Computer-implemented method and apparatus for sensor positio ning for leak detection in a water network

The present invention relates to a computer-implemented me thod, an apparatus and a computer program product for sensor positioning for leak detection in a water network.

Drinking water is one of the most important commodities. Considerable water losses due to leakage can occur in water distribution networks. Water losses can have a significant percentage even in developed countries. Therefore, it is a major challenge to detect leaks for avoiding and reducing wa ter losses. Efficient leakage detection requires sensors, e.g. flow meters, placed at favorable positions. For the sen sor placement, topological as well as hydraulic consideration need to be considered in order to install sensors at relevant positions in the network. More sensors at relevant positions increase the quality of the leakage detection, however, it is for example desired to avoid redundancies without decreasing the detection quality.

It is therefore an objective of the present invention to im prove the sensor positioning for leakage detection in a water distribution network.

The object is solved by the features of the independent claims. The dependent claims contain further developments of the invention.

The invention provides according to the first aspect a compu ter-implemented method for leak detection in a water network, comprising the steps:

(a) receiving a hydraulic model for the water network, wherein pipes and nodes of the water network are modelled ba sed on a water network model,

(b) performing a hydraulic simulation for the water network using the hydraulic model, (c) based on the hydraulic simulation, determining total sen sor sensitivities to modelled leaks for modelled leak detec tion sensors wherein

- for a respective modelled leak, sensor sensitivities to said leak are determined, these sensor sensitivities are sor ted and respective modified sensor sensitivities are genera ted based on the respective sorting index, and,

- for a respective leak detection sensor, a total sensor sen sitivity is calculated by combining the respective modified sensor sensitivities, and

(d) selecting at least one modelled leak detection sensor ba sed on the respective total sensor sensitivity and exporting its respective sensor position for positioning of a real leak detection sensor in the water network.

For executing the inventive method, a computer program pro duct may be provided. The inventive method may be implemented by means of one or more processors.

It is an important advantage of the proposed computer- implemented method that sensitivities of leak detection sen sors as well as the spatial distribution of leak detection sensors are considered for optimized sensor positioning. Hence, the method advantageously enables well-distributed po sitioning of leak detection sensors avoiding redundancies. Therefore, an improved leakage detection can be achieved. In particular, the optimized position for at least one sensor, e.g. a flow meter, is determined such that said sensor enab les leak detection of multiple leaks distributed over the wa ter network.

According to a preferred embodiment, a ranking list for the respective modelled leak detection sensors can be exported, wherein the ranking is based on the respective total sensor sensitivities, and wherein the ranking list can comprise the corresponding sensor positions. Preferably, based on the provided information in the ranking list, real sensors can be placed at corresponding favorable positions and/or previously placed real sensors can be re placed or removed. Therefore, the ranking based on total sen sor sensitivities enables optimized and efficient sensor placement in a water network.

According to a preferred embodiment, a sensitivity threshold can be applied to the total sensor sensitivities listed in the ranking list and at least one leak detection sensor can be selected from the ranking list based on the sensitivity threshold, and the sensor position of the selected leak de tection sensor can be provided by means of a user-interface for sensor positioning in the water network.

Preferably the ranking list can be sorted in a decreasing or der and a sensitivity threshold is applied to select that at least one sensor which is sensitive enough to detect at least one leak of a specific size. The sensitivity threshold is therefore preferably defined based on the size of the at least one leak which shall be detected. Hence, by introducing the sensitivity threshold to the ranking list, those sensors can be selected which are able to detect leaks of a defined size, resulting preferably in well-distributed sensor positi ons.

According to an advantageous embodiment, the at least one se lected leak detection sensor can be removed from the ranking list and the ranking list can be updated by recalculating the total sensor sensitivities.

By removing, blanking or hiding the respective entry of the selected sensors in the ranking list, the ranking can be up dated such that the next most sensitive sensor can be deter mined and positioned. Preferably, all leaks having a size above the sensitivity threshold and which may therefore be detected by the previously selected sensors are removed from the simulation and/or not further considered. This process can be reiterated until enough modelled sensors are deter mined to detect the modelled leaks.

According to a further embodiment, the hydraulic model may be based on a load model for the water network, wherein the load model may comprise inflow and outflow rates of the water net work.

The load model for the water network can comprise information about inflows and outflows, wherein outflows comprise throug hput and consumption. The inflow and outflow rates can be de duced from flow measurements and/or estimated from annual flow quantities.

According to a further embodiment, the load model may be ge nerated comprising the steps:

- calculating the distribution of inflow shares for a respec tive inlet pipe based on measured and/or estimated annual flow quantities for the water network,

- calculating the outflow rates for a respective node based on measured annual consumption of consumers, and

- calculating the inflow rates based on the inflow shares and the outflow rates.

Usually, merely annual flow values for a water network are provided. Therefore, the load model can be generated as described in order to determine flow rates as input for the hydraulic simulation.

According to a preferred embodiment, the hydraulic model can be a steady state hydraulic model.

The calculation of a steady state hydraulic model is advan tageously simple and/or fast.

