Computerized Power Demand Control in an Electrochemical Plant

The invention relates in general to power demand control for switching electrical loads according to a priority schedule and more particularly to a power demand control system featuring cell line control, in an electrochemical plant facility supplied with energy from the utility company, so as to keep the demand at the end of a period of control within a preassigned demand limit while maximizing plant production.

Power demand control is known in which loads are controlled in accordance with a priority schedule which is changed, or fixed, in accordance with the dynamics of current consuming production units and of sheddable loads. Adaptive load priority determination with a power demand control system is generally known from U.S. Pat. No. 4,204,127, issued May 20, 1980. The system and method disclosed in the aforementioned patent application have been described in the context of an arc furnace installation, since in such installations the decision to switch, or not to switch, should not only take into account the requirements of power demand limit but also the economics and requirements of production. For the purpose of the present description, U.S. Pat. No. 4,204,127 is hereby incorporated by reference.

This invention is directly applicable to a computerized control system for the heavy chemical industries which are using an electrolysis process, such as the chlor-alkali industries. Thus, chlorine and caustic soda are produced almost entirely by electrolytic methods from fused chlorides, or aqueous solutions of alkali metal chlorides. In such a plant the energy consumed by the electrolyzer cells represents the large majority of the overall energy consumption from the utility company power line. The energy consumed in the electrolysis of the brine is the product of the current flowing and the potential of the cell. The load factor, e.g. the ratio average to maximum demand is high, namely of the order of 94% to 96%. Despite such a high load factor, any increase by a few percentages, say up to 98 or 99%, represents a sizable yearly profit increase to the manufacturer.

Accordingly, an object of the present invention is to provide a power demand control system which, besides some sheddable and reconnectable loads, operates mainly on electrolytic cells which can be controlled so as to maximize productivity without exceeding a power demand limit.

Another object of the present invention is to provide, in accordance with a priority schedule, for mininal switching and adjusting of loads mainly including electrolytic cells while closely meeting a power demand limit not to be exceeded at the end of any of successive demand periods.

A further object of the present invention is to control with a computer, during a given demand period, the power rate of a plurality of electrolytic cells in an electrolytic plant in order to attain an ideal rate of consumption before the subsequent demand period, thereby to minimize switching and adjusting of loads under the constraint of an assigned power demand limit.

Still another object of the present invention is to maximize the production from electrolytic cells in a chemical plant by cell selection and cell adjustment while minimizing the selection and adjustment control steps, and maintaining each cell between a maximum and a minimum amperage.

In a power utilization system including loads drawing energy from a utility company, electrolytic cells are adjusted in order to draw prescribed variable amounts of power as part of a power demand control system having an assigned demand limit. In order to provide regulation of the manufacturing process involved with the cells and at the same time meet the power demand constraints, energy is supplied to each electrolytic cell line through an individual tapping transformer controlled by a computer and each cell line has an individual power meter and an individual amperemeter feeding back information to the computer. Current is controlled by tap regulation under command of the computer in order to achieve toward the end of each demand period, an ideal rate of consumption, whereby the overall plant totalizes energy consumed in such amount to meet the assigned target. To the extent that the achieved ideal rate of consumption maintains itself in the user's facility, electrolytic cell line regulation is no longer required right after the end of the particular demand period. Therefore, excessive tap changing is avoided.

Corrective action in response to a demand error is effected by the control system according to the invention through changing the demand error by an amount sufficient to allow re-establishment of the ideal rate of consumption toward the end of each demand period.

The invention permits regulatory action on several electrolytic cell lines, the consumption at a given time being a function of an assigned power demand limit and a function of the availability of tap regulation during integrated portions of the demand period of control, thereby to maximize the utilization of the overall cell line capacity within the entire industrial plant.

Control is performed on units of production having a priority order for control, so as to bring the overall production to ideal rate before each new demand period. Power demand control may involve other types of loads, for instance loads that can only be switched ON or OFF, or loads having constraints making them available for control only when the attached constraints have been lifted as a result of fulfillment of assigned operative requirements, such as the expiration of prescribed times ON or OFF, or by any externally lifted prohibition, or which have in any form regained their degree of freedom.

The technique of computerized power demand limit is applied to an electrochemical facility supplied with electrical power from the utility company and this is achieved here without adversely affecting the output by the cell lines of the chemical plant.

As generally known in power demand control, the synchronizing ("DEMAND") pulse from the plant KWH meter resets the computer demand period timer and actual timing of the period is from the computer real-time clock. KWH pulses from the plant KWH meter are received by the controller as interrupts (or as frequently scanned contact closure inputs (CCI's)) which represent KWH consumed. These interrupts initiate the calculation of the total KWH in the period in relation to the elapsed time of the period. From these data an actual point on the demand curve is determined, and the actual slope of the curve is known preferably by using a weighted average of the last three pulse periods (last four pulses). This actual slope is used to predict final demand at the end of the period. In accordance with the present invention, depending on the demand prediction, on the time into period, and the availability of cell lines which can be tapped up or down, the computer may either lower or raise the taps of an electrolytic cell.

Each cell line is supplied with DC current derived by rectification from the AC power network and through an individual transformer. The transformer has taps for regulation purpose. Typically, a cell line is used in the production of chlorine by electrolytic dissociation of a brine solution in a vat. The main purpose is to increase the load factor without increasing electrical demand charges.

Another requirement is to hold each cell line within specified electrical current limits in order to keep peak efficiency for the process. The electrical current is sensed and derived as an analog input. The cell line individual power and the total plant power are measured by pulsing individual and total plant KWH meters.

The control system determines how much to regulate a selected cell line by reference to previous tap adjustments effected on the associated transformer upward, or downward. A minimum (Lo A) and a maximum (Hi A) current per cell line have been set by the operator. Therefore, the transformer taps are adjusted without exceeding such limits. In this regard, tap changes are performed by reference to the changes in the analog input observed from a preceding tap move. Upon each regulating step, the number of taps necessary to meet the demand limit and the number of taps possible within the permissible range are determined. When the required demand correction is not possible, or is limited, another cell line is selected and adjusted in the same fashion. The sensed value of the current increment .DELTA.KA in the cell line and the resulting power increment .DELTA.KW are taken from past history, namely at the last step of regulation.

