This invention relates to process for operating an electrochemical desalination device (“ED device”) for desalinating liquids. ED devices and their use for desalinating liquids have been known for a number of years and include electrodialysis (ED) devices, electrodeionization (EDI) devices and electrodialysis metathesis (EDM) devices. ED devices have been used, for example, to remove salts from brackish and sea water and from various industrial solutions or for substituting one ion for another. Typical ED devices comprise at least one anode, at least one cathode, concentrating compartments and desalting compartments, wherein the concentrating compartments and desalting compartments are located between the anode and the cathode. The concentrating compartments and desalting compartments comprise a wall made from an anion exchange membrane (AEM) which is permeable to negatively charged ions and wall made from a cation exchange membrane (CEM) which are permeable to positively charged ions. In some ED devices the walls of compartments are kept open by so-called spacers. The AEMs and CEMs alternate, thereby providing alternate concentrating compartments and desalting compartments. When liquid requiring desalination is passed through the desalting compartments and a current is applied across the electrodes, anions pass from the desalting compartments through the AEM walls and into the concentrating compartments. Similarly cations pass through the CEM walls and into the concentrating compartments. In this way, liquid passing through the desalting compartments of the ED device can be desalinated. Over time, salts present in the liquid being desalinated (particularly calcium salts, magnesium salts and bicarbonates) can precipitate and form scale (insoluble deposits) within the ED device. The scale reduces the ability of the ED device to function. Scale formation can result in an increased electrical resistance, decreased efficiency and eventually in breakdown of the ED device.

A number of techniques are used to slow down scale formation or remove scale from an ED device. One such technique is called polarity reversal. In polarity reversal, the direction of the current applied across the ED device is reversed periodically. The electrode once serving as the cathode becomes the anode and the electrode once serving as the anode becomes the cathode. Also when the polarity is reversed, the concentrating compartments become desalting compartments and the desalting compartments become concentrating compartments

While polarity reversal is useful to slow down undesirable scale formation in an ED device, it is generally not sufficient, on its own, to fully prevent scale from building up and reducing efficiency of the ED device. Therefore the additional step of periodically flushing the ED device (e.g. with an acidic solution to dissolve the scale) is typically used in addition to polarity reversal. Frequent flushing of the ED device is generally necessary to enable ED devices to operate reliably for a good lifetime. However flushing is a slow and laborious process taking typically 4 to 24 hours each time, due to the need to dissolve scale without damaging the membranes within the ED device, resulting in significant down time and loss of productivity. Periodic flushing of the ED device system is generally done at a much lower frequency than polarity reversal, e.g. at a fixed interval in the range of once every 20 to 600 hours. Often the flushing frequency is determined once for a certain ED device at a site and is never changed afterwards.

W02020/023698 describes a method of operating an electrochemical separation device which is operated until the electrical resistance in the electrochemical separation device reaches a first predetermined threshold, after which regeneration is started.

It is the aim of the present invention to provide a more efficient process for desalinating liquids using an ED device than simply flushing the ED device at a set frequency. Furthermore, the present invention seeks to reduce the down time lost to flushing of ED devices.

According to the present invention there is provided a process for operating an electrochemical desalination device for desalination of a liquid, the process comprising the steps of:

(a) monitoring one or more parameters over time during desalination of the liquid;

(b) calculating the average relative change versus time (S) of the one or more parameters;

(c) generating a trigger when the average relative change versus time (S) of the one or more parameters exceeds a predetermined threshold value; and

(d) flushing the electrochemical desalination device in response to the trigger.

Brief description of the drawings

Fig. 1 shows the measured values for several parameters (ED device current, conductivity of the desalted liquid exiting the device within a single desalination period and ED device resistance) at the start of desalination, i.e. using a clean, previously unused ED device

Fig. 2 shows the measured values of the same parameters as Fig. 1 after the ED device has been desalinating river water for 400 hours and polarity reversal being performed every 16 minutes.

Fig. 3 shows the measured values of the same parameters as Fig. 2 (after the 400 hours of operation) after a flushing of the ED device in accordance with the process of the present invention.

Preferably the ED device is or comprises an electrodialysis (ED) device, an electrodeionization (EDI) device or an electrodialysis metathesis (EDM) device.

