EP3662210A1 - Method for characterizing a storage water heater and for learning the drawing profile - Google Patents

Abstract

Object of the present invention is a method for learning, value and timing of the physical and thermal characteristics of the hot water drawings profile in a storage water heater, where said profile repeats cyclically at given time intervals and where there are sensors of temperature (S.loc.i) which read local temperatures (T.loc.i) whence it is possible to devise an average local temperature T.loc approximating the average storage temperature (T.acc) only in the absence of turbulence in the storage tank (S). According to the invention, the amount of each water withdrawal or cluster of withdrawals is considered represented by the reduction (AT.tap) of the storage temperature (T.acc) caused by the water withdrawal and is calculated a posteriori, once the withdrawal is terminated at a time t3 during or at the end of a subsequent heating phase triggered by the reduction of the local temperature (T.loc) caused by the withdrawal and is considered equal to the difference between storage temperature (T.acc) a time before the start of withdrawal and the storage temperature (T.acc) at time t3 plus the increase (ΔΤ) of storage temperature (T.acc) caused by the heating itself. The main advantage of the method of the invention is that the determination of the extent of the withdrawals is not affected by the fact that said local temperatures (T.loc.i), the only ones that can be measured directly, are not generally representative of the temperature (T.acc) average of water in the storage tank.

Description

METHOD FOR CHARACTERIZING A STORAGE WATER HEATER AND FOR LEARNING THE DRAWING PROFILE

DESCRIPTION

Object of the present invention, in a generic storage water heater controlled by an electronic control, a new method for learning the user's hot water consumption habits as well as a new method for managing the water maintenance temperature aimed at heating the water only in the quantity and at time foreseen based on said habitual consumptions.

An instantaneous water heater can deliver a hot water flow rate strictly proportional to the installed thermal power. Generally, there is difficulty in installing high powers and this sets a limit on the maximum allowed flow rate. Advantage of storage water heaters is that they can deliver very high water flows with limited installed thermal power. The amount of water which can be delivered at the temperature of use Tu during a single tapping may be larger than the volume of the storage tank because this is specifically maintained at a storage temperature T.acc greater than said temperature of use Tu and the water withdrawn is then used by mixing it with cold water.

In the present description, frequent reference will be made to the storage temperature T.acc, meaning a fictitious temperature, representative of the enthalpy content of the water in the storage tank, not necessarily directly measurable, and equal to the average water temperature in the tank. In other words, given C the thermal capacity of a mass of water in a storage tank and E the thermal energy that such mass can deliver to an environment at 0 °C, by storage temperature T.acc (expressed in °C) it is meant the E / C ratio.

Since the storage tanks are expensive and cumbersome, it is common to have a volume as much as possible small by maintaining, however, the storage temperature T.acc high (generally 60 - 75 °C) whereas the effective temperature of use Tu, normally included between the 35°C and 40 °C, is obtained just upstream from the points of use by mixing with cold water; however, water is often distributed at temperatures higher than use temperature Tu to compensate for cooling along the distribution pipes.

Generally, the volume V of the tank is selected in order to satisfy the largest of the withdrawals foreseeable for a specific user by maintaining the storage temperature

T.acc at the maximum possible value while the installed thermal power must be such as to restore a reserve of sufficient water for the following withdrawal.

In conclusion, several different models of storage water heaters are needed to accommodate each different category of users.

In order to satisfy the highest rated hot water flow, which is the largest foreseeable withdrawal, it is obvious that the water heater is kept for most of the time at a storage temperature T.acc which is unnecessarily high for most of the remaining withdrawals.

As a result, as known, in storage water heaters the main cause of inefficiency are thermal losses which can also be very relevant and often useless throughout the day, also away from withdrawal timings.

Therefore, several methods have been developed which vary for accuracy and ease of use, in order to limit the heat losses by keeping the temperature of the water heater at the minimum values able to satisfy user needs.

The minimum requirement to guarantee the service is always met is that the water heater is maintained, at least for a part, at a minimum temperature not lower than the usage temperature Tu, in order to withstand unexpected minor withdrawals and that the tank volume is large enough to guarantee the largest water withdrawal required for that user, keeping the temperature at the maximum allowed value. Generally, withdrawals have a very uneven pattern during the day, both for times and amount of consumption, tending to gather at specific timings. From here on, said tapping pattern, consisting of the times and amount of withdrawals, will be called drawing profile.

While the drawing profile is uneven during the day, this is highly repetitive during specific time cycles that are repetitive between them: in particular, for a weekly time cycle. User habits, in fact, are so stable that it is recognizable a specific drawing profile for Monday, Tuesday and so on with, in particular, significant differences between working days and holiday days as well as between midweek holyday days and vacation times.

This cyclicity of the water drawing profile allows, therefore, to foresee them with reasonable certainty and it is possible, then, to implement the methods of controlling the temperature of the water heater so that it is variable during the day. Each of said repetitive time intervals is herein said as a withdrawal cycle.

Although the cycle of withdrawals generally lasts one week, where each day of the week can be understood as a sub-period of the cycle having its characteristics of withdrawal that differentiate it from the other days, for particular users, such as in the working environments where any difference in behaviour is not correlated to the day of the week, the cycle of withdrawals, however repetitive, may have a duration other than seven days and the sub-periods different from 24 hours. In order to reduce the heat losses, the simple method in use has always been to activate and deactivate the heating element by means of a clock so that the desired temperatures are guaranteed only during the period in which withdrawals are expected.

Another simple method, less efficient from the point of view of energy but economically more advantageous for the same, is that of activating the heating element only during lower time tariffs periods; the water may be unnecessarily too hot well in advance of the needs, but in any case, it has been obtained at relatively low costs.

These are methods in which simply the offset temperature T.off of the thermostat is set to a fixed value; nevertheless, the storage temperature T.acc is reduced because the heating element is forcibly deactivated.

The most effective methods for reducing consumption are those methods which allows the storage temperature T.acc to vary over time in a programmed manner. For this to be possible, the drawing profile must be known.

The document EP 0 866 282 describes a device in which it is possible to program the desired sequence of withdrawals, that is, the drawing profile. The size of the n withdrawals foreseen in the time sequence t.l, t.2, ... t.k, ... t.n is recorded by setting for each time t.k the temperature T.off.k that it is believed can satisfy the k-th withdrawal Pk. A limitation of the method consists in the difficulty of a correct setting, since the user may not be aware of the actual hot water withdrawal times and the actual T.off.k setting values to obtain the desired amount of hot water at use temperature Tu. The set-up method, therefore, involves a series of adjustments for tests and errors with a high probability that the user will quit adjusting the set up as soon as the heating needs are satisfied without knowing if he could have achieved this more efficiently. Another difficulty lies in the fact that the actual time when the desired temperature is reached depends on the heating time, which is difficult to evaluate and however variable over the time for the same water heater for various reasons such as calcareous deposits, seasonal variations of the room temperature in which the water heater is housed or the temperature of inlet water to the storage tank, reduction over the time of the effective heat output of the heating element.

The prior art document GB 2 146 797, on the other hand, detects information on the timing and amount of each withdrawal using flow sensors and sets, for each withdrawal, the storage temperature T.acc at a value that is intermediate between the minimum and the maximum allowed and proportional to the expected withdrawal volume. The method has the drawback of requiring the presence of flow sensors to detect the withdrawals; moreover, it has no self-adapting capability, in the sense that it learns the variability of the withdrawals but, by assigning to each size of withdrawal an unchangeable temperature because generated by a pre-set formula, it does not have the possibility to correct it if it is too high or too low.

According to the document EP 0 356 609, instead, the sequence of the timing of the withdrawals and the corresponding desired storage temperatures T.acc are preset in an electronic processor; the computer consequently establishes the control values that the adjustment temperature for the thermostat have to assume for each time interval. Subsequently these adjustment temperatures are adjusted by raising them for the intervals in which the desired storage T.acc have not been reached and decreasing them in the opposite case. A limitation of the method, as in the first document mentioned, is the necessity of having to pre-set the foreseen withdrawals; another limitation, as in the second document cited, is that it works on a fixed pre-set storage temperature T.acc which, however, is not guaranteed to be the best to ensure the desired performance in the most efficient manner.

The document US 2003/0194228, in a storage water heater, detects if there are water withdrawals in progress by calculating the speed of the temporal variation of the water temperature in the storage; if withdrawals are detected, it records the time, duration and heating methods put in place to satisfy it. Each subsequent withdrawal is compared to the previous memorized ones and, if a similar one is found, the same power supply methods applied to the previous one are applied to the ongoing withdrawal. This method is ineffective for storage water heaters since the power supplied as the withdrawal starts is a late intervention since in a storage water heater this power is insufficient to provide the required thermal energy instantaneously. Consequently, the requested withdrawal cannot be satisfied even by drawing on information relating to a similar prior withdrawal. The thermal power to be supplied is calculated taking into account the water temperatures read at intervals of time but also environmental parameters (such as ambient temperature) and / or construction parameters (such as the thermos-physical and / or geometric characteristics of the water heater itself) so that the control software must be customized for each model of water heater.

Document US 5 526 287 A detects the beginning and the end of a withdrawal by checking, from the outside, the temperature of the cold water inlet pipe in a storage tank. This method of control must be considered inadequate to detect both the size and the effective duration of the water withdrawal: the external temperature of the pipeline, in fact, varies with the passage of cold water in ways strongly influenced by the external temperature, the thermal inertia as well as the hot water temperature in the storage which is transmitted along the pipe. The target temperature for hot water is calculated considering continuously both said external temperature to the inlet pipe and the external temperature to the outlet pipe temperature towards the users as well as considering the amount of water withdrawal in progress, the thermal heat loss and the thermal power available based on formulas containing constants (R, B, C) empirically predefined and characterizing the water heater. This method is also not suitable for guaranteeing the performance for storage water heaters because, as in the previous document, it intervenes belatedly.