Should the characteristics of the cell lines change, productivity maximization might require the intervention of an operator, for instance setting new limits (Hi A and Lo A). In order to protect the system against unwarranted intervention by the operator, provision is made in the system, according to the invention, for different degrees of freedom in several tiers. Thus, human interfacing is broken into four tiers: (1) display only (without panel key), (2) protected process and control operator functions, (3) password regulated engineer's functions and (4) password programmer's functions for detailed debugging, memory modification and simulating malfunctions. For instance, the engineer will have entered boundaries so that upon any setting of Hi A, or Lo A, by the operator outside those limits, the system will reject the entry. When equipment malfunctions, an alarm is triggered, the operator is informed of the cause, and advised of possible solutions and reset procedures.

As a result, according to the present invention, power demand limit control in a plant is integrated with control of production units in the nature of electrolytic cells, as part of an adaptive control process.

The invention will now be described in the context of the supply of electrical energy by a utility company to an industrial plant including a plurality of electrolytic cell lines, and reference shall be made to the accompanying drawings in which:

FIG. 1A illustrates diagrammatically the principle of calculation of the demand error for power demand control;

FIGS. 1B and 1C illustrate diagrammatically cell line control for the two successive phases of control according to the present invention;

FIG. 2 is an overview of a plant including the computerized control system according to the invention;

FIG. 3 is a circuit diagram illustrating the steps used in the generation of an error signal conditioning regulatory action with the computerized control system according to the invention;

FIG. 4 shows how, by tap changing, the system is brought to operate under the ideal rate of consumption toward the end of the demand period, in accordance with the present invention;

FIG. 5 illustrates in diagram form the method of controlling the tap of the individual transformer of a particular cell line;

FIG. 6 is a diagram illustrating tap and deadband control for a selected individual cell line with the computerized control system according to the present invention;

FIGS. 7A-7G are flow charts showing routines and subroutines accomplishing the various tasks for handling CCI inputs in digital form, and for calculating errors and parameters leading to tap changing for maximizing productivity within the limit of an assigned power demand.

For the sake of clarity and for the purpose of illustration only, the present invention will be described in the context of the power demand control system of U.S. Pat. No. 3,872,286, issued to R. E. J. Putman, and U.S. Pat. No. 4,204,127, issued to W. C. Carter. The descriptions of the aforementioned patent and patent application are hereby incorporated by reference.

FIG. 1 illustrates the principle of calculation of the error for any point M along the trajectory during a demand period of 30 minutes, for instance. A clock installed by the power supply company determines the initial time of each demand period, (or the final time of a preceding demand period). The watthour meter provides a "KWH pulse" which represents the magnitude of the power which has been consumed during a certain time .DELTA.t corrsponding to a full rotation of the disc of the meter, representing a constant increment or unit of energy (KWH). Thus, the .DELTA.t interval appearing along the time axis is essentially variable. This time interval is detected as a representation of the slope at point M and it represents the power P.sub.t in KW hour/hour. If the load of the plant is maintained in the same condition until the end of the 30 minute period (for instance), the energy curve will follow the tangent MW. However, the Demand Limit is set at T, therefore there is an error TW by default, which in terms of power is: ##EQU1## as is evident from the geometry of triangles MNT and MWT. Having determined Slope=P.sub.t /60 by triangulation, and the sign of the error (depending upon whether the intersection point W is above (-) or below (+) the target T) power demand control is effected in accordance with general practice, by selectively adding, or shedding, suitable loads in the plant.

However, as explained hereinafter, besides power demand control during successive demand periods, the present invention proposes to effect cell line control within a user's facility in order to reach an ideal rate of consumption before the end of the demand period. The ideal rate curve is shown in FIG. 1B by line OT, a straight line from the origin at time t=0 to the target T at the end of the period. Besides the cell lines which are adjustable current consuming loads, the plant includes other loads, for instance the air conditioning units, the fans, the light, elevator banks, for instance. The major current consuming loads may represent 80% of the overall consumption in the plant. Under such conditions, the target, or assigned demand limit (DEM LIM) is such that when such major units of production are at their best load factor, e.g. 100%, the power rate is constant and the operative point M.sub.t follow ideal rate (IR) line OT. Nevertheless, during the demand period, the loads other than such major adjustable loads, and also some discrepancies in the unities of production themselves, will bring the operative point to deviate from line OT. For the purpose of illustration, it is assumed that at time t=10 minutes, for a demand period of 30 minutes, the operative point is at M.sub.t, rather than at m on line OT. At the same time tangential vector MW defined by the actual power P.sub.t at this point M indicates an inticipated error TW relative to the target. Instead of correcting this error under conventional power demand control, the present invention proposes to change the anticipated error by an amount sufficient to return on the ideal rate (IR) curve OT. To this effect, a point M.sub.IR is arbitrarily chosen, five minutes before the end of the period in the instance of FIG. 1B, on the ideal rate curve OT. Control, during a field of control ab, of the power rate at point M.sub.t deflects MW toward M.sub.IR. If no disturbance occurs and point M.sub.IR has been reached the power rate is changed again to M.sub.IR T, so that the loads will now follow the ideal rate line OT to the end of the period. Therefore, the target T will be reached under the ideal power rate so that no further control will be necessary at the beginning of the subsequent period. However, as shown in FIG. 1C, M.sub.t could at time t=25 minutes, be either too high (M.sub.1) or too low (M.sub.2) relative to M.sub.IR. If at M.sub.2, the slope M.sub.2 T to reach the target will be larger than the ideal rate slope (B) parallel to OT, whereas at M.sub.1 slope M.sub.1 T is smaller than B. In accordance with the present invention, once in the second phase of cell line control, namely, at the ideal rate start time (t=25 minutes in the illustration given for a 30 minute demand period), it is ascertained whether the slope (M.sub.1 T, or M.sub.2 T) to the target T is larger, or smaller, than the ideal rate slope B, e.g. whether M is below, or above, the operative point M.sub.IR intended in the first phase of cell line control. If M.sub.t is above M.sub.IR (at M.sub.1), the power rate is changed from the current value E to M.sub.1 T, thereby reaching the target, although at a slope less than ideal rate since the latter would bring the demand at the end of the demand period at D.sub.E ' in excess of the assigned demand limit. If M.sub.t is below M.sub.IR (at M.sub.2), the power rate is changed from the current value E to the ideal rate B. The demand at the end of the demand period will be D.sub.E, thereby missing somewhat the target from below, but the loads will be at the ideal rate, for the subsequent demand period. It appears that the present invention allows (1) to substantially reach the target T, while (2) substantially following the ideal rate before the end of the demand period, e.g. at the beginning of a new demand period, it being understood that, for the sake of clarity, FIG. 1C shows with exaggeration the discrepancies at M.sub.1 and M.sub.2 about the intended point M.sub.IR.