The ED device typically comprises one or more electrodialysis stacks and optionally a housing for the one or more electrodialysis stacks. The ED device typically comprises desalting compartments and concentrating compartments. Each compartment is separated from adjacent compartments by an AEM on one side and a CEM on the other side.

In this specification, the combination of desalting compartment and a concentrating compartment is referred to as a cell pair. Thus a cell pair is typically bounded by two outer membranes of the same charge (e.g. both are CEMs or both are AEMs) and an inner membrane of opposite charge to the two outer membranes. The inner membrane of opposite charge to the two outer membranes provides an ion- permeable barrier between the compartments on either side.

Preferably the ED device comprises a number of cell pairs (e.g. 10 or more cell pairs, especially 10 to 1200 cell pairs, preferably 20 to 600 cell pairs) and two electrode compartments wherein the cell pairs are located between the electrode compartments. One of the electrode compartments comprises an anode and the other a cathode.

The ED device optionally comprises several hydraulic and electrical stages, e.g. two hydraulic and two electrical stages.

The parameters that are monitored are not particularly limited but in general the parameters are parameters which indicate whether the rate of desalination is changing. For example, the parameters include one or more of the following: the conductivity of the desalted liquid exiting the electrochemical desalination device, the current flowing through the electrochemical desalination device, the (electrical) resistance of the electrochemical desalination device, the voltage across the electrochemical desalination device, the pH of the desalted liquid exiting the electrochemical desalination device, the pH of the concentrated liquid exiting the electrochemical desalination device, the flow rate of the liquid passing through the desalting compartments, the flow rate of the liquid passing through the concentrating compartments, the pressure in the desalting compartments, and the pressure in the concentrating compartments

How the parameter is measured depends on the parameter. However conductivity may be measured using a conductivity meter, current using an ammeter, resistance using an ohmmeter, voltage using a voltmeter, pH using a pH meter, flow rate using a flow rate meter and pressure using a pressure meter. A multimeter may be used to measure several of the electrical parameters.

When current is the monitored parameter it is preferred that the ED device is operated at constant voltage.

When voltage is the monitored parameter it is preferred that the ED device is operated at constant current.

When pressure is the monitored parameter (i.e. liquid pressure), the pressure may be monitored at, for example, an inlet to the ED device, at an outlet of the ED device or at both an inlet and an outlet of the ED device. The inlet/outlet may be for the liquid to be desalinated or a liquid which is concentrated. When pressure is the monitored parameter, preferably the feed liquid (i.e. liquid requiring desalination which enters the ED device) is flow rate-controlled (i.e. the flow is kept constant and the pressure varies).

When flow rate is the monitored parameter, preferably the feed liquid (i.e. liquid requiring desalination which enters the ED device) is pressure-controlled (i.e. the pressure is kept constant and the flow rate varies).

Preferably the monitoring of the one or more parameters comprises measuring the parameter at least 5 times to give a least 5 measurement points, more preferably at least 10 times to give at least 10 measurement points. The measurement points may be taken at a suitable time interval, e.g. on average once every 0.2 to 2 minutes, e.g. once every 0.5 minutes or once every minute.

In contrast to prior processes where the ED device is flushed after a fixed period of time or when a predetermined threshold value of a parameter itself has been reached, the method of the present invention requires calculation of the average relative change versus time (S) of the one or more parameters and flushing occurs when (S) reaches a predetermined value. The present process allows for flushing to occur with more flexibility and can accommodate changes with respect of the feed composition and/or conditions such as temperature without the need to adjust the period of time between flushes. Thus the present process avoids the need to determine a new fixed period of time between flushes to take account of changes in the feed composition and/or the conditions. Furthermore, the flushing starts only when the value of (S) reaches the predetermined value indicating that flushing is necessary, thereby providing a more efficient process with the downtime lost due to flushing kept to a minimum

Preferably calculating the average relative change versus time (S) is based on a number of measurements of a parameter at different times (measurement points) within one (or more) desalination period (i.e. before the first polarity reversal or between two polarity reversals).

The period between the start of the process (when the current is switched on) and the first polarity reversal and the period(s) between each polarity reversal and the period after the final polarity reversal until the current is switched off are called desalination periods.

The present process preferably comprises polarity reversal (i.e. a reversal in the direction in which current flows in the ED device) on average once in every 5 to 60 minutes, more preferably an average of once in every 10 to 30 minutes, e.g. every 20 to 30 minutes.