The document EP 2 362 931 Bl is the first among the documents listed here that identifies and records the water drawing profile exclusively by monitoring the accumulated water temperatures with one or more sensors placed in areas of the tank most likely to be affected by the entry of cold water which occurs at each withdrawal. A rate of decrease of such temperatures over a certain threshold at a- given time indicates that a withdrawal has begun at that moment while the amount of temperatures decrease, indicates the amount of the withdrawal itself. In this way it is possible to build the withdrawal cycle. This document therefore indicates how to identify the timing and the amount of the withdrawals exclusively based on the monitoring of water temperatures in the storage. The timing estimate of each withdrawal is very precise, thanks to said positioning of the sensors near the cold water inlet, however the same position makes it difficult to assess, from the temperatures read by the sensors, the real average water temperature at the end of the water withdrawal. This may cause an underestimation of this average water temperature with consequent overestimation of the amount of the withdrawals and therefore a subsequent management of the water heater at higher temperatures than necessary. Acquired the drawing profile in a first cycle of withdrawals in which the T.off temperature of the thermostat is kept high enough to guarantee a quantity of hot water equal to the maximum expected for that model of water heater, in subsequent cycles, substantially all the previous documents bring the T.off temperature to the maximum value necessary for the first of the next expected withdrawals and activate the heating element in advance on the withdrawal time only of the time necessary to reach such T.off temperature coinciding with said first expected withdrawal.

The activation of the heating element with the correct advance involves the knowledge of the heating speed that most of the aforementioned documents provide for estimating through learning.

A common drawback to all the methods described above is that they only aim to satisfy the first of the next scheduled withdrawals, and afterwards they may fail to satisfy a consistent withdrawal occurring shortly after the withdrawal just fulfilled due to lack of sufficient time to restore the temperature T.off to the new needed value.

The document EP 2 366 081 Bl can construct a profile of "fictitious water withdrawals" that allow the water heater to prepare in advance for one or more important withdrawals close to a first withdrawal. For the rest, the document identifies and records the profile of the withdrawals in a very similar way as the previous EP 2 362 931 Bl of which therefore has the same merits and limits for these aspects.

The document EP 2 328 046 Bl assigns to the T.off temperature only four possible predetermined values corresponding to an expected "important", "normal", "weak" or "minimum" estimated withdrawal. The extent and the time of each withdrawal is not directly detected but by the measurement of the activation time of the heating element, triggered by the decrease of the storage temperature T.acc, within sliding time windows; naturally, longer or shorter activation times for the T.off temperature recovery were caused by more or less important withdrawals. The method has the advantage of acquiring data on withdrawals without additional sensors in addition to that which drives the thermostat but, by its nature, not providing direct and immediate measurements of the occurrence of withdrawals and their amount, nor knowing the speed of heating, requires a recursive learning, by successive approximations, for the development of which many withdrawal cycles are necessary and can necessarily decide and discriminate only among a few predetermined values for T.off. Therefore, the maximum energy saving that the method allows is achievable with delay compared to previous documents and only in a less accurate way. In addition, even a marked change in user behaviour, immediate adaptation to the following cycle is not possible.

A general problem in determining the energy content and the energy consumption of a water heater is that these are strictly correlated to the value of the storage temperature T.acc while the temperature probes can only measure the local temperature (herein called local temperature T.loc) which is very far from the storage temperature T.acc if the water in the water heater is not in a steady state. A general object of the present invention is at least partly to overcome said these drawbacks.

In particular, an object of the present invention is that of acquiring, in a more accurate manner than is known today, the time and the size of a water withdrawal with the aid of temperature sensors but without a direct measurement of the storage temperature T.acc at the end of a withdrawal.

A further object of the present invention is to detect the heating speed of the water by the heating elements, minimizing the errors of assessment that the temperatures actually and locally read can induce.

A further object of at least some variants of the present invention is to detect said water heating speed separately for each of the possible different types of heating elements present and / or for groups thereof

A further object, at least for some variants of the present invention, is to construct the drawing profiles based on said acquisitions of times and entities of each withdrawal or groups of small withdrawals and on said heating speed.

Another possible aim, at least for some variants of the present invention, is to store said drawing profiles in synthetic form, preserving only the essential data for a possible management method of the storage temperature T.acc in the time which minimizes the heat losses while satisfying the needs of users.

Further objects, features and advantages of the present invention will be better evidenced by the following description of a basic version of the water drawing profiles acquisition method according to the main claims and of some preferred variants according to the dependent claims, all illustrated, purely by way of a non- limiting example, in the attached drawings, in which:

- figures 1.a and 1.b show, schematically, in section, a storage water heater with the essential elements for the objects of the invention; in fig. l.a, the water heater is heated by an electrical resistance while in fig. 1.b from a coil fed by a generic heat transfer fluid;

- figure 2 shows, in a graph, the temporal pattern of a local temperature during a complete heating phase of the water from an initial phase, at room temperature, up to the switching off of the heating element to reach the desired temperature in the absence of withdrawals;

- figure 3 shows, in a graph, the cyclical time pattern of a local temperature that oscillates between a minimum temperature reached by cooling from thermal losses to a maximum reached at the switching off of the heating element following the restoration of the target temperature and always in the absence of withdrawals;

- figure 4 shows, in a graph, the pattern of the storage temperature and of a local temperature, directly measured, during a period consisting of a first phase of mild cooling by thermal losses followed by a sudden cooling due to the effect of a withdrawal in turn followed by a heating when a heating element switches on;

- figure 5 shows, in a graph, the pattern of the storage temperature and of a local temperature, directly measured, during a period consisting of phases of mild cooling due to thermal losses interspersed / interrupted by a few phases of sudden cooling by withdrawals in turn followed by heating phases up to the temperatures determined from time to time on the basis of an acquired drawing profile;

- figure 6 shows, in a graph, the pattern of a directly measured local temperature during a realistic period consisting of phases of mild cooling due to thermal losses interspersed / interrupted by many phases of sudden cooling by withdrawals, even very small, in turn followed by heating phases up to the temperatures determined from time to time on the basis of an acquired drawing profile;

- figure 7 shows, in a graph, the pattern of the storage temperature and of a local temperature, directly measured, during a cooling phase for thermal losses, followed by restoration of the initial temperature and in the absence of water withdrawal;

- figure 8 shows, in a graph, during a heating step by a heating element with a thermal power slowly decreasing as the temperature increases, a rising slope of the same temperature with curvilinear pattern which can be approximated to a linear slope;

- figure 9 shows, in a graph, during a heating step by a heating element with thermal power more markedly decreasing than in fig. 8 as the temperature increases, a rising slope of the same temperature with curvilinear pattern, which can be approximated to two consecutive linear sections.

- figure 10 shows, in a graph, a first phase of mild cooling for thermal losses followed by a sudden cooling due to a first withdrawal in turn followed by a heating phase which stops before the average temperature has returned to the value had before the first withdrawal and then followed by a second withdrawal;

- figure 11 shows, in a graph, a first phase of mild cooling by thermal losses followed by a sudden cooling due to a first withdrawal in turn followed by a heating step which continues during a second withdrawal.

The aforesaid graphs, all showing the time scale in the abscissa and temperature in ordinates, are mostly on different scales to highlight from time to time certain or other aspects relevant to the invention and are only examples of possible situations non-binding for the purposes of the invention itself.

For graphic clarity, in the formulas, as symbol of multiplication, it has been used, the asterisk "*" instead of the usual "."

Unless otherwise specified from time to time, all the variables that appear in the following mathematical expressions are used in their absolute value.

For local temperatures we mean measurements actually sensed locally by a temperature sensor.

The storage temperature T.acc, as it has been defined, can substantially coincide with the temperatures that can actually be measured locally only when the water is in "steady state", i.e. not subject to turbulence due to, for example, water inlet when the temperature is substantially homogeneous in the storage tank and therefore knowable with a good approximation, anywhere measured.

In particular, it has been experimentally found that there is a sufficient correlation, for the purposes of the invention, between the storage temperature T.acc and the local temperatures read by the sensors in at least two circumstances: no withdrawal in progress with correlated turbulences and / or starting from a few minutes after the end of a warming phase.

In other cases, instead, the actual local temperatures can be very different from the storage temperature T.acc, especially because the temperature sensors are usually placed near the cold water inlet and often also in the heating element HE.

By heating element HE, it is meant any known heat source such as a group of one or more electric heaters, exhaust discharge pipes, heat exchanger of the condenser in a heat pump, heat exchanger of a hot water space heating system, etc.

The description of the invention is limited initially to the case in which said heating element HE is unique and capable of delivering a thermal power P substantially constant and independent of the level of the target water temperature. The invention will then be generalized to the case in which there are more types of heating elements and / or it is appropriate to take into account a variability of the thermal power when the water temperature varies. Figures 1.a and 1.b show a diagram of the tank S of a storage water heater, in the example of the vertical type, with a cold water inlet IN and a hot water outlet OUT and provided with a heating element HE which can be switched from the OFF state to ON and vice versa by a thermo-regulator TR.

The thermo-regulator TR is of electronic type, suitable for switching said heating element HE from the OFF state to ON and vice versa when a temperature sensor STR detects the reaching respectively of the switch off temperature T.off and the switch on temperature T.on = T.off. - Aist, where Aist is a default or adjustable hysteresis value.

Moreover, the thermo-regulator TR is of the type suitable for communicating the OFF and ON states of said heating element HE and the current values of at least said switch off temperature T.off to a control unit comprising a microprocessor MP.

Advantageously, said thermo-regulator TR is integrated in said microprocessor MP and, even more advantageously, can receive by the latter values of the switch off temperature T.off and of switch on T.on to be set at each time in agreement, for instance, with was decided by any control program for managing the water heater temperature operating while the methods according to the present invention are also active. As it will be shown, such possible variation of the values of said switch off T.off and switch on T.on temperatures does not prevent achieving the goals of the invention.

Furthermore, s local temperature sensors S.loc.i are provided (with i from 1 to s and with s > 1), preferably located near the cold water inlet IN and the heating element HE.

Preferably, said s local temperature sensors S.loc.i are of the NTC type, which ensure reading accuracies far higher than those necessary for the purposes of the invention.

Advantageously, one of the said s temperature sensors S.loc.i can coincide with the temperature sensor STR of the thermo-regulator TR.