Indeed, should during the first phase of control the correction made for operative point M.sub.t, an error TW develop which no longer matches the effort to reach point M.sub.IR, a new correction of the error TW will again take place by cell line control, during the same first phase of control, so as to point again toward M.sub.IR. At the end of the period of control ab, M.sub.IR may be reached while progressing further toward the end of the field of control ab. As the system becomes closer to the target there is a tighter orientation toward M.sub.IR. Once the ideal rate has been established toward the end of the demand period there will no longer be any control in the next demand period because of IR'.

It appears from the above that the invention provides for a no control field (Field #1) initially in the period (for instance, up to 5 minutes at a, according to FIG. 1B) followed by a first phase of control from a to b (field #2) and finally a second phase, e.g. an ideal rate control field of 5 minutes (field #3). If the system does not detect any disturbance following regulation to ideal rate consumption from M.sub.IR in the preceding period, an equilibrium is present in which no control at all is required for the next period (field #1). Therefore tap changes, at points such as M.sub.t or M.sub.IR, will be minimized in the overall succession of demand periods.

In order to closely follow the target, while not overreacting to rapid transient conditions, a deadband is provided on either side of the trajectory (see FIG. 1A) by defining two angles which should not be exceeded. The deadband will permit effective demand control while avoiding excessive control activity as long as the projected tangent remains within the two limits MU and MV defined by such angles. The upper limit MU will represent the "decrease vector" and the lower limit MV of the deadband will represent the "increase vector" for control. Such deadband, preferably, is made variable throughout the period of control becoming smaller and smaller as the error calculation becomes more accurate while advancing into the demand period.

Referring to FIG. 2 a plant facility is shown to include n cell lines CL.sub.1 . . . CL.sub.n of an electrolysis production system. The utility company lines L.sub.1, L.sub.2, L.sub.3 supply energy via bus lines l.sub.1, l.sub.2, l.sub.3 in the plant to individual transformers T.sub.1 . . . T.sub.n which have taps at their secondary. The tapped secondary output is fed in a corresponding converter (CNV.sub.1 . . . CNV.sub.n) generating direct current on bus lines to the electrodes of the electrolytic cell line (CL.sub.1 . . . CL.sub.n). The plant also includes loads such as LD.sub.1 . . . LD.sub.n which are energized from the main power lines L.sub.1, L.sub.2, L.sub.3.

The main KWH meter M of the utility company is used to generate KWH pulses on line 1 as earlier stated. Individual wattmeters M.sub.1 . . . M.sub.n and amperemeters A.sub.1 . . . A.sub.n provide indication of the individual power consumptions and of the individual ampere consumptions, respectively.

The computerized control system 10, according to the present invention receives input signals: on line 1 from watthour meter M; on line 2 from M.sub.1 . . . M.sub.2 indicating watts W.sub.1 . . . W.sub.n ; on line 5 from A.sub.1 . . . A.sub.n indicating the amperage A.sub.1, . . . A.sub.n. It generates error signals in accordance with FIGS. 1A, 1B and 1C, and causes changes in the power rate of one or more of the cell lines CL.sub.1 . . . CL.sub.n in accordance with FIGS. 1B and 1C, by lines 4 carrying command signals to selected tap changers TP.sub.1 . . . TP.sub.n. Also in accordance with general power demand control technique, the computerized control system 10 may cause certain loads such as LD.sub.1 . . . LD.sub.n to be switched ON or OFF by lines 6 which carry commands to switches SWL.sub.1 . . . SWL.sub.n of these loads.

Provision is also made in the system for protective Auto-Manual switching, ON or OFF, of any of the cell lines. Individual switches SW.sub.1 . . . SW.sub.n are thus inserted at the energy input to individual cell lines CL.sub.1 . . . CL.sub.n. The system also receives from the operator the assigned demand limit DEM LIM which fixes the target C to be reached in each demand period as well as the Hi A and Lo A settings of the ampere limits mentioned earlier.

The technique of power demand control of loads such as CL.sub.1 . . . CL.sub.2 and LD.sub.1 . . . LD.sub.n is well known and explained at length in the aforementioned patent and patent application incorporated by reference. It will suffice to recall that in order to control the power demand by shedding or adding loads, the control system is responsive to the power consumption continuously recorded by the kilowatthour meter of the power supply company. Referring to FIG. 3, the computerized control system 10 receives over line 1 a KWH pulse which as a .DELTA.t characterizes the consumption at any particular instant within the demand period. The power supply company also provides timers which by blocks 63, 65 and 67 determine the beginning and the end of each demand period, and the time left in the demand period. In the instant case it is assumed that each such demand period lasts 30 minutes. For each turn of the disc of the meter there is a pulse generated which will be hereinafter called "KWH pulse". The succession of these pulses represents on a time scale the energy consumed for one turn of the disc. The computer system, through its interrupt (contact closure input) unit, assesses the status of the status contact interrupts, and more generally, monitors all the input data fed into the computer system regarding the individual loads in the plant with their constraints, effectuates calculations, makes decisions which are converted, after each of the above-mentioned KWH pulses, into whatever load control action is necessary through the controlled operation of relays.

Included as part of the control system, is a computer system typically known as a W-2515 under its trade name. It can be a digital computer system, such as a Prodac 2000 (P2000) sold by Westinghouse Electric Corporation. A descriptive book entitled "Prodac 2000 Computers Systems Reference Manual" has been published in 1970 by Westinghouse Electric Corporation and made available for the purpose of describing in greater detail this computer system and its operation. The input systems, associated with the computer processor are well known and include a conventional contact closure input system to effectuate scanning of the contacts or other signals representing the status of the equipment. Also, operator controlled and other information input devices and systems are provided such as a teletypewriter. The contact closure output system is also conventional and part of the Prodac 2000 general purpose digital computer system sold.

