MONITORING A CLOSED WATER SYSTEM

Abstract

The present invention relates to a system and method for continuous monitoring of system health in closed water systems. Sensors are provided to measure a plurality of system parameters. The measurements are compared to threshold ranges. A diagnosis of system health, specifically in relation to corrosion, is derived from the comparison of at least a first parameter to its threshold range. The diagnosis is further refined by comparison of a further parameter to its threshold range. Also disclosed are example sensors and methods, including galvanic sensors, optical corrosion sensors and methods for monitoring the effectiveness of inhibitors in the water system using conductivity measurements.

Claims

1-105. (canceled)

106. An optical sensing apparatus for mounting in a water system and for monitoring corrosion in the water system, comprising: a metal sample having a uniform thickness and a first planar surface and a second planar surface opposite the first planar surface, wherein the first planar surface is arranged to be in contact with water within the water system; a light source configured to emit light towards the second planar surface of the metal sample; and a light sensor configured to receive light reflected by the second planar surface of the metal sample, and output a signal indicative of the intensity of the reflected light.

107. An optical sensing apparatus according to claim 106, further comprising a transparent element disposed at least partly between the light source and the metal sample.

108. An optical sensing apparatus according to claim 107, wherein the transparent element has a third planar surface arranged adjacent to the second planar surface of the metal sample.

109. An optical sensing apparatus according to claim 106, further comprising a seal for protecting the second planar surface from water in the water system.

110. An optical sensing apparatus according to claim 109, wherein the seal is located adjacent to the first planar surface and wherein the metal sample is provided with a corrosion resistant coating on the first planar surface in the vicinity of the seal.

111. (canceled)

112. An optical sensing apparatus according to claim 106, wherein the metal sample is a film having a thickness of 1 mm or less.

113-118. (canceled)

119. An optical sensing apparatus according to claim 106 wherein the or each metal sample is replaceable.

120. An optical sensing apparatus according to claim 106, further comprising a processor configured to receive the signal indicative of the intensity of the reflected light from the light sensor.

121. (canceled)

122. An optical sensing apparatus according to claim 120, wherein the processor is configured to relate the signal indicative of the intensity of the reflected light to corrosion of the metal sample, and wherein the light sensor is configured to output a plurality of signals indicative of the intensity of the reflected light over time and wherein the processor is configured to determine a rate of corrosion of the metal sample from the plurality of signals indicative of the intensity of the reflected light over time received from the light sensor.

123. (canceled)

124. An optical sensing apparatus according to claim 120, wherein the processor is configured to receive a plurality of signals indicative of an amount or a rate of corrosion of a corresponding plurality of metal samples; and wherein the plurality of metal samples are formed from the same metal as one another, wherein each of the plurality of metal samples has a different thickness and wherein the processor is configured to determine a range of maximum pinhole corrosion depths in the water system from the plurality of received signals.

125. An optical sensing apparatus according to claim 106, further comprising an optical element configured to direct light emitted by the light source towards the second planar surface of the metal sample, and/or configured to direct the reflected light towards the light sensor.

126-133. (canceled)

134. A method of monitoring corrosion in a water system, the method comprising: mounting a metal sample in a water system, the metal sample having a uniform thickness and including a first planar surface and a second planar surface opposite the first planar surface, wherein the first planar surface is arranged in contact with water of the water system; emitting light towards the second planar surface of the metal sample; receiving light reflected by the second planar surface of the metal sample at a light sensor; generating a signal indicative of the intensity of the reflected light; and correlating the intensity of the reflected light to corrosion of the metal sample.

135. A method according to claim 134, further comprising generating a plurality of signals indicative of the intensity of the reflected light over time, and determining a rate of corrosion from the plurality of intensities of the reflected light over time.

136-139. (canceled)

140. The method according to claim 134, further comprising a second light sensor for directly sampling the light emitted from the light source to provide a reference value, wherein correlating the intensity of the reflected light to corrosion of the metal sample includes comparing the reference value to the received light reflected from the second planar surface.

141-147. (canceled)

148. The method according to claim 134, further comprising controlling one or more of: power; intensity; and/or spectral weight of the emitted light.

149. A sample for use in an optical sensor for monitoring corrosion in a water system, the sample comprising: a metal element having: a first surface for exposure to the water of the water system; and a second surface opposite the first surface for receiving and reflecting light; wherein a portion of the first surface is provided with a corrosion-resistant coating for providing a location for forming a seal between the sample and the sensor.

150. The sample of claim 149, wherein the metal element is planar and/or has a uniform thickness.

151. The sample of claim 149, wherein the corrosion-resistant coating is applied to the edges of the metal element.

152. The sample of claim 151, wherein the corrosion-resistant coating extends between 0.5 mm and 5 mm inward from the edge of the first surface.

153-155. (canceled)

156. The sample of claim 149, wherein the metal element is formed from stainless steel, copper, brass, aluminium, or other materials representative of metals in the water system.

157-159. (canceled)

Description

[0267] A selection of specific examples will now be described in detail to illustrate some of the effects of the system and method described herein, with reference to the Figures, in which:

[0268] FIG. 1 shows a flow chart representing the method disclosed;

[0269] FIG. 2 shows a flow chart representing an example of the method refining a corrosion diagnosis caused by low pressure, according to one embodiment;

[0270] FIG. 3 shows a flow chart representing an example of the method refining a corrosion diagnosis caused by a leak, according to one embodiment;

[0271] FIG. 4 shows a flow chart representing an example of the method refining a corrosion diagnosis caused by high pressure, according to one embodiment;

[0272] FIG. 5 shows a flow chart representing an example of the method verifying the occurrence of planned maintenance events, according to one embodiment;

[0273] FIG. 6 shows a flow chart representing an example of the method identifying a maintenance event, according to one embodiment;

[0274] FIG. 7 shows a graph representing changes in dissolved oxygen levels over time due to maintenance events, according to one embodiment;

[0275] FIG. 8 shows a graph representing changes in pH levels over time due to maintenance events, according to one embodiment;

[0276] FIG. 9 shows a drawing of a system configured to carry out the method disclosed, according to one embodiment;

[0277] FIG. 10 shows a schematic of a sensor as described herein;

[0278] FIG. 11 illustrates the relationship between conductivity, dissolved oxygen and galvanic current;

[0279] FIG. 12 shows a flow chart of a method of determining inhibitor concentrations from the conductivity of the water, compensating for temperature effects;

[0280] FIG. 13 shows a schematic of an example embodiment of an optical sensor for detecting and/or monitoring corrosion in a water system;

[0281] FIG. 14 shows a schematic of another example embodiment of an optical sensor for detecting and/or monitoring corrosion in a water system, having a plurality of metal samples of different thicknesses;

[0282] FIG. 15 shows a schematic of another example embodiment of an optical sensor for detecting and/or monitoring corrosion in a water system, having a replaceable metal sample;

[0283] FIG. 16 shows a schematic of another example embodiment of an optical sensor for detecting and/or monitoring corrosion in a water system, having a light source and light sensor housed in a separate unit;

[0284] FIG. 17 shows a schematic of another example embodiment of an optical sensor for detecting and/or monitoring corrosion in a water system, having a baffle between the light source and the light sensor;

[0285] FIG. 18 shows a schematic of an example embodiment of an optical sensor for detecting and/or monitoring corrosion in a water system, having a second light sensor for directly sampling light emitted by the light source;

[0286] FIG. 19 shows a schematic of another example embodiment of an optical sensor for detecting and/or monitoring corrosion in a water system, having a second light sensor for directly sampling light emitted by the light source;

[0287] FIG. 20 shows a flow chart illustrating an exemplary method according to the present disclosure;

[0288] FIG. 21A shows a schematic of a plan view of a metal sample for use in the sensor;

[0289] FIG. 21B shows a side view of the metal sample for use in the sensor; and

[0290] FIG. 22 shows a schematic of an arrangement of multiple metal samples in a sensing apparatus.

[0291] The Figures will now be described in more detail. In each case, similar elements are labelled with the same number. The devices presented operate along the same broad principles, and consequently the overall operation will not be described in detail in each case. Instead, the differences between each Figure will be emphasised, and it is to be understood that operational principles are generally transferrable between each Figure, except where this would cause a contradiction.

[0292] FIG. 1 shows a flowchart representing the method disclosed. The first step 101 comprises receiving, from a first sensor, a value of a first parameter selected from the plurality of parameters. The plurality of parameters is based on at least one of the following: pressure; make-up water flow rate;

[0293] dissolved oxygen; cumulative dissolved oxygen; inhibitor dosing levels; biofilm accumulation; temperature; conductivity; galvanic current; cumulative galvanic current; crevice corrosion rate; and/or pH. This may involve measuring the value directly, or obtaining it through proxy means and inferring the result. The second step 102 comprises comparing the received value of the first parameter to a threshold range for the first parameter. The third step 103 comprises providing a diagnosis of a corrosion state based on the comparison of the value of the first parameter to the threshold range for the first parameter. The fourth step 104 comprises receiving, from a further sensor, a value of a further parameter selected from the plurality of parameters. This may involve measuring the value directly, or obtaining it through proxy means and inferring the result. As discussed above, this further parameter provides a more targeted approach to diagnose system abnormalities. The fifth step 105 comprises comparing the received value of the further parameter to a threshold range for the further parameter. The sixth step 106 comprises refining the diagnosis of the corrosion state based on the comparison of the value of the further parameter to the corresponding threshold range.

[0294] FIG. 2 shows an example embodiment of the method involving pressure as a determined parameter, and an example process for if the pressure was too low. In this scenario, first step 201 involves receiving, from a pressure sensor, the pressure of the system. The step 202 involves comparing the value obtained from the pressure measurement to a predefined threshold range of acceptable pressure values. If the measured value of the pressure is outside the threshold range, step 203 occurs. Step 203 involves providing a diagnosis based on the comparison. The diagnosis may comprise that the pressure is outside the threshold range, and in this case further comprising that it is too low, and below the lower limit of the threshold range. However, if the measured value is within the threshold range, step 203a occurs. Step 203a involves providing a diagnosis that the pressure is normal and within the threshold region. If the method takes this branch, then the system is behaving as normal. However, following this, further measurements of the same or different parameters may be taken in order to verify that the system is behaving normally.