The already said microprocessor MP is capable to perform the following reading, recording, and processing functions, foreseen for the different variants of the invention, including:

- knowing said switch off temperature T.off

- receiving from said temperature sensors S.loc.i, the related signals representing local temperatures T.loc.i;

- calculating the local temperature T.loc equal to the average value, possibly suitably weighted, of said local temperatures T.loc.i,

- measuring the passing of time,

- writing and reading in a specific status registry HE-ON/OFF the current status [ON] or [OFF] of said heating element HE,

- saving the time length Δί.οη of the ON states of the HE heating element,

- saving said local temperatures T.loc.i associated with the time of their reading;

- writing and reading in a specific status registry TAP the current states [NO TAPPING], [TAPPING-ON], [TAPPING-ALERT] representing, respectively, absence, occurrence or probable occurrence of withdrawals;

- processing and saving in stable or temporary way data according to the possible methods of the invention which will be shortly described, and which essentially consist of:

• reading/writing the detected local temperatures T.loc.i arranged to the time of reading/writing;

• reading/writing states in said status registries HE-ON/OFF and TAP

• execution, generally conditioned by the states read in said HE-ON/OFF and TAP status registries, of calculations based on the local temperatures T.loc.i detected,

• writing of the results of said calculations;

- possibly emit system status signals that can be perceived or detected by the user and/or by a technical assistance technician;

- possibly keep in memory a calendar containing information on holidays or pre-holiday days, concerning a user behaviour, similar to holidays or pre- holydays on a weekly basis.

According to the invention, said local temperatures T.loc.i can be used to calculate at least:

• the withdrawal start times,

· the heating speed by the heating element,

• the cooling speed due to thermal losses.

Other durable or temporary memory saving, and processing capabilities of the microprocessor MP may be foreseen and will be apparent with the description of a basic method and many variants according to the invention.

For the purposes of the invention, the mean value T.loc of said local temperatures T.loc.i is generally relevant, (possibly suitably weighed to give greater relevance to one or the other of them) or, as will be seen, the single T.loc.i value of each of them or of only one of them.

The choice of possible weights to be attributed to said local temperatures T.loc.i, with the exception of preferred weights in the cases listed below, does not play a decisive role for the purposes of the invention but rather affects the degree of precision of the method and is therefore easily performed by a skilled person depending on the location of the local temperature sensors S.loc.i and on the water heater model.

The size of each withdrawal is considered represented by the reduction AT.tap of the storage temperature T.acc caused by the withdrawal; however, this reduction is not directly detectable because the position and the quantity of said sensors S.loc.i is not able to provide valid information for the whole storage tank when there are turbulences and temperature stratifications due to withdrawals in progress or just terminated. In other words, the storage temperature T.acc is not directly measurable in these circumstances.

According to the invention, then (see in particular Figures 4, 10 and 11), the reduction AT.tap, caused by a withdrawal occurred at a time t2, is calculated a posteriori, once the water withdrawal is completed, at a subsequent time t3 preferably at the end or in any case during a subsequent uninterrupted heating phase triggered by the reduction of the local storage temperature T.loc below the switch on temperature T.on caused by the water withdrawal.

Said reduction AT.tap is considered equal to the difference between the storage temperature T.acc.2 at a time before the starting of the withdrawal and the storage temperature T.acc.3 at the time t3 plus the increase ΔΤ of the storage temperature T.acc caused by the same heating.

In fact, for the storage temperature at the end of the withdrawal T.acc.23, non- directly measurable, the following holds:

T.acc.23 = T.acc.2 - AT.tap

but also:

T.acc.23 = T.acc.3 - ΔΤ

from which is obtained:

ΔΤ-tap = T.acc.2 - T.acc.3 + ΔΤ = T.acc.2 - (T.acc.3 - ΔΤ)

where is it:

· T.acc.2 = temperature T.loc.2 read at the start of the withdrawal (the water heater was at rest)

• T.acc.3 = T.loc.3 - AT.loc

. ΔΤ = ΔΕ / C

By the way, the temperature T.loc.3, if read at the time of the heater switch off, coincides substantially with the threshold temperature for switch off T.off.3 which can be substantially different from T.acc.2 in particular if is active any known management program that assigns different values during the day to the switch off temperature.

ΔΤ-loc, here called "local temperature decrease" (described in detail below), is a temporary deviation between the temperature T.loc.3 read and the storage temperature T.acc.3 which can optionally be taken into account for more accurate calculations. In this case this is read in a memory and may have a pre-defined experimental value (and also null) or, preferably, determined according to a procedure that will be described later.

The increase ΔΤ is in turn proportional to the energy ΔΕ supplied during the entire heating phase which at least partially recovers the energy lost with the withdrawal. In fact, it is simply ΔΤ = ΔΕ / C, where C is the heat capacity of said storage tank S full of water.

According to a simple variant of the invention, this energy ΔΕ can be calculated directly by knowing a priori the nominal thermal power P of the heating element HE and the duration 5t.HE.on of said heating step from time t2 to t3 according to the formula ΔΕ = P * 6t.HE.on. However, this direct method is often unsatisfactory because the effective thermal power Pe and the thermal capacity C can be different from the nominal ones and also change over time for various factors such as for example: voltage fluctuations, degradation, scale build up, etc. Therefore, more advanced methods are preferred which indirectly and implicitly take into account the actual values of said effective thermal power Pe and heat capacity C as well as any disturbance factors.

Consequently, the preferred method for estimating ΔΤ is to place

ΔΤ = v.T.rise * 8t.HE.on

where v.T.rise is defined as the speed of the variation of the storage temperature T.acc, i.e. as the increase in temperature T.acc, in the unit of time by the heating element HE.

The formula ΔΤ = v.T.rise * 5t.HE.on is energetically equivalent to the previous ΔΤ = ΔΕ / C but has the advantage of not requiring knowledge of the quantities P and C while v.T.rise is measurable with good precision according to procedures that will be later on described and then stored in a special memory and possibly periodically updated.

Ultimately, you have the formula:

AT.tap = T.acc.2 - (T.acc.3 - v.T.rise * 5t.HE.on)

The meaning of this formula is very clear in fig. 4; less evident, but will be soon explained, in figs. 10 and 11.

Obviously, this procedure for calculating the reduction AT.tap is applicable only for those withdrawals which are sufficiently important to activate the heating bodies HE, i.e. to bring the local temperature T.loc below the switch on temperature T.on; otherwise, for each smaller consecutive withdrawal, the time when they occur can be detected if they cause a reduction of T.loc, while, their sizes are assimilated to that of a single withdrawal which becomes detectable only when the progressive decreases of T.loc finally trigger the activation of the heating body.

This procedure for calculating the reduction AT.tap is also more reliable in the case in which, from the timing of withdrawal, at the time t2, to the time t3 in which the temperature T.loc.3 is read, the water has reached a "steady" state which ensures the substantial accuracy of said formula T.acc.3 = T.loc.3 - AT.loc.

However, there are at least two cases in which it is necessary to read the temperature T.loc.3 in a time t3 in which said "steady" state has not been reached. The first one (see Fig. 10) is when the time t3 coincides with the end of the heating phase following the withdrawal which causes the temperature reduction AT.tap.1 but this phase is short because the variable switch off temperature T. off is at the time lower than the temperature T.acc.2 at the start of withdrawal and therefore easily reachable, before the steady state.

The second one (see Fig. 11) is when, during said heating phase triggered by the temperature reduction AT.tap.l, a further withdrawal occurs, represented by the temperature reduction AT.tap.2, at a time t3 earlier than the one it would have been necessary to reach again the steady state. It is therefore necessary to take into account the temperature T.loc.3 read at this time t3 in order to ignore the effects of said further withdrawal.

According to the invention, the steady state at the time t3 is considered reached if for the corresponding measured temperature T.loc.3, it is:

T.loc.3 - T.loc.2 > AT.q

Where AT.q, here called index stability, can also have a negative value. The physical meaning of this condition is clear: the more the temperature T.loc.3 is higher than T.loc.2, the longer it will be necessary to reach it and therefore the more time the water heater will have to reach the steady state.

The quiet index AT.q is an empirical parameter, predefined and pre-saved, depending on the model of the water heater and set by the expert of the field, substantially representing the degree of precision accepted in assuming the local temperature T.loc read as usable to get T.acc.3 according to the already presented formula T.acc.3 = T.loc.3 - AT.loc.

An acceptable and preferred value is AT.q = 0, that is: if T.loc.3 - T.loc.2 > 0 the water heater is considered in steady state at time t3 otherwise it is considered still in a state of turbulence.

Returning to the formula AT.tap = T.acc.2 - (T.acc.3 - v.T.rise * 5t.HE.on) shown in Figure 4 and in reference to Figures 10 and 11, we see how at time t3, in the situations illustrated here, the local temperature T.loc.3 is very far from T.acc.3 that it was assumed to calculate through the formula T.acc.3 = T.loc .3 - AT.loc because the usable t3 time is too close to t2 in order to consider the T.loc reading representative of the T.acc of the moment. Therefore the aforementioned formula AT.tap = T.acc.2 - (T.acc.3 - v.T.rise * 5t.HE.on) loses physical significance. However, the same formula returns to have its general value to provide at least an estimate of AT.tap with the following assumptions and definitions.

V.T.rise.loc is defined as the angular coefficient of the line tangent the local temperature T.loc rise curve at time t3, where T.loc is detected during the heating phase.

As for physical meaning, V.T.rise.loc represents the temperature rise speed of the water heater "seen" by the S.loc.i. temperature sensors.

For the determination of V.T.rise.loc any computational mathematical technique can be used which allows to calculate the angular coefficient of the tangent at a given point of a continuous curve known for algebraic mathematical function or known by points. The angular coefficient V.T.rise.loc is calculated at predetermined time intervals (e.g. 5 minutes), replaces the previously calculated value in a memory and is considered valid at the current time (between t2 and the instant it will be assumed as t3).

δί.οη.Ι is defined as the time interval between the already defined t2 and t3. δΐ.οηΐ .fict is defined as the fictitious time interval that would be necessary for the local temperature T.loc, to reach the storage temperature of fictitious end of heating T.acc.4 from T.loc.3, where T.acc.4 = T.acc.2.

From both Figures 10 and 11 follows:

v.T.rise.loc = (T.acc.2 - T.loc.3) / δΐ.οηΐ .fict, from which

δί.οηΐ .fict = (T.acc.2 - T.loc.3) / v.T.rise.loc

From both Figures 10 and 11 then follows:

for the first withdrawal AT.tapl

AT.tapl = T.acc.2 - (T.acc4 - v.T.rise * (δί.οη.Ι + 5t.onl.fict)) =

= T.acc. 2 - (T.acc.2 - v.T.rise * (δί.οη.Ι + 5t.onl.fict)).