The controller is a Westinghouse W2500 based mini-computer with between 32 and 64K 16 bit words. The inputs to the controller are:

The outputs are:

The computer system used in the control system according to the invention includes both Hardware and Software. For instance, the interrupt unit is associated with an interrupt handler. Software is being used as a convenient means of quickly and efficiently performing operations as required in monitoring data, doing calculations, making decisions and translating treatment of information into control action within the short time intervals determined by the recurrent transmission of KWH pulses from the power supply company meter.

It is observed that the inputs consist of interrupts (internally generated via frequently scanned CCI's or external EI's) which are successively handled by the process interrupt handler. One interrupt will receive the 48 V DC pulse generated by the external clock and is used to reset the associated registers in the computer when it is received. Another interrupt will receive a train of 48 V DC pulses transmitted by the meter, each pulse representing KWH consumed. Another interrupt could be reserved for a second KWH meter if needed. Three other interrupts could be used to receive a status which corresponds to the status of one load contact in the plant and belongs to one group of three associated with one scan contact output.

The normal operator interface with the system will be via a CRT. A log of system performance could also be provided together with any other messages that may be required. Via the CRT keyboard the operator will also be able to change the values of various constants relating to the system as a whole or to individual items of equipment. The time and date and on-peak and off-peak demand levels can also be changed using the same keyboard.

In response to a clock pulse the program puts out data to be printed out for the preceding demand period. All registers are initially cleared in which accumulated values are stored including time into period and KWH during the period. Prorated values of time and KWH are stored in those registers when the KWH pulse does not coincide with the clock pulse.

The plant and each cell line is checked for faulty KWH or KA metering, trip or other unusual condition. Parts or all of the control may be suspended and the trouble alarm activated under these circumstances. If any cell line is found to be out of prescribed limits, the system adjusts the cell by tapping in order to bring the load within limits.

A KW error signal ERR is calculated (circuit 71 of FIG. 3) from the time between KWH meter pulses by predicting the energy usage at the end of the demand period assuming the rate remains constant from that time on.

The demand period is divided into four sections (not necessarily equally). A deadband, CD, and a hysteresis, H, are provided which are used during four successive control fields as follows:

As earlier mentioned, ideal rate control serves to minimize tap changes over the long run.

The cell lines are classified in accordance with a priority schedule set by the operator into the cell line control system 10. Limits are set on the amperage (KA) which is drawn by the particular cell line. Cell line control is responsive to an individual KWH meter and an analog input derived from an individual ampere meter. Each tap move is estimated from the previous tap change and its effects as increments .DELTA.KW and .DELTA.KA on KW and KA. These estimations are done after the tap has had time to settle into position.

The controller periodically checks whether taps need to be moved to control demand. If movement is required, the lowest priority cell line will be moved first as far as possible toward its limit. If not enough compensation can be made by this line, the next priority line will begin moving simultaneously. Each control period moves only one tap per cell line. The next control period re-calculates errors from the start, including new estimates of KW/TAP and KA/TAP. The object is to choose the minimum change between the necessary tap changes to control the KW and those possible within the KA limit. When not enough of the expected control is available with one cell line, the next cell line in the order of priority is used.

The invention will now be described in detail by reference to FIGS. 3, 5 and 6 which indicate in block diagram form how the energy management control system performs the required control steps, by reference to FIG. 4 illustrating the effect of tap changing in bringing the system to operate in the ideal rate, and by reference to FIGS. 7A-7G, which are flow charts characterizing the computer process in all modes of operation which are material to the present invention.

The generation of the error signal in the field of control of the demand period will first be explained by reference to FIGS. 1B, 1C, 3 and 4. FIG. 1B shows the ideal rate line OT which the system attempts to reach from demand period to demand period and where it would stay when the electrolytic cell lines CL.sub.1 . . . CL.sub.n consume the right current and in the absence of any major disturbance from the overall demand in the plant. Thus, no control would be necessary since the ideal rate will automatically lead, within the deadband allowance, to the assigned demand limit or target T at the end of each demand period. However, due to discrepancies in the cell lines, or to other loads in the plant, like LD.sub.1 . . . LD.sub.n in FIG. 2, the operative point at time t is at M.sub.t rather than at m on the IR characteristic. FIG. 3 shows circuitry which leads to bringing M back on the ideal rate line OT before reaching the target T. A first correction in the power rate is effected at time t in order to bring, as illustrated in FIG. 1B, the operative point to M.sub.IR which is typically the ideal rate point at time 25 minutes, e.g. 5 minutes before the end of the demand period, thereby changing the power rate of point M.sub.t from MW to MM.sub.IR. A second correction is effected when point M.sub.IR has been reached. At this point, as illustrated in FIG. 1C, the power rate is changed to become the ideal rate slope OT, so that, from M.sub.IR on, the system follows M.sub.IR T to the target. Thus, the system according to the invention provides a first adjustment at M.sub.t followed by a second adjustment at M.sub.IR.

It is noted, however, that the system provides for power demand control concurrently with cell line control. An error between MW and MT is detected which will amount to TW in KWH, short of the target at the end of the period. Demand control by the technique known and described in the U.S. Pat. No. 3,872,286 of R. Putman, or in the U.S. Pat. No. 4,204,127 of W. C. Carter, when applied alone would bring the operative point at the target T via line MT. It is to a such control action that the present invention applies the two above-mentioned successive control actions, first at M.sub.t, to follow M M.sub.IR, secondly at M.sub.IR, to follow M.sub.IR T.

Considering now FIG. 4, the power level in the period when the operative point is at M.sub.t, namely time t in the period, is shown by a horizontal line passing by point A, which represents a constant level relative to the target. If M.sub.t is on target the level is zero. The error is thus the area under the level A. In order to nullify this error, control of the power rate of the cell lines--in the absence of any external disturbance--would conventionally consist in bringing progressively the consumption below the horizontal axis and as far as the level B so that within the period the two areas above and below the axis be equal. The same process during the subsequent period CD will keep the overall consumption from demand period to demand period on the required target. In accordance with the present invention, however, control is extended beyond AB down to B' by artificially changing the error to obtain the compensation level B'. After following the power rate B' until the instant marking 5 minutes before the end of the demand period, namely at C', a second control action brings the actual consumption to a demand limit level D' which is close to the target level at the end of the demand period. Therefore, with the subsequent demand period the ideal rate being reached (which is shown in FIG. 1B by a new line IR' effectively followed by M.sub.t) no control is necessary (D', E' on FIG. 4).