[0295] In the instance that the pressure is too low, air may be drawn into the system, possibly eventually causing corrosion. At this point, an alert/message may be sent to the user that the pressure is too low, and optionally that further checks may be made to check for any signs of potential corrosion. To check if this is the case, further parameters may be measured in order to refine the diagnosis. If the pressure is too low, the process proceeds to step 204, where a further parameter value is received, in this case the dissolved oxygen level. Step 205 compares the dissolved oxygen level with a threshold range of acceptable dissolved oxygen levels. If the dissolved oxygen level is outside the threshold range, then step 206 occurs. Step 206 involves refining the diagnosis by determining that the dissolved oxygen levels are higher than the upper limit of the threshold range. If the measured dissolved oxygen level is within the threshold range, step 206a occurs in refining the diagnosis and stating that the oxygen levels are normal. In this case, the pressure may be checked again to ensure that it wasn't an anomalous result, and the dissolved oxygen levels may be monitored and measured again at a later time or at regular time intervals to check for any increase. Optionally, other parameters may be measured to ensure the diagnosis is correct.

[0296] If the oxygen levels are too high, then this suggests that air has been drawn into the system as a result of the system pressure being too low. This may lead to corrosion if left unattended. At this point, a further alarm/message may be sent to the user to bring it to their attention that the diagnosis has been confirmed. Optionally, other parameters may then be determined in order to further verify the diagnosis, or take a measurement of the rate of corrosion. Corrective action to move each parameter value back to within the threshold range may occur at any stage, preferably after the corresponding diagnosis is provided. For example, the pressure may be adjusted to within the threshold range after the diagnosis of the pressure being too low. Optionally, this may occur at another stage after further diagnosis.

[0297] For example, the method may progress to step 207 where a value of the galvanic current is received. This is then compared to a threshold range in step 208. Following this, if the measured value of the galvanic current is outside the threshold range, then step 209 occurs in refining the diagnosis and confirming that corrosion is taking place. However, if the galvanic current is within the threshold range, then step 209a in refining the diagnosis in that no corrosion has taken place yet is performed. Following step 209a, values of other parameters may be received to further determine if corrosion is taking place. If step 209 takes place, further parameters may also be measured such as crevice corrosion rate.

[0298] This embodiment provides an example of how a low pressure state can be detected and corrected as a preventative method to stop corrosion. The combination of measurements allows the system to identify a low pressure state which has led to air being drawn into the system, and may provide a diagnosis that corrosion is likely to occur before it has happened. This holistic overview of the system health can be used to take preventative corrective action.

[0299] This example also shows how the method can be used in a reverse sequence to identify the cause of a positive corrosion state instead of confirming corrosion. For example, if dissolved oxygen levels are above the threshold range, then the further parameter may be pressure to check if the pressure is too low, causing air ingress. Depending on this comparison, the diagnosis suggests a potential cause of the positive corrosion state. For example, if the pressure is normal, then the refined diagnosis may suggest that there may be a leak. Receiving a value of the make-up water flow rate may help refine the diagnosis and narrow down the corrosion cause.

[0300] FIG. 3 is another example embodiment showing an example process of detecting a leak in a closed water system. In this example, first step 301 involves receiving a value of the make-up water flow rate which could be measured by a water meter or flow sensor on the water make-up line. Step 302 compares the measured make-up water flow rate with a threshold range. If the value is outside the threshold range, and make-up water is being drawn into the system, then step 303 occurs. Step 303 involves providing a diagnosis that make-up water is being drawn into the system, which may be due to a leak in the closed water system, causing the system to be topped up with make-up water. This make-up water may be aerated, which may bring oxygen into the water system, and consequently lead to corrosion. However, if the measured make-up water is within the threshold range, then a substantial amount of make-up water has not been drawn into the system, and the diagnosis can be provided to state that the flow rate is normal during step 303a. In this example, the flow rate may be measured again at a later stage, along with other parameters.

[0301] If the water make-up flow rate is too high, the process proceeds to step 304, where a value of a further parameter is received, in this case the dissolved oxygen level in order to establish if the make-up water has brought oxygen into the system which in turn may lead to corrosion. It can also be used to further verify that make-up water has been drawn into the system. Step 305 compares the measured dissolved oxygen level with a threshold range of acceptable dissolved oxygen levels. If the dissolved oxygen level is outside the threshold range, then step 306 occurs. Step 306 involves refining the diagnosis by determining that the dissolved oxygen levels are higher than the upper limit of the threshold range. If the measured dissolved oxygen level is within the threshold range, step 306a occurs in refining the diagnosis and stating that the oxygen levels are normal. In this case, the make-up water flow may be checked again to ensure that it wasn't an anomalous result, and the dissolved oxygen levels may be monitored and further values used at a later time or at regular time intervals to check for any increase. Optionally, other parameters may be measured to ensure the diagnosis is correct. This checking process may be carried out at any step that involves providing a diagnosis or updating the diagnosis such as correcting or confirming the diagnosis.

[0302] If the oxygen levels are too high, then this suggests that aerated make-up water has been drawn into the system as a result of a leak in the system. This may lead to corrosion if left unattended, while the leak may have disastrous consequences depending on the function and location of the system. At this point, a further alarm/message may be sent to the user to bring it to their attention that the diagnosis has been confirmed and a leak may be occurring. Optionally, additional parameters may then be determined in order to further verify the diagnosis, or take a measurement of the rate of corrosion.

[0303] For example, the method may progress to step 307 where values of galvanic current are received from a galvanic current sensor. This is then compared to a threshold range in step 308. Following this, if the measured valued of the galvanic current is outside the threshold range, then step 309 occurs in refining the diagnosis and confirming that corrosion is taking place. However, if the galvanic current is within the threshold range, then step 309a in refining the diagnosis in that no corrosion has taken place yet is performed. Following step 309a, values of further parameters may be used to further determine if corrosion is taking place. If step 309 takes place, further parameters may also be measured such as crevice corrosion rate.

[0304] FIG. 4 is another example embodiment of the method disclosed, showing an instance where the pressure may be too high. This embodiment comprises steps that have previously been described, but are used to determine a different problem. The holistic overview of using measurements of set parameters can be used to identify the root cause of a problem. For example, in this case all the parameters of FIG. 3 are measured, and if they are all outside the threshold, the cause may be that a leak has occurred in the closed water system. However, by measuring the pressure, it may be determined that a high pressure has led to water loss through automatic air vents or other components, causing make-up water to be drawn into the system and consequent increase in dissolved oxygen.

[0305] In this example, first step 401 involves receiving a value of the pressure of the system. The step 402 involves comparing the value obtained from the pressure measurement to a predefined threshold range of acceptable pressure values. If the measured value of the pressure is outside the threshold range, step 403 occurs. Step 403 involves providing a diagnosis based on the comparison. The diagnosis may comprise that the pressure is outside the threshold range, and in this case further comprising that it is too high, and above the upper limit of the threshold range. However, if the measured value is within the threshold range, step 403a occurs. Step 403a involves providing a diagnosis that the pressure is normal and within the threshold region. If the method takes this branch, then the system is behaving as normal. However, following this, further measurements of the same or different parameters may be taken in order to verify that the system is behaving normally.

[0306] In the example that the pressure is too high, water may be forced out through automatic air vents (AAVs), pressure relief valves (PRVs) or other components, leading to aerated make-up water being drawn into the system, possibly eventually leading corrosion. At this point, an alert/message may be sent to the user that the pressure is too high, and optionally that further checks may be made to check for any signs of potential corrosion. To check if this is the case, further parameters may be measured in order to refine the diagnosis. If the pressure is too high, the process proceeds to step 404, where values of a further parameter are usedin this case the make-up water flow which could be measured by a water meter or flow sensor on the water make-up line. Step 405 compares the measured make-up water flow rate with threshold range. If the value is outside the threshold range, and make-up water is being drawn into the system, then step 406 occurs. Step 406 involves refining the diagnosis, confirming that the pressure is too high and has caused make-up water to be drawn in due to system water loss through automatic air vents. However, if the measured make-up water is within the threshold range, then a substantial amount of make-up water has not been drawn into the system, and the diagnosis can be refined correspondingly during step 406a. In this example, the flow rate may be measured again at a later stage, along with other parameters in order to ensure that the high pressure does not cause an uptake of make-up water at a later stage. A corrective action such as adjusting the pressure to within the threshold range may be taken.

[0307] If the water make-up flow rate is above a threshold (in some examples greater than zero), the process proceeds to step 407, where values of an additional, third parameter are receivedin this case the dissolved oxygen level in order to establish if the make-up water has brought oxygen into the system which in turn may lead to corrosion. It can also be used to further verify that make-up water has been drawn into the system. Step 408 compares the measured dissolved oxygen level with a threshold range of acceptable dissolved oxygen levels. If the dissolved oxygen level is outside the threshold range, then step 409 occurs. Step 409 involves refining the diagnosis by determining that the dissolved oxygen levels are higher than the upper limit of the threshold range. If the measured dissolved oxygen level is within the threshold range, step 409a occurs in refining the diagnosis and stating that the oxygen levels are normal. In this case, the pressure may be checked again to ensure that it wasn't an anomalous result, and the dissolved oxygen levels may be monitored and measured again at a later time or at regular time intervals to check for any increase. Optionally, other parameters may be measured to ensure the diagnosis is correct. This checking process may be carried out at any step that involves providing a diagnosis or refining the diagnosis such as correcting or confirming the diagnosis, confirming the cause, or re-assessing the cause of the positive corrosion state.

[0308] If the oxygen levels are too high, then this suggests that aerated make-up water (i.e. fresh water having more dissolved oxygen than is desirable for system water) has been drawn into the system as a result of the system pressure being too high causing water loss through automatic air vents, pressure relief valves, or other components. This may lead to corrosion if left unattended. At this point, a further alarm/message may be sent to the user to bring it to their attention that the diagnosis has been confirmed. Optionally, further parameters may then be determined in order to further verify the diagnosis, or take a measurement of the rate of corrosion.

[0309] For example, the method may progress to step 410 where values of the galvanic current are received. This is then compared to a threshold range in step 411. Following this, if the measured valued of the galvanic current is outside the threshold range, then step 412 occurs in refining the diagnosis and confirming that corrosion is taking place. However, if the galvanic current is within the threshold range, then step 412a in correcting the diagnosis in that no corrosion has taken place yet is performed. Following step 412a, further parameters may be measured to further determine if corrosion is taking place. If step 411 takes place, further parameters may also be measured such as crevice corrosion rate.