Each desalination period preferably lasts for 5 to 60 minutes, more preferably 10 to 30 minutes, e.g. 20 to 30 minutes. The direction of the current is reversed between each desalination period.

In order to increase accuracy when calculating the average relative change versus time (S) of the one or more parameters, one may combine measurements of parameters from more than one desalination period, i.e. base the calculation in step (b) on the average value of the average relative change versus time (S) of the one or more parameters for two or more desalination periods.

When polarity reversal takes place, not only does the direction in which current flows across the ED device change but the flow paths of the feed streams are also switched such that what was the concentrating compartment becomes the desalting compartment and what was the desalting compartment becomes the concentrating compartment. As a consequence of the desalting compartment and concentrating compartment switching, the corresponding sensors (e.g. voltmeter, ohmmeter, ammeter etc.) for a particular stream of liquid also change and hence there may be a difference between the measured values of parameters from one desalination period to the next. For this reason it is preferred that the calculation in step (b) uses parameter measurements obtained from desalination periods where the current flows in the same direction, e.g. parameter measurements from every other desalination period.

For a short period of time after a polarity reversal, inevitably there is some cross contamination of the liquids flowing respectively through the desalting compartments and the concentrating compartments. Also when the process first begins it can take a few minutes for parameter measurements to stabilise. Therefore in step (a), it is preferred to disregard the parameter measurements taken in the first few minutes after polarity reversal (i.e. in the first few minutes of each desalination period), or to just not measure the parameters in this period of time. For the same reason, when calculating (S) in step (b) it is preferred not to take into account the parameter measurement in the first few minutes after the process first begins or after a polarity reversal. Thus in step (b) it is preferred that the calculation of (S) is based on parameters measured at least 1 , 2, 3, 4 or 5 minutes after each desalination period has begun (e.g. at least 1 , 2, 3, 4 or 5 minutes after the process has started (for the first desalination period) and e.g. at least 1 , 2, 3, 4 or 5 minutes after each polarity reversal). In this way, one may base the calculation in step (b) on parameters measured when the desalination has stabilised.

There are several methods which may be used to calculate the relative change versus time (S) of a parameter. Such methods include rate of change, ordinary least squares regression (OLS), reduced major axis method (RMA), median slope regression, total least squares or Deming regression, least absolute deviations (LAD) regression, moving average slope, and combinations thereof. An example of a combination of methods is the calculation of the slope based on an average of the first few data points and the average of the last few data points within a desalination period. Preferably the method used to calculate the relative change versus time (S) of a parameter in step (b) and the method used to determine the predetermined threshold value used in step (c) are the same method.

Preferably the average relative change versus time (S) is calculable or calculated using Formula (I): Formula (I) wherein:

DR is the relative change in a parameter (P) at a time t calculated using Formula (II) below;

DR is the average of all DR values for the parameter (P) at several different times; t is the time at which parameter (P) was measured; and t is the average of all of the times at which the parameter (P) was measured;

DR = (Pn/po -1 ) x 100% Formula (II) wherein po is the first measurement point of the parameter (P); and p _n is the n ^th measurement point of the parameter (P).

The above calculation of (S) may be performed using commonly available computer software such as Microsoft Excel or similar.

The predetermined value(s) which generate a trigger in step (c) may be determined by computational modelling or based on field test data at the site of the desalination plant.

In one embodiment the trigger in step (d) may be generated in response to the value of (S) reaching a predetermined threshold for one parameter (e.g. for the parameter of the current flowing through the electrochemical desalination device). In another embodiment, the trigger in step (d) may be generated only in response to values of (S) being reached for two or more parameters (e.g. the (S) value being met for at least two of current, resistance and conductivity).

This latter embodiment requiring at values of (S) being reached for two or more parameters has the advantage over using only one parameter (S) that if one sensor used to measure a particular parameter malfunctions the trigger will not be generated unless at least one other parameter also exceeds its predetermined threshold value. In this way, downtime due to premature flushing caused by one faulty sensor can be avoided.

In one embodiment the trigger generated in step (c) causes the flushing in step (d) without human intervention. For example, steps (a) to (d) are controlled by a computer program, e.g. in a PLC, in which step (d) is performed as a consequence of the trigger generated in step (c). In another embodiment, the trigger generated in step (c) causes the flushing in step (d) with human intervention. For example, step (c) activates an alarm and in response to that alarm a human starts step (d), e.g. manually.