Where, in turn

T.acc.2 - v.T.rise * (δί.οη.Ι + 6t.onl.fict ) = T.acc.23.

In essence, according to the method, the size of a withdrawal is considered represented by the reduction , of storage temperature AT.tap given by the formula AT.tap = T.acc.iniz - (T.acc.fin - v.T.rise * 6t)

Where:

T.acc.iniz is the start of withdrawal temperature so far called T.acc.2 and assumed to be equal to T.loc.2,

and where, again

• if at time t3 it is T.loc.3 - T.loc.2 > AT.q (water heater in a steady state) then T.acc.fin = T.acc.3 and 5t = 6t.HE.on

• else T.acc.fin = T.acc.2 e 8t = (δί.οη.Ι + δί.οηΐ .fict).

For the purposes of the present invention, the time of withdrawal t2 is considered the same as the time of consequent switch on of the heating element, that is to say that the temperature decrease from T.acc.2 to T.on is considered instantaneous. Note that the formula AT.tap = T.acc.iniz - (T.acc.fin - v.T.rise * δί) can always be calculated in the given conditions using the reading of the local temperature T.loc because for T.acc.2 the value read T.loc.2 was considered as valid while T.acc.3 is used only in the situation of steady state in which the formula T.acc.3 = T.loc.3 - AT.loc is considered valid.

Before proceeding with the description of the invention, it is appropriate to describe what has been found on the variations of the local temperatures T.loc.i and storage T.acc inside the tank S depending on its state (heating, cooling, water withdrawal in progress, mixed situations).

The present invention is based on the following assumptions, experimentally found to be valid with the approximation sufficient for the stated purposes:

- temperatures, wherever detected, are substantially uniform between them after an appropriate time when disturbing actions due to withdrawals have ceased and which last few minutes as easily monitored and, in any case, shown below;

- a few minutes after the end of a heating phase not disturbed by withdrawals, the temperatures, wherever detected, are equal to the storage temperature T.acc or if they deviate by quantities negligible, or in any case small, detectable and measurable.

For greater clarity, in the absence of heating and turbulence, the temperature inside the storage follows a regular pattern typical of the water heater model and therefore known experimentally. The temperatures read by the local temperature sensors S.loc.i are therefore representative of the enthalpy content of the stored water because they allow to assess the storage temperature T.acc.

Fig. 2, as mentioned, shows the trend of the local temperature T.loc during a complete water heating phase, in the absence of withdrawals, from an initial phase, at aqueduct temperature, until the switch off of the heating element once the target temperature is reached.

For the purposes of the invention, the curve, which is strictly of the exponential type, or, more generally, to an asymptotic trend towards a limiting temperature T.lim, can be considered to have a linear slope from the beginning to the end of the ON state of the heating element HE at least if this is able to supply a constant power P, substantially independent of the operating temperature, as certainly in the case of electrical resistances.

However, also for other heating elements HE, such as at least the heat pump capacitors, the power delivered P can most often be satisfactorily considered constant in the range of temperatures of interest.

Likewise, the cooling curve shown e.g. in figure 3 can be considered linear. At the end of the heating phase, i.e. when T.loc has reached said switch off temperature T.off, a small cusp may, but not necessary, occur due to an already mentioned sudden decrease AT.loc of the local temperature T.loc (this decrease AT.loc is better seen in Fig. 3). This is due to the fact that, if the local temperature sensors S.loc.i, as almost always happens, are near the heating element HE, the local temperature T.loc is affected by such proximity when the heating element HE is in ON state. This decrement AT.loc of the local temperature T.loc therefore depends on the architecture of the water heater and may even be non-existent or irrelevant. When the HE heating element is in OFF state, this effect, as long as it exists, ceases in a few minutes and the local T.loc temperature is uniform with that of areas furthest from the local temperature sensors S.loc.i. Even more, in the absence of turbulence and stratification of cold water, in moments far from the withdrawals with relative turbulence and considering that the convective motions due to the heating are very weak, it can be considered, for the purposes of the invention, that the whole water heater arrives at a uniform temperature and therefore that the local temperature T.loc becomes either equal to the storage temperature T.acc or in any case a direct correlation can established, so that T.loc is anyhow representative of the enthalpy content of the water.

In the previous heating phase, however, once the effect of these turbulences and stratifications has ceased, the storage temperature T.acc is equal to the local temperature T.loc minus AT.loc but both increase at the same speed. For the same reason, it can be stated that the storage temperature T.acc reached at the end of the heating is substantially equal to the switch off temperature T.off of the thermostat minus AT.loc since also the temperature sensor STR of the thermo-regulator TR is affected by its proximity to the heating element HE and is one of the s sensors S.loc.

The said sudden decrease AT.loc, much faster than the subsequent cooling due to heat losses, is a decrease by "stabilization" that could be defined as physiological. It should not be confused with a decrease in temperature due to a withdrawal, less than ever to said heat losses. Therefore, usefully, according to a preferred variant of the invention, it must be identified but in this case, since it can change from water heater to water heater and over time, it must be pre-stored then self-learned and updated.

It should be noted that the AT.loc decrement is much less evident or absent if the S.loc.i sensors are distant from the HE heating element, as for example, in the event that this is a coil of hot heat-carrying fluid HE immersed in the tank S or wound around it (see Fig. 1.b)

Some preferred procedures for self-learning

Once described the phenomena which, if present, must be taken into account to interpret the values of the local temperatures T.loc, the following section will describe in detail some preferred methods and procedures for the self-learning of the water drawing profile and the features of the water heater according to the possible variants of the invention.

The main goal is to detect all the essential data for the management of the water heater aimed at minimizing energy consumption with the same performance delivered.

Several procedures are devised all autonomous and executable simultaneously in parallel for learning the various parameters needed.

Each procedure can end with the storage of data that replace previously saved values.

Identification of the start and end time of withdrawals

Also for this identification the variation of the local temperatures T.loc.i. is checked.

According to prior art, it is possible to generically identify the start of a withdrawal by detecting a significant and sudden decrease of at least one of the local temperatures T.loc.i and its end, when the same local temperatures T.loc.i tend to stabilize.

According to the invention, however, the following method is preferred as less prone to errors:

- at recalculation time intervals 6.t.ric, the local temperatures T.loc.i(t) are recorded at the general current time t;

- if for at least one of the local temperatures T.loc.i there is a decrease at least over than the stabilization decrease AT.loc by a predefined threshold value T.thr,

- then this indicates a withdrawal in progress and the timing of start of withdrawal t.in.tap is saved,

- the time of the end withdrawal t.fin.tap saved is the one when said decrease is lower than a speed ν.δΤ of temperature variation.

The reference to the stabilization decrease AT.loc is very appropriate in order not to confuse, as indicated above, between the "physiological" drop at the end of a heating phase with a small withdrawal; the reference to the speed of decrease is then appropriate in order not to confuse with the decrease due to thermal losses. A preferred method for such verification can follow the steps below:

(a) once the water heater is installed or when it is switched on after an idle period, the status [NO_TAPPING], which the method interprets as a signal of no withdrawal in progress, is saved in a status registry TAP;

(b) at recalculation time intervals 6.t.ric, the local temperatures T.loc.i(t) are recorded at general current time t;

(c) after a time interval for detecting withdrawals 6.t.tap i.e. at the following time slot t + 5.t.tap, the local temperatures T.loc.i(t + 5.t.tap) are saved again;

(d) for each local temperature sensor S.loc.i, the change AT.loc.i.tap = T.loc.i(t) - T.loc.i (t + 5.t.tap) is calculated;

• if for one of the variations AT.loc.i.tap has AT.loc.i.tap / 6.t.tap> ν.δΤ (where ν.δΤ as mentioned, is the variation speed of the temperatures T.loc.i),

• then this is interpreted as a possible but not certain withdrawal in progress and the status [TAPPING_ALERT] is stored in a TAP status registry,

• else the status [NO_TAPPING] is saved in the TAP status registry;

(e) if for one of the variations AT.loc.i.tap (for i from 1 to s) > (T.thr + AT.loc) (where by T.thr is meant a predefined threshold value and the value AT.loc is the last saved),

• then this is interpreted as a certain water withdrawal in progress, the start time t.in.tap is saved and the status [TAPPING ON] is written in said registry,

· otherwise in said status registry TAP the status [NO_TAPPING] is saved.

As for the identification of the end of withdrawal time,

• when for each of the variations AT.loc.i.tap is valid AT.loc.i.tap / 6.t.tap <ν.δί,

· and in said TAP status registry the status [NO_TAPPING] is not already registered,

• then this is interpreted as the end of the water withdrawal that was in progress and:

• the time t.fin.tap of such positive verification as the time of the end of the withdrawal that was in progress,

• and, in said TAP status registry, the status [NO_TAPPING] are saved.

Ultimately, since ν.δΐ, although chosen as small, can be large enough that, once a withdrawal has ended, it is certain that AT.loc.i.tap / 6.t.tap <v.6t, since this recursive process with time interval of recalculation 8.t.ric, the TAP status registry stores the status [NO TAPPING] as soon as the water withdrawal has been completed, which enables the calculation of a new stabilization decrement AT.loc as will be described shortly.

With regard to the time interval for detecting water withdrawals 5.t.tap, a preferred value is 10 sec. As regards the predetermined threshold value T.thr a preferred value is 5 °C. Regarding the velocity ν.δΤ of temperature variation T.loc.i.6t a preferred value is 0.1 °C / sec.

Self-learning and updating of stabilization decrement AT.loc

If the local temperature sensors S.loc.i are very far from the heating element HE, the stabilization decrement AT.loc is, as already mentioned, substantially zero otherwise it can be quantified by calculating the decrease of the local temperatures T.loc.i in a 5-minute interval at the end of a heating phase and after having ascertained that this decrease is not due to withdrawals in progress and that the heating element is OFF.

Basically, there is a continuous monitoring of the local temperatures T.loc.i in narrow time intervals. When 5 minutes have elapsed at the end of a heating phase and in the meantime no withdrawals have occurred (i.e. when in the status registry TAP the status [NO_TAPPING] has been registered), the stabilization decrease AT.loc is equal to the decrease in the local temperature T.loc in those 5 minutes and its value is stored in place of a previous one in memory.