The generation of an error signal under demand control, and the corrections to effectuate control in two steps at such operative points like M.sub.t and M.sub.IR, will now be explained by reference to FIGS. 1B, 1C and 3 while keeping in mind FIG. 4.

Referring to FIG. 3, the kilowatthour meter of the utility company provides at 60 a KWH pulse which is received at 61 and at 62. At 61 the kilowatthours consumed A since the beginning of the demand period are totalized continuously and outputted on line 81. The total is reset to zero by 63 at the end of the demand period in response to the total count of counter 67 counting D=30 minutes (illustratively) and to a clock signal from 65 counting C (in seconds)=the time left in the period. Circuit 62 derives from the KWH meter the time .DELTA.t which between two pulses indicates how many kilowatts are being used by the plant. Thus, the present power P.sub.t, or slope E at the operative point M.sub.t of FIGS. 1A, 1B, 1C, is outputted by circuit 62. This output, or power consumed P.sub.t at time t, is at 64 multiplied by C the time left in the period, known from circuit 65, in order to indicate the amount of KWH predicted to be used in the time left. This amount from 64 is added to the KWH already consumed of circuit 61, thereby deriving at 66 the predicted KWH amount due to MW (FIGS. 1A, 1B) the projection of the tangent at M.sub.t to the end of the period. The average is derived by dividing at 68 by the demand period D. TW, the error ERR from the target is the difference between the KWH demand limit (B of circuit 70 ) and the output of circuit 68. In order to make power demand control in response to such error compatible with the intended cell line control according to the present invention, both the KWH demand limit (ordinate of T on FIGS. 1A, 1B) and the expected demand (ordinate of W on FIGS. 1A, 1B if MW of FIG. 1B were made to follow MT) are converted into units of average power in the demand period. Accordingly, the expected energy consumption given at the output of circuit 66 is in circuit 68 divided by the demand period D (Dem Per) and subtracted in circuit 71 from the KWH limit at the target directly given by circuit 70 as a limit on average power. The output of circuit 71 is the power demand control error, ERR e.g. the expected demand error caused by the angle between MT and MW on FIG. 1B. Since, according to the present invention, cell line control should bring the slope of the power rate at M.sub.t from MW to M.sub.IR, rather than MT, cell line adjustment due to the error of circuit 71 will be replaced by cell line adjustment in response to a new error obtained by a modification of the demand control error of circuit 71. Such modification is determined by the aforementioned condition that when time will be 25 minutes, the power point should be at M.sub.IR.

The error predicted (ERR at 71) under power demand control (e.g. along MT) is the Demand Limit (DEM LIM at 70) less the sum of the current energy usage (A at 81) and the consumption in the time left (C at 84) at the present power rate (E at 82 or 87).

In order to change MT to MM.sub.IR, the system takes into account the fact that during the last 5 minutes, the ideal rate will be consumed rather than 5 minutes at the present rate. The error should then be reduced by the latter amount and increased by the former amount. Since the ideal rate is Demand Limit/Demand Period followed during 5 minutes (or D from 77 and G from 75), the correction is as calculated by 73; (Demand Limit-PW).times.G/D. The output of multiplier 73 is the correction applied by circuit 72 to the error ERR of circuit 71 in order to effect via line 89 the required target change during the first phase of control.

In order to determine when time instant t=25 minutes has been reached within the period, operational amplifier OA.sub.1, used as a comparator, detects when the time left C from circuit 65 has become equal to 5 minutes, or rather 5.times.60 seconds. At such time by diode D.sub.2 the circuit 78 is enabled while, due to diode D.sub.1, circuit 72 is disabled. Circuit 78 is responsive to value A from circuit 61 (on line 81), to value B from circuit 70 (on line 83), to value C from circuit 65 (on line 84), and to value D from circuit 67 (on line 85). Circuit 78 computes the slope of M.sub.1 T, or M.sub.2 T, of FIG. 1C by the formula: ##EQU2## Referring to FIG. 1C, this formula first expresses that the amount of energy left to consume to go to the target T is the target amount (namely the slope B of OT at ideal rate multiplied by D the demand period) less the amount of energy consumed at M.sub.1 (or M.sub.2), e.g. A as derived from circuit 61 on line 81, once operational amplifier OA.sub.1 has flipped over (t=25 minutes). Moreover, the slope M.sub.1 T (or M.sub.2 T) is determined by dividing the KWH amount by the time, namely C (5 minutes, or rather 5.times.60 seconds in the instance of FIG. 1C) as derived from line 84. It is realized that while the ideal rate slope B is desired from M.sub.1, M.sub.2 or M.sub.IR to T, so that the system emerges at the beginning of the next period with the ideal rate IR', if M.sub.t is at M.sub.1, e.g. above M.sub.IR, the slope B would overshoot the target. To prevent this eventuality, the system tests whether M.sub.t is above or below M.sub.IR by comparing F to the ideal rate slope B. This is done by operational amplifier OA.sub.2. If F<B, by diode D.sub.4, the system gives up the ideal rate slope expectation, and follows M.sub.1 T. If F>B, e.g. below at M.sub.2, the choice by diode D.sub.3 is to follow the ideal rate B, thereby ending at D.sub.E below the target. Correction of the power rate is by line 90 to compensate for the angle between the present slope E (derived from circuit 62 and line 82) and M.sub.1 T (if diode D.sub.4 controls), or M.sub.2 D.sub.E (if diode D.sub.3 controls).

In conclusion, during the first phase of control, by line 89 for any point M.sub.t before reaching M.sub.IR vector MW is forced to orientate itself toward M.sub.IR. Thus, at any instant of decision during the field of control ab within the demand period, e.g. every minute, due to control by line 89 vector MW is constantly oriented toward a point M.sub.IR situated on the ideal rate line (t=25 minutes). A second phase of control takes place at time t=b (when t=25 minutes is reached). Cell line control is via line 90, in this second phase. The power rate is altered by control of the cell lines so as to follow substantially the ideal rate along line M.sub.IR T to the target.