The trigger generated in step (c) optionally triggers the flushing in step (d) immediately or there may be a delay, e.g. to the end of the desalination period in which the trigger is generated.

When performing the process of the present invention a number of different strategies may be used. For example, one may choose a low predetermined threshold value for step (c) to ensure that only a low amount of scaling occurs in the ED device and only a short flushing period and/or mild flushing conditions (e.g. using a flushing liquid having a pH which is not very acidic) are required. Alternatively one may choose a higher predetermined threshold value for step (c) which results in a higher degree of scaling in the ED device and requires more aggressive flushing conditions (e.g. a longer flushing period and/or harsher flushing conditions (e.g. using a flushing liquid having a pH which is more acidic). Ultimately there will be a trade-off between the using a high predetermined threshold value which results in higher scaling and the requirement for more aggressive flushing and a lower predetermined threshold value which results in less scaling and less aggressive flushing but requires flushing more often.

Advantageously the process is flexible and may be tuned as desired to provide the option between, for example, a high frequency of short duration flushing or a lower frequency of long duration flushing. Of course a combination of these strategies is also possible within the process e.g. the process optionally comprises a number of short flushing periods alternated with a single long flushing period. One may also perform the process such that that if the interval between two flushing procedures (step (d)) becomes shorter than a predetermined time period then an adapted flushing scheme is included in the process, e.g. using a longer flush and/or a more aggressive flushing liquid (e.g. fluid of lower pH).

The process of the present invention provides efficient desalination where flushing is done only when it is actually needed rather than at fixed intervals as is current practice. As a result the productivity of the ED device is optimized and time spent flushing the ED device is reduced.

In one embodiment, when the parameter being monitored in step (a) is conductivity of the desalted liquid exiting the electrochemical desalination device, the predetermined threshold value of that parameter used in step (c) is preferably at least 0.04%/minute (i.e. when the conductivity of the desalted liquid exiting the electrochemical desalination device changes by 0.04%/minute the trigger is generated in step (c)). A predetermined threshold value of 0.04%/minute for the conductivity of the desalted liquid exiting the electrochemical desalination device generally provides flushing in step (d) at a point when there is a low level of scaling within the ED device. In another embodiment, when a higher degree of scaling is acceptable, the predetermined threshold value of the conductivity of the desalted liquid exiting the electrochemical desalination device (as used in step (c)) is preferably at least 0.1 %/minute, more preferably at least 0.2%/minute and especially at least 0.4%/minute.

In order to avoid excessive scaling of the ED device, when the parameter being monitored in step (a) is conductivity of the desalted liquid exiting the electrochemical desalination device, the predetermined threshold value of that parameter is preferably lower than 2.0%/minute.

In one embodiment, when the parameter being monitored in step (a) is the current flowing through the electrochemical desalination device, the predetermined threshold value of that parameter used in step (c) is preferably at most -0.2 %/minute. Generally this results in a low level of scaling within the ED device. In another embodiment, when a higher degree of scaling is acceptable, the predetermined threshold value of the current flowing through the electrochemical desalination device is preferably at most -0.4%/minute, more preferably at most -1.0%/minute, especially at most -1 .5 %/minute.

In order to avoid excessive scaling of the ED device, when the parameter being monitored in step (a) is the current flowing through the electrochemical desalination device, the predetermined threshold value of that parameter is preferably higher than -3.0%/minute. For the avoidance of doubt, -2.0 is higher than -3.0 due to the negativity of the value.

The current flowing through the electrochemical desalination device may be measured between the power supply and the device.

In one embodiment, when the parameter being monitored in step (a) is the voltage across the electrochemical desalination device, the predetermined threshold value of that parameter used in step (c) is preferably at least 0.2%/minute. Generally this results in a low level of scaling within the ED device. In another embodiment, when a higher degree of scaling is acceptable, the predetermined threshold value of the voltage across the electrochemical desalination device is preferably at least 0.4%/minute, more preferably at least 1.0%/minute, especially at least or 1 .5%/minute.

In order to avoid excessive scaling of the ED device, when the parameter being monitored in step (a) is the voltage across the electrochemical desalination device, the predetermined threshold value of that parameter is preferably lower than 3.0%/minute.

The voltage across the electrochemical desalination device may be measured between the power supply to the cathode and the power supply to the anode.

