EP4639436A1

EP4639436A1 - Methods, apparatuses and computer programs for assessing an energy efficiency of warm water tanks

Info

Publication number: EP4639436A1
Application number: EP23837282.5A
Authority: EP
Inventors: Moritz HOEFERT; Joerg Krogmann; Andreas Eichhorn; Dennis Patrick SKOPP; Matthias WOHLMUTH; Bruno SCHIEBLER; Christian Raphael Stephan EIGL
Original assignee: BASF SE
Current assignee: BASF SE
Priority date: 2022-12-20
Filing date: 2023-12-20
Publication date: 2025-10-29
Also published as: MX2025007219A; CN120380492A; WO2024133572A1

Abstract

A computer-implemented method for assessing an energy efficiency of a tank, in particular a warm water tank, said tank being suitable for holding a warm liquid/water comprises: receiving tank data indicative of geometrical and material properties of the tank; Determining, by a trained artificial neural network (ANN) and based on the received tank data, a heat loss of the tank; generating an energy efficiency score value (EESV) based on a preset method for calculating the energy efficiency according to a test standard for energy efficiency and as a function of the determined heat loss; and outputting/displaying the EESV. Aspects include apparatuses, pro- cessing devices and computer programs for implementing/executing the method. The method allows for virtually generated energy efficiency labels (VELA) for insulated tanks.

Description

Methods, apparatuses and computer programs for assessing an energy efficiency of warm water tanks

This disclosure relates to a method for assessing an energy efficiency of a tank, in particular an isolated tank for a fluid. Further, apparatuses and devices involved in carrying out the method are disclosed. Often, warm water needs to be stored in an isolated tank at a specific temperature and can be used for heating purposes or as warm water itself. The disclosure, in particular, relates to aspects of assessing an energy efficiency of an isolated warm water tank.

Energy-related products and devices such as heating apparatuses may be ranked according to their efficiency in maintaining their function based on an energy loss or a dissipation during their operation. For example, hot water storage tanks, boilers or collectors are considered more energy efficient if they show a reduced heat loss compared to other similar or same appliances.

Some governments define energy rating labels to facilitate the consumer's choice for less energy consuming appliances. An energy rating label or energy rating is a label affixed to various appliances prior to retail sale. This allows consumers to compare the energy efficiency of products and may influence their purchasing decision. For example, the European union provides for Regulation No. 812/2013 of February 18, 2013 (Ell energy label) which is hereby incorporated by reference. The Ell energy label specifies how fluid tanks are classified or rated into energy efficiency classes and defines the standard test methods to be used. The class can be seen as a score value.

Determining the energy efficiency of a product, in particular, according to a standard test defined in a regulation, conventionally requires carrying out a preset method involving experimental set ups under standard environmental conditions. Generally, any product needs to be tested and classified with regard to its energy loss under those standard conditions. The rating applies to the final product.

It is desirable to develop products, such as tanks for warm fluids, that are eligible for a good energy efficiency rating. Various factors during the development process have an impact on the energy efficiency so that the design, prototyping, testing, pilot production and manufacturing stages may need to be adapted to obtain a better performance. Until the testing stage for energy efficiency is reached resources are consumed in several stages, in particular the design and prototyping stages. For example, choosing a specific insulation material, a wall thickness or form of a tank has an impact on the resulting energy efficiency. Conventionally, the develop- merit stages are repeated iteratively with adapted product characteristics to arrive at the desired energy efficiency rating.

It is, however, desirable to determine or estimate its energy efficiency in an early development stage of a product, e.g. already while specifying the design of a tank.

It is therefore an object of the present disclosure to provide an efficient method for assessing an energy efficiency of a tank.

According to one aspect of this disclosure, a computer-implemented method for assessing an energy efficiency of a tank comprises: receiving tank data indicative of geometrical and material properties of the tank; determining, by a trained artificial neural network (ANN) and based on the received tank data, a heat loss of the tank; generating an energy efficiency score value (EESV) based on a preset method for calculating the energy efficiency according to a test standard for energy efficiency and as a function of the determined heat loss; and outputting/displaying the EESV.

A tank may be a container suitable for holding a fluid, e.g. a liquid and/or gaseous phase of a substance. The tank is preferably suitable for holding a warm liquid, wherein the liquid may be water. In embodiments, the tank is part of a water heater, a hot water storage tank and/or a package of a water heater and a solar device. For example, the tank is a home appliance. The method may be applied to tank data relating to tanks/containers suitable for any fluid media, such as supercritical media. Other applications can be contemplated.

In alternative embodiments, the tank is configured to store or to hold a fluid. The fluid may be air, a gas, a heating medium, and/or coolant. One may contemplate of containers/tanks that isolate the inner volume containing the fluid with respect to the outer environment. An energy efficiency of the isolated storage of the fluid can be obtained by the suggested method.

In embodiments, the fluid is a gas or liquid at a temperature between -100°C and 100°C. The tank is, e.g. an isolated container or box, implemented to store a gas or liquid at temperatures between -30°C and 15°C, preferably between -30°C and 0°C, and more preferred between - 20°C and -4°C. In embodiments, the fluid is water at a temperature between 20°C and 100°C. The tank is, e.g. an isolated warm water tank, implemented to store water at temperatures between 0°C and 100°C, preferably between 4°C and 100°C, and more preferred between 20°C and 100°C.