A preferred detailed procedure for calculating the stabilization decrement AT.loc is as follows:

a) a plausible stabilization decrease AT.loc based on laboratory tests on the specific model of the water heater (e.g. 5 °C if the sensors of local temperature

S.loc.i are near to the heating element HE) is saved at each start-up of the water heater (that is, at the first start-up and at every restart of the water heater after an idle period) as a predetermined value to be replaced with updated values after which the calculation procedure of the decrement AT.loc continues with the following steps:

b) at recalculation time intervals 5.t.ric, the local temperatures T.loc.i(t) are recorded at the current general time t;

c) after a detection time interval S.t.ril, i.e. at the following timing t + 5.t.ril the local temperatures T.loc.i (t + 5.t.ril) are again recorded; if for 5 minutes there is no decrease in the local temperatures T.loc.i then the decrement AT.loc is not considered calculable and the procedure returns to step (b), otherwise it proceeds with step (d),

d) for each local temperature sensor S.loc.i, the difference AT.loc.i = T.loc.i(t) - T.loc.i (t + S.t.ril) is calculated;

e) if the algebraic values AT.loc.i

• are negative (temperature rising instead of decreasing) either because the heating element HE is still or has in the meantime passed into ON state or by thermal inertia of the heating element which, even in the OFF state, is still hot;

· or have markedly increased due to an ongoing withdrawal, confirmed by the presence in the TAP status registry of the state [TAPPING_ON] or [TAPPING ALERT] ,

• then the decrement AT.loc is not considered calculable and the procedure goes back to step (b),

· otherwise, the stabilization decrement AT.loc is calculated by setting it equal to the average, possibly weighted, of the differences AT.loc.i = T.loc.i(t) - T.loc.i (t + δ.ΐ. ril) for i from lto s (where the average weights can take into account the greater relevance of the signals received from one or other local temperature sensor S.loc.i).

f) the calculated stabilization decrement AT.loc is then saved to replace the previously stored value.

For the detection time interval 5.t.ril between the values of the local temperatures T.loc.i to be compared, a possible value is 5 minutes.

As regards the aforementioned recalculation time interval 6.t.ric, it can be very short: the current microprocessors allow to set recalculation time intervals S.t.ric also in the order of 1 sec.

The same recalculation time 6.t.ric can also be used as a samples interval for data collection and processing of the further sections of the method according to the invention which must be described.

Since, as stated, the essential characteristic of the invention is to estimate the amount of each withdrawal based on the restoration of the energy taken by the same withdrawal, it is very appropriate that the heating process by the HE heating element is well analysed.

Preferred method for calculating the rising slope of the storage temperature T.acc heated by the heating element HE

v.T.rise indicated the variation speed of the storage temperature T.acc (i.e. the rising slope of the graph of the same temperature) when the heating element HE is in the ON state and no withdrawals are in progress.

In the absence of such withdrawals and considering that the cooling due to heat losses has negligible effect, it is obviously represented in a graph by a linear line if the heating element HE provides a substantially constant power, certainly true at least if the heating element HE is an electrical resistance but, for the purposes of the invention, such linearity is an adequate approximation also for other types of heating element HE.

It has already been noted that, in the absence of perturbations and withdrawals turbulences, the rising slope of the local temperature T.loc is substantially identical to said rising slope v.T.rise of the storage temperature T.acc and therefore said slope can be calculated in a predetermined time interval t.samp in which the heating element HE is in the ON state and it has been verified that the local temperature T.loc is growing and in a way substantially conforms to the theoretical rise curve (which, in particular, is substantially linear at least if the heating element HE delivers constant thermal power P as at least in the case of an electrical resistance).

After these conditions, the value of said rising slope v.T.rise is equal to the ratio between the difference of the local temperature T.loc values at the end and at the beginning of said interval t.samp divided by the duration of the interval itself.

Examples of significant rising slopes are the whole slope of Figure 2 and, in Figure

6, at least the circled portion of the last illustrated heating step.

Since the value of said rising slope v.T.rise is subject to variations over time also depending on environmental conditions and its calculation, as mentioned, is only possible in particular conditions of the water heater, preferably it is continually recalculated at predefined time intervals of few minutes.

Once a new value of v.T.rise considered valid (that is, in the absence of turbulence from withdrawals in tank S) is obtained:

· either this replaces the previously stored value v.T.rise.prec, which, at the beginning, is a predefined experimentally plausible value,

• or, preferably, a weighted average of the rising slope v.T.rise.med is calculated between the previous value v.T.rise.prec and the new value v.T.rise and this weighted average v.T.rise.med is stored in place of the previous value v.T.rise.prec.

For the preferred values to be attributed to the average weights, see below.

A preferred method for this verification can be provided in the following initial conditions and recursively can develop in the following steps:

• in a number n.r of memory registries MR for detecting rising slope are recorded in a timing order corresponding local temperature values

T.loc(t.int) detected at the same time intervals 6.t.rise of detecting rising slope where:

• in the first memory registry MR.(t.in) a first value is registered, T.loc(t.in), the one read at an initial time t.in,

· in the last memory registry MR.(t.fin ) a last value T.loc(t.fin) is recorded at the final time t.fin which, of course, is equal to t.in + (n.r - 1) * 6.t.rise (where obviously t.fin - t.in = t.samp);

• in the intermediate memory registries MR.(t.int), T.loc(t.int) values of the generic intermediate timings t.int are recorded;

· the n.r memory registries, at the first start of the water heater or whenever it switched back on, contain predefined values, e.g. also null.

(a) At intervals of time 5.t.rise

• the T.loc value is measured,

· in said n memory registries, which constitute a sliding window, the time sequence of all the stored T.loc(t) values is scrolled one step abandoning the older T.loc(t.in) value,

• in the last memory registry MR.(t.fin) the last value T.loc(t.fm) read at the final timing t.fin, the T.loc value just read is registered;

(b) if said last value T.loc(t.fin) read results <T.loc(t.in),

• then go back to step (a);

(c) the offset scost.rise from the line identified by the points [T.loc(t.in), t.in] and [T.loc(t.fin), t.fin] is calculated for each intermediate value T.loc(t.int) measured at its intermediate time t nt;

(d) if the offsets scost.rise for one or more intermediate values T.loc(t.int) exceeds a predetermined threshold value scost.rise.max,

• then go back to step (a);

• otherwise it is calculated, for said rising slope,

v.T.rise = [T.loc(t.fin) - T.loc(t.in)] / (n.r-1) * 6.t.rise;

(where (n.r-1) * 6.t.rise = t.fin - t.in = t.samp)

(e) the weighted average v.T.rise.med = (v.T.rise.prec * w.l + v.T.rise * w.2), where w.1 and w.2 are suitable weights, is calculated and saved replacing the previous v.T.rise.prec in;

(f) the procedure returns to step (a).

Preferred method for calculating the rising slope of the storage temperature T.acc when several heating elements HE of different types are present

As already anticipated and explained further on, the case where water heater has several types of heating elements HE is not uncommon. The procedure described above can then be repeated for the calculation of the rising slopes v.T.rise. i specific for each combination of heating elements HE simultaneously in ON state and then at step (e) is stored also for which group of heating elements HE in the ON state and for which local temperature range T.loc the calculation has been carried out. The calculation is interrupted if one of the heating elements HE simultaneously in ON status ceases to be in that state.

Then there are heating elements HE, whose rise speed cannot be considered constant throughout the local temperatures range T.loc where these can be operational. For them, then, the rising slope can be represented by a sequence of several consecutive linear sections (see Figure 9). The second section, and possibly further again, those following could be identified, by way of example, as follows:

• after a predefined minimum time t.samp, considered sufficient to determine the v.T.rise relevant to the first section, it is checked whether the following T.loc(t.int) values, read in generic subsequent t.int timings, continue to remain within the predetermined threshold value scost.rise.max

· as soon as this is no longer verified, the procedure described is repeated to detect a possible value v.T.rise.i.2 relating to the second segment, also memorizing the local temperature range T.loc of validity.

Of course, the calculations and subsequent updates of all possible rising slopes v.T.rise.i are only possible starting from the timing when, in the water heater, the necessary ON / OFF status combinations of the heating elements HE available are occurring and in the specific local temperature ranges T.loc of validity.

Preferably said appropriate weights w.l and w.2 are respectively equal to ¾ and

¼.

Said equal time intervals 5.t.rise may preferably be of the duration of 1 minute and the number n.r of local temperature values T.loc detected may be equal to 25. Thus a preferred sample time is t.samp = 25 minutes

Said threshold value scost.rise.max may be as small as the said local temperature sensors S.loc.i. E.g. it can also be equal to 0.1 °C although much higher values, e.g. 2 - 3 °C are more than enough.

This results in a very accurate calculation of the v.T.rise rising slope.

The procedure assimilates to a linear slope any curved rising slope provided that the deviation from the linearity does not exceed a predefined value; e.g. the aforementioned 2 - 3 °C.

Calculation of heat losses

Also the thermal energy lost to heat losses, herein called decrement velocity for thermal losses 6T.loss, is calculated with criteria similar to those used for the withdrawal temperature drop AT.tap.

In fact, even the temperature T.acc.2 to which the water heater drops by cooling before the heating element HE is reactivated, can be calculated indirectly.

With reference to Figure 7, at the end of a heating phase in which the heating element HE goes into the OFF state (in a time tl known to the microprocessor), the storage temperature T.acc has the value T.acc.l (also known). Then a cooling phase starts up to a time t2, (also known to the microprocessor), in which the HE heating element changes back to ON state.

At the end of the ON state, after a known interval At. on and once the storage temperature T.acc.3 (also known) reached, the temperature T.acc.2 which the water heater had reached by cooling is finally calculated as T.acc.2 = T .acc.3 - v.T.rise * Δΐ.οη. As for the velocity (or rate) of the storage temperature T.acc, indicative of the heat losses, given (T.acc.l- T.acc.2) = AT.loss, it is v.AT.loss = (T.acc.l- T.acc.2) / (t2 - tl).

Also said rate of reduction of the storage temperature 6T.loss is calculated continuously because it is subject to variations also for environmental reasons and the new value is stored in place of a previous one either as such or after being averaged with the previous one.

The calculations according to the process just mentioned are valid if during the whole period there have been no perturbations from withdrawals, i.e. if the status [NO_TAPPING] is stored in the TAP status registry during the whole process otherwise the procedure stops and starts again at a subsequent time in which the heating element HE goes into the OFF state and the water heater is in the status [NO_TAPPING].

