en) 2017-06-28 2021-09-28 Uop Llc Process and apparatus for dosing nutrients to a bioreactor
US11194317B2 (en) 2017-10-02 2021-12-07 Uop Llc Remote monitoring of chloride treaters using a process simulator based chloride distribution estimate
US11365886B2 (en) 2017-06-19 2022-06-21 Uop Llc Remote monitoring of fired heaters
US11396002B2 (en) 2017-03-28 2022-07-26 Uop Llc Detecting and correcting problems in liquid lifting in heat exchangers
US11676061B2 (en) 2017-10-05 2023-06-13 Honeywell International Inc. Harnessing machine learning and data analytics for a real time predictive model for a FCC pre-treatment unit
US20230368590A1 (en) * 2022-05-13 2023-11-16 Regents Of The University Of Minnesota System and method for controlling a compression ignition engine

Families Citing this family (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10235481B2 (en) 2014-02-05 2019-03-19 Yokogawa Corporation Of America System and method for online measurement of vapor pressure in hydrocarbon process streams
US10983982B2 (en) * 2016-03-02 2021-04-20 Accenture Global Solutions Limited Method and system for approving a submission of information
WO2017214311A1 (en) * 2016-06-07 2017-12-14 Yokogawa Corporation Of America System and method for online measurement of vapor pressure in hydrocarbon process streams
US10570733B2 (en) * 2016-12-05 2020-02-25 Baker Hughes, A Ge Company, Llc Synthetic chromatogram from physical properties
GB2571687B (en) * 2017-02-22 2021-12-22 Halliburton Energy Services Inc Slurry design process
US10794401B2 (en) 2017-03-28 2020-10-06 Uop Llc Reactor loop fouling monitor for rotating equipment in a petrochemical plant or refinery
US10816947B2 (en) 2017-03-28 2020-10-27 Uop Llc Early surge detection of rotating equipment in a petrochemical plant or refinery
JP6928119B2 (en) * 2017-06-02 2021-09-01 アスペン テクノロジー インコーポレイテッド Computer systems and methods for building and deploying predictive inference models online
ES2947044T3 (en) * 2017-11-14 2023-07-31 Xylem Vue Inc System and method that uses probabilistic forecast for sewer control
US10739763B2 (en) * 2018-11-13 2020-08-11 Rockwell Automation Technologies, Inc. Industrial control system with predictive machine learning module
CN109633059B (en) * 2018-12-14 2020-02-21 浙江省检验检疫科学技术研究院 Method for predicting flash point of flammable liquid mixed solution based on headspace gas chromatography-mass spectrometry
WO2020210034A1 (en) * 2019-04-10 2020-10-15 Exxonmobil Research And Engineering Company Dynamic quality control in petrochemical, chemical, and pharmaceutical manufacturing processes
CN110210000A (en) * 2019-04-18 2019-09-06 贵州大学 The identification of industrial process efficiency and diagnostic method based on Multiple Non Linear Regression
KR20200137103A (en) 2019-05-29 2020-12-09 삼성전자주식회사 Apparatus and method for updating bio-information estimation model
WO2021019551A1 (en) * 2019-07-26 2021-02-04 Tata Consultancy Services Limited System and method for real-time self-optimization of manufacturing operations
CN111523779A (en) * 2020-04-13 2020-08-11 湖州同润汇海科技有限公司 Method for determining distillation range curve of atmospheric and vacuum straight-run diesel oil on line
CN111592907B (en) * 2020-05-18 2021-11-09 南京富岛信息工程有限公司 Online analysis and detection method for oil dry point of atmospheric tower top
JP2022173863A (en) * 2021-05-10 2022-11-22 株式会社東芝 Modular type time series data predictor, modular type time series data prediction method and program