[0310] By measuring certain parameters, the root cause of a system positive corrosion state can be identified, which may be rectified before corrosion takes place. The thresholds may be adjusted to new maintenance thresholds wherein the maintenance thresholds represent new limits within which normal operation of the maintenance event occurs. For example a parameter may be outside the normal threshold for a positive corrosion state during a maintenance event such as the dissolved oxygen concentration rising sharply during a mains water flush. However, this may be expected during the maintenance event, and hence a new maintenance threshold is used. If a parameter is outside this maintenance limit then there is an issue with the maintenance event being carried out, and the user can be alerted in the usual manner for corrosion detection mode.

[0311] FIG. 5 shows a flowchart of an example embodiment of the method disclosed, used for the assessment of a planned maintenance event. For example, if the system is configured in a maintenance mode, and the system is expecting specific maintenance events to occur, then this allows the diagnosis to confirm that the event has been successful by monitoring certain parameters. For example, if the system is expecting a water flush event 501, then values of the dissolved oxygen levels may be received in step 502. If there is a steep rise in dissolved oxygen levels 503 then this can be used to confirm that the water flush has occurred. This maintenance event may then be labelled on a graph displaying the monitored dissolved oxygen levels. Values of other parameters such as pH may be used such as in step 504 in order to confirm the water flush. Other parameters such as conductivity in step 506 may also be performed to further validate the measurements of other parameters in determining a maintenance event. However, in some embodiments only two measurements may be used to provide the diagnosis that an event such as a water event has occurred. Clearly it is preferable to monitor several parameters to provide a more detailed overview of the system, and multiple parameters can be used to inform the user about changing conditions which may be indicative of various events.

[0312] If a change in pH in step 505 is detected at the same time as the dissolved oxygen increased, this further confirms the diagnosis that the water flush occurred. For example, this may be as a result of the parameter (e.g. pH) reaching an expected value. In addition, if a change in conductivity is detected in step 507, this also indicates a water flush in combination with the measurements of other parameters. The values obtained and associated changes in parameters can be used to inform the diagnosis in step 508 which may be that a water flush has occurred. This can also comprise informing the user of a suspected water flush occurring.

[0313] Furthermore, the overall time of the water flush can be determined by recording the time of the deviation from normal. This can be performed by an additional step after diagnosis 508, which is not shown in FIG. 5. For example, the flowchart may comprise determining the duration of the water flush. Upon a comparison of this time to a required pre-set time for a mains water flush to occur successfully, this will confirm or otherwise that the water flush was carried out for the required length of time.

[0314] Further parameters may be monitored that are not shown in FIG. 5. In other examples, fewer than three parameters may be monitored and used to diagnose the maintenance event.

[0315] Further maintenance events may be monitored in this way. For example, in the pre-commissioning process it is typical that inhibitor may be added in step 509 shortly after the mains water flush occurs. This can be confirmed by receiving further values of conductivity after addition of inhibitor in step 510 and observing a sudden increase at 511. This may be further confirmed by receiving values of the pH in 512 and observing the increase due to the water flush eventually plateauing in step 513. These indications allow the diagnosis 514 to confirm that the inhibitor has been successfully added. Of course, if the expected changes to parameters are not detected, this can indicate that each maintenance event has not occurred successfully, and an alert may be sent to the user.

[0316] Further parameters may be monitored to ensure that the corrosion inhibitor is being effective. For example, the galvanic current or conductivity may be monitored to observe the levels of inhibitor and ensure it is effective in suppressing corrosion. Other maintenance events may be monitored in this way such as dynamic flushing, observing a cleaning chemical added to the water system, and draining the system. In some cases, this may be used to identify unplanned maintenance events, which may also be detected even if the system is not configured in the maintenance mode. For example, it may detect errors such as equipment failure or unplanned events such as a heating event that should not be occurring.

[0317] FIG. 6 is a flowchart showing a method of identifying or confirming a maintenance event using measurements of multiple parameters. For example, given a detected rise in a parameter such as dissolved oxygen at 601, values of various parameters may be used to determine what has caused this sudden rise. For example, values of the conductivity may be received at 602, wherein if a change is detected then this is indicative of a mains water flush event (for example flushing with mains water usually will result in a decrease in conductivity due to the aerated water). Other parameters can be measured to confirm this, for example the pH may be used at 604, wherein if the pH changes to an expected value at 605, this is also indicative of a mains water flush event. For example, the pH will increase if the water flush follows an acid clean, but will decrease if it follows normal operation in which alkaline inhibitors have been used. Furthermore, values of the flow rate may be used at 606, wherein if a rise in flow rate is also detected it indicates a mains water flush. By using multiple parameters the origin of the detected change 601 can be identified. In this case the diagnosis 608 may comprise that a mains water flush event has occurred. This process can be used to further verify that a particular maintenance event has occurred.

[0318] FIG. 7 is a graph showing an example of monitoring dissolved oxygen levels during maintenance events. The dissolved oxygen levels are shown in parts per million (PPM), while the measurements are shown over a period of 10 days. In other cases, the dissolved oxygen may be expressed in other units such as mg/L. The threshold upper limit for normal operation can be seen at approximately 0.5 PPM, an example threshold for this example. Above this value, a positive corrosion state occurs which may lead to corrosion due to high oxygen levels. Under normal circumstances the oxygen levels would ideally be kept below this. However, during a maintenance event the levels may far exceed this threshold. As such, a new maintenance threshold is required to ensure the smooth undertaking of each maintenance event. This can be seen at 10 PPM, where the maintenance threshold upper limit is shown. This is an example limit that may allow the system to monitor the success of maintenance events without unduly triggering a warning to a user that the normal system parameters have been exceeded.

[0319] Correspondingly, if the value exceeds this parameter, this is indicative of a positive corrosion state during a maintenance event, or some other malfunction. Since dissolved oxygen tends to saturate at the maximum possible value (e.g. around 10 PPM at 20 C. and 1 bar pressure), the maintenance mode threshold can be set around this level, so as to suppress alarm signals relating to positive corrosion states, when such saturation events occur. In the unlikely event that the dissolved oxygen nevertheless exceeds this value, an alarm mechanism exists to detect a malfunction (inlet water too aerated, wrong temperature, sensor malfunction, pressure too high, air ingress into the system, etc.). In some cases, the cumulative dissolved oxygen may be monitored, and a separate threshold will be provided for this. Therefore, during an event such as a water flush, if the cumulative dissolved oxygen exceeds a pre-set threshold, then an alert may be triggered. This may be useful to monitor the total amount of oxygen in the system, especially in situations where the instantaneous threshold will not be exceeded unless in a malfunction. For instance, if a big spike is detected, but this is brought under control in a certain length of time e.g. 1 day, then the cumulative dissolved oxygen may not exceed its threshold.

[0320] After dynamic flushing at day 0, a chemical cleaner is added which causes a dramatic drop in the dissolved oxygen down to 0 PPM. On day 1 a mains water flush occurs as the dissolved oxygen rises sharply up to approximately 9 PPM. After no more fresh water is being added to the system, the dissolved oxygen drops down to 0 PPM due to oxygen scavenging events taking place, but this occurs over a time interval of approximately 1 day. Also at the time when fresh water is no longer being added to the system, inhibitor is added, which may be detected by changes in conductivity and pH (seen in FIG. 8). After day 2, the oxygen level is back to approximately 0 PPM. Accordingly, the new maintenance threshold may comprise an upper limit of around 10 PPM, as shown in FIG. 7, meaning that if the dissolved oxygen increases up to 9 PPM as in FIG. 7, then the system is still behaving normally. However, if the level increases above 10 PPM, a positive corrosion state is triggered and the same procedure may occur as previously described when the system is in corrosion detection mode. When the parameter measurement reaches a certain value, it may be used to trigger an alert to perform the next event, or in some cases the next event may automatically occur. For example, if the dissolved oxygen level reaches 9 PPM, a message alert may appear containing information that the inhibitor should be added to passivate metal surfaces in light of the elevated dissolved oxygen levels, and/or scavengers may be added to reduce the oxygen levels in the system water. In another example, the inhibitor may be automatically added to the system when the dissolved oxygen level reaches a certain value e.g. 9 PPM. For this event, it may be appropriate to monitor other parameters such as conductivity and pH, and provide maintenance thresholds on these parameters in order to better provide an alert system.

[0321] In this example, the dissolved oxygen levels increase due to a mains water flush bringing aerated water into the system. The dissolved oxygen levels of the inlet water drawn into the system may be known, for example this may be tap water. As such the maintenance mode threshold range may be adjusted based on this. Therefore the threshold limit would not be exceeded for normal operation of this planned maintenance event.

[0322] FIG. 8 is another graph showing the pH levels over time for the same event sequence as FIG. 7. The pH drops when an acidic cleaner is added. The pH then increases and plateaus as the cleaner is consumed. During the mains water flush the pH rises sharply as residual acid from the cleaner is removed. When inhibitor is added, the rise begins to plateau around pH 9.5. Some example threshold limits have been indicated on FIG. 8. For example, the upper and lower threshold limits for normal operation are shown at pH 8.5 and 6.5 respectively. Under normal operation, the pH is desired to be within these limits to prevent corrosion. However, clearly in this instance, when a mains water flush occurs, the pH exceeds the normal upper threshold limit of pH 8.5. Accordingly, the limit is adjustable such that a maintenance upper limit can be provided to prevent an alarm being triggered when the pH exceeds the normal upper limit. For example, instead of an alarm being triggered, and the potential for automatic corrective action, the limit can be used as an indication that the event is occurring as planned. In addition, the maintenance threshold upper limit may be used to ensure that the pH does not deviate from that expected during the maintenance event. For example, this is shown as pH 10 in FIG. 8. Correspondingly, the lower limits are shown for normal and maintenance conditions, wherein these would work in a corresponding way as the upper limits.

[0323] These graphs in FIGS. 7 and 8 can be used to explain the processes of FIGS. 5 and 6, where events can be identified from the measurements of various parameters, and can be labelled accordingly.

[0324] FIG. 9 is a drawing showing an example embodiment of the apparatus for detecting corrosion in a closed water system disclosed. It comprises an inlet 901 through which water in a closed water system can enter a measuring unit 902. The measuring unit 902 allows the flow of water from the inlet 901 to the outlet 903. The measuring unit 902 comprises a plurality of sensors 904. In this embodiment, a first sensor 904a, a second sensor 904b, a third sensor 904c, and one additional sensor 904d are shown connected to the measuring unit 902. The number of sensors may be different to this, for example the apparatus may comprise more or fewer than shown in FIG. 9, which is only shown as an example embodiment of the apparatus. Each of the sensors is configured to determine values of the corresponding parameter selected from the plurality of parameters. For example, the first sensor is configured to determine values of a first parameter.