How the error signal of line 89, or line 90, is used to alter the consumption of the cell lines CL.sub.1 . . . CL.sub.n will now be explained by reference to FIGS. 5 and 6.

Referring to FIG. 5, the KWH meter is used by circuit 24 to calculate the projected demand and derive the error signal which may be as derived on line 89 or 90 of FIG. 3. The error signal ERR on line 29 is compared by comparator circuit 111 to the deadband. The decision is whether .vertline.ERR.vertline..gtoreq.DEADBAND? If yes, cell selector (22, 112) selects cell CL.sub.i in accordance with the order established by the operator. Each cell line CL.sub.1 . . . CL.sub.n has a watt meter W.sub.1 . . . W.sub.n and an amperemeter A.sub.1 . . . A.sub.n providing values such as W.sub.i, A.sub.i for cell line CL.sub.i, as shown. A correction determination circuit (25, 116, 118) which is also part of the cell line control circuit 10 has a transfer function as shown in block (25, 116, 118). Adaptive priority in response to the correction determination circuit 24 for the selected cell line CL.sub.i (in accordance with line 31) is effected via tap changer TP.sub.i on the taps of the associated transformer T.sub.i in accordance with the expected number of tap moves given by the correction determination circuit. The transfer function is characterized by deadbands A and D and hysteresis B and C in the up tapping as well as in the down tapping direction about the target symbolized by the horizontal axis. Once the tap begins moving, the deadband A or D is reduced temporarily to B or C until the error lies again within the new band, and movement ceases. This new deadband is called hysteresis. Thus, for upward tapping there is no control until the error of line 29 reaches the magnitude OD and the deadband is immediately reduced to OC. This avoids controls back and forth about the same operative point. The deadband is made variable by reducing OD, or OA at each control stage during the demand period. As the period proceeds, more past history is used relative to current rate projections in estimating whether the projection overshoots or undershoots the target at the end of the period. This makes the error calculation by circuit 24 more stable towards the end of the period, giving a slower response to short perturbations in the demand. By reducing the deadband, more accurate control to target is possible without increasing the number of tap changes unnecessarily.

FIG. 5 shows across the interface, between the computer system 10 and the hardware, the tap changer TP.sub.i (26) with its control characteristic.

A problem has to be solved, however, when tapping a cell line. The number of taps available to get so many kilowatts required by the correction determination circuit should be effective without exceeding a maximum and a minimum ampere magnitude permissible with the cell line. In order to keep all cell lines within their designated kiloampere operating limits (Hi A and Lo A), upon each control cycle the computer will immediately begin moving any lines found to be out of limits and compensate the error signal. It will then compare the number of tap changes required e.g. necessary for demand control with the number of taps available up to the limit for each cell line by priority e.g. possible taps. The minimum between these two numbers is taken as the probable control available from that line. If more is necessary, the next priority line is similarily tested. This procedure is repeated until the error signal is expected to be reduced below the hysteresis deadband point. By allowing more than one cell line to change at any control period (1 minute apart), the controller is responsive to large error signals. Thus, it responds slowly to noise within the deadband, but quickly to larger perturbations outside the deadband.

Since there is a 30% difference in the change of KW and KA for tap moves high or low in the operating band, it is necessary to approximate the expected changes by the observed value of the previous change. Before any tap move is made, the current value of the cell line's power W.sub.i and amperage A.sub.i are saved. After the move, the new values are subtracted for an estimate of the changes when taps are moved again in either direction. Therefore, in response to a previous A.sub.i, circuit 114 calculates the increment .DELTA.KA for the reset control cycle, and in response to the previous W.sub.i, circuit 224 calculates the increment .DELTA.KW for the next control cycle. These .DELTA.KA and .DELTA.KW are used by the correction determination circuit (26,116,148) in ascertaining the necessary and possible number of taps which lead to a definite number on line 33. After each tap move, circuit 119 reduces the error according to the expected tap moves so that comparator 31 can proceed with the left over, either on the same cell, or an another cell, through cell selector (22,112).

Referring to FIG. 6 the overall control operation is schematically shown in block diagram.

The error determination circuit 24 is shown to respond to DEM LIM of circuit 70, to the KWH pulse meter 60 upon the start of a control cycle (circuit 102) in accordance with the time into the demand period (65) like in FIG. 3. Also, like in FIG. 3, the compensation of the error by circuits 69, 73, 74 and 72 for the first phase of control leading to line 89, and the adjustment by circuits 78, 76 and 77 for the second phase of control leading to line 90, are included in circuit 24. Circuit 101 provides for the adjustment when circuit 103 indicates t=25, e.g. a 5 minute time interval before the end of the demand period has been reached. From circuit 24, the control signal (of lines 89 or 90 in the circuit of FIG. 3) goes to the determination at 104 whether the cell line in question is within or outside the ampere limits HiKA and LoKA set by the operator at 105. If the cell line is out of limits, the system by 106 moves the tap to reach the limit, and at 107 the error is adjusted accordingly. By 108 the tap changer is locked out from control. If the cell line is within the limits, from 104 the system goes to 109 where in accordance with the time left in the demand period (65) the deadband is established, while it is assumed that it is progressively reduced in the demand period. Then, at 110 the deadband is adjusted to provide the hysteresis (OC or OB in FIG. 5) if taps have been moved in the previous control cycle.

At this point, a determination is made, at 111, whether the .vertline.ERR.vertline. exceeds the deadband. If the error is smaller than the deadband no control is made. Otherwise, the system goes to 112, e.g. for the determination of the cell line which is next as priority cell line. If it is locked at 113 the operation is not successful with this cell line and the loop looks for another cell line. If the cell line is not locked, the system establishes how many taps are necessary within the error ERR. To this effect, the individual meter provides for every change. If a tap is changed, this is translated into (.DELTA.KA) the amount of KW (.DELTA.KW) generated or taken away in the system, and such present change is taken as an indication for the next move. Accordingly, the system knows how many taps are left for correction and how many taps are necessary to effectuate the correction of the error .vertline.ERR.vertline..

On the other hand, the limits HiKA and LoKA are limitations to such corrective moves. Therefore, at 115 is established what is the number of taps possible by the formula .vertline.LIMIT-ERR.vertline. /.DELTA.KA/TAP. The number of taps possible may be also limited by the number of control cycles left in the period. This determination is also made at 115.