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100191361A1 (en) * 2009-01-23 2010-07-29 Mks Instruments, Inc. Controlling a Manufacturing Process with a Multivariate Model
US20100204925A1 (en) * 2005-09-01 2010-08-12 Tareq Abduljalil Albahri Method for measuring the properties of petroleum fuels by distillation
US20100228485A1 (en) * 2009-03-09 2010-09-09 Soraya Betancourt Method for integrating reservoir charge modeling and downhole fluid analysis
US20100224365A1 (en) * 2009-03-06 2010-09-09 Carlos Abad Method of treating a subterranean formation and forming treatment fluids using chemo-mathematical models and process control
US20120022670A1 (en) * 2006-09-13 2012-01-26 Rockwell Automation Technologies, Inc. System and method for utilizing a hybrid model

Family Cites Families (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4886590A (en) 1987-11-02 1989-12-12 Man-Gill Chemical Company Chemical process control system
US5260865A (en) 1991-04-01 1993-11-09 Beauford Martin H Nonlinear model based distillation control
US5402333A (en) 1992-06-15 1995-03-28 E. I. Du Pont De Nemours & Co., Inc. System and method for improving model product property estimates
US5771861A (en) * 1996-07-01 1998-06-30 Cummins Engine Company, Inc. Apparatus and method for accurately controlling fuel injection flow rate
JPH1082883A (en) * 1996-09-06 1998-03-31 Hitachi Ltd Plant diagnostic device
BR9803848A (en) 1998-10-08 2000-10-31 Opp Petroquimica S A Inline system for inference of physical and chemical properties, inline system for inference of process variables, and inline control system
SE515570C2 (en) 1999-10-05 2001-09-03 Abb Ab A computer-based process and system for regulating an industrial process
US6751527B2 (en) 2000-02-11 2004-06-15 Stone & Webster Process Technology, Inc. Heat and material balance method of process control for petrochemical plants and oil refineries
US7157046B2 (en) * 2001-02-16 2007-01-02 Steris Inc. High capacity flash vapor generation systems
NZ530763A (en) * 2001-08-08 2006-07-28 Shell Int Research Process to prepare a hydrocarbon product having a sulphur content of below 0.05 wt%
US8200072B2 (en) * 2002-10-24 2012-06-12 Shell Oil Company Temperature limited heaters for heating subsurface formations or wellbores
US7246023B2 (en) 2004-01-26 2007-07-17 Ranko, Llc Flexible process optimizer
US20060074598A1 (en) 2004-09-10 2006-04-06 Emigholz Kenneth F Application of abnormal event detection technology to hydrocracking units
US7584024B2 (en) * 2005-02-08 2009-09-01 Pegasus Technologies, Inc. Method and apparatus for optimizing operation of a power generating plant using artificial intelligence techniques
US7292899B2 (en) * 2005-08-15 2007-11-06 Praxair Technology, Inc. Model predictive control having application to distillation
US8036760B2 (en) 2005-10-04 2011-10-11 Fisher-Rosemount Systems, Inc. Method and apparatus for intelligent control and monitoring in a process control system
US7917240B2 (en) * 2006-09-29 2011-03-29 Fisher-Rosemount Systems, Inc. Univariate method for monitoring and analysis of multivariate data
US9764314B2 (en) * 2006-11-07 2017-09-19 Saudi Arabian Oil Company Control of fluid catalytic cracking process for minimizing additive usage in the desulfurization of petroleum feedstocks
US20090093892A1 (en) * 2007-10-05 2009-04-09 Fisher-Rosemount Systems, Inc. Automatic determination of the order of a polynomial regression model applied to abnormal situation prevention in a process plant
US8055479B2 (en) * 2007-10-10 2011-11-08 Fisher-Rosemount Systems, Inc. Simplified algorithm for abnormal situation prevention in load following applications including plugged line diagnostics in a dynamic process
US8209045B2 (en) * 2008-04-07 2012-06-26 Honeywell International Inc. System and method for discrete supply chain control and optimization using model predictive control
US9110462B2 (en) * 2010-05-04 2015-08-18 Frito-Lay North America, Inc. Batch control using bang-bang control
JP2011253275A (en) * 2010-06-01 2011-12-15 Yokogawa Electric Corp Plant simulator
JP5696385B2 (en) * 2010-07-13 2015-04-08 横河電機株式会社 Calorific value calculation system for liquefied natural gas and calorific value calculation method for liquefied natural gas