In more rigorous terms, a possible procedure is as follows:

(a) as said heating element HE has gone into the OFF state, a time tl and the corresponding value T.acc.l of said storage temperature (T.acc) are recorded;

(b) as said heating element HE switches back to the ON state, the time t2 is recorded;

(c) as said heating element HE switches again to the OFF state, the time t3 and the corresponding value T.aec.3 of said storage temperature T.acc are recorded;

(d) by the formula T.acc.2 = T.acc.3 - v.T.rise * (t3 - 12) T.acc.2 is calculated where T.acc.2 is the value assumed for said storage temperature T.acc at the time t2;

(e) it is set v.AT.loss = (T.acc.1- T.acc.2) / (t2 - tl) where v.AT.loss is the value assumed for the cooling speed of said storage tank S;

(f) a weighted average is calculated between the value v.AT.loss just calculated and the value in memory, the new value is stored in place of the previously homologous value in memory;

(g) during the whole process, if the status [NO_TAPPING] is not stored in the status registry (TAP), go back to step (a).

Saving data on the water drawing profile

In principle, for each collected withdrawal could be stored separately at least the time of start of withdrawal t.in.tap and the corresponding drop in temperature of withdrawal AT.tap if not also the time of end of the t.fin.tap withdrawal, but according to the invention, the following method is preferred which, by aggregating more information, takes up much less memory space while recording sufficient data for any management method aimed at reducing thermal losses while ensuring the performance required by the user. The method foresees to divide the withdrawals cycle into a number nr.h of consecutive time intervals Int, all of same predetermined duration A.t.int. Thus the j-th time interval Int starts at the timing tj = (j -1) * A.t.int after the start of the cycle of withdrawals. Since the cycle has almost always the duration nr.d = 7 days, the preferred duration of said predetermined time intervals is A.t.int = 1 h. According to the method, for each of the nr.h time intervals Int of the entire water withdrawals cycle, both a fictitious withdrawal representing the total of the withdrawals found in the same interval and a corresponding fictitious time of the beginning of the same water withdrawal is calculated, after which said data can be stored as such.

Preferably, the process of clustering and storage of data according to this last variant takes place in the following way.

For each interval of time Int, during which k withdrawals were recorded, each reported by storage temperature reductions AT.tap.i occurred at the corresponding times Δ-t.i from the beginning of the Int interval itself (with i from 1 to k), the following values are calculated:

• a fictitious withdrawal AT.tap.tot equal to the sum of all the storage temperature reductions AT.tap.i, i.e.

[AT.tap.tot =∑,(AT.tap.i) for i from 1 to s] ;

• a withdrawal fictitious timing A.t.fict taken as representative of the individual timings Δ-t.i in which the aforementioned k withdrawals occurred starting from the beginning t.j of said time interval Int and equal

• to the weighted average value of the actual timings Δ.ί.ί in which each withdrawal AT.tap.i has been recorded if there have been withdrawals, i.e.

[A.t.fict =∑,(AT.tap.i * A.ti) /∑(AT.tap.i ) for i from 1 to s if∑(AT.tap.i

≠ )l

• to half of the time interval A.t.int of duration of the interval Int if there have been no withdrawals, i.e.

[A. fict = A.th / 2 if∑,(AT.tap.i = 0)].

The data thus aggregated can be stored as such and possibly continuously updated during one or more cycles following a first one, then storing the same as such but preferably it is also possible to take into account the consumptions found in the homologous intervals of one or more previous cycles by means of weighted averages or filtering operations so as to attenuate variations of user behaviour that could be occasional and non-definitive. In this case it is envisaged to keep the data relating to a number of cycles immediately preceding the current cycle plus the data of the current cycle stored in a M.cyc memory. The M.cyc memory has a sliding window in the sense that at the end of each cycle all the data flow in the memory registries; the data of the older cycle are lost while the data of the other cycles take the place of those of the cycle to each of them preceding. Preferably the number n.cyc.prec of previous cycles reaches up to 5.

In the following way, it is possible to take into account the data recorded in one or more previous cycles in the homologous interval Int:

• if the calculated fictitious withdrawal AT.tap.tot is higher than that present in the memory of the just preceding cycle for the same interval Int, both the fictitious withdrawal AT.tap.tot and the fictitious timing A.t.fict are stored as such,

· else if the calculated fictitious withdrawal AT.tap.tot is lower than that present in the memory of the just preceding cycle, the following is saved in memory:

• as size of the withdrawal, the fictitious value AT.tap.tot as filtered with the values stored in the homologous intervals Int of the most recent n.cyc (where the quantity of cycles n.cyc used in the filter is equal to 0 if the cycle in question is the first after start-up and increases by one unit for each subsequent cycle up to the maximum value n.cyc.prec);

• as a fictitious timing A.t.fict, the weighted average between the new fictitious timing A.t.fict and the one that was in the memory of the cycle just before.

It is clear how, according to this preferred method of registration:

• the consumption data are recorded as such during the first cycle after startup, which finds all the memories empty;

· an increase in consumption is immediately accepted to satisfy users at the following cycle;

• a reduction in consumption is taken into account gradually, thanks to filtration, and is accepted as a new habit only if it is actually repetitive.

The set of procedures described so far for the self-learning method is able to get all the information on the water drawing profile and to characterize the water heater as regards heating and cooling speed by the only reading of the temperatures of one or more local temperature sensors S.loc.i, associated with the timings of said readings and with appropriate processing of such data by the microprocessor MP.

These procedures can be implemented simultaneously and cyclically to update the data detected.

Detection of the minimum temperature of use

Although temperature sensors are not provided for the hot water outlet OUT, it is useful to detect and store the minimum usage temperature T.acc.min accepted by the user; it can be considered equal to the minimum value found for the already defined storage temperature value T.acc at the end of water withdrawal T.acc.23. In fact, it is reasonable to assume that a withdrawal is interrupted by the user when the water begins to exit at unsatisfactory temperature. In this way it is possible to periodically replace in memory any predefined value T.acc.min (for example = 40 °C) with a value actually measured.

It has been seen that many of the procedures described provide for the use of data stored in memory registries previously at the beginning of the same procedures; in some cases, these can have predefined values once and for all, in others they are modified and stored in place of the previous ones or by the same procedure or by others that take place sequentially or in parallel. In any case, when the water heater is switched on, plausible values are stored, i.e. compatible with experimental measurements and which make the various procedures described feasible and reliable, while recursive recalculation of the stored data refines the results of the various procedures that use data calculated from the others.

Detection of rising slopes

So far, in the discussion it has been made for the most part only reference to heating bodies HE whose rising slope v.T.rise are substantially linear, as certainly it is for a group of electric resistances or for exhaust pipes of a storage gas water heater; these in fact deliver a constant thermal power whatever the local temperature T.loc of the water that heat, if anything, may lead to potentially damaging overheating.

Often, however, said heating elements HE can be of type more energy efficient; e.g. it can be the coil of a space heating system or, much more widely, the condenser of a heat pump HP.

Typically, since the heat pump HP has limits, at least from the practical point of view, at the maximum condensation temperature attainable and the thermal power delivered P decreases as the same condensation temperature rises, the rising slope of the storage temperature T.acc it is not linear, but it is a regularly increasing curve with an asymptotic trend towards this maximum condensation temperature to be understood as limiting temperature T.lim.

Likewise, also a heating element HE consisting of the coil of a room heating system produces a rising slope of the same type towards a temperature limit T.lim which is the one, established in the boiler, of the heat transfer fluid and also here the deliverable thermal power P decreases directly with the temperature difference T.lim - T.acc.

But in reality, for control improving opportunities, even a group of electrical resistances can have a variable thermal power P, for example, if it is considered appropriate to reduce this power P as soon as the programmed T.off switch off temperature is approached.

Said heating elements HE can be referred to as "heating elements HE with thermal power P decreasing as the temperature rises" or, more briefly, "with an asymptotic rising slope".

With known mathematical techniques more or less sophisticated it is possible to assess whether the asymptotic rising slope has a regular growth, and therefore reliable and apt to be saved, or shows discontinuities or abrupt slope variations (i.e. discontinuity in the first derivative which is the velocity v.T.rise) attributable to perturbations due to withdrawals in progress.

In a general sense, if the rising slope has such regular growth, a function of the type T.loc = f (t) can be obtained (from which we derive T.acc = T.loc - AT.loc) which establishes the T.acc storage temperature which is reached at any time t. By way of example, a general procedure, herein summarily indicated, could provide the following steps whose development in detail is within the reach of computer technicians:

(a) at equal time intervals 8.t.rise, for a time preferably of at least 25 minutes, the status registry TAP is monitored and the local temperature values T.loc are measured and stored in corresponding n.r memory registries MR;

(b) if during the process the status [TAPPING-ON] or [TAPPING-ALERT] is detected, the procedure returns to step (a) in fact the local temperatures T.loc read are considered not significant because they are affected by withdrawals else parameters of a function T.loc = f (t) best approximating the value pairs (T.loc; t), where t is the timing of detection, is estimated;

(c) these parameters are saved until the following measurement occurs;

Since the trend of the asymptotic rising slope is characteristic for each type of heating element HE (e.g. comparable to a simple linear slope or a curve that has an exponential component, this facilitates the definition of the mathematical form of the function T = f (t) reducing the parameters to be calculated.

Most preferably, however, an asymptotic rising slope is also characterized and saved with criteria similar to those already indicated for the linear case.

In fact, in most cases (see fig. 8) the heating element HE with asymptotic rising slope may simply be equated with a heating element HE with linear slope already dealt with because the range of the storage temperature T.acc in which it is used allows such simplification.

To this goal it is sufficient that the already defined predetermined threshold value scost.rise.max is within predetermined values such as the already indicated 2 - 3 °C. In other words, the method simply ignores that the rising slope is curved and assimilates it to a linear slope.

In other cases (see Fig. 9), the heating element HE with asymptotic slope is used in such a wide range of temperatures and / or time that the slope cannot be accurately represented by a single line but can still be represented by two or more consecutive linear segments: the first one valid within a first local temperature range T.loc from Tloc.l to T.loc.2, the second one from Tloc.2 to T.loc.3 and so on. In this situation, determined and saved a first value v.T.rise.l within said sample interval, t.samp, the procedure continues to verify up to which value T.loc.2 the local temperature T.loc rises continuing to stay within the said predetermined threshold value scost.rise.max. As soon as this condition is no longer verified, the procedure for calculating the rising speed v.T.rise is repeated and a second value T.rise.2 valid starting from this value T.loc.2 is calculated, and so on.