According to a further embodiment, the hydraulic model may be simplified compared to the real water network by modelling multiple pipes as single pipes and/or removing single taps and/or projecting a consumer load into a neighboring node.

According to a preferred embodiment, the water network model can be computer-readable and generated based on measured wa ter network data and for at least a closed subnet of the wa ter network.

According to a further embodiment, a real leak detection sen sor can be positioned in a water network based on the ex ported position of the selected modelled leak detection sen sor.

Preferably, the real leak detection sensor is positioned at or close to the exported position.

According to a second aspect, the invention provides an appa ratus comprising a digital computer, configured to perform the computer-implemented method according to the invention.

According to an advantageous embodiment, the apparatus com prises a user-interface which is configured to provide the ranking list of the modelled leak detection sensors for sen sor positioning in the water network.

The invention further comprises a computer program product directly loadable into the internal memory of a digital com puter, comprising software code portions for performing the steps of the said method when said product is run on a compu ter.

A computer program product, such as a computer program means, may be embodied as a memory card, USB stick, CD-ROM, DVD or as a file which may be downloaded from a server. For example, such a file may be provided by transferring the file compri sing the computer program product from a wireless communica tion network. Embodiments of the invention will be explained in more detail by reference to the accompanying figures.

Fig. 1 shows a first flow chart including method steps in volved in an embodiment of a method for sensor po sitioning for leak detection in a water network ac cording to the invention;

Fig. 2 shows a second flow chart including method steps involved in an embodiment of a method for sensor positioning for leak detection in a water network according to the invention; and

Fig. 3 shows a schematic diagram of an embodiment of an apparatus for sensor positioning for leak detection in a water network according to the invention.

Equivalent parts in the different figures are labeled with the same reference signs.

Figure 1 shows a flow chart including method steps involved in an embodiment of a method for sensor positioning for leak detection in a water network according to the invention.

The first step 10 involves receiving a hydraulic model for the water network. The hydraulic model is based on a water network model, wherein the water network model comprises mo delled pipes and nodes of the real water network. Preferably the water network model is computer-readable and generated based on measured water network data such as topology infor mation. Preferably the water network model and therefore the hydraulic model comprises information for at least a subnet of the real water network, wherein the subnet can be defined by calculable inflow and outflow shares for a closed net of pipes. The advantage of modelling merely a closed subnet of the real water network lies in predictability of flows resul ting in a higher leak detection accuracy. The hydraulic model is preferably provided as a computer- readable data structure comprising parameters which specify the hydraulic properties of the modelled part of the water network. Preferably the hydraulic model is based on a load model specifying the inflow and outflow rates of the real wa ter network.

The next step 20 involves performing a hydraulic simulation for the water network based on the hydraulic model. Prefe rably the hydraulic simulation is executed at least twice, wherein firstly the hydraulic simulation is executed without introducing leaks and secondly at least one leak is simula ted. Leaks are preferably modelled at position of nodes, e.g., consumer nodes. The size or strength of a leak is pre ferably determined by a defined consumption. It is further preferably assumed, that leak detection sensors, e.g. flow meters, can be positioned at positions of pipes. Therefore, by determining the effect of a leak on the flow through a pipe, the required sensor sensitivity of a sensor placed at the position of that pipe can be determined. The effect of a leak on a pipe results in a difference of water flow through the pipe. Based on said difference in water flow, a required sensitivity of a potential sensor placed at the position of said pipe can be deduced.

The next step 30 involves the analysis of the hydraulic simu lation. Based on the hydraulic simulation, total sensor sen sitivities to modelled leaks are determined. For example, a comparison of the executed simulation without modelled leaks to the executed simulation comprising modelled leaks, the effect of leaks on the water flows through pipes can be eva luated. The quotient, difference quotient or difference of such a flow through a pipe shows the effect of a leak on said pipe and can be considered for example as the potential sen sitivity of a sensor to said leak. In other words, based on the hydraulic simulation, the quotient, difference quotient or difference in water flow due to at least one leak on a respective pipe can be determined yielding a potential sensor sensitivity for a sensor placed at the position of that pipe. Preferably, a number of potential leaks are modelled and in cluded in the hydraulic simulation. For a respective modelled leak, sensor sensitivities for a number of pipes to the respective leak are determine. Subsequently, the determined sensor sensitivities are sorted for example in a decreasing order and multiplied with their respective sorting index, ge nerating modified sensor sensitivities. Hence, the modified sensor sensitivities result from weighted sensor sensitivi ties. The weighting of the sensor sensitivity is based on the sorting index.

For a respective modelled leak detection sensor, a total sen sor sensitivity is subsequently calculated by combining, e.g. multiplying, the respective modified sensor sensitivities for that sensors.

In the next step 40 the total sensor sensitivities for all modelled sensors are provided and at least one modelled leak detection sensor is selected based on its respective total sensor sensitivity. Preferably, a ranking list of the total sensitivities is generated, step 50, and the most sensitive sensor is selected from that ranking list based on a predefi ned sensitivity threshold, step 60. The position of that sel ected sensor can further be provided, step 70, for sensor po sitioning in the real water network. The sensor position can for example be provided by means of a user-interface such that a real leak detection sensor with a corresponding sensi tivity can be positioned in the real water network, step 80.

The entry of the selected sensor is removed from the ranking list and the ranking list is updated, step 90, by repeating the steps 30 to 70, wherein preferably only remaining leaks which fall below the sensitivity threshold are modelled and sensor sensitivities are determined only for these remaining sensors. In other words, preferably, all leaks having a size above the sensitivity threshold, and which may therefore be considered to be detected by the selected sensors, are remo- ved from the simulation and/or not further considered when updating the ranking list.

Figure 2 shows a second flow chart including method steps in volved in an embodiment of a method for sensor positioning for leak detection in a water network according to the inven tion.

First, a computer-readable water network model WNM is provi ded. The water network model WNM preferable represents the real water network or at least a subnet of the real water network. The water network model WNM is preferably simplified with respect to the real water network, wherein for example pipes without intersections are modelled as one pipe and/or single tap offs are removed. Consumers and leaks can be mo delled in nodes.

Further, a load model LM for the water network is provided. The load model LM can be generated based on measured and e- stimated annual flow quantities.

For example, the load model LM can be generated comprising the following steps. Firstly, based on measured annual flow quantities of the inflows of the water network the distribu tion, i.e. a percentage value, of the water inflow is calcu lated. The percentage value is used as the mean value for a normal distribution of the inflow distribution. The according standard deviation can be decided depending on specific net work situations, e.g. such that 1 sigma results to 5 percen tage points. This formulates the distribution to calculate the split of water inflow. If measured annual values are not available, a typical actual distribution can be used as a re placement for the mean value. Secondly, based on measured an nual consumption values of all consumers, the actual water outflow rate for each node is calculated in e.g. liter per second (1/s). This value is used in a further normal distri bution, with the actual water outflow rate as a mean value. The standard deviation can be set depending of specific net work situations, e.g. such that 1 sigma results to 5% of the actual value. This setup formulates the distribution to cal culate the water consumption.

The inflow rates are computed in liter/second by using the percentage values of the inflow and the total sum of all con sumption values.

Based on the water network model WNM and the load model LM a steady state hydraulic model HM is generated, and as hydrau lic simulation HS is executed based on the hydraulic model HM. Preferably, at least two hydraulic simulations are set up and executed, wherein one comprises modelled leaks and the other is modelled without leaks in order to determine water flow difference in the pipes due to a leak. In other words, the different flow through a pipe due to a leak somewhere in the water network specifies the required sensitivity of a leak detection sensor at the position of that pipe for detec ting that leak. Therefore, in order to determine a sensor sensitivity, the flow difference is determined based on the hydraulic simulation.

The hydraulic simulation results are further evaluated to de termine optimal positions for leak detection sensors. For each modelled leak LI, L2, ..., LN, sensor sensitivities SI,

S2, ..., SN for respective sensors positioned at the position of a pipe are determined. For example, a larger effect, i.e. flow difference, is expected for pipes close to a leak than for pipes farther away, resulting in higher sensor sensitivi ties for sensors placed closer to the leak. The resulting sensor sensitivities SI, S2, ..., SN for the respective leak LI are sorted in a decreasing order and subsequently weighted, e.g., multiplied, by the respective sorting index, generating modified sensor sensitivities mSl, mS2, ..., mSN. A correspon ding procedure is performed for the other modelled leaks L2, L3, ..., LN.

The resulting modified sensor sensitivities mSl, mS2, ..., mSN are combined to total sensor sensitivities TS1, TS2, ..., TN for each respective sensor. The total sensor sensitivity therefore specifies the required sensitivity of a respective sensor to detect modelled leaks of a specific size. Prefe rably a ranking list RL is generated comprising the total sensor sensitivities TS1, TS2, ..., TN and respective positions of the sensors, i.e. the position of the respective evaluated pipe. The ranking list RL can for example be ordered in de creasing order based on the total sensor sensitivity values. Preferably, a real sensor can be placed at the position cor responding to the highest total sensitivity of a modelled sensor.

The ranking list RL can preferably be recalculated when at least one sensor is selected from the list based on a sensi tivity threshold. The recalculation of the list comprises de termining the sensor sensitivities by considering only leaks not well-detected by the sensors placed so far. Preferably this process is repeated until sensors are placed to detect all modelled leaks.

Figure 3 shows a schematic diagram of an embodiment of an ap paratus 100 for sensor positioning for leak detection in a water network according to the invention. The apparatus com prises a digital computer P which is configured to perform a computer-implemented method according to the invention. The apparatus 100 preferably comprises an interface IF configured to receive for example a hydraulic model for a water network. The apparatus further comprises preferably a storage unit S configured to store and provide for example a hydraulic model for a water network. The apparatus 100 comprises preferably a user-interface UI configured to provide the sensor position ing information, e.g., a ranking list for the leak detection sensors for positioning real sensors in a water network.

Although the present invention has been described in detail with reference to the preferred embodiment, it is to be un derstood that the present invention is not limited by the disclosed examples, and that numerous additional modifica tions and variations could be made thereto by a person skil- led in the art without departing from the scope of the inven tion.