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100204925A1 (en) * 2005-09-01 2010-08-12 Tareq Abduljalil Albahri Method for measuring the properties of petroleum fuels by distillation
US20120022670A1 (en) * 2006-09-13 2012-01-26 Rockwell Automation Technologies, Inc. System and method for utilizing a hybrid model
US20100191361A1 (en) * 2009-01-23 2010-07-29 Mks Instruments, Inc. Controlling a Manufacturing Process with a Multivariate Model
US20100224365A1 (en) * 2009-03-06 2010-09-09 Carlos Abad Method of treating a subterranean formation and forming treatment fluids using chemo-mathematical models and process control
US20100228485A1 (en) * 2009-03-09 2010-09-09 Soraya Betancourt Method for integrating reservoir charge modeling and downhole fluid analysis

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP2954427A4 *

Cited By (32)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107430398A (en) * 2015-03-30 2017-12-01 环球油品公司 System and method for tuning process modeling
US10534329B2 (en) 2015-03-30 2020-01-14 Uop Llc System and method for improving performance of a plant with a furnace
CN107430398B (en) * 2015-03-30 2020-12-15 环球油品公司 System and method for tuning a process model
US10839115B2 (en) 2015-03-30 2020-11-17 Uop Llc Cleansing system for a feed composition based on environmental factors
WO2017062416A1 (en) * 2015-10-05 2017-04-13 Honeywell International Inc. Hybrid analyzer for fluid processing processes
US11022963B2 (en) 2016-09-16 2021-06-01 Uop Llc Interactive petrochemical plant diagnostic system and method for chemical process model analysis
US10754359B2 (en) 2017-03-27 2020-08-25 Uop Llc Operating slide valves in petrochemical plants or refineries
US10678272B2 (en) 2017-03-27 2020-06-09 Uop Llc Early prediction and detection of slide valve sticking in petrochemical plants or refineries
US10752845B2 (en) 2017-03-28 2020-08-25 Uop Llc Using molecular weight and invariant mapping to determine performance of rotating equipment in a petrochemical plant or refinery
US11396002B2 (en) 2017-03-28 2022-07-26 Uop Llc Detecting and correcting problems in liquid lifting in heat exchangers
US10752844B2 (en) 2017-03-28 2020-08-25 Uop Llc Rotating equipment in a petrochemical plant or refinery
US11130111B2 (en) 2017-03-28 2021-09-28 Uop Llc Air-cooled heat exchangers
US10663238B2 (en) 2017-03-28 2020-05-26 Uop Llc Detecting and correcting maldistribution in heat exchangers in a petrochemical plant or refinery
US10794644B2 (en) 2017-03-28 2020-10-06 Uop Llc Detecting and correcting thermal stresses in heat exchangers in a petrochemical plant or refinery
US11037376B2 (en) 2017-03-28 2021-06-15 Uop Llc Sensor location for rotating equipment in a petrochemical plant or refinery
US10844290B2 (en) 2017-03-28 2020-11-24 Uop Llc Rotating equipment in a petrochemical plant or refinery
US10670027B2 (en) 2017-03-28 2020-06-02 Uop Llc Determining quality of gas for rotating equipment in a petrochemical plant or refinery
US10670353B2 (en) 2017-03-28 2020-06-02 Uop Llc Detecting and correcting cross-leakage in heat exchangers in a petrochemical plant or refinery
US10962302B2 (en) 2017-03-28 2021-03-30 Uop Llc Heat exchangers in a petrochemical plant or refinery
US10695711B2 (en) 2017-04-28 2020-06-30 Uop Llc Remote monitoring of adsorber process units
US11365886B2 (en) 2017-06-19 2022-06-21 Uop Llc Remote monitoring of fired heaters
US10913905B2 (en) 2017-06-19 2021-02-09 Uop Llc Catalyst cycle length prediction using eigen analysis
US10739798B2 (en) 2017-06-20 2020-08-11 Uop Llc Incipient temperature excursion mitigation and control
US11130692B2 (en) 2017-06-28 2021-09-28 Uop Llc Process and apparatus for dosing nutrients to a bioreactor
US10994240B2 (en) 2017-09-18 2021-05-04 Uop Llc Remote monitoring of pressure swing adsorption units
US11194317B2 (en) 2017-10-02 2021-12-07 Uop Llc Remote monitoring of chloride treaters using a process simulator based chloride distribution estimate
US11676061B2 (en) 2017-10-05 2023-06-13 Honeywell International Inc. Harnessing machine learning and data analytics for a real time predictive model for a FCC pre-treatment unit
US11105787B2 (en) 2017-10-20 2021-08-31 Honeywell International Inc. System and method to optimize crude oil distillation or other processing by inline analysis of crude oil properties
US10901403B2 (en) 2018-02-20 2021-01-26 Uop Llc Developing linear process models using reactor kinetic equations
US10734098B2 (en) 2018-03-30 2020-08-04 Uop Llc Catalytic dehydrogenation catalyst health index
US10953377B2 (en) 2018-12-10 2021-03-23 Uop Llc Delta temperature control of catalytic dehydrogenation process reactors
US20230368590A1 (en) * 2022-05-13 2023-11-16 Regents Of The University Of Minnesota System and method for controlling a compression ignition engine

