Statistical process control for air separation process

• US 5,257,206 A
• Filed: 04/08/1991
• Issued: 10/26/1993
• Est. Priority Date: 04/08/1991
• Status: Expired due to Fees

• Current Assignee: Praxair Technology Incorporated (Linde Plc)
• Original Assignee: Praxair Technology Incorporated (Linde Plc)
• Inventors: Hanson, Thomas C.
• Primary Examiner(s): Cosimano, Edward R.
• Application Number:
  US07/663,595
• Time in Patent Office: 932 Days
• Field of Search: 364/141, 364/148, 364/149, 364/150, 364/152, 364/469, 364/502
• US Class Current: 700/273
• CPC Class Codes: F25J 2290/10 Mathematical formulae, modeling, plot or curves; Design methods
• CPC Class Codes: F25J 3/04303 Lachmann expansion, i.e. expanded into oxygen producing or low pressure column
• CPC Class Codes: F25J 3/04412 in a classical double column flowsheet, i.e. with thermal coupling by a main reboiler-condenser in the bottom of low pressure respectively top of high pressure column
• CPC Class Codes: F25J 3/04678 cooled by oxygen enriched liquid from high pressure column bottoms
• CPC Class Codes: F25J 3/04848 Control strategy, e.g. advanced process control or dynamic modeling
• CPC Class Codes: G05B 21/02 electric

Abstract

Statistical process control system for an air-separation plant using off-line and on-line computer means to define control chart limits, trend thresholds and dynamic models for use with on-line process variable data for determination of statistical process control.

Claims

1. Statistical process control method for an air separation process for the production of one or more constituents of the air feed to the process wherein historical data for each of a plurality of pre-determined process variables of said process is electronically available, said method comprising:

(i) providing computer means under program control;

(ii) computer generating with said computer means control chart parameters and alarm thresholds for a distribution type based on the probability density function (PDF) of said historical data for at least one of said process variables;

(iii) providing an electronic data base interactive with said computer means and storing said control chart parameters in a data file of said data base;

(iv) storing in a data file of said data base continuous data for at lest one of said process variables in a time series of predetermined data sample groups;

(v) computer generating with said computer means alarm criteria based on said control chart parameters;

(vi) storing said alarm criteria in a data file of said data base;

(vii) computer generating with said computer means a stationary auto-regressive integrated moving average (ARIMA) type series model based on said time series of predetermined data sample groups stored in a data file of said data base for calculating a forecast of the variable data;

(viii) storing said ARIMA model in a data file of said data base;

(ix) computer generating on-line with said computer means the residuals of real-time process variable data from said air-separation process on-line for said at least one of said process variables and the forecast of the variable data from said ARIMA model;

(x) computer generating on-line with said computer means a control chart based on said residuals, said alarm criteria and said forecast;

(xi) computer generating the trend of a predetermined number of said residuals of consecutive variable data values on-line with said computer means and displaying said trend on said control chart; and

(xii) computer generating a visual display of an alarm indication on-line with said computer means based on said alarm criteria when the trend for a predetermined number of said residuals for consecutive variable data values continuously increases or continuously decreases.

2. Improvement in accordance with claim 1 wherein said computer means generates a control chart which displays said parameters and values based on observations of real-time data for at least one of said process variables.

3. Improvement in accordance with claim 1 wherein said control chart parameters are stored in said electronic data base.

4. Improvement in accordance with claim 1 wherein historical data for each of a plurality of pre-determined process variables is electronically collected for at least one (1) twenty-four hour period for one or more of each such variable.

5. Improvement in accordance with claim 1 which comprises the additional steps of computer generating a histogram of said time series with said computer means and determining from said histogram a designation for a distribution form of the time series of said histogram.

6. Improvement in accordance with claim 1 wherein said air separation process is under control of a process control computer which utilizes at least one electrically controlled set point to adjust at least one control device of said process and wherein said method includes the step of computer generating an electrical signal which is communicated to said process control computer to adjust at least one said electrically controlled set point when an alarm threshold is attained.

7. Improvement in accordance with claim 1 wherein an electronic data base interactive with said computer means is provided for electronically storing in a data file thereof continuous historical data for said plurality of pre-determined process variables.

8. Improvement in accordance with claim 7 which includes the step of electronically storing in a data file of said data base continuous data for one or more air separation process variables in a time series of predetermined data sample groups.

9. Improvement in accordance with claim 8 which includes the step of computer generating a histogram of said time series with said computer means.

10. Statistical process control method for an air separation process for the production of one or more constituents of the said feed for the process wherein the process has been running acceptably for a relatively long period of time so that historical data for each of a plurality of pre-determined process variables of said process is electronically collectable, the said method comprising:

(i) providing computer means for on-line and off-line computation under program control and an electronic data base interactive with said computer means;

(ii) electronically storing in a data file of said data base continuous historical data for one or more air-separation process variables in a time series of predetermined data sample groups;

(iii) computer generating with said computer means control chart parameters corresponding to a distribution type based on the probability density function (PDF) of said time series of said historical data stored in a data file of said data base and a designated confidence level;

(iv) storing said control chart parameters in a data file of said data base;

(v) observing and processing real-time variable data from said air-separation process on-line with said computer means and displaying observations of said real-time data on a control chart together with said control chart parameters;

(vi) computer generating the trend of a predetermined number of such observations for consecutive, on-line variable data values with said computer means and displaying said trend on said control chart; and

(vii) computer generating a visual display of an alarm indication on-line with said computer means when said trend for a predetermined number of real-time values for consecutive variable data values continuously increases or continuously decrease beyond said control chart parameters.

11. Statistical process control method for a separation process for the production of one or more constituents of the air feed to the process wherein the process has been running acceptably for a sufficiently long period of time so that data for each of a plurality of pre-determined process variables is electronically collectable, said method comprising:

(i) providing computer means for off-line and on-line computation under program control and an electronic data base interactive with said computer means;

(ii) electronically storing in a data file of said data base continuous data for one or more air separation process variables in a time series of predetermined data sample groups;

(iii) computer generating a histogram of said time series of predetermined data sample groups with said computer means;

(iv) determining from said histogram a designation for a distribution form of the time series of said histogram;

(v) computer generating with said computer means control chart parameters corresponding to said distribution form based on the probability density function (PDF) of said time series of predetermined data sample groups stored in a data file of said data base and a designated Type I error;

(vi) storing said control chart parameters in a data file of said data base;

(vii) computer generating with said computer means alarm criteria based on said control chart parameters stored in a data file of said data base;

(viii) storing said alarm criteria in a data file of said data base;

(ix) computer generating with said computer means a stationary auto-regressive integrated moving average (ARIMA) type series model based on said time series of predetermined data sample groups stored in a data file of said data base for calculating a forecast of the variable data;

(x) storing said ARIMA model in a data file of said data base; and

(xi) observing and processing real-time variable data from said air separation process on-line in said computer means and displaying observations of said real-time data on a control chart together with said alarm criteria based on said control chart parameters and said forecast of the variable data from said ARIMA model.