The values v.T.rise.l, v.T.rise.2, etc. are then saved together with the temperature ranges T.loc.1 ÷ T.loc.2, T.loc2 ÷ T.loc.3, etc. within which they are valid.

Many storage water heaters then provide the co-presence of at least two types of heating elements HE, one of which is usually always a group of electrical resistors, to be used simultaneously and/or sequentially according to various methods established by the control program and aimed at savings (energy or economic), others to assure the service in case of urgency.

In this case, it is necessary to calculate the function T.loc = f (t) to be applied when the heating is by both or any other further heating elements HE. Since the heating curve of an electric resistance is linear, it will be similar to a linear one also the slope given by the combination electric resistance plus heating element HE with asymptotic curve in the temperature fields where the latter can be considered to a linear slope.

A complete characterization of the heating process is then obtained by memorizing the various values of rising slopes v.T.rise.l associated with the ON / OFF states of the heating bodies HE and at the temperature ranges for which they have been measured and considered valid.

In conclusion a table of v.T.rise values can be stored, as shown below, to be considered as an example and not exhaustive: v.T.rise elements HE in ON condition validity v.T.rise.l electrical resistance all temperatures v.T.rise.2 condenser PC from T.loc.l to T.loc.2 v.T.rise.3 condenser PC from T.loc.2 to T.loc.3 v.T.rise.4 resist. Electr. + cond. PC from T.loc.l to T.loc.2 v.T.rise.5 resist. Electr. + cond. PC from T.loc.2 to T.loc.3

v.T.rise.i generic combination of HE from T.loc ... to T.loc ...

Then for the increase of temperature ΔΤ during the heating period of duration At. on the formula

ΔΤ =∑i [v.T.rise.i * ( t_i+, - 1)] with i from 1 to k

where

· k is the total number of combinations of heating elements HE which, alone or in combination with others, are in ON status at certain intervals T.loc; • each of said k rising speeds v.T.rise.i is specific for those of said heating elements HE in ON status and for the range of said local temperatures T.loc simultaneously read;

· the continuous time interval from da ti to ti+i , is equal to the entire period of duration of heating At.on.

is valid.

The above mentioned formula has a general value. Obviously not all the possible k combinations are activated in each heating phase; for those that do not activate it is simply t = t₊ i, that is (t_i+, - ) = 0.

Each v.T.rise.i is a value pre-stored and fixed or updated with data measured subsequently, for example starting from an initial learning phase before the water heater becomes operational to the user's service; in this case it is preferable to first characterize v.T.rise.i for the heating bodies with lower operating temperature ranges; the rising curves v.T.rise.i due to the combination of two or more heating bodies HE may simply be the sum of the individual rising curves v.T.rise.i relevant to each HE heating body HE when individually in ON state.

This laborious evaluation of the ΔΤ, a parameter necessary to appreciate the amount of a withdrawal just taken, depends on the fact that the management method, whatever it is, which in the meantime regulates the storage temperature T.acc according to the drawings profile can decide autonomously which and / or how many of the available heating elements HE it must activate according to the circumstances, favouring at his discretion the urgency of heating or the saving. Therefore, at least starting from when the water heater is controlled by the management method, aimed at satisfying the performance according to its own current criteria, the learning method according to the invention cannot but adapt to it. However, once it knows, moment for moment, the ON / OFF status of each heating element, it is able to calculate the ΔΤ according to the previous formula. Of course, some of the described characterization processes may fail due to the passage in the state registry TAP to the [TAPPING ON] or [TAPPING_ALERT] states, as provided for in the described procedures, but the problem is momentary since very preferably the procedure may be iterative, so that it will be successful in a subsequent situation; at least in night periods when withdrawals are substantially absent.

To summarize some basic characteristics of the invention,

• the storage temperature T.acc, only one representative element of the energy content of the tank S, is considered equal to the local temperature T.loc in time periods not disturbed by withdrawals or heating whereas in the presence of such perturbations it is calculated from what has been learned in the periods without perturbations;

• the methods described:

• are preferably recursive so as to continuously update the characterization of the water heater and / or detect modifications to the drawings profile;

• use also pre-stored values such as the stabilization decrement AT.loc or the rising slopes v.T.rise.i but can in turn correct and refine at least some of these values by self-learning;

• can be executed simultaneously;

• make available to any management method suitable for this type of water heater the data necessary for the best management of the management method.

The self-learning method described can be used for any optimized management method, which is sufficient to know the extent of the withdrawals (expressed as a reduction in the storage temperature), the time in which they start and the available energy resources. Advantageously, optimized management methods derived from what has been described in the cited documents EP 2362 931 Bl or EP 2 366 081 Bl could be used with which it is possible to establish when and for how long the heating element HE should be set to ON and which it must be the T.off switch off temperature to satisfy the following withdrawal or group of withdrawals.

Finally, there are also models of storage water heaters that provide more than one storage tank S; the methods of the invention remain always applicable considering the behaviour inside each storage tank S one by one.

Claims

Claim 1. Method for learning, in value and time, the profile of hot water withdrawals in a storage water heater comprising a storage tank (S) of the water to be heated

where said profile cyclically repeats in predetermined time intervals and where the thermal energy content is considered represented by the storage temperature (T.acc) to be intended as the average temperature of the water in storage tank

and said water heater is equipped with:

- one or more heating elements (HE);

- a thermo-regulator (TR) of the electronic type able to switch said one or more heating elements (HE) from the state of OFF to ON and vice versa when a temperature sensor (STR) senses the attainment respectively of the switch off temperature (T.off) and switch on (T.on) with T.on = T.off - Aist,

- one or more local temperature sensors (S.loc.i; S.loc.i, STR) placed preferably near the cold water inlet (IN) of said storage water heater and of said heating element (HE),

- a microprocessor (MP) able at least of:

- knowing the said switch off temperature (T.off)

receiving the corresponding signals representative of the local temperatures (T.loc.i) from said temperature sensors (S.loc.i; S.loc.i, STR);

calculating the local temperature (T.loc) equal to the average possibly weighed of said local temperatures (S.loc),

where the weights of said average possibly weighed are set by a skilled person in the art based on the position of said local temperature sensors S.loc.i and the model of the water heater

- measuring the passing of time, saving the durations At. on of said ON states of said heating elements (HE),

saving said local temperatures (S.loc.i; S.loc.i, STR) in association with the time of their reading;

characterized in that it detects and writes

in a special status registry (HE-ON / OFF) the current states [ON] or [OFF] of said heating elements (HE),

and in a special status registry (TAP) the status [NO TAPPING] or [TAPPING-ON] or [TAPPING- ALERT], indicative respectively of the absence, occurrence or probable occurrence of withdrawals, state detectable from the reduction of one or more of said local temperatures (S.loc.i; S.loc.i, STR) beyond a predetermined threshold (T.thr + AT.loc) and / or of a predetermined speed (v.5t);

and by:

the reduction (AT.tap) of the storage temperature (T.acc) considered representative of the amount of a withdrawal (or of a group of small consecutive withdrawals),

- which brings the storage temperature (T.acc) at a value lower than the current value of said switch on temperature (T.on) at the time t2, which is considered the beginning time of said withdrawal,

- and which consequently triggers a heating step with the switching of one or more of said heating elements (HE) to the state of [ON] substantially in said same beginning time t2 and continuously at least until the subsequent time t3,

is calculated according to the formula

AT.tap = T.acc.iniz - (T.acc.fin-v.T.rise * 6t)

where for the values T.acc.iniz, T.acc. fm,v.T.rise, 5t in the formula, the following relations are valid:

- T.acc.iniz = local temperature T.loc.2 read at the time t2 of the beginning of withdrawal,

- if T.loc.3 - T.loc.2 > AT.q then

T.acc.fm = T.acc.3 and 5t = 5t.HE.on

and v.T.rise * 8t is the increment ΔΤ of the storage temperature T.acc due to the heating lasting 5t.HE.on

- if T.loc.3 - T.loc.2 <AT.q then

T.acc.fm = T.acc.2 and 5t = (δΐ.οη.Ι + 5t.onl.fict)

- v.T.rise is the speed of variation of the storage temperature T.acc by said heating elements (HE) in the [ON] state read in a memory registry

and where, in turn

- T.acc.3 = T.loc - AT.loc

- 6t.HE.on is the duration of said heating phase, between said instants t2 and t3 when T.loc.3 - T.loc.2 > AT.q

- δΐ.οη.1 is the duration of said heating phase between said instants t2 and t3 when T.loc.3 - T.loc.2 <AT.q

- δΐ.οηΐ .fict = (T.acc.2 - T.loc.3) / v.T.rise.loc

- T.acc.2 is assumed equal to T.loc.

2

- T.loc.3 is the local temperature read at time t3

- AT.loc is a stabilization decrement read in a specific memory;

- v.T.rise.loc is the angular coefficient of the tangent line at time t3 the curve of the rising slope of the local temperature T.loc,

- AT.q, index of stability, is an empirical parameter, predefined and pre-recorded, depending on the model of the water heater and set by a skilled person in the art

said calculation being considered valid and executable provided that during said whole phase of [ON] status, the status [NO_TAPPING] is always stored in said status registry (TAP).

Method for learning the drawings profile according to the previous claim characterized in that said index of stability AT.q is equal to 0.

3. Method for learning the drawings profile according to at least Claim. 1 characterized in that

said increase ΔΤ of the storage temperature (T.acc) in said heating phase of duration 6t.HE.on is calculated by the sum

ΔΤ =∑i [v.T.rise.i * ( n - ti)] with i from 1 to k where

- k is the number of combinations of said heating elements (HE) which can be simultaneously in ON state;

- by v.T.rise.i is meant each of the k rising speeds v.T.rise.i specific for the combination of those among said heating elements (HE) in ON state and for the range of said local temperatures T.loc simultaneously read;

- the sum of consecutive intervals (ti+i - ti) is equal to the entire heating period of with duration Δί.οη;

- said rising speed values v.T.rise.i are read from a memory;

and where said sum ΔΤ =∑i [v.T.rise.i * (ti+i - ¾)] is simplified to the formula ΔΤ = v.Trise * Δί.οη if k = 1 , i.e. if only one of said heating elements (HE) is present or active.