Also Published As

Publication number Publication date
US20140309756A1 (en) 2014-10-16
US10175661B2 (en) 2019-01-08
CN105229627B (en) 2019-06-21
EP2954427A4 (en) 2017-05-31
CN105229627A (en) 2016-01-06
EP2954427A1 (en) 2015-12-16
JP2016507113A (en) 2016-03-07

Similar Documents

Publication Publication Date Title
US10175661B2 (en) System, method and apparatus for determining properties of product or process streams
JP2016507113A5 (en) System and method for determining characteristics of a product or process stream
US10235481B2 (en) System and method for online measurement of vapor pressure in hydrocarbon process streams
EP3289420B1 (en) Computer system and method for causality analysis using hybrid first-principles and inferential model
US6897071B2 (en) Topological near infrared analysis modeling of petroleum refinery products
RU2498274C2 (en) System and method for alkylation process analysis
US8512550B2 (en) Refinery crude unit performance monitoring using advanced analytic techniques for raw material quality prediction
US20140278141A1 (en) Direct match spectrographic determination of fuel properties
US8481942B2 (en) Direct match spectrographic determination of fuel properties
KR20110086706A (en) Method for predicting hydrocarbon process stream stability using near infrared spectra
He et al. A novel adaptive algorithm with near-infrared spectroscopy and its application in online gasoline blending processes
He et al. Online updating of NIR model and its industrial application via adaptive wavelength selection and local regression strategy
Nespeca et al. Rapid and simultaneous prediction of eight diesel quality parameters through ATR-FTIR analysis
WO2013096240A9 (en) Global crude oil quality monitoring
WO2006126978A1 (en) Topological near infrared analysis modeling of petroleum refinery products
Leal et al. Using spectroscopy and support vector regression to predict gasoline characteristics: a comparison of 1H nmr and nir
de Souza et al. PAT soft sensors for wide range prediction of key properties of diesel fuels and blending components for the oil industry
CA3026377C (en) System and method for online measurement of vapor pressure in hydrocarbon process streams
US20170097330A1 (en) Hybrid analyzer for fluid processing processes
ShahnOvSky et al. Integrated monitoring for optimising crude distillation
US11402323B2 (en) Systems and processes for performance property determination using optical spectral data
RU2750060C2 (en) Method and system for industrial process control using trained machine learning algorithm (mla)
Gourvénec et al. 38 NIR Spectroscopy Applied to the Oil and Gas Industry
Fan et al. Online Determination on the Properties of Naphtha as the Ethylene Feedstock Using Near-Infrared Spectroscopy
Aparicio et al. Bromine number prediction for Colombian naphthas using near-infrared spectroscopy and chemometric methods

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 201480017536.2

Country of ref document: CN

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 14748750

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2014748750

Country of ref document: EP

ENP Entry into the national phase

Ref document number: 2015556250

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE