METHOD AND SYSTEM FOR MONITORING THE FUNCTIONALITY OF ELECTROLYSIS CELLS

Abstract

Method and system for monitoring the functionality of electrolysis cells for use in chlor-alkali electrolysis. The method uses the analysis of the current/voltage characteristic of the cells in relation to an AC voltage overlaid on the operating voltage

Description

The invention relates to a method for monitoring the functionality of electrolysis cells. Furthermore, the invention relates to a system for monitoring the functionality of electrolysis cells and in particular the use thereof in chlor-alkali electrolysis.

The invention is directed to monitoring methods known per se for electrolysis cells, which conventionally make use of the detection of an averaged current-voltage characteristic of the cell.

BACKGROUND OF THE INVENTION

The monitoring of electrolysis facilities and the diagnosis of defective cells are fundamentally based on the measurement of process-relevant parameters such as temperatures, differential pressures, or also gas concentrations, but above all on the measurement of cell voltages and cell current. The cell voltage is very sensitive with respect to changes in the electrolyzes and is easily detectable, also at the individual cells.

Outotec describes, in WO 2005/052700 A1, a monitoring system for copper electrolysis.

In such a monitoring system, the data of typically several hundred single cells, distributed onto multiple so-called stacks, are processed. Depending on the complexity of the monitoring system, it consists of a plurality of modules which communicate with one another, and which assume different data (partial) processing steps (cf., for example, WO 2007/087728 A1, WO 2001/078164 A3). In general, the tasks of the modules disclosed therein can be described as follows:

1. Detecting and bundling the measured values

2. Optical preparation of the measured values for the display in the control room, display of alarms, etc.

3. Storing and managing the measurement data in databanks

4. Preparing (filtering, averaging, normalization, etc.) and analyzing the data

5. Diagnosis of errors and anomalies

6. Reporting defects and taking countermeasures

Decisive steps of all of these systems are the data analysis and the subsequent diagnosis about the cell status.

Simple systems check in this case whether the measured single cell voltages are within established limiting value ranges, for example, as described in DE 2652774 C2, whether they do not exceed an upper limiting value and they do not fall below a lower limiting value. Otherwise, a warning message is displayed in the control room. In WO 2007/087728 A1, load-dependent limiting values are established with the aid of reference i-U curves. For the parameters U₀ and k of the characteristic curve ascertained by linear regression, upper (U_0,max, k_max) and lower limiting values (U_0,min, k_min) are determined. The limiting values for the cell voltage U_min and U_max can then be calculated continuously according to:

U=k*i+U₀

The chronological change of the voltage is also permanently monitored and compared to a reference value of the permissible change k*di/dt.

Both systems analyze the measurement data only with respect to established limiting values. These can also be designed dynamically and can even be optimized on the basis of historical data and learning-capable structures, however, they do not permit the qualitative or quantitative determination of errors in individual electrodes.

A system is known from EP 00002006418 B1, which compares the single cell voltages to one another and assigns a degree of damage (not damaged, not severely damaged, severely damaged) via two limiting values. In this manner, the intention is above all for defects (so-called pinholes) on ion exchange membranes to be discovered. These are only detectable early when the NaCl electrolyzer is put into operation (i.e., at low current density), since the transfer of NaOH increases the pH value of the anolyte, and then the formation of oxygen is preferred. This runs at a cell voltage between 1.2 V and 1.5 V and is therefore approximately one volt less than the voltage for chlorine formation (in combination with hydrogen-developing cathodes). With increasing current density, the chlorine formation supersedes the oxygen formation and the cell voltage jumps to the chlorine potential. If the voltage of a cell only increases slowly, however, a defect can be presumed. Inferences about the size and position (top or bottom half of the cell) can then be drawn by analyzing further measured variables such as differential pressure or fill level in the cell, or also by adapting a parameter model to the i-U curves (for small current densities, according to the patent figure up to approximately 0.5 kA/m²).

The method according to WO 2001/078164 A2 permits finer damage classification. The focal point here is the extraction and identification (pattern recognition) of events within the continuously recorded input data and the linkage thereof to diagnoses, which are stored in a databank. Measured physical variables or also, as described as preferred in WO 2001/078164 A2, variables already calculated from measured values such as data from analyses in the time or frequency range or adaptation parameters from the regression of i-U curves, as are ascertained according to WO 2006/133562 A1, can be used as input data. The diagnosis then runs in two steps:

• • In a learning phase, the input data for an event in the case of a specific operating state are assembled and linked to a diagnosis, which is established by an expert, and stored in an intelligent event-diagnosis databank on the basis of a neuronal network.
  • In the diagnosis phase, the system then checks in the respective operating state whether the continuously incoming data sufficiently correspond to stored input data. If the similarity to a known event is sufficiently large, an alarm and the corresponding diagnosis are reported. If only inadequate correspondences are found, this indicates an event which is not yet known, which can be linked by an expert to a new diagnosis.

A precise description of the above-mentioned regression of i-U curves by means of three-parameter adaptation can be found in WO 2006/133562 A1. As described, by means of a parameter model of the form U=U₀+S*log(i)+R*i, the profile of i-U curves can be simulated and items of information both about the overvoltages and also about the ohmic components of electrolysis cells can be obtained. The corresponding measured values for the characteristic curves are obtained during startup and shutdown processes or in the event of a load change. The characteristic curves must first be extracted from the process data for this purpose and, inter alia, filtered, smoothed (removal of the rectifier ripple), and normalized (compensation for deviations in the temperature and concentration). After the parameter determination, the quality of the adaptation is checked. If there is a sufficient coefficient of determination and confidence interval, the ascertained adaptation parameters are stored in a databank and compared to reference parameters (for example, from cell ageing models) or assigned to so-called previously established “operation classes”. Every “operation class” applies for a specific parameter value range and characterizes a cell status. A similar parameter model is used according to DE 10217694 A1 for the dynamic determination of the i-U characteristic of a fuel cell having a motor as a consumer. Current and voltage are continuously recorded in the event of a load change and the adaptation parameters are determined, which can be used for the torque control and regulation of the drive.

A further approach for identifying anomalies is the use of predictive models, using which the normal operating state is calculated. In contrast to previously proposed systems, in which errors were identified on the basis of repeating combinations of input data (pattern recognition), the deviation of the measured values from the normal, calculated state is now detected and defined as an anomaly. The better data base is advantageous to represent the normal operating state; the difficulty of developing a good model which can image various operating states is shown to be disadvantageous.

WO 2005/052700 A1 explains the functionality of an online monitoring system for copper electrolysis. With the aid of a predictive parameter model, which was ascertained in the laboratory, the theoretical cell voltage is calculated as a function of other operating parameters such as temperature, current density, or concentration. The deviation between measured and theoretical cell voltage is ascertained and the trend thereof is analyzed by means of a model. With the aid of a fuzzy logic model, this trend is converted into a status index (number between 0 and 1), which characterizes the instantaneous state of the cell. With application of a further fuzzy logic model and use of the status index, older condition indexes, and in consideration of appearances of ageing, a new condition index which characterizes the cell status over a longer period of time is then ascertained.

WO 2007/087729 A1 discloses a system and a method, respectively, to simulate the normal operation of a facility by means of predictive models and detect errors on the basis of the deviation between measured process parameters and the modeled normal operation.

In the first step, within a learning phase, a model is prepared and validated, or the validity range and the precision are determined, on the basis of reference data and expert knowledge for respectively one operating state (for example, putting into operation and taking out of operation, load change). Thus, deviations which lie outside the precision can be identified later as an anomaly. Using the model parameters, deviations between model and measured values, which occur due to errors which are already known, can also be stored as a so-called signature or as an anomaly pattern (sequence of signatures) in a databank.

In operation, depending on the operating state, the process variables calculated using the respective model are then compared to the measured process variables and the signature thereof is calculated from the deviations. If the deviations of the signature are sufficiently large, an anomaly can be presumed. If the calculated signature then corresponds to a signature stored in the databank, the ascertainment of an error is possible.

Prior art of monitoring and diagnosis systems based on impedance spectroscopy:

The patent application US 2005/0287402 A1 describes a system for monitoring and regulating the water economy of a fuel-cell stack on the basis of its impedance.

The construction, which is expressly described as an example, comprises the fuel-cell stack, an inverter, a control unit, a water metering system, and a load (motor).

In principle, a voltage having a harmonic wave (ripple) is applied to the fuel cell by the inverter, which also results in ripple in the current at the output of the fuel cell. By comparing both variables, the impedance of the fuel cell is determined If the value is excessively high, water is metered to the cells. For this purpose, only the voltage of one cell is measured as representative. A rectangular harmonic wave having an amplitude in the mV range and a frequency of 8 kHz is typically applied via the controller of the inverter.

However, harmonic waves having arbitrary shape and arbitrary amplitudes and also frequencies can also be generated. These can be used for more extensive monitoring (for example, to infer the status of the cell components), also in conjunction with more complex analysis methods (Fourier transform).

The system is to be applicable to the monitoring of arbitrary current generators, but in particular fuel cells, which are coupled to an arbitrary current converter (an “AC/DC rectifier” is also listed as an example).

A system is explained in patent application US 2010/0216043 A1, which monitors and controls a facility consisting of multiple fuel-cell stacks, in that electrochemical impedance spectroscopy is carried out at individual reference cells distributed over the stacks. Via special “monitoring circuits” (Wheatstone bridge circuit), alternating currents having various frequencies are output and the voltages of the cells are measured. An analysis system generates from these data the corresponding impedance curves (Bode plot and Nyquist plot) and determines polarization and ohmic resistances. The characteristics obtained in this case are “extrapolated” to the remaining cells and used to regulate a wide variety of operating parameters (for example, gas volume streams, humidity), to operate the fuel cells as optimally as possible.

A reliable method for identifying faulty individual electrolysis cells from a plurality of electrolysis cells of an electrolyzer has heretofore not been known. The object of the present invention, proceeding from the above-described prior art, is to provide an improved monitoring method for electrolysis cells, in particular for use in chlor-alkali electrolysis, which overcomes the above disadvantages and allows reliable identification of electrolysis cells which are defective or impaired in their function from a plurality of electrolysis cells of an electrolyzer, in particular in the case of chlor-alkali electrolysis.

SUMMARY OF THE INVENTION

The subject of the invention is a method for monitoring the functionality of electrolysis cells of an electrolysis facility, in particular a membrane electrolysis facility, preferably of multiple electrolysis cells operated simultaneously in production, characterized in that the current/voltage curve of an AC voltage overlaid on the electrolysis voltage is measured and compared to the predefined characteristic values of a functional electrolysis cell and the comparison value is detected (see FIGS. 1 and 2).

DETAILED DESCRIPTION

The advantage and novelty of this monitoring system in relation to existing and commercially available systems is the acquisition of additional items of information about the electrochemical behavior by measuring current and voltage time curves with higher resolution than usual, in order to therefore use the harmonic waves in the current and voltage for the diagnosis of the cell status. These items of information are not available in conventional systems, since their measured value detection at conventional sampling rates (at most 100 Hz) cannot sufficiently resolve the harmonic waves having basic frequencies of typically 300 Hz or 600 Hz (depending on the rectifier construction, 6 or 12 pulses). In addition, in conventional systems the effort is made to obtain current and voltage signals which have as little interference and are as free of harmonic waves as possible using integrating elements and filters. These are necessary for the analysis of stationary current-voltage characteristics (i-U curves), on which presently all known electrolysis monitoring systems are based. However, these characteristics are only accessible during putting into operation or taking out of operation and in the event of load changes, i.e., in the event of a disturbance of running operation.

During constant, stationary operation, the systems can only access averaged values for current and voltage. It is therefore possible to register the change of the cell voltage over a longer operating time (drifting of the cell voltage) and possibly to diagnose errors by incorporating data of other sensors or chemical analyses. However, there is no access to an instantaneous current-voltage characteristic, which reacts particularly sensitively to damage and even allows the assignment of damage to individual cell components (for example, a pinhole in the membrane).

The measuring unit (8 in FIG. 1) must have a sufficient resolution both for the level of the electrical signal and also for the time (for example, for 100 measurement points per half wave of a 600 Hz harmonic wave from a 12 pulse rectifier, a sampling rate of 120 kHz results; the measurements during the development of the method were performed using a measuring card having 2 MHz sampling rate and 14 bit resolution).

The time curve of the current I (6 and 7) must always be detected simultaneously with the voltage measurement U (5) at the same resolution. The signal from the shunt resistor 2, which is installed in the normal case in an electrolysis facility, (for example, voltage drop in the range of 50 mV at currents of approximately 15,000 A) indicates the mean value of the current strength I_DC using integrating measuring instruments very precisely. For the novel monitoring system, which requires exact recording of the current harmonic waves (at 15,000 A, only a few hundred amperes amplitude of the ripple, see FIG. 2), the signal from a shunt resistor 2 is unsuitable solely because of excessively large overlaid interference, however. A sufficiently sensitive current measuring unit is required here, which also records the alternating current component I_AC in the frequency range generated by the rectifier 1 with low phase and amplitude errors. It is preferably based on the induction by way of the magnetic field of the current, for example, as in a Rogowski coil 3 having a corresponding amplifier. This only detects the alternating current component (AC component I_Ac) of the current. The total current including the ripple can then be ascertained by addition of the AC component I_Ac to the direct current component (DC component I_DC), which is generally accessible from the measurement on a shunt resistor 2 in the control room.

To ensure short signal paths, the cell voltages U are to be detected as close as possible to the electrolyzer 4 (5) and transmitted therefrom in digital form to the control room. The use of multiple measuring units (for example, per single cell), or one measuring unit which queries the measured value successively via multiplexer, is conceivable. The permissible voltage values within the bipolar switched electrolyzer and the potential difference in relation to earth are to be considered with respect to metrology and the safety regulations (for example, by potential-separated data transfer). Voltage U (5) and current I (6 and 7) of the cells must be detected synchronously, wherein the current measuring unit is only required once in a circuit made of rectifier 1 and electrolyzer 4.

To illustrate the data analysis of the monitoring system according to the invention, it is to be compared to the classical path for the analysis of dynamic current-voltage characteristics, the electrochemical impedance spectroscopy (EIS), which is proven at a laboratory scale. Particularly low-interference excitation signals are used in this case, typically solely sinusoidal harmonic waves of different frequencies having amplitudes in the millivolt range, which are overlaid with a clean direct current.

The novel monitoring system for large-scale industrial electrolyzers not only uses an excitation signal which can be influenced, but rather it uses the existing ripple of the rectifier 1, which in principle does not have a simple, precisely defined oscillation form. Because of the superposition of the three sinusoidal phases of the three-phase current and due to the power control in the rectifier by way of phase angle control by means of thyristors, the ripple, depending on the rectifier construction (6 pulse or 12 pulse), consists of a frequency spectrum having a base frequency of 300 Hz or 600 Hz and the multiples of these base frequencies. The amplitude is variable and is dependent on the rectifier operating point (i.e., on the fraction of the full load power). If this is not in the optimum range, for example, upon setting of smaller powers by phase angle control from a large rectifier, the ripple can greatly exceed the typical 5%.

In addition, there is an array of types of interference which can additionally be superimposed on the ripple. These can originate from the rectifier 1 itself, on the one hand, for example, as a regulation artifact of the thyristor controller, or can arise by way of interaction with the environment, for example, by induction from strong magnetic fields. The individual pulses also vary as a result of relatively small differences in the thyristors.

This makes it necessary to firstly separate the items of information from the interfering components and prepare them for analysis. This preparation unit can fundamentally be embodied in this case as a hardware or software solution. Inter alia, (bandpass) filters are conceivable in this case, using which the current and voltage components in the rectifier base frequency may be isolated for analysis. During the measurements carried out for this invention in the laboratory and also on a testing facility electrolyzer, an averaging algorithm worked out for this purpose was used. It automatically recognizes the limits of the individual pulses in a current or voltage time curve, cuts apart the pulses, overlays them, and forms the average values from the respectively overlaid data points of all pulses.

It was surprisingly shown in this case that in spite of the strong interference due to overlaid oscillations on the testing facility electrolyzer, the results for a 50 Hz network frequency period were very well reproducible, but the individual pulses and the individual electrolysis cells could have significant differences.

The analysis unit 8 finally produces the relationship between the current and voltage harmonic waves and the state of the electrolysis cell. The foundation of this analysis is formed by the relationship between the current and voltage harmonic wave amplitudes and the phase shift between both harmonic waves.

The simplest and most effective possibility for analyzing this is offered by plotting as a dynamic i-U curve for the detected small current density range, as shown in FIG. 2 (the current I is converted for this purpose into the current density i, i.e., in relation to the active surface of the electrolysis cell). The curved profile then contains as cell-relevant items of information:

• • the mean value of the cell voltage, as is also detected in the conventional systems,
  • the slope, which can be used similarly to the k factor in stationary i-U curves,
  • a hysteresis, which is only very small but is still significantly recognizable in cells having hydrogen-developing cathodes, while it has a strong effect in cells having oxygen depolarized cathodes (ODC) due to the high capacitances in these gas diffusion electrodes.

Since not all components of an electrolyzer (anode, cathode, membrane) influence these variables in the same manner, the analysis—in particular also of chronological changes and of differences in relation to the standard state—allows a statement about the state of the components or a diagnosis of faulty components. A first analysis step is the estimation of a complete i-U curve for the entire current density range from the data obtained for the small current density range. From this, the following may be differentiated:

• • changes in the ohmic resistance of the cell, for example, increase of the membrane voltage drop due to contaminants or changes in the transition resistances upon the contact of an ODC, which primarily affect the slope of the curve,
  • changes in the thermodynamic potential difference of the electrochemical reaction, as arise, for example, due to mixing of anolyte and catholyte upon the occurrence of a pinhole (breakthrough) in the membrane, which primarily is shown in the extrapolated axis section of the curve for the current density zero,
  • damage to an ODC, for example, reduction of the effective active ODC surface area due to a lack of oxygen access as a result of partial coverage of the ODC gas side with sodium hydroxide solution, which is primarily recognizable from changes of the hysteresis.

These items of information permit, for example, a targeted alarm to be triggered, to avoid costly subsequent damage in a timely manner.

The novel monitoring system can additionally also incorporate further analysis methods known in principle from the literature for finer analyses. Thus, the items of information contained in the measurement results, which go beyond the data detected in the existing and commercially available systems, can optimally utilize:

• • Use of models, the parameters of which are adapted by means of regression calculation to the measured data. For example, electrical engineering equivalent circuit diagrams are known, as are used in electrochemical impedance spectroscopy (EIS), to ascertain values for reaction resistances, ohmic resistances, or double-layer capacitances. The use of nonstationary reaction models is also conceivable, however, which also consider material transport phenomena in addition to the electrochemical reaction.
  • Another analysis possibility is to first convert the time curves of current and voltage harmonic waves into the frequency range, for example, by Fourier transform, and to analyze them on the basis of the impedance curve, the Bode plot, or the Nyquist plot, as is conventional in impedance spectroscopy. The calculation of model parameters is usually performed in this case on the basis of electrical engineering equivalent circuit diagrams.
  • As already described in WO 2001/078164 A3 and WO 2006/133562 A1, the parameters ascertained in the time or frequency range could then be used to classify events or anomalies, in that the obtained parameter combination is compared to combinations of parameters which are linked to a diagnosis or damage and are stored in databanks (pattern recognition).

The following preferred embodiments of the invention result from the above description.

The novel method is preferably carried out in an electrolyzer, in which the electrolysis cells are provided having bipolar interconnection of the electrolysis cells.

In a preferred embodiment, the harmonic wave AC voltage of the rectifier is used as the AC voltage for the generation of the electrolysis voltage, for example, from a network AC voltage.

In a particularly preferred embodiment, the possible interfering components in the AC voltage signal and/or alternating current signal (for example, measurement noise) are filtered out before or after the detection of the signal.

Furthermore, a preferred form of the novel method is advantageous, in which the alternating current/AC voltage components are measured at a sampling rate of at least 10 kHz, preferably at least 100 kHz.

In a particularly preferred embodiment of the novel method, the measurement of the alternating current component at the current supply to the electrolysis cell to be measured is performed inductively, in particular with use of a Rogowski coil.

To identify electrical contacting faults and/or membrane damage, in a preferred embodiment of the method, the slope in a derived current density-voltage characteristic curve is used as a characteristic value for the functionality of the individual electrolysis cells (i-U curve).

To identify leaks in the ion exchange membrane in membrane electrolyzers or to identify electrode flaws, in another preferred embodiment of the method, the extrapolated axis section of the characteristic curve for the current density zero in a derived current density-voltage characteristic curve is used as a characteristic value for the functionality of the individual electrolysis cells.

Furthermore, the change of the hysteresis of the characteristic curve can be used in particular as a characteristic value for the functionality of the individual electrolysis cells in a derived current density-voltage characteristic curve.

It is also advantageous to combine the method with methods known per se for monitoring the single cell voltage of the electrolysis cells. The characteristic values ascertained by the novel method can be used further, for example, with the above-described analysis methods (learning-capable structure, predictive model) for event-dependent controllers of electrolysis facilities.

In a particularly advantageous implementation of the novel method, if a predefined number of electrolysis cells which are impaired in their functionality is exceeded based on the measured value detection, a warning signal is generated, which is used for informing the operating personnel or for automatically taking the individual electrolysis cells or the entire electrolyzer out of operation.

The novel method is applied in particular in electrolysis facilities for electrolysis of alkali chloride solutions, in particular of lithium chloride, sodium chloride, or potassium chloride solutions, preferably of sodium chloride solutions or of hydrochloric acid. However, the novel method is not fundamentally restricted to these electrolysis methods. Application in water electrolysis is also conceivable.

A further object of the invention is also a system for monitoring the functionality of electrolysis cells of an electrolysis facility having multiple electrolysis cells, in particular a membrane electrolysis facility, preferably having multiple electrolysis cells operated simultaneously in production, at least comprising a voltage generator for generating an AC voltage overlaid on the electrolysis DC voltage, a voltage measuring unit connected to measure the DC voltage component and the AC voltage component over the individual electrolysis cells, at least one current measuring unit for measuring the direct current component and the alternating current component of the electric current, which flows to the electrolysis cells, and a data processing unit, which records the measured values of the DC voltage components, the AC voltage components, the direct current component, and the alternating current component, generates a current/voltage curve and compares the current/voltage curve of the individual electrolysis cells to the predefined characteristic values of a functional electrolysis cell.

A monitoring system is preferred which has current and voltage measuring units for electrolysis cells in an electrolyzer having bipolar interconnection of the electrolysis cells.

A monitoring system is preferred in which the AC voltage generator is formed by a rectifier for generating the electrolysis voltage from AC voltage, which has a harmonic wave AC voltage in operation.

A novel monitoring system is particularly preferred which comprises electronic filters for possible interfering components in the AC voltage signal upstream of the detection unit of the data processing unit.

The current measuring unit of the monitoring system contains, in addition to the conventional direct current measurement with the aid of a shunt resistor as described above, in particular an inductive alternating current measuring unit and comprises in particular a Rogowski coil as a measured value sensor.

In a particularly preferred embodiment of the monitoring system, the data processing unit has an output unit having signal generator.

In a particularly preferred variant of the monitoring system, the signal generator is electrically connected to an optical and/or to an acoustic warning device and/or to a facility controller for the operation of the individual electrolysis cells or for the operation of selected cell stacks or for the operation of the entire electrolyzer.

As described, the system is particularly preferably connected to an electrolysis facility for the electrolysis of alkali chloride solutions, in particular lithium chloride, sodium chloride, or potassium chloride solutions, preferably of sodium chloride solutions or hydrochloric acid.

The invention will be explained in greater detail hereafter on the basis of the figures by way of the examples, which do not represent a restriction of the invention, however.

BRIEF DESCRIPTION OF THE DRAWINGS

In the figures:

FIG. 1 shows a schematic diagram of the monitoring system according to the invention

In FIG. 1, the reference signs have the following meanings:

1 transformer and rectifier for electrolysis voltage

2 shunt resistor

3 Rogowski coil

4 electrolysis cell

5 measurement of the cell voltage U (AC voltage measurement)

6 measurement of the direct current component I_DC in the cell current I

7 measurement of the alternating current component I_AC in the cell current I

8 measured value detection

FIG. 2 shows an example of the chronologically detected cell voltage U and current density i and the detail generated therefrom from the current density-voltage characteristic curve (i-U curve)

FIG. 3 shows a comparison of the i-U curves before and during membrane damage due to calcium addition

FIG. 4 shows a time curve of the membrane damage due to calcium addition: rise of the cell voltage only due to the current-dependent component of the i-U curve (=slope b times mean current density, resulting from the ohmic resistance), while the axis section (i.e., the electrochemical reaction) remains nearly constant

FIG. 5 shows a time curve of the membrane damage due to a pinhole: collapse of the cell voltage predominantly due to the axis section of the i-U curve (change of the electrochemical reaction) but hardly due to its current-dependent component (the slope b, i.e., the ohmic resistance, remains substantially constant)

EXAMPLES

The examples describe the novel method for monitoring industrial electrolysis single cells for two experimentally simulated malfunctions in detail.

In this case, the dynamic current-voltage characteristic of the single cells is analyzed for the monitoring. It results from the reaction of the cell 4 to the periodic alternating current signal, which is present as a harmonic wave of the direct current as a result of the ripple of the rectifier 1 in many industrial electrolysis facilities.

FIG. 1 shows the measurement construction in principle. Due to the ripple, the current I, with which the electrolysis cell 4 is supplied by the rectifier 1, is not constant but rather oscillates periodically by a small amount. This minimal current change also has an effect on the electrolysis cell 4, which reacts with periodic changes of the cell voltage U. The idea of the measurement concept is that the time curve of both the current and also the cell voltage is detected and by comparison of the periodic changes of both dimensions, inferences can be drawn about the status of the cell and its components (see FIG. 2).

As examples, membrane damage was experimentally simulated: The goal of the experiments was to detect the fault of the functionality of a chlor-alkali electrolysis cell by way of the monitoring method according to the invention after intentional damage of the membrane:

a) by calcium contamination

b) by perforation (pinhole).

By applying this novel system, the various types of damage are not only to be generally recognized, but rather also are to be identified or differentiated from one another.

For this purpose, a chlor-alkali laboratory cell (anode: expanded metal dimensionally-stable anode (DSA), cathode: oxygen depolarized cathode (ODC), membrane: Flemion F 8020 Sp, finite gap arrangement) having a membrane surface area of 21 cm² was continuously operated under typical industrial conditions (T_catholyte=T_anolyte=80° C., W_NaCl=19 wt.-%, W_NaOH32 wt.-%, slight basic solution overpressure) at a mean current density of 4 kA/m². A 6 pulse laboratory rectifier, which corresponds to the construction of an industrially used rectifier, was used as the power supply.

The membrane damage was carried out as follows:

• • a) membrane contamination:

after reaching the stationary state, the metering of a calcium-containing brine was performed (W_NaCl=19 wt.%, W_Ca2+=2.5 wt.-%) directly into the anode chamber by means of a syringe pump (delivery rate: 0.5 ml/h) over the entire further experimental time, so that a calcium concentration of W_Ca2+=240 ppm (weight proportion) resulted in the anode chamber (beginning of the experiment identified in FIG. 4).

• • b) Membrane perforation (pinhole):

After reaching the stationary state, the membrane was pierced using a titanium wire and an approximately 0 5 mm hole (pinhole) was generated (beginning of the experiment identified in FIG. 5). The wire had been installed in the rear side of the anode chamber together with a feed-through before the cell was put into operation. For the experiment, it could be moved up to the membrane from the outside without touching the DSA grating.

During the entire operation, ripple measurements were carried out at intervals of 15 minutes (during the experiment even at significantly shorter intervals of up to 10 seconds), in that the ripple cell voltage U and the ripple cell current I (voltage drop at a shunt) were detected at sampling rates of 500 kHz via a measuring card connected to a computer. By plotting the measured cell voltage U against the cell current density i, the ripple i-U curves according to FIG. 2 were obtained, which were analyzed by means of linear regression. Two ripple i-U curves are shown before and during calcium contamination as examples in FIG. 3 (the measurement times are plotted in FIG. 4). The dashed straight line shows the linear regression of the curve. The circles correspond to the mean values of the ripple cell voltage and of the ripple current density.

The time curves of the mean cell voltage and of the axis section and of the current-dependent component (slope b multiplied by mean current density, here: 4 kA/m²), ascertained from the linear regression, are shown for the membrane contamination in FIG. 4 and for the membrane perforation in FIG. 5.

The prior art is heretofore the tracking of the time change of the cell voltage, which does indicate a malfunction, but does not permit further diagnosis. The novel system provides additional items of information for diagnosis in the form of the time change of the axis section and of the current-dependent component of the ripple i-U curves. The analyses result in the following:

• • a) membrane contamination (FIG. 4):

Shortly after the beginning of the calcium addition, the cell voltage increases continuously. It is known that calcium forms poorly-soluble deposits in the membrane and therefore obstructs the sodium ion transport, so that the membrane resistance increases. As a result, the current-dependent component increases simultaneously with the mean cell voltage, while the axis section remains constant.