In one embodiment, when the parameter being monitored in step (a) is the resistance of the electrochemical desalination device, the predetermined threshold value of that parameter used in step (c) is preferably at least 0.2%/minute Generally this results in a low level of scaling within the ED device. In another embodiment, when a higher degree of scaling is acceptable, the predetermined threshold value of the resistance of the electrochemical desalination device is preferably at least 0.4%/minute, more preferably at least 1.0%/minute, especially at least or 1 .5%/minute. In order to avoid excessive scaling of the ED device, when the parameter being monitored in step (a) is the resistance of the electrochemical desalination device, the predetermined threshold value of that parameter is preferably lower than 3.0%/minute.

The resistance across the electrochemical desalination device is generally determined by calculating the ratio of voltage and current.

The predetermined threshold values mentioned above are mere examples and may be adapted to the local situation such as feed liquid characteristics and membrane properties.

The flushing in step (d) preferably comprises passing a flushing fluid through the ED device. The flushing fluid may be passed through the desalting compartments, the concentrating compartments or through both the desalting compartments and the concentrating compartments.

Each flushing in step (d) is preferably performed for a period of 1 to 24 hrs, depending on the chosen threshold and the conditions applied.

Preferred flushing fluids include dilute acid, e.g. dilute hydrochloric acid, citric acid, sulphamic acid solution or a mixture comprising two or more thereof.

The preferred pH of the flushing solution will depend on the properties of the feed liquid and in some cases on the sensitivity of the AEMs and CEMs to acid and the properties of spacers used (if any). Preferably the flushing liquid has a pH of 0 to 6.5; in many cases a pH of between 1 and 2 is considered appropriate.

Step (d) optionally further comprises introducing air bubbles into the flushing liquid to create turbulence and/or flushing. Step (d) optionally further comprises flushing with alkaline and/or hypochlorite solutions as a further flushing fluid, e.g. to remove biological fouling.

Preferably step (d) comprises the simultaneous flushing of the desalting compartments and the concentration compartments.

During the process steps (a) to (d) may be performed a number of times, from 10 times to more than 1000 times for desalination processes that run for several years or decades.

The invention will now be illustrated by the following non-limiting Examples in which all parts and percentages are by weight unless otherwise specified.

Examples

An ED device was constructed as described in Table 1 below. The ED device contained a stack of 300 cell pairs (each pair comprising one desalting compartment and one concentrating compartment) between an anode compartment comprising the anode and a cathode compartment comprising the cathode:

River water having the composition described in Table 2 was fed through the desalination compartments of the ED device at a flow velocity of about 10 cm/s. River water was circulated through the concentration compartments of the ED device at a flow velocity of about 10 cm/s. A constant voltage of about 1 V/cell pair was applied across the ED device (via the electrodes). The polarity across the electrodes was reversed every 16 minutes to give 16-minute desalination periods and simultaneously the feeds into the desalting compartments and concentrating compartments was also reversed.

Table 2. Composition of River Water Steps (a) and (b) - Monitoring and Calculating the average relative change versus time

(S 1

The following parameters were measured over time during the desalination of the river water: (i) current flowing through the electrochemical desalination device (abbreviated below to “Current”); (ii) conductivity of the desalted river water exiting the electrochemical desalination device (abbreviated below to “Product Conductivity”); and (iii) the resistance of the electrochemical desalination device (abbreviated below to “Resistance”); and the results are shown in Table 3 below along with the relative change of each parameter (DR, calculated using Formula (II) described above while ignoring the values until five minutes had elapsed to allow for stabilisation):

Fig. 1 to Fig. 3 show plots of the monitored data described in Tables 3a, 4a and 5a below. In Tables 3a to 3d, 4a to 4d and 5a to 5d the average relative change versus time (S) of each parameter was calculated starting with the measurement point 5 minutes after the process had started or a polarity switch to ensure that the values were stable. In each desalination period 12 data points were taken (5 min to 16 min) with one minute time interval. Data and calculated average relative change versus time S are given in Tables 3 to 5.

In Table 3a to 3c the situation at the start of the desalination is shown, i.e. clean compartments without scaling. In Table 3d the average results are shown.

Table 3a. Monitored data at the start of desalination: desalination period 1

To following illustrates how the relative change versus time (S) of -0.12 was calculated for the current parameter:

The measurements began after 5 minutes when the desalination had stabilised (t=5 min). 5 minutes was taken as first measurement point. DR for the 2 ^nd measurement point (t=6 min) was for the parameter current (24.2/24.3 - 1) x100% = -0.412. This same calculation was performed for all measurement points using Formula (II) described above.