Then at 116 the minimum between the taps necessary of 114 and the taps possible of 115 is recognized to ascertain the number of taps expected to be moved. Although at 117 the possibility that a tap move be not an integer is taken into account by the evaluation "TAPS.gtoreq.1?" If yes, the tap decision to move the tap by so many steps is taken at 118. For each move the error is adjusted to take into account the operation, and the system goes back to 111 for assuming the possibility of further control by 111, and so on.

The overall operation of the computerized control system according to the invention will now be reviewed from a consideration of the flow charts of FIGS. 7A-7G.

FIGS. 7A-7G illustrate the operative steps under power demand control and cell line control according to the present invention. FIG. 7A is a condensed representation of the operations of the circuit of FIG. 3. Starting at 200 the system calculates at 201 the demand error DCERR by establishing:

DCERR=Demand Limit--(KWH used this period+present kwt33 time left)/demand period, which is the demand control process leading to circuit 71 of FIG. 3. Then, at 202, the question is like with comparator OA.sub.1 of FIG. 3 "whether ideal rate control (at M.sub.IR and at time set=5 minutes from the end, in the example) is in effect?" If the answer is NO, the system proceeds at 203 by calculating a correction by adding to DCERR the expression: (DEM LIM-Present KW).times.ideal rate time/DEM PER. This compensation yields the corrected error at line 89 of FIG. 3.

If the answer is YES, the system goes to 204 where the maximum possible kilowatt demand possible is calculated with the expression: ##EQU3## This is like circuit 78 of FIG. 3. Then, like at OA.sub.2 in FIG. 3, the system compares the slopes characterized by the actual demand of KW and DEM LIM. If it does not exceed DEM LIM (by D.sub.4 in FIG. 3) the error to be compensated is (DCERR=XDEM-Present KW) at 206, with the result outputted via 90 (like from circuit 77 of FIG. 3). If, however, DEM LIM is exceeded at 205, the system sets XDEM at DEM LIM and uses B-E instead of F-E to be outputted by circuit 77 of FIG. 3, or at 90 from 206 in FIG. 7A.

FIG. 7B relates to cell line selection. While this subroutine is periodically started at 209, the question at 210 is whether trip of a possible cell line is detected. If yes, then at 204 the decision is to clear the "tap moving" flags. This ends the chain. If at 210 the answer is NO, at 211 is ascertained whether the controller is on "Auto". If NO is the answer, the system returns to 214. If YES, the ratio of the time left to the period is determined at 212, then ERRKW=DCERR is initialized at 213 which leads to a loop through all cell lines at 214. For the selected cell line at 215 the question becomes "whether the tap was moved in the previous control cycle on this cell line 8". If yes, the .DELTA.KA is calculated for this tap move, and the system goes to the chain of FIG. 7C.

FIG. 7C starts with two questions: whether the change .DELTA.A of current due to the move is too low (at 217) or too high (at 218). If yes at either one, by 219, or 220, an indication of the trouble is set and by 221, or 222 the change of current .DELTA.KA is set to the low limit (Lo.DELTA.KA) at 221, or to the high limit (Hi.DELTA.KA) at 222, in the second instance. Thereafter, the system goes to 223 where at least four pulses are counted from the particular cell line after the taps settle in order that the KWH pulses come to a steady state. At 224 the .DELTA.KW resulting from the tap move is calculated. If it is found too low at 225, trouble indication is set at 226, while at 227 .DELTA.KW is set to the low limit. If it is found too high (at 228) steps are taken at 229 and 230 which are similar to steps 226, 227. Then, the system goes to 231 where the "tap moving" flag is reset, and the remote switch is checked at 232 to see whether it is "Manual" , upon which the system asks (at 240 in FIG. 7D) whether all cell lines are done. If it is "Auto", the system goes to 233 in FIG. 7D, where the question is whether the AUTO/MANUAL is in "Manual" at the computer end. If so, all blocks are bypassed up to 240. If NO, the system checks at 234 whether KA is above the Hi/Hi limit, or at 236 whether KA is below the low limit. In the first instance, the system starts tapping down at 235. In the second instance, the system starts tapping up. Consequently at 238 adjustment is made correlatively to such tap moves in terms of the ERRKW error signal for the change in .DELTA.KW (positive or negative) and at 239 the tap movement lockout is set. Then, when all cell lines are done, the system enters the chain of FIG. 7E.

FIG. 7E shows the deadband effect on cell line control. In a first position of the field of control, a deadband constant C.sub.1 is calculated at 241. If (at 242) the taps are moving, the hysteresis is subtracted at 270. If Not, the system goes directly from 242, as well as from 270 in the alternative, to 243 where the determination is made whether operation is for the second position of the field of control. If YES, a deadband constant C.sub.2 is used and applied at 244. If No from 243 or if Yes through 244, the system goes to 245 for a check "whether in the ideal rate control phase of the demand period?" If YES, at 246, the ratio is made equal to 1. If No, the system goes directly to 247 where is ascertained "whether in the no control phase or not?" If YES at 247, the system goes to 269 for "save permissive status" and "output tap changes" at 268 in FIG. 7G. If NO, at 248 all priorities are scanned through and at 249 the comparison .vertline.ERRKW.vertline. >Deadband is made. If the answer is YES the system goes to the entry of the chain of FIG. 7F; if NO, the way is again at the entry of 269 in FIG. 7G.