4. Method for learning the drawings profile according to the previous claim characterized by the fact that

the value of said k temperature variation speeds v.T.rise.i is continuously recalculated at predetermined time intervals S.t.rise a few minutes long starting from when the corresponding ON/OFF status combinations of said heating elements (HE ) are fulfilled and in the status register (TAP) the status [NO TAPPING] is registered and according to the following procedure:

- the most recent n.r local temperature values (T.loc) measured at equal predefined time intervals 6.t.rise are recorded in the same number of memory registries (MR)

- it is checked that said local temperature values (T.loc) in the selected sampling time interval t.samp = 6.t.rise * (n.r - 1) grow linearly within a predefined deviation scost.rise.max,

in the status register (TAP) the status [NO TAPPING] continues to be registered,

the ON / OFF status of said heating elements (HE) remains unchanged;

- if this check is positive,

the ratio between the increase of said local temperature (T.loc) and said sampling time interval t.samp = 5.t.rise * (nr - 1) is taken as the value v.T.rise.i for said temperature variation speed, the last stored value v.T.rise.i for said combination i of said heating elements (HE) in ON state is replaced with a weighted average of the same with the new calculated value,

- if this check is negative

the procedure goes back to the beginning.

Claim 5. Method for learning the drawings profile according to Claim 3 characterized in that

said rinsing speeds v.T.rise.i are directly detected exclusively for each of said heating elements (HE) in the ON state while the remaining said rising speeds v.T.rise.i of any other specific combination are obtained from the sum of said v.T.rise.i speeds directly detected.

Claim 6. Method for learning the drawings profile according to previous Claims

4 or 5 characterized by the fact that

- after said minimum preset time t. samp, considered sufficient for the determination of a first heating speed v.T.rise.1 relevant to a first section of the heating curve, it is continued to check up to which value T.loc.2 the increase of said local temperature (T.loc) continues to remain within the predetermined threshold value scost.rise.max, - as soon as this is no longer verified, the procedure described is repeated to detect a new value v.T.rise.2 relevant to a second section, and so on,

- the values v.T.rise.l, v.T.rise.2, etc. are saved together with the temperature intervals T.loc.1 ÷ T.loc.2, T.loc2 ÷ T.loc.3, etc. within which they are valid.

Claim 7. Method for learning the drawings profile according to any previous

Claim from 4 onwards characterized by the fact that

said time intervals S.t.rise are equal to 1 minute and said sampling time interval t.samp is at least equal to 25 minutes.

Claim 8. Method for learning the drawings profile according to any previous claim characterized by the fact that

to identify the start time of each withdrawal,

- at recalculation time intervals 6.t.ric, the local temperatures T.loc. i(t) are recorded at the current general timing t,

- if for at least one of the local temperatures (T.loc.i) there is a decrease higher of said stabilization decrement (AT.loc) by at least a certain predefined threshold value T.thr,

- then this indicates a withdrawal in progress and the moment of start of withdrawal t.in.tap is stored

where

- the values of said stabilization decrement (AT.loc) and the predefined value of said threshold T.thr are read from specific memory registries.

Claim 9. Method for learning the drawings profile according to the previous claim further characterized by the fact that

(a) at the installation of the water heater or any re-start after an idle period, in a status register (TAP) the status [NO_TAPPING] is stored

(b) at recalculation time intervals of duration 5.t.ric, the said local temperatures T.loc.i(t) are registered at the current general instant t,

(c) at a subsequent time t + 6.t.tap, the local temperatures T.loc

T.loc.i(t + δ-t.tap) are recorded again;

(d) for each of these local temperature sensors (S.loc.i) the variation AT.loc.i.tap = T.loc.i (t) - T.loc.i (t + 5.t.tap) is calculated;

if for one of the above variations AT.loc.i.tap, AT.loc.i.tap/5.t.tap > ν.δί,

then this is interpreted as a possible but not certain withdrawal in progress and the status [TAPPING_ALERT] is stored in said status register (TAP),

else the status [NO_TAPPING] is stored in the status register (TAP);

(e) if, at least for one of the said variations AT.loc.i.tap (for i from 1 to s), AT.loc.i.tap> (T.thr + AT.loc);

then this is interpreted as a certain withdrawal in progress, the start time t.in.tap is stored and the status [TAPPING_ON] is written in said status registry (TAP)

else the status [NO_TAPPING] it is stored in said status register (TAP).

Claim 10. Method for learning the drawings profile according to any previous claims characterized in that

the end time t.fm.tap of each withdrawal is considered to be the one in which the decrease of each of said local temperatures (T.loc.i) is less than a predetermined speed ν.δΤ of variation.

Claim 11. Method for learning the tapping profile according to previous claims characterized by the fact that

when for each of the aforementioned variations AT.loc.i.tap there is AT.loc.i.tap / 8.t.tap <v.6t and in said TAP status register, it is not already registered the state [NO_TAPPING] then are stored:

- the timing t.fin.tap of this positive check as the end of the withdrawal that was in progress

- and, in said status register (TAP), the status [NO_TAPPING]

Claim 12. Method for learning the drawings profile according to any previous claim, excluding 2 characterized in that said stabilization decrement (AT.loc) is calculated and updated according to the following steps:

(a) storing of a predetermined value of said stabilization decrement (AT.loc) at start-up of the water heater;

(b) at recalculation time intervals 5.t.ric, recording of the T.loc.i(t) value of said local temperatures (T.loc.i) to the current general timing t;

(c) after a withdrawal time interval 5.t.ril, recording of the T.loc.i value (t + 5.t.ril) of said s local temperatures (T.loc.i) at the current general timing t +5.t.ril;

(d) for each of these local temperature sensors (S.loc.i) calculating the difference AT.loc.i = T.loc.i (t) - T.loc.i (t + 5.t.ril );

(e) if

- the algebraic AT.loc.i values are negative or said status register (TAP) indicates the states [TAPPING_ON] or [TAPPING_ALERT] , returning to step (b),

- else, calculating, for said stabilization decrement (AT.loc) of the weighted average of the differences AT.loc.i = T.loc.i (t) - T.loc.i (t + δ-t.ril) for i from 1 to s;

(f) saving the new value of said stabilization decrement (AT.loc) in place of the homologous value previously in memory.

Claim 13. Method for learning the drawings profile according to any previous claims characterized in that

for the calculation of the cooling speed (v.AT.loss) of said storage temperature (T.acc) considered representative of the thermal losses, the following steps are carried out

(a) saving a timing tl wherein said one or more heating elements (HE) have gone into the OFF state and the corresponding T.acc.l value of said storage temperature (T.acc);

(b) saving the timing t2 wherein at least one of said heating elements (HE) returns to ON state;

(c) saving the timing t3 wherein said heating element (HE) returns to OFF again and the corresponding T.acc.3 which is read for said storage temperature (T. acc);

(d) calculating T.acc.2 = T.acc.3 - v.Trise * (t3 - 12) where T.acc.2 is the value assumed for said storage temperature (T.acc) at time t2;

(e) calculating v.AT.loss = (T.acc.l - T.acc.2) / (t2- tl) where v.AT.loss is the value assumed for the cooling speed of said storage tank S;

(f) calculating a weighted average between the value v.AT.loss just calculated and the value in memory, and storing the new value in place of the homologous value previously in memory;

(g) during the whole process, if the status [NO_TAPPING] is not stored in the status register (TAP), going back to step (a).

Claim 14. Method for learning the drawings profile according to any previous claims, characterized in that

- said cycle of withdrawals with a duration of nr.d days is divided into a number nr.h of predetermined consecutive time intervals (Int) of equal length A.t.int;

- for each of said nr.h hour time intervals (Int) for the entire withdrawal cycle a fictitious withdrawal (AT.tap.tot) is calculated and saved, representing the totality of the withdrawals found in the same interval and a corresponding fictitious timing (A.t.fict) of the beginning of said fictitious withdrawal;

Claim 15. Method for learning the drawings profile according to at least the previous claim, characterized in that said fictitious withdrawal (AT.tap.tot) is set equal to the sum of the storage temperature reductions

(AT.tap.i) of all said k withdrawals

and said a fictitious timing (A.t.fict ) is equal

- to the weighted average value of the actual timings (A.ti) in which each of said withdrawals (AT.tap.i) was recorded, if there were withdrawals,

- to the half of duration of said time interval (A.t.int) if there have been no withdrawals.

Claim 16. Method for learning the drawings profile according to any previous claims characterized in that said learning of said withdrawals continues during one or more cycles subsequent to a first.

Claim 17. Method for learning the drawings profile according to at least the , previous claim, characterized in that

the data on said withdrawals found in the cycle in progress are saved taking into account also the data found in the homologous intervals of previous cycles by weighted averages or filtering operations between the data of said cycle in progress and the corresponding ones of said one or more previous cycles.

Claim 18. Method for learning the drawings profile according to at least the previous claim, characterized in that

- a number n.cyc.prec of previous memories (M.cyc.prec) is provided where are saved the data on said withdrawals related to n.cyc.prec consecutive cycles just preceding the one in progress are stored;

- if any of the said fictitious withdrawals (AT.tap.tot) calculated is greater than that present in the memory at the time (M.cyc.corr) in the homologous interval Int, both the fictitious withdrawal AT.tap.tot and the fictitious timing A.t.fict replace the homologous values present in the memory at the time (M.cyc.corr); if instead said fictitious withdrawal AT.tap.tot is lower than the one in the memory from the just previous cycle, it is registered in the memory:

as size of the withdrawal, the fictitious value AT.tap.tot but filtered with the values stored in the homologous intervals Int of the most recent cycles n.cyc,

as a fictitious timing A.t.fict, the weighted average between the new fictitious timing A.t.fict and that was found in the memory of the cycle just before,

while the n.cyc quantity of cycles used in the filter is equal to 0 if the cycle in question is first after power up and increases by one unit for each subsequent cycle up to the maximum value n.cyc.prec.

Claim 19. Method for learning the drawings profile according to any previous claims characterized in that for each of said withdrawals are recorded at least the value of said withdrawal timing t.in.tap and the corresponding said withdrawal temperature drop AT.tap

Claim 20. Method for learning the drawings profile according to previous claim characterized in that the value of said end-of-withdrawal time t.fm.tap is further memorized by each of said detected withdrawals.