S was then calculated using Formula (I) t was the average value of the time (from t=5 to t=16) calculated using Microsoft Excel and was found to be 10.5 minutes. The average of the DR values for current was similarly calculated and was found to be -0.926%. Then, for each measurement point, the difference between the calculated value DR and the calculated average value was calculated, e.g. for the 2 ^nd measurement point (DR - DR ) was (-0.412 - -0.926) = 0.514. For this measurement point (t - 1 ) = (6 - 10.5) = -4.5. Subsequently (DR - DR )(t - 1 ) is calculated as well as (t - t ) ² . For the 2 ^nd measurement point these values were respectively -2.31 and 20.25. The same was done for all measurement points. S was then the ratio of the sum of all values obtained for (DR - DR )(t - t ) and the sum of all corresponding values obtained for (t - t ) ² . For current the value for S was found to be -0.122, rounded off to -0.12.

The relative change versus time (S) for the product conductivity and resistance were calculated in an analogous manner and were found to be 0.02%/m in and 0.10%/min respectively, as shown in the penultimate row of Table 3a.

Desalination continued, the subsequent polarity reversal periods were monitored and of two of them the corresponding measurements are as shown in Table 3b and 3c below:

Table 3b. Monitored data at the start of desalination, desalination period 2

Table 3c. Monitored data at the start of desalination: desalination period 3

Table 3d. Average values for desalination periods 1 to 3, at the start of desalination

After 400 Hours of Operation After about 400 hours of operation a moderate amount of scaling had formed as can be concluded from the monitored parameters. These results are shown in Table 4a to 4d.

The average relative change versus time (S) of all three monitored parameters when measured after 400 hours of operation (Tables 4a to 4d) were significantly higher value than the corresponding values measured initially (Tables 3a to 3d).

Table 4a. Monitored data after 400 hours of operation: desalination period 1

Table 4b. Monitored data after 400 hours of operation: desalination period 2

Table 4c. Monitored data after 400 hours of operation: desalination period 3

Table 4d. Average values for desalination period 1 to 3, after 400 hours of operation

The average relative change versus time (S) of the current had a value of -1.24 %/m in, the average relative change versus time (S) of the product conductivity had a value of 0.44 %/min and the average relative change versus time (S) of resistance had a value of 1.29 %/min.

Step (c) - generating a trigger when the average relative change versus time (S) of the one or more parameters exceeds a predetermined threshold value

The predetermined threshold values for generating a trigger were estimated based on model calculations and tested under real life conditions. The predetermined threshold values for current, product conductivity were set on -1.20, 0.45 and 1.20 %/minute respectively. The trigger for flushing (step (d)) required two out of the three predetermine values to be met. Based on the data it was decided to manually start the flushing procedure.

Step (d) - flushing the electrochemical desalination device in response to the trigger. The flushing of the ED device was executed for 10 hours at room temperature with a flushing solution (hydrochloric solution having a pH of 1.5). After flushing, the process was continued and the values of the parameters (post-flushing) and the values of (S) are shown in Tables 5a-5d below.

Table 5a. Monitored data after 400 hours of operation and flushing desalination period 1

Table 5b. Monitored data after 400 hours of operation and flushing, desalination period 2

Table 5c. Monitored data after 400 hours of operation and flushing, desalination period

3

Table 5d. Average values for desalination period 1 to 3, after 400 hours of operation and flushing The results in Tables 5a to 5d show that the flushing was successful in descaling the ED device: after flushing the average relative change versus time (S) of all three parameters (current, product conductivity and resistance) had returned to low values similar to those in Tables 3a to 3d when the desalination first begun with fresh, previously unused membranes. The results show that for desalination of the river water described in Table 2 using the ED device described in Table 1 the following three predetermined threshold values are effective to be used to trigger flushing:

(S) for current flowing through the electrochemical desalination device: -1.25

%/m in; (S) for the conductivity of the desalted liquid exiting the electrochemical desalination device: 0.45 %/min; and

(S) for resistance of the electrochemical desalination device: 1.25 %/min. Preferably either two out of the three parameters or all three parameters may be taken as trigger for starting the flushing procedure.