Referring to FIG. 7F, at 250 the question is whether .vertline.ERRKW.vertline.>Deadband was achieved at the last control period and control was permissive. If No at 250, "permissive time" is set and the system goes to 269 (FIG. 7G). If YES, the search is at 251 for a cell line with this priority, and at 252 is checked whether the tap lockout has been set for such cell line. If YES at 252, there is a return to the entry of FIG. 7D. If NO, the system calculates at 253 the number of taps necessary to compensate for the error ERRKW. Then at 254, the necessary number of taps is found by relating .vertline.ERRKW.vertline. to the (.DELTA.KW).times.Ratio, following which, at 255, is ascertained "whether it is the first tap to move" and "whether ERRKW is negative" and "whether control is in the third phase." If the answer is YES, one tap is added to the number of necessary taps so as to force the compensated ERRKW positive. If the answer is NO to 255, as well as after 256, the system goes to the entry of the chain of FIG. 7G where at 257, the number of possible taps is sought for KA between Hi and Lo limit. Then at 258 the number of taps possible is calculated by .vertline.Limit-KA.vertline./KA and at 259 the number of taps which is the lowest between the necessary and possible taps is found, when there is a number of control periods left. This is followed at 260 by the determination whether such minimum number of taps is .gtoreq.1. If NO, there is a return to 248 on FIG. 7E. If YES, at 261 tap change is initiated, while at 262 the present .DELTA.KA and .DELTA.KW are saved for the following .DELTA.KA and .DELTA.KW calculations. At 263 ERRKW is corrected by the expected .DELTA.KW.times.Ratio adjustment with the taps. Then, at 264 is checked whether all priorities are done. If No, there is a return to 248 of FIG. 7E. If YES, at 264, the question becomes, at 265, "whether any taps are initiated." A YES goes to 269. A NO passes first through the test whether .vertline.ERRKW.vertline.>Deadband since control is necessary. Then (by 267) the trouble flag is set indicating no further control is available. If NO, 267 is by-passed. The flow chart ends by 268 for "output tap changes", after saving "permissive status" at 269.

The philosophy adapted is: "No control is better than erroneous control." Therefore, protection is provided for three types of occurrences which would lead to erroneous control: metering malfunctions, cell line trip, and human error.

The computer controller is constantly checking for faults in the pulsing KWH meters and the analog kiloampere signals. Every pulse is limit checked for being too fast or too slow. The pulse times are software filtered and converted to a KW rate. These rates are again checked for being too high or too low. Likewise filtered KA values are checked. Any deviation results in the offending cell line being put in a manual (non-automatic tap changing) mode and/or the controller itself being switched to manual. An annunciator sounds and the operator is informed of the source of the trouble, possible corrective action and instructions for resetting the control into automatic.

Failure of the controller results in a special alarm. While the computer automatically restarts after a power failure, restart of the control itself is always in the manual mode. Control cannot start until the operator decides that everything is ready.

After a tap change, the resulting differentials of power and current are limit checked. This traps taps which are stuck, at the physical limit of their range, or move more than one step, all dangerous conditions. Again these cells are put in manual with a trouble alarm to the operator.

A power trip of a cell is more difficult to detect because of the use of pulsing KWH meters. A trip is construed to be the drop of the analog signal to near zero and the rapid drop of the pulse rate. Either condition alone indicates a meter malfunction instead. The controller is put in manual because it is not desirable for the remaining lines to compensate the loss by driving up to their maximums.

The above malfunction conditions are all tested by simulation of the fault via software since in the present instance the use of actual hardware is not feasible. Such simulation is added to a normal simulation of signals used to test the above stated control philosophy before installation.

Protecting the controller from most human errors has been accomplished by two means:

First, any critical numbers being entered by an operator are put into temporary storage, for visual checking and computer limit checking, before being used for actual control. For example, if the operator wishes to change KA operating limits for a particular line, he enters these numbers (a Hi Hi limit, beyond which the controller taps down immediately, a Hi limit used as the calculational limit, and the Lo limit) to temporary storage. This has no effect on actual control. After he visually checks the new entries, he requests that they be moved to permanent storage and used for control. At this time, the computer limit checks each entry for (a) correct range, and (b) difference from neighboring entry. Only after these criteria are met will the numbers be used for actual control.

Secondly, a four-tiered interfacing scheme is used. The first two tiers are separated by a key lock on the computer panel. If the console is locked out, only the display buttons are active, so no unauthorized person can manipulate anything in the controller. With the key in and unlocked, an operator can perform the following functions from the CRT and keyboard:

change the controller's time and date

put a cell line into auto or manual

change the demand limit

display and change KA limits on cell lines

change control section time boundries

put the controller in auto or manual

change cell line priorities

request an explanation of the trouble alarm

list the available functions.

The next higher level is accessed through a password available to the plant engineer only. The engineer has access to all the above functions, and in addition he can do the following:

change the control deadband

change the voltage input conversion to KA

display and change bounds on the cell line KA limits to be entered by the operator

change max and min differences between KA limits

change the tap relay closing time

change the tap settling time

change meter KWH/pulse factors

list the engineer's functions

A fourth level is available through another password to the system designer (programmer). At this level the functions are:

start/stop an offline plant simulation for testing

simulate a malfunction for training or testing

request debug displays each control and demand period

read/write directly into memory.

The following CRT displays are made available to the operator by the system:

In Appendix A is included an instruction program listing prepared to control a process operation of the cell line control system in accordance with the present invention. The instruction program listing is written in the assembly language of the Intel 8080 microprocessor computer system. Many of these computer systems have already been supplied to customers, including customer instruction books and descriptive documentation to explain to persons skilled in this art the operation of the hardware logic and the executive software of this particular digital computer system. The instruction program listing is included to provide an illustration of the controlling steps involved and also of the protective measures such as alarm and password. This instruction program listing can be fully understood by persons skilled in this art. ##SPC1## ##SPC2## ##SPC3## ##SPC4## ##SPC5## ##SPC6## ##SPC7## ##SPC8## ##SPC9## ##SPC10## ##SPC11## ##SPC12## ##SPC13## ##SPC14## ##SPC15## ##SPC16## ##SPC17## ##SPC18## ##SPC19## ##SPC20## ##SPC21## ##SPC22## ##SPC23## ##SPC24## ##SPC25## ##SPC26## ##SPC27## ##SPC28## ##SPC29## ##SPC30## ##SPC31## ##SPC32## ##SPC33## ##SPC34## ##SPC35## ##SPC36## ##SPC37## ##SPC38## ##SPC39## ##SPC40## ##SPC41## ##SPC42## ##SPC43## ##SPC44## ##SPC45## ##SPC46## ##SPC47## ##SPC48## ##SPC49## ##SPC50## ##SPC51## ##SPC52## ##SPC53## ##SPC54## ##SPC55## ##SPC56## ##SPC57## ##SPC58## ##SPC59## ##SPC60## ##SPC61## ##SPC62## ##SPC63## ##SPC64## ##SPC65## ##SPC66## ##SPC67## ##SPC68## ##SPC69## ##SPC70##