[0325] The measuring unit 902 is illustrative of a device to mount and position the sensors 904 in such a way that they intersect the water flowing from inlet 901 to outlet 903. It is also provided to ensure an air-tight seal such that the sensors 904 can monitor the parameters while preventing oxygen from entering the closed system, which allows representative measurements to be taken. In other examples, the measuring unit 902 may not be present, and the sensors may be directly connected to the water flow. A connection 905 is provided to allow transfer of data from the sensors 904 to a monitoring station 906. This connection may be an electrical connection, or may be fibre optic. The monitoring station 906 may comprise the processor to receive data from the sensors and interpret this data. For example, this processor may be an FPGA or may be a PC located locally or remotely. The monitoring station 906 may comprise the memory for storing a threshold range for each of the first, second, and third parameters, and in this embodiment the threshold range corresponding to the additional parameter as well. However, the memory may be located elsewhere, separate from the monitoring station and separate from the processor. It may further comprise a data logger or data recording means.

[0326] Alternatively, the processing may be performed elsewhere, and the data may be transferred from the monitoring station 906 to a processor in a remote location. For example, this may occur via an internet connection. In some cases it may be uploaded to the cloud. In the embodiment in FIG. 9, the monitoring station 906 comprises a display screen 907. In some embodiments, a display screen is not located adjacent to the sensors, and instead the data is transferred for example to a remote screen or accessible via a web interface, for user convenience. In some embodiments, the monitoring station may be connected to a wireless network, to which it may upload sensor data, for example to a web interface. The display screen 907 may display the sensor readings from sensors 904 in real-time or it may display information relating to how one or more sensor measurements change over time. This screen may be interactive in order for a user to obtain details on measurements from each sensor.

[0327] More additional sensors may be present in the apparatus and the processor may be configured to receive data from each of these additional sensors. A sensor (not shown in FIG. 9) may be included between the inlet 901 and the outlet 903, for example the one described in more detail below.

[0328] Although not shown in FIG. 9, the apparatus may comprise a user input for manually selecting the parameters from the plurality of parameters. The user input may also adjust the apparatus configuration from corrosion detection mode to maintenance mode and vice versa. For example this may comprise a switch or interface such as a keyboard or touchpad.

[0329] Although not shown in FIG. 9, the apparatus may comprise means for adjusting a parameter such as control of a pressurisation unit, control of water flow rate, control of make-up water flow rate, control of automatic air vents or pressure relief valves, addition of corrosion inhibitor, addition of an anti-biofilm agent, a heating and/or cooling unit, and/or pH control. These elements may be present within measuring unit 902, or may exist at specific points along the water flow path, depending on the requirements of the means to adjust each parameter.

[0330] The measuring unit 902 allows the sensors 904 to be immersed in the water flowing from inlet 901 to outlet 903 whilst not interrupting the flow. In other words, the flow is continuous and measurement of the sensors 904 does not require a disruption to the flow and samples taken of the water. This ensures the water being measured has not become aerated during the process, unlike many previous systems. In some embodiments, the sensors 904 may be distributed around the system, or a plurality of monitoring stations 906 may be present at different locations around the system. As described above the sensors 904 may comprise means for measuring one or more parameters such as pressure; make-up water flow rate; dissolved oxygen; cumulative dissolved oxygen; inhibitor dosing levels; biofilm accumulation; temperature; conductivity; galvanic current; cumulative galvanic current; crevice corrosion rate; and/or pH. Although four sensors are shown in FIG. 9, the system may comprise more than four, or fewer than four. In some examples, fewer than all of the sensors may be actively monitoring and outputting data concurrently. In some cases, as few as two sensors may be present or actively monitoring and outputting data. In other cases, all the sensors present are actively monitoring data at the same time.

[0331] Some sensors may not require immersion in water, such as temperature. The measuring unit 902 provides a means for positioning each sensor in an appropriate location to perform the determining of the corresponding parameter. For example, in some cases the thermal conductivity of pipes (particularly copper pipes) may be generally high enough such that the measurement of the temperature of the exterior of the pipe is a good approximation to the temperature of the water within the pipe.

[0332] FIG. 10 shows an example embodiment of the sensor described above. Inlet 1001 is for receiving water from a closed water system, for example a heating, ventilation and air conditioning systems, allowing the water to pass through the sensor to outlet 1002, where the water is returned to the closed water system. The pipes connecting the sensor to the closed water system, such as at the inlet 1001 and outlet 1002, may be for example made of copper or other metals typically used in such systems. The sensor comprises a sensing chamber 1003, located between the inlet 1001 and the outlet 1002, such that the water passes through the sensing chamber. The sensing chamber comprises an outer chamber wall for retaining water in the sensing chamber. The sensing chamber also comprises a first measurement surface 1004 formed from a first metal. For example the first metal may be copper. The sensor also comprises a second measurement surface 1005 mounted at least partly within the sensing chamber, and formed from a second metal, the second metal being different from the first metal. For example, the second metal may be steel. In other examples the first and/or second metals are selected from: brass, steel, copper and alloys thereof or other metals or alloys typically used in such systems, wherein the first and second metals are different. In the example shown in FIG. 10, the first sensor 1004 is the outer chamber wall. The second measurement surface 1005 is typically chosen to be the anode by selecting a metal lower (i.e. less noble) in the galvanic series than the first measurement surface 1004. This prolongs the life of the sensor, since it will be the less noble metal which corrodes, and consequently (in the design shown) the first measurement surface 1004, which doubles as the outer wall) will not be corroded, thereby reducing the likelihood of leaks. The second surface can be replaced if needed, either to change the metal, to provide information on the corrosion of a different metal/alloy, or to replace a corroded inner measurement surface. In any case the metal/alloy which will be the anode is advantageously made from a metal which is representative of metals in the closed water system to which the sensor is connected. In some cases, a variety of versions of the sensor having anodes made from different metals may be provided for use in different systems having predominantly those metals exposed to water.

[0333] The sensor is configured such that water flowing in through the inlet 1001 flows through region 1006 between the inner surface of the first measurement surface 1004 and the outer surface of the second measurement surface 1005. For example, the first measurement surface 1004 may be a pipe of circular cross-section, and the second measurement surface 1005 may be a pipe or rod of circular cross-section, with a diameter smaller than the first element, such that the second measurement surface 1005 is disposed within the first measurement surface 1004. In this example, the water flows through annular region 1006 between the two surfaces.

[0334] The first and second measurement surfaces include electrical connection points for connecting to current sensing means 1008 (also known as current measuring device). The presence of the current measuring device 1008 means that a preferential flow path between the first 1004 and second 1006 measurement surfaces exists. This causes charge build-up on the measurement surfaces to flow though the current measuring device. The resulting current is measured, and an indication of the amount of corrosion occurring with respect to time is provided by the current measuring device 1008. The current measuring device 1008 in some cases has a local memory for storing the measured current as a function of time. In other cases, the current v time information is communicated elsewhere, for example as part of the overall system health monitoring described above. FIG. 10 shows the first and second measurement surfaces including electrical connection points 1007a and 1007b, respectively, for connecting to a current sensing means 1008 via an electrical connection 1009. In some cases, the electrical connection 1007 points may allow the current sensing means 1008 to be reversibly connected, e.g. for replacement. In other examples, the current sensing means 1008 may be permanently connected (e.g. hard-wired) to the sensor. For example, the current sensing means 1008 may be an ammeter, perhaps especially configured to measure small currents such as on the milliamp scale. In some examples, the current sensing means is configured to integrate the current over time.

[0335] As water flows over the two surfaces, galvanic corrosion of the anode occurs. This causes charge to build up on the surfaces. Since the surfaces are connected by an ammeter, the charge flows along this current path and the ammeter registers a signal, proportional to the rate of corrosion. The total amount of corrosion which has occurred can be derived from this by integration, as described above.

[0336] In some examples, the sensor is configured to send the output from the current sensing means to a processor. This processor may be the same processor of the apparatus for monitoring a plurality of parameters for detecting corrosion described above. In this case, the sensor may be one of the sensors of the apparatus.

[0337] In some examples, there are two measurement surfaces within the sensing chamber 1003, and the outer wall of the sensing chamber 1003 is not used as a measurement surface. In other respects such a sensor operates in much the same way as that described above. Although the measurement surfaces 1004, 1006 are shown as smooth, flat surfaces, corrugations, ridges, or other complex shapes may be used to increase the surface area.

[0338] The current sensing means 1008 can be configured to relate the current it is measuring to the degree of effectiveness of the inhibitors in the water system, as described in detail above. Additionally or alternatively, the current can be related to a rate of loss of thickness of exposed metal surfaces in the system, or the current can be integrated with respect to time and the result used to give a measure of the total loss of metal thickness in the system during the period over which the integration occurred. This can be done either as an integrated system, or by passing the current data to an external processor which performs the steps of integration and/or relating the current to a degree of effectiveness of the inhibitor or to the rate of loss or total loss of metal in the system. In some cases, the current sensing means 1008 may be configured to send alerts to a user that the measurement indicates a positive corrosion state (non-zero current), and suggest a corrective action such as addition of inhibitor. The sensor may even be configured to take corrective action automatically and/or to monitor the current while corrective action is being taken to iteratively arrive at the correct concentration of inhibitor to fully passivate exposed metal surfaces in the system.

[0339] Turning now to FIG. 11, an illustration of the effect of inhibitors in a closed water system is shown. More specifically, the graph 1100 demonstrates how the galvanic current in a system changes as inhibitor dosing levels change (as determined by measurement of conductivity), in this case at a fixed temperature of 60 C. This figure shows the effects in an oxygenated water system. As discussed in detail above, the galvanic current is a direct representation of the amount of corrosion occurring in the system. An arrow 1102 points towards low inhibitor concentrations. The conductivity of pure tap water (i.e. with no inhibitor added) is around 300 S/cm in this example. However, the conductivity of mains water varies throughout the UK and is lower in soft water regions, and higher in hard water regions. As inhibitor is added to the water, the conductivity increases. At the full recommended dose of inhibitor, the conductivity reaches approximately 850 S/cm.

[0340] It can be seen from this arrow 1102 that the conductivity and the concentration of inhibitor are correlated with one another. Moreover, the correlation is positive, in that low concentrations correspond to low conductivities and vice-versa. It is clear that if the correlation between the inhibitor concentration and the conductivity is known in advance, then the concentration can be directly related to the measured conductivity value to a reasonable degree of accuracy. This provides a convenient way of determining the concentration. Indeed, as is clear, seven data points have been measured, corresponding to seven inhibitor concentrations and their corresponding conductivity. These can be stored together, for example in a tabular format. The correspondence between these values can be used, e.g. to change the scale on the x-axis from conductivity to inhibitor concentration (e.g. in ppm, percentage [by volume or weight], percentage of recommended dose, etc.) for ease of reference by a user.

[0341] As noted above, the graph 1100 shows data at a constant temperature of 60 C. In some cases, each data point may be stored with a corresponding temperature at which that data point was measured. By storing temperature data in this way, the determination of concentration from conductivity can be improved, since the effect of temperature can be removed. In some cases, two or three of the parameters of conductivity, inhibitor concentration and temperature can be fit to a generalised equation with variable coefficients. The set of coefficients which gives the best fit to the data can be stored and used to determine correlations between conductivity and concentration where no data points exist. In some cases, the stored data can be boosted with further calibration measurements, further refining the accuracy of the fit. A different correlation may be determined for different inhibitor types, as set out above.

[0342] As shown in the graph 1100, the galvanic current is high at low inhibitor concentrations and vice-versasee the stepped plot line 1104. This relationship is to be expected since the purpose of the inhibitor is to prevent or reduce corrosion. Therefore when there is no inhibitor, corrosion continues largely unchecked, and the corresponding galvanic current is high. As inhibitor approaches its recommended value, the current is reduced due to the effectiveness of the inhibitor. The stepped nature of the graph 1100 naturally leads to considering there to be two different regimes for the inhibitor. A first regime 1106 in which galvanic currents are high because the exposed surfaces of the metal in the system are not passivated, or at least the passivation is not completely effective. This spans approximately the range of conductivities between approximately 300 S/cm and 650 S/cm in this example.

[0343] The second regime 1108 ranges from approximately 650 S/cm to 850 S/cm. Here, the galvanic current plateaus at a low value and does not change much with further inhibitor being added to the system. This is a sign that the exposed metal surfaces of the system have been passivated effectively. Clearly the system should be operating in the second regime 1108 and not the first regime 1106, in order to minimise corrosion when dissolved oxygen is present.

[0344] Overall, the graph 1100 in FIG. 11 illustrates that the inhibitor concentration and the conductivity are inherently linked; that the temperature is an important effect to account for in the conversion between these two parameters; and that the effectiveness of the inhibitor in the system can be directly seen from the dramatic effect it has on galvanic currents.

[0345] As noted above, the correlation between the conductivity and the inhibitor concentration is important to determine accurately, so that the concentration can be determined simply and accurately. In FIG. 12, a flowchart 1200 of a method for determining the correlation and/or using a correlation so determined to convert between these two parameters. The method starts at a first step 1202, in which a correlation between conductivity and inhibitor concentration is stored. This may be by virtue of a system being provided with this information pre-installed, or there may be various calibration steps, e.g. adding inhibitor at a known concentration to the water in the system and measuring the conductivity. As noted above, this can include storing data in a tabular format or by reference to an equation with variable coefficients (in which the best fit coefficients are stored). Any of these calibration steps may be performed as often as necessary to ensure that adequate accuracy can be achieved. Additional readings taken at any time while the device is operating (and for any reason, e.g. calibration, or normal operation) can be added to the stored correlations, so that they may be retrieved in future. In some cases, more recent data overwrites older data.

[0346] Next, at step 1204, a value for the conductivity is determined. This is typically performed by a suitable sensor which forms part of the overall system. These measurements may be made continuously or periodically, for example. A central control and processing unit may be configured to control such a sensor, and request that a reading be taken at times when this is needed, for example.

[0347] At step 1206, the temperature of the water is determined, so that temperature effects can be accounted for. In some cases, the temperature may be determined before the conductivity in order for the conductivity measurement to be actively compensated. In other cases, the order is not important and the compensation is done at a later stage when the conductivity is related to the correlation. In part, the measurement of temperature may include using a heater or cooling device to hold the temperature constant (e.g. at a value for which data exists). Alternatively, different correlations may be stored which correspond to different temperatures. In some cases, the temperature measurements may be used to actively compensate for the temperature effects in the conductivity sensor. The correct correlation for the current temperature can be used, or where this is not available, the closest may be used (possibly including an extrapolation or interpolation).

[0348] At step 1208, the determined conductivity value is compared to the appropriate stored correlation. Next, at step 1210, the comparison of the conductivity to the inhibitor takes account of the temperature measurement. For example, the conductivity measurement may be compensated based on the temperature. Finally, in step 1212, the corresponding inhibitor concentration is provided based on the comparison. This provides a simple way to determine the inhibitor concentration (which is difficult to determine accurately) by use of a proxy measurement.

[0349] The strict order of the steps in FIG. 12 need not be adhered to. For example, the temperature may be measured prior to the conductivity being measured. In some cases, the conductivity sensor has a temperature-compensated output.

[0350] As will be clear, the specific sensing systems for the inhibitor concentration and galvanic currents provide improved sensing of some of the parameters which the general system uses to determine system health.

[0351] FIG. 13 shows an example embodiment of an optical sensor 2100 for detecting corrosion in a water system. The optical sensor 2100 comprises a sensor housing 2102, which, in use, is partially positioned within a water system. As shown, the sensor 2100 is mounted to a pipe 2104 of the system, which is filled with system water 2106. The sensor includes a metal sample 2112 (for example a thin film or foil of metal) in contact with the system water 2106. In particular, a first surface of the metal sample 2112 is in contact with the water 2106 of the water system. The contact between the first surface of the metal sample 2112 and the system water 2106 leads to corrosion of the metal sample 2112. As the metal sample 2112 corrodes, corrosion debris and/or tarnishing will appear on a second surface opposite surface to the first surface (and separated from the first surface by the thickness of the metal sample 2112), which will decrease the reflectivity of the surface. In some cases, the metal sample 2112 will corrode away entirely in parts, for example due to pinhole corrosion progressing through the entire thickness of the sample 2112, or due to uniform corrosion of such severity that the parts of the metal sample 2112 are entirely corroded away.

[0352] In some examples, and as shown in FIG. 13, the second planar surface of the metal sample 2112 is not arranged in contact with the system water 2106. In some examples, to ensure that pinhole corrosion occurs through the thickness of the metal sample 2112 from the first planar surface in contact with the system water 2106 towards the second planar surface, and not in the other direction, water is prevented from contacting the second planar surface.

[0353] The optical sensor 2100 is configured to illuminate the metal sample 2112 with a beam of light 2110 emitted from a light source 2108. The optical sensor 2100 is also provided with an optical sensor 2116 to receive the reflected light 2114 and determine the intensity of that reflected light beam 2114. The reflectivity of the metal sample 2112 can be determined from the intensity of the reflected light beam 2114 received, which is related to the amount of corrosion debris on the second surface of the metal sample 2112, tarnishing due to contact with the system water 2106 and outright missing metal. Consequently, the change in reflected light intensity is related to the amount (and in some cases, the type) of corrosion of the metal sample 2112. Note that the optical sensor 2116 can detect changes in reflected light in this way whether or not the reflection is diffuse or specular. In either case, the amount of reflected light changes and a change in received light intensity can be detected.

[0354] A common cause of corrosion in water system is dissolved oxygen within the system water 2106. This causes corrosion of metals within the water system, as described above. The metal sample 2112 is used as a sacrificial sample to monitor and/or detect corrosion. If the water 2106 in the water system 2102 is corrosive, for example it contains high levels of dissolved oxygen or other corrosive components (acids, bases, organic chemicals, etc.), then corrosion of the metal sample 2112 can occur. This may take the form of pinhole corrosion, also known as pitting corrosion, where pinhole-sized holes form in a metal. Due to surface irregularities and imperfections, a small pit forms in the first surface in contact with the water due to pitting corrosion. Pinhole corrosion is particularly troubling because it results in very little loss of metal, so is hard to assess severity using weight loss or cumulative galvanic current studies, yet it causes damage to the deep structure of metal, and can corrode holes entirely through metal (e.g. through pipe walls, causing leaking). Worse still, pinhole corrosion is often obscured by corrosion debris such as tubercles or tarnishing, meaning that it is often not clear whether a tarnished portion represents mere surface corrosion or if one or more pinholes have caused much deeper problems.

[0355] The sensor 2100 has an O-ring seal 2118 to prevent system water 2106 from entering the interior of the housing 2102. This seal also helps to ensure that the system water 2106 cannot contact the second surface of the metal sample 2112 by leaking around the outer edge of the metal sample 2112. The O-ring seal can be formed from any suitable material for making a watertight seal which can withstand the conditions found in the water system. For example, temperatures up to around 100 C., high pressures, and exposure times of 5 to 10 years. In addition, a transparent plate 2120 is provided on the second surface of the metal sample 2112, thereby ensuring that the system water 2106 cannot contact the second surface, other than by corroding through the metal sample 2112. The transparent plate 2120 can be formed from plastics, glasses, etc. and be suitable for withstanding the temperatures and pressures set out above for the timescales of intended use. Since the metal sample is relatively thin (often thinner than the thickness of pipes, etc. in order to provide advanced warning of a corrosion problem), it may not be sufficiently strong to withstand the pressures in the water system. The transparent plate 2120 can provide structural support to the metal sample 2112. The transparent plate 2120 does not affect the operation of the system, since (being transparent) light emitted 2110 from the light source 2108 travels through the transparent plate 2120, reflects from the metal sample 2112, travels back through the transparent plate 2120 and is received by the light sensor 2116. In any case, the sensor can be calibrated to ensure the range of output is indicative of 0% to 100% loss of material. In some cases, there may be no need for a transparent element 2120 in this location, for example, where the metal sample 2112 is strong enough to resist the system pressures, and where there is nothing on the interior of the housing 2102 which would be adversely affected by system water 2106 leaking into the housing 2102, as would be the case if the metal sample 2112 is penetrated by corrosion and there is no transparent element 2120.

[0356] Since only the first surface of the metal sample 2112 is exposed to the system water 2106, the only way for the second surface to corrode (i.e. for tarnishing or loss of metal to show on the second surface) is for the corrosion to travel through the whole thickness of metal. As shown, the metal sample 2112 has a uniform thickness, which means that detection of a pinhole corroding through the sample 2112 is not dependent on the location of the corrosion. In other words, pitting corrosion occurring anywhere on the metal sample 2112 must travel through the same thickness of metal in order to form a pinhole, and has the same effect on the reflected light beam 2114.

[0357] As the metal sample 2112 is chosen to be thin, pitting can form a small pinhole through the metal sample 2112. Corrosion debris spreads from pinholes on the second surface of the metal sample 2112. The corrosion debris decreases the reflectivity of the surface, which provides a detectable decrease in intensity of reflected light 2114.

[0358] The metal sample 2112 is chosen to be a metal typically found elsewhere in the water system 2102. Typically, the metal sample 2112 may be formed from carbon steel, copper, stainless steel, brass, or aluminium. By choosing a metal that is present in the water system, determining the amount of corrosion or rate of corrosion of that metal within the optical sensor 2100 provides an indication of the corrosion of that metal elsewhere in the system.

[0359] The metal sample 2112 is chosen to be a thin film such that corrosion occurs quicker than at sections of metal in the rest of the system 2102. This allows a fast diagnosis of corrosion without waiting for corrosion of parts of the actual system to progress to unsafe levels. This can prevent significant damage to the system, savings costs by identifying corrosion without waiting for catastrophic failure.

[0360] The metal sample 2112 is sealed by seals 2118 such that the first surface of the metal sample 2112 is exposed to the water while the second surface is not. The seals 2118 prevent water from leaking around the metal sample 2112 while it is held in position. For example, the seals 2118 may be in the form of rubber seals, or a glue/resin between the metal sample 2112 and the walls of the sensor housing 2102. The seals 2118 prevent the water from being in contact with the electronics and sensing apparatus, which will now be described.

[0361] The optical sensor 2100 comprises a light source, for example the light source 2108 in FIG. 13 may be a light-emitting diode (LED) 2108. The LED 2108 is configured to emit light onto the metal sample 2112. In particular, the LED 2108 is configured to emit light onto the second surface of the metal sample 2112. The second surface is the opposite surface to the surface in contact with the water. The emitted light is shown in FIG. 13 by arrows 2110. The metal sample 2112 is reflective, and therefore reflects the light from the LED 2108. The reflected light is shown by arrows 2114.

[0362] In other examples, the light source 2108 may be a laser, or a fluorescent light bulb or any other suitable light source.

[0363] The optical sensor 2100 also comprises a light sensor 2116, for example, the light sensor 2116 shown in FIG. 13 may be a photodetector. The photodetector 2116 is configured to detect the light emitted from the LED 2108. In particular, the photodetector 2116 is configured to detect the reflected light 2114 from the second surface of the metal sample 2112.

[0364] The photodetector 2116 is positioned adjacent to the LED 2108 such that light from the LED 2108 directed towards the metal sample 2112 (light beam 2110), which is then reflected (light beam 2114) by the second surface of the metal sample 2112, and is received by the photodetector 2116. The photodetector 2116 and the LED 2108 are on the same side of the metal sample 2112, facing the second surface of the metal sample 2112. This allows the light source and the light sensor to be isolated from the water within the water system 2102, for example by the housing 2102 and the seals 2118.

[0365] In some examples, the photodetector 2116 is configured to not detect light emitted directly from the LED 2108 that has not been reflected by the metal sample 2112, that is no direct light path between the LED 2108 and the photodetector 2116 exists. For example, an opaque barrier may be placed between the LED 2108 and the photodetector 2116 to prevent stray light affecting the readings, as set out in more detail with regard to FIG. 17. In other examples, the output of the photodetector 2116 photodiode is calibrated to take into account the inner optical properties of the cavity including the reflectivity of the metal sample 2112. In other words the photodetector 2116 may be calibrated to remove the effects of the direct LED light on the readings, and only detect changes due to changing reflectivity of the metal sample 2112. In yet further examples, the light source 2108 and light sensor 2116 may have relatively narrow angular ranges of emission and/or detection respectively, so that there is no direct transfer of light from the light source 2108 to the light sensor 2116.

[0366] The photodetector 2116 is configured to transmit a signal corresponding to the detected light reflected by the metal sample 2112. For example, the photodetector 2116 outputs a signal which is related to the intensity of the detected light, in the form of an analogue signal. For example a current or voltage signal with its amplitude related to the received intensity. In other examples the sensor 2100 may output a digital signal in a standardised format, such as Modbus.

[0367] The sensor 2100 may further comprise a processor (not shown), configured to receive the signal from the light sensor 2116 corresponding to the detected light reflected by the metal sample 2112. For example, the light sensor 2116 is configured to measure the intensity of the received light, and transmit the signal corresponding to the intensity towards the processor (which may be remote, for example for collating many measurements from a variety of sensors). The signal is transmitted to the processor via an electrical, wireless, fibre-optic, etc. connection.

[0368] The processor may comprise a data acquisition system for storing the received data from the light sensor 2116. For example, the data acquisition system may be a data logger or other memory controllable by the processor.

[0369] The processor may be configured to process the data received from the light sensor 2116. For example, the processor receives periodic measurements of the intensity from the light sensor 2116. The processor is configured to convert the measurement signal from the light sensor 2116 into intensity of light. The processor is configured to store the periodic intensity values in a table, and output to a user. For example, the data may be outputted to a display device for user interaction by displaying the table of values, or displaying a graph of the intensity values over time.

[0370] The processor is configured to detect changes in the values of intensity. For example, if the intensity of light decreases substantially over time, this may correspond to a decrease in the reflectivity of the metal sample 2112. This decrease in reflectivity is caused by corrosion of holes through the metal sample 2112 and/or subsequent spread of corrosion debris over the surface reducing the intensity of light reflected by the surface. This data may be extrapolated to estimate the time remaining until one or more pipes in the system corrode through and start to leak.

[0371] Upon detecting a decrease in intensity, the processor may be configured to trigger an alert. A threshold may be present to indicate corrosion of the metal sample 2112. For example, a decrease from an initial high value of intensity by a certain amount, below a threshold, may trigger an alert. In another example, the threshold may be related to the gradient of the decreasing intensity, or a percentage decrease from a highest value of intensity. If a threshold is exceeded, an alert such as a message to a user may be triggered.

[0372] The processor may be configured to control the light source 2108 and/or the light sensor 2116. For example, the intensity or spectral range of light emitted by the light source 2108 or the spectral range at which the light sensor 2116 is most sensitive may be controlled electronically to gain more information about the type and severity of corrosion in the system.

[0373] Consider now FIG. 14, which shows another example embodiment of an optical sensor, 2200 having a plurality of metal samples 2112. The sensor 2200 is similar to that shown in FIG. 13, but has three metal samples 2112a, 2112b, 2112c. Each of these is formed from the same metal in FIG. 14, and has a different (uniform) thickness. Each sample has its own respective light source 2108a, 2108b, 2108c and light sensor 2116a, 2116b, 2116c for detecting corrosion in the manner set out above. Since each metal sample 2112 has a different thickness, knowledge of which sample has corroded provides information on the depth of metal which has corroded in the system and also the depth of metal which has not corroded. For example, if the thinnest metal sample 2112a is 0.025 mm thick, with the next sample 2112b being 0.05 mm thick and the thickest sample 2112c being 0.075 mm thick, then if the first sample 2112a shows tarnishing or other signs of corrosion, while the other two 2112b, 2112c do not, then it can be inferred that the corrosion depth is between 0.025 mm and 0.05 mm. This determination can be performed locally using a processor (not shown), or the readings can be transmitted to a remote location for processing.

[0374] In some cases, instead of metals of different thicknesses, the multiple samples 2112 may be formed from different metals and/or alloys. This can provide a measure of how different metals present in the system are faring under exposure to system water 2106. Of course, in some cases, there may be a combination of different metals and different thicknesses to provide an estimate of the maximum pitting corrosion depth in several metals at once.

[0375] While each metal sample 2112 is shown with a corresponding light source 2108 and light sensor 2116, in some cases a single light source 2108 may be configured to shine light onto a plurality of metal samples 2112, and/or a single light sensor 2116 may be configured to receive light from a plurality of metal samples 2112. Thus, associating many samples 2112 with a single housing 2102 as shown in FIG. 14 may allow a reduction in the amount of electronics in the system, thereby saving costs.

[0376] The plurality of metal samples 2112 may be formed from the same metal as one another, wherein each of the plurality of metal samples 2112 has a different thickness to the thickness of the other metal samples 2112. This can allow corrosion of different types of metal to be monitored by providing samples 2112 of different metals. In other cases, a series of samples of different thicknesses may be provided. This can provide a more compact sensor than providing a separate sensor for each thickness of metal and/or each different metal type. Of course, an alternative arrangement is to provide a separate sensor 2100 such as that in FIG. 13 for each metal and/or thickness. Where a single sensor 2200 has a plurality of metal samples 2112, the sensor 2200 may share various measurement components. For example, a single light source 2108 (or in general fewer light sources 2108 than there are samples) can be arranged to shine onto the second surface of multiple metal samples 2112, thereby ensuring that each sample 2112 is illuminated with the same light, so allowing a meaningful comparison of the received light 2114 intensity, as the light received 2110 by the second surface of each sample 2112 can be controlled to be the same in each case. In other cases, there may be a single light sensor 2116 and multiple light sources 2108. For example, each light source 2108 may emit light 2110 in a relatively narrow wavelength range and the sensor 2116 may be configured to detect the received intensity broken down by received wavelength, thereby allowing the sensor 2116 to correlate intensity in a wavelength with a particular sample (wavelength division multiplexing), and thus monitor several samples 2112 using only one sensor 2116, or more generally using fewer sensors 2116 than there are samples. In yet another case, other aspects such as processors for interpreting the received intensity data or housings 2102 for enclosing the sensors 2116 may be shared between multiple metal samples 2112. An alternative method of sharing light sources 2108 and light sensors 2116 is to use a highly directed light source 2108 and/or light sensor 2116 and illuminate the second planar surface of different samples 2112 at different times (time division multiplexing), and synchronise the sensor 2116 with the light source 2108, so that light received 2114 at a given time is correlated with a given metal sample 2112. In yet further examples, baffles such as that shown in FIG. 17 (element 2134) may be used to prevent direct transmission from a light source 2108 to a light sensor 2116, and also to prevent transmission of light from one sensing unit (light source 2108, metal sample 2112 and light sensor 2116) to another sensing unit, i.e. to prevent cross talk between measurements on each sample 2112.

[0377] Turning now to FIG. 15, a further example sensor 2300 is shown, which has a bypass arrangement. The sensor 2300 operates in a broadly identical manner to that shown in FIG. 13. In this case, however, the sensor 2300 is mounted on a section of pipe 2104 having two flow paths in parallel. The main pipe 2104 branches into a sensor section 2104a having the sensor 2300 mounted to it, and a bypass section 2104b, having no sensor. System water 2106 is able to flow through the main pipe, and then either through the sensor section 2104a or the bypass section 2104b. The water 2106 is controlled by two valves, each having a first position 2122 where they block water from flowing into the bypass section 2104a and a second position 2124 where they block water from flowing into the sensor section 2104b. Thus, where water is intended to flow through the sensor section 2104a, the valves are each set to their first positions 2122, and water is blocked from flowing into the bypass section 2104b and instead flows past the sensor 2300 via the main section 2104a of pipe. Similarly, the water 2106 can be arranged to miss out the sensor section 2104a of pipe by diverting water 2106 through the bypass section 2104b or pipe, which is achieved by setting the valves to their second positions 2124. In some cases, the two valves can be connected such that triggering one valve to change between its first 2122 and second 2124 positions causes the other one to change. That is, rather than having each valve being independently controllable (leading to four distinct configurations, two of which are effectively to block flow through the element entirely), there are simply two configurations: bypass and sensor.

[0378] After traversing the sensor 2104a or bypass 2104b section of pipe the system water 2106 re-joins the main pipe 2104. The sensor may be supplied with such a pipe section, in which case the main pipe 2104 may be provided with solderable joints, screw threads, etc. for connecting into a main water system.

[0379] In any case, the purpose of this arrangement is to provide a means for installing the sensor and for replacing the metal sample 2112. For example such a pipe arrangement may be fitted to the water system, and the valves set to their second positions 2124 so that system water 2106 flows through the bypass section 2104b, and not through the sensor section 2104a. In some cases, no sensor 2300 need be fit to the sensor section during installation. For example, this pipe arrangement may be installed with a view to fitting a sensor at a later date. Since the system water flows through the bypass section 2104b there is no danger that the system water will leak out of a hole in the sensor section 2104a of the pipe, since there is no water flow through that section. In other cases, the sensor section 2104a may be provided with a cover for plugging the hole where the sensor 2300 is intended to be installed.

[0380] In any case, when a sensor is to be fit to the system, the valves are set to their second positions 2124 to isolate the sensor mounting hole. It may be necessary at this stage to drain the sensor section 2104a of any system water 2106 in the sensor section. In other examples, it may be desirable not to drain the sensor section 2104a, e.g. to avoid introducing air into the system, which can reduce pumping efficiency and increase the rate of corrosion. In any case, once the sensor section 2104a no longer contains water under pressure, any cap can be removed from the hole for mounting the sensor 2300 and the sensor 2300 slotted into place. The sensor can be sealed in place using any suitable means such as using O-rings and clamps, screw threads, etc. In some cases, solder or welding may be used, but this can make removal of the sensor 2300 e.g. for replacement of the metal sample 2112, difficult.

[0381] Once the sensor 2300 has been mounted to the sensor section 2104a, the system water can be directed back though the sensor section 2104a by setting both valves to their first positions 2122. This causes system water 2106 to flow past the metal sample 2112 and thereby provides an in-situ measurement of corrosion in the water system.

[0382] In case the metal sample 2112 needs replacing, for example if it corrodes entirely away, or simply to the point where it is difficult to gain any further information on the corrosion state of the sample 2112, a corresponding process can be followed: [0383] 1. Set both valves to their second positions 2124 to direct system water 2106 down the bypass pipe 2104b. [0384] 2. Remove the sensor 2300 from the sensor pipe 2104b. [0385] 3. Replace the metal sample 2112 (and/or clean the sensor 2300, perform maintenance, etc.) [0386] 4. Replace the sensor 2300 in the sensor section 2104a and seal the sensor 2300 in place. [0387] 5. Set both valves to their first positions 2122 to direct system water 2106 down the sensor pipe section 2104a.

[0388] The sample 2112 may be held to the sensor 2300 using a screw thread, which can help to press the seal 2118 to form a watertight fit.

[0389] Consider now FIG. 16. This shows a sensor arrangement 2400 similar to that shown in FIG. 13. Here, however, the housing 2102 does not contain the light source 2108 or the light sensor 2116. Instead, the light source 2108 and the light sensor 2116 are housed in a second housing 2126, separated from the housing 2102. A fibre optic cable 2128 transmits the light from the light source 2108 to the interior of the housing and, once the light has reflected from the metal sample 2112, another fibre optic cable transmits the light back to the light sensor 2116. This arrangement means that even in the event that system water leaks into the housing 2102, the electronic parts of the sensor 2400 are protected from damage due to system water 2106.

[0390] This design is sealed from the system water 2106 by double O-ring seals 2118. One seal is located in contact with the metal sample 2112 and the system water 2106. The second seal is located behind the transparent plate 2120. This means that even in the case where the system water has corroded through the entire thickness of the metal sample 2112, system water 2106 is prevented from entering the housing 2102 by the second seal 2118 forming a seal with the transparent plate 2120 and the housing 2102. It should be borne in mind that the Figures are only schematic. For example, in order to provide uniform light intensity on the second surface of the metal sample 2112, it may be preferable for the fibre optic cable 2128 which transmits light from the light source 2108 to the interior of the housing 2102 to occupy substantially all of the upper surface of the housing 2102, and the other fibre optic cable 2128 to take up less of the upper surface of the housing 2102, or to exit the housing 2102 from a different surface. In other words, the fibre optic cables 2128 may be differently sized relative to each other and the housings 2102, 2126 than they appear in FIG. 16, which is not to scale.

[0391] In FIG. 17, yet another sensor apparatus 2500 is shown, in this case using a lens 2130 to focus light on the second surface of the metal sample 2112. Similarly to the sensing apparatuses described above, the sensor 2500 in FIG. 17 has a housing 2102 containing a light source 2108 and a light sensor 2116. The light 2110 emitted from the light source 2108 is directed not directly towards the metal sample 2112, but instead to a lens 2130. The lens 2130 directs the emitted light 2110 towards the second planar surface of the metal sample 2112. The lens can be configured to redirect the emitted light 2110 in such a way as to provide a homogenous light intensity over the entire second surface of the metal sample 2112. In other words, the lens can be used to smooth out any inhomogeneities in the emitted light 2110 to ensure that the sensor detects approximately the same loss of reflected light intensity irrespective of the location at which corrosion occurs on the metal sample 2112.

[0392] The light output 2132 from lens 2130 is directed to the metal sample 2112, whereupon it is reflected 2132 and enters the lens 2130 again (in some cases, there may be a second lens for receiving the reflected light). The lens 2130 directs the light towards the light sensor 2116. This can help to ensure that as much of the reflected light 2114 as possible is directed to the light sensor 2116, thereby improving the sensitivity of the sensor 2500 to small changes in reflectivity. In some cases, the lens 2130 may be replaced with one or more prisms, fibre optic arrangements, mirrors, etc. or a combination of these to achieve the same effect.

[0393] Additionally, the sensing apparatus 2500 has a baffle 2134 to block direct transmission of light 2110 emitted from the light source 2108 to the light sensor 2116. This reduces the occurrence of false negatives, as it prevents situations in which the reflected light 2114 is reduced due to corrosion, but this is not detected by the light sensor 2116 because the signal is swamped by directly transmitted light.

[0394] In some cases, it may be advantageous to combine the features shown in FIGS. 16 and 17, for example to provide a sensor with both a lens 2130 and fibre optic cables 2128 running to a second housing 2126. As noted above the fibre optic cable 2128 transmitting light to the housing 2102 should be relatively large to ensure that the light received by second surface of the metal sample 2112, is broadly uniform. An alternative way to achieve this is to mount a lens 2130 inside the housing 2102, as shown in FIG. 17, to provide a more even distribution of light emitted by the fibre optic cable 2128 onto the second surface of the metal sample 2112.

[0395] In FIG. 18, a further example of a sensing apparatus 2600 is shown. This example has a light source 2108 which is housed in a mounting 2140 attached to the internal walls of the housing 2102. The mounting 2140 has a recess housing the light source 2108, so that the mounting 2140 operates a little like the baffle 2134 shown in FIG. 17.

[0396] The light source 2108 in FIG. 18 is arranged to emit light into a fibre-optic bundle 2136. A first portion 2136a of the fibre optic bundle is directed towards a reference sensor 2138, the reference sensor also being housed in the mounting 2140. The first portion 2136a may be only a single fibre of the fibre-optic bundle 2136, or it may be a plurality of fibres. The first portion 2136a directs a portion of the light emitted by the light source 2108 to the reference sensor 2138. A second portion 2136b of the fibre optic bundle (in this case, the entire remainder of the bundle 2136) transmits the light towards the metal sample 2112. In other words, the end of second portion 2136b of the bundle 2136 which directs light towards the metal sample 2112 acts a little like the light source 2108 in other examples, in the sense that the second portion 2136b of the bundle emits light 2110 towards the metal sample 2112. This light is reflected and received and the intensity interpreted in the manner described above.

[0397] The reference sensor 2138 receives light directly from the light source 2108, so detects any variations in the emitted light intensity and/or spectral range without the emitted light 2110 being affected by environmental factors, such as the change in reflectivity of the metal sample 2112. This means that a drop in intensity of the reflected light 2114 detected by the light sensor 2116 can be correlated with the intensity measured by the reference sensor 2138. Where the intensity of the emitted light 2110 drops, the reference sensor 2138 and the light sensor 2116 will both detect a drop of approximately the same magnitude as each other, indicating a false positive which can be discounted. In cases where the light sensor 2116 detects a change in the intensity of reflected light 2114 which does not correlate with a drop in emitted light 2110 detected by the reference sensor 2138, or where the light sensor 2116 detects a drop larger than would be expected for a corresponding drop detected by the reference sensor 2138, then the event can be logged as representing corrosion.

[0398] FIG. 19 shows another example of a sensing apparatus 2700 having a reference sensor 2138. In this case, however, rather than using a fibre-optic bundle, the reference sensor 2138 is provided with a reference beam of light 2110a by virtue of a beam splitter 2142. As before, the light source 2108 emits a light beam 2110. This is directed towards the beam splitter 2142 which splits the emitted light beam 2110 into two beams 2110a and 2110b. The first of these beams 2110a is directed to the reference sensor 2138 and the reference sensor 2138 operates in much the same manner as discussed above in relation to FIG. 18. The second beam 2110b is directed towards the second surface of the metal sample 2112 and the beam is then reflected as a reflected beam 2114 and enters the light sensor 2116 in the manner set out above. Similarly to the situation described in respect of FIG. 18, the use of a reference sensor 2138 allows variations in the intensity and/or spectral range of the emitted light 2110 to be accounted for, so improving the reliability of measurements provided by the sensing apparatus 2700.

[0399] The beam splitter 2142 can be selected so that it splits any proportion of the light to the reference sensor 2138 as desired. Typically, only a small proportion of light (e.g. no more than 20%) should be diverted for reference sampling, so that the actual measurement is performed with a reasonable intensity of light. Naturally, this reduced absolute magnitude of the reference measurement can be adjusted to account for the smaller proportion of light being received by the reference sensor 2138.

[0400] Consider now FIG. 20 which shows a flow chart 2800 describing the operation of the method as set out herein. The method starts at step 2150 where a metal sample is mounted in a water system, the metal sample having a uniform thickness and including a first planar surface and a second planar surface opposite the first planar surface, wherein the first planar surface is arranged in contact with water of the water system. As noted above, this exposes only one surface of the metal sample to the system water, and consequently allows a determination of the extent and sometimes the type of corrosion by monitoring corrosion which penetrates throughout the uniform thickness of the metal sample.

[0401] The method continues at step 2152 in which light is emitted towards the second planar surface of the metal sample. As noted above, a light source may be supplied to emit the light, and is arranged to direct the emitted light towards the second planar surface. The light source may be adjustable in the sense that the intensity or spectral composition of the emitted light may be adjusted or controlled as part of the method. Indeed, the light source may be controlled in the sense that it emits no light most of the time, and is switched on periodically to emit light and provide a reading.

[0402] Next, at step 2154, the light reflected by the second planar surface of the metal sample is received, for example at a light sensor. The light reflects from the second planar surface in a known way, for example changing intensity and/or spectral range in a known way in response to an entirely clean surface, and changing intensity and/or spectral range in a known way, different to the first known way, in response to tarnishing or other signs of corrosion. The received light is analysed by factoring in the difference between the known composition of reflected light from a clean surface and the known composition of reflected light from a tarnished and/or otherwise corroded surface to arrive at an indication that the surface is showing signs of corrosion. Optionally, the sensor can determine the extent and/or type of corrosion. The sensor can be configured to synchronise with the light source to only detect reflected light at times when the light source is on, thereby reducing power consumption. Indeed, the signals from the sensor can also be analysed only when the light source is on, in order that resources are not used analysing data when no light is supplied to the metal sample. In other examples, the spectral range and/or intensity of the sensor are/is selected to conform to the expected intensity and/or spectral range of light emitted by the light source and reflected from the metal sample. Optionally, as set out above, the method may make use of a reference sensor to normalise the measurement and help rule out false positive results.

[0403] Next, at step 2156 the system (e.g. the light sensor) generates a signal indicative of the intensity of the reflected light. This may be an analogue signal, e.g. where the magnitude of a current or voltage output by the sensor is representative of the received light intensity. In other cases, the output may be digitised, for example to encode light levels as a digital signal, optionally, wherein different wavelength bands are separately encoded to allow for a spectral analysis. In one example, this may include sending digitised intensity values for each of a red, green and blue (RGB) band, similarly to how digital images are stored. In the RGB system, it is common for the peak intensity of the bands to be located at wavelengths of approximately: Red: 650 nm; Green: 525 nm; Blue: 440 nm, for example. In other cases, different spectral bands may be used, as appropriate, including more than three bands, for example or extending beyond the visible range of the electromagnetic spectrum.

[0404] In step 2158 the intensity of the reflected light is correlated with corrosion of the metal sample. This step relates, as set out above, a change in light intensity with the onset or progression of corrosion in the metal sample. Such a determination can be used to e.g. provide an alert to a maintenance team that pipes of a certain thickness and made from a particular metal are likely to fail soon and should be replaced.

[0405] Modifications to the general method 2800 set out in FIG. 20 may be made, for example by using some of the features of the examples set out respect of the other Figures.

[0406] Consider now FIGS. 21A and 21B, which show a plan and side view respectively of a metal sample system 2900 for use in the sensors described herein. The sample system 2900 comprises a metal disc 2112, having a layer of corrosion-resistant material 2144 having an annular shape around the edges of the metal disc 2112. As shown in e.g. FIG. 13, where a metal sample 2112 is mounted on a sensing apparatus 2100, there is a portion at the edges where a seal 2118 contacts the metal sample 2112. It has been found that the presence of the seal disproportionately increases the rate of corrosion near to the point of contact between the seal 2118 and the metal sample 2112. The corrosion-resistant coating 2144 is positioned to align with the region where the seal 2118 contacts the metal sample 2112. The corrosion-resistant coating may be wider than the footprint of the seal 2118 to allow a user to position the sample 2112 and the seal 2118 with a reasonably low degree of accuracy. There is a balance between allowing a user to be inaccurate in their positioning of the sample and ensuring that a reasonable area of metal is exposed to the system water.

[0407] As shown in FIG. 21B, the metal sample 2112 may be supplied with a corresponding transparent disc 2120. This may, for example, be adhered to the metal disc (using a compatible and transparent adhesive), or it may be supplied loose for holding in place by the clamping action of the seals 2118. Indeed, since the metal sample systems 2900 are intended to be replaceable, once a sensing apparatus has been sold, users may find it useful to buy replacement metal sample systems only. Since the transparent element 2120 does not usually become corroded or damaged, the metal sample 2112 with its corrosion-resistant coating 2144 may be supplied as a separate element that is the metal sample 2112 may be supplied without a transparent element 2120.

[0408] As an example, the metal disc 2112 may be formed from a metal representative of metals in the water system into which the metal disc 2112 is to be mounted, for example, carbon steel, aluminium, brass, copper, stainless steel, etc. The disc 2112 may be approximately 10 to 24 mm in diameter, and the corrosion-resistant coating 2144 may extend inwardly from the edge by approximately 3 mm. The metal disc 2112 may be supplied in various thicknesses, for example 0.025 mm, 0.05 mm, 0.075 mm, 0.01 mm, etc.

[0409] Turning to FIG. 22, which shows an arrangement for multiple metal samples 2112 in a sensing apparatus 2200. Here a plurality of metal samples 2112 is arranged in a grid. Horizontal rows are all formed from the same metal, and are shown separated by dashed lines simply to guide the eye. A first set of metal samples 2112a is formed from a first metal representative of metals in the system, for example copper. A second set of metal samples 2112b is formed from a second metal representative of metals in the system, for example brass. A third set of metal samples 2112c is formed from a third metal representative of metals in the system, for example aluminium. A fourth set of metal samples 2112d is formed from a fourth metal representative of metals in the system, for example stainless steel. An arrow (A) shows the direction in which thickness increases for each set of metal samples 2112a-2112d. In some cases, for example, the first metal sample 2112 (the leftmost sample) in each set of samples 2112a-2112d may have the same thickness, e.g. 0.025 mm. The next samples moving to the right in the direction of arrow (A) may have respectively thicknesses of 0.05 mm, 0.075 mm and 0.1 mm. In other examples, the metals in each set of metal samples 2112a-2112d may have different thicknesses, for example based on the likelihood of corrosion of that metal type, the typical thicknesses of components made from that metal, and the degree of accuracy required for that metal. In general, metals which corrode quicker may have thicker metal sample 2112 sizes than metals which corrode slower. Similarly, where components tend to be made thicker (or tend to be able to withstand severe pitting), thicker metal samples 2112 may be chosen. As noted above, the difference in thickness between adjacent metal samples 2112 is related to the accuracy with which the pitting depth can be determined. Therefore, the progression of thickness of samples 2112 can be selected to provide the desired resolution for that metal.

[0410] While four thicknesses of each type of metal are shown in the Figure, some cases may have more or fewer samples 2112. Indeed, in some case each metal type may have a different number of thicknesses, depending on the information a user wishes to obtain. While four metal types are shown in this example, different examples may provide more or fewer metal types.

[0411] While not shown here, the sensing apparatus has the features described above for making the measurement, such as at least one light source 2108, at least one light sensor 2116, etc. Multiplexing may be used as described above to provide fewer light sources 2108 and/or light sensors 2116 than the number of metal samples 2112 (i.e. fewer than 16 light sources 2108 and/or sensors 2116 in the example shown in FIG. 22).

[0412] In any of the above examples of sensing apparatuses, internal walls of the housing may have diffuse reflective inner surfaces for providing an optical integrating cavity. Optical integrating cavities are chosen so that all light emitted by the light source eventually reaches the light sensor. They have the effect of smoothing out the light intensity in the interior of the housing so that each portion receives approximately the same light intensity. This prevents inhomogeneities in the angular distribution of light emitted from the light source from causing different parts of the sample to be illuminated with different intensities of light and ensures that the background light levels are highly consistent. Such a situation could lead to different parts of the sample showing different effects when corrosion occurs, if the inhomogeneities are severe enough. Of course, providing a highly homogeneous light source is another solution to this. Where the internal walls of the cavity are said to be diffuse reflective surfaces, this does not necessarily apply to the second planar surface of the metal sample, and certainly does not apply to the transparent element, where such an element is present.

[0413] It will be appreciated from the above description above that many features of the different examples are interchangeable with one another. The disclosure extends to further examples comprising features from different examples combined together in ways not specifically mentioned. Indeed, there are many features presented in the above examples and it will be apparent to the skilled person that these may be advantageously combined with one another. Examples of features which are combinable with other features in this way include: the baffle 2134 shown in FIG. 17; the use of multiple metal samples 2112 in a single sensing apparatus as shown in FIG. 14; the use of multiplexing with multiple samples to reduce the number of light sources and/or sensors in an apparatus; the use of the double O-ring seal arrangement of FIGS. 16 to 19 for improving the seal; the use of fibre optic cables 2128 as in FIG. 16 to mount the light source 2108 and light sensor 2116 in a second housing 2126; and the use of a lens 2130 as shown in FIG. 17.