• • b) membrane perforation (FIG. 5):

Basic solution passes through the pinhole into the anode chamber, increases the pH value, whereby the anodic oxygen formation is preferred and the chlorine production comes to a stop, i.e., a strong change of the electrochemical reactions occurs. Since the oxygen formation occurs at lower equilibrium potential, the cell voltage decreases abruptly, as does the axis section. The current-dependent component remains nearly unchanged.

Thus, various types of membrane damage can be identified using the system according to the invention by way of the different behavior of axis section and current-dependent component.

Claims

1. Method for monitoring the functionality of multiple electrolysis cells of a membrane electrolysis facility, which multiple electrolysis cells are operated simultaneously in production, wherein the current/voltage curve of an AC voltage overlaid on the electrolysis voltage is measured and compared to the predefined characteristic values of a functional electrolysis cell and the comparison value is detected.

2. Method according to claim 1, wherein the electrolysis cells are provided in an electrolyzer having bipolar interconnection of the electrolysis cells.

3. Method according to claim 1, wherein the harmonic wave AC voltage of the rectifier is used as the AC voltage for the generation of the electrolysis voltage.

4. Method according to claim 3, wherein the possible interfering components in the AC voltage signal and/or alternating current signal are filtered out before or after the detection of the signal.

5. Method according to claim 1, wherein the alternating current/AC voltage components are measured at a sampling rate of at least 10 kHz.

6. Method according to claim 1, wherein the measurement of the alternating current component at the current supply to the electrolysis cell is performed inductively, with use of a Rogowski coil.

7. Method according to claim 1, wherein the slope in a derived current density-voltage characteristic curve is used as a characteristic value for the functionality of the individual electrolysis cells to identify electrical contact faults and/or membrane damage.

8. Method according to claim 1, wherein the extrapolated axis section of the characteristic curve for the current density zero in a derived current density-voltage characteristic curve is used as a characteristic value for the functionality of the individual electrolysis cells to identify leaks in the ion exchange membrane in membrane electrolyzers or to identify electrode flaws.

9. Method according to claim 1, wherein the changes of the hysteresis of the characteristic curve are used as a characteristic value for the functionality of the individual electrolysis cells in a derived current density-voltage characteristic curve.

10. Method according to claim 1, wherein the method is combined with the monitoring of the single cell voltage of the electrolysis cells.

11. Method according to claim 1, wherein, if a predefined number of electrolysis cells which are impaired in their functionality is exceeded based on the measured value detection, a warning signal is generated, which is used for informing the operating personnel or for automatically taking the individual electrolysis cells or the entire electrolyzer out of operation.

12. Method according to claim 1, wherein the method is operated in an electrolysis facility for electrolysis of alkali chloride solutions or hydrochloric acid.

13. System for monitoring the functionality of electrolysis cells of an electrolysis facility having multiple electrolysis cells, comprising a voltage generator (1) for generating an AC voltage overlaid on the electrolysis DC voltage, a voltage measuring unit (5) connected to measure the DC voltage component and the AC voltage component over the individual electrolysis cells (4), at least one current measuring unit (6, 7) for measuring the direct current component and the alternating current component of the electrical current, which flows to the electrolysis cells, and a data processing unit (8), which records the measured values of the DC voltage components, the AC voltage components, the direct current, and the alternating current component, generates a current/voltage curve and compares the current/voltage curve of the individual electrolysis cells to the predefined characteristic values of a functional electrolysis cell.

14. System according to claim 13, having current and voltage measuring units for electrolysis cells in an electrolyzer having bipolar interconnection of the electrolysis cells.

15. System according to claim 14, wherein the AC voltage generator is formed by a rectifier for generating the electrolysis voltage from AC voltage, which has a harmonic wave AC voltage in operation.

16. System according to claim 13, comprising electronic filters for possible interfering components in the AC voltage signal upstream of the detection unit of the data processing unit.

17. System according to claim 13, wherein the unit for measuring the alternating current component is an inductive alternating current measuring unit and comprises a Rogowski coil as a measured value sensor.

18. System according to claim 13, wherein the data processing unit has an output unit having signal generator.

19. System according to claim 18, wherein the signal generator is electrically connected to an optical and/or acoustic warning device and/or to the facility controller for the operation of the individual electrolysis cells, or selected cell stacks, or the entire electrolyzer.

20. System according to claim 13, wherein the electrolysis facility is a facility for the electrolysis of alkali chloride solutions or hydrochloric acid.