In embodiments, thus, a computer-implemented method for assessing an energy efficiency of a warm water tank comprises: receiving tank data indicative of geometrical and material properties of the tank; determining, by a trained artificial neural network (ANN) and based on the received tank data, a heat loss of the tank; generating an energy efficiency score value (EESV) based on a preset method for calculating the energy efficiency according to a test standard for energy efficiency and as a function of the determined heat loss; and outputting/displaying the EESV.

In embodiments, the warm water tank complies with Regulation No. 812/2013 of February 18, 2013 (Ell energy label), in particular Art. 1 , 2 and/or 5. The heat loss may be considered a standing loss.

In embodiments, the warm water tank is a water heater comprising a heating element, the water heater having a rated heat output of < 100 kW, preferably < 100 kW, and more preferred < 20 kW.

In embodiments, the warm water tank has a storage volume of < 1000 liters, preferably < 500 liters, and more preferred < 200 liters.

The method can be carried out at the design stage of the tank where the tank data is defined, and a product performance regarding its energy efficiency can be predicted using the method. This allows for altering the tank data at an early development stage and potentially improving the EESV. Tank data is preferably tank design data, e.g. diameter, height, insulation thickness and material, and a number of connections of the tank to be manufactured. Hence, the presented method facilitates a product development process for a tank, in particular a hot water storage tank.

In embodiments, the dependent energy efficiency score value and/or the determined heat loss are/is independent from the operation of the tank or container. In particular, the energy efficiency score value and/or the determined heat loss only dependent/s on structural or material properties of the container/tank. In embodiments, the energy efficiency score value corresponds to a respective energy efficiency class, e.g. according to a governmental regulation as identified above or below.

Using an ANN allows for an efficient determination of the expected heat loss of the tank without having to execute experiments on a prototype. In particular, an ANN is useful, because it may be trained by appropriate training data. Training data may be real world data (ground data) available from prior development processes for insulated tanks. Alternatively, or additionally, training data may include synthetic or virtual tank data and associated data indicative of heat/energy flows in connection with the "virtual" tank. Synthetic or virtual tank data is, for example, generated by a computational simulation. The ANN may determine heat loss paths within the designed tank and/or with its environment. Determining may also include simulating the thermal properties of the tank, e.g. as a digital twin, wherein the trained ANN is involved.

In particular, in the step of determining, by a trained ANN and based on the received tank data, a heat loss of the tank is directly determined by the ANN, i.e. the ANN outputs the heat loss in response to the input tank data. It is an advantage that no further calculational resources are necessary.

It is an advantage that the method can be carried out in real-time due to the efficient ANN heat loss determination.

In embodiments, the ANN is a feed-forward ANN comprising: an input layer having input nodes for receiving input data; an output layer having output nodes for outputting output data; and intermediate layers each having a plurality of nodes; and wherein the tank data is the input data, and heat flows being indicative of the heat loss, the heat loss and/or the EESV is the output data.

Investigations of the applicant show that feed forward (FF) AN Ns are in particular useful for determining, predicting, assessing, calculating, estimating heat losses in tank systems based on tank data.

Nearest neighbor classifiers (kNN) or support vector machines (SVM) exhibited inaccuracies for tank systems with high heat losses. However, kNNs and SVMs may still be useful if small to intermediate heat losses are to be predicted. For example heat losses between 0 and 300 W were predicted through an FF ANN and a kNN trained on the same data. It was observed that the kNN exhibited significant deviations from the observed heat loss (training data) with respect to the FF ANN predictions for values above 200 W.

In embodiments, nodes of the pluralities of nodes have a rectified linear activation function. Studies show that four, five or six neurons per intermediate layer provide accurate and reliable heat loss predictions based on tank data. If larger ANNs are used, e.g. 8 - 32 neurons per intermediate layer overfitting may take place, so that correlations between input data and specific heat losses are less reliably predicted.

In embodiments, the ANNs are computer-implemented deploying software libraries, e.g. available from tensorflow.org and/or https://github.com/keras-team/keras at time of filing this application. Various programming languages may be used to implement the respective ANN, e.g. python. One may contemplate of other software libraries or repositories that are readily available in order to build ANNs to be trained with tank data.

In embodiments, the input layer is a normalization layer. This avoids over-weighing the influence of the number of connections. Tank data, on which the prediction by the ANN is based rather correlates with other factors, such as geometry or material parameters.

Embodiments include feed forward ANNs with a normalizing input layer, two five-neuron or six- neuron intermediate layers, and one output node, wherein a rectified linear activation function is used.

In embodiments, training the ANN comprises: providing a set of training input data and an associated set of training output data; receiving the set of training input data; generating output data by the ANN; calculating a loss function value based on a comparison between generated output data and the associated training output data; and adapting the weights of the nodes as a function of the generated output data and/or the loss function value.

According to an aspect of this disclosure, training may encompass repeating the steps of receiving, generating, calculating and adapting until the loss function value remains constant within a predetermined tolerance interval.

In embodiments, the computer-implemented method further comprises: generating training data by a computational simulation process implemented to compute a heat loss and/or heat flows based on a set of tank data and a selected test standard for energy efficiency, wherein a set of training input data is tank data and the associated set of training output data is the computed heat loss and/or heat flows.

Preferably, a step of storing the generated training data in a training database is carried out.

Training data alternatively or additionally may be generated by a design of experiments (DoE) using a simplified model of the tank system to be assessed. The computational model then employs an engineering formula model.

In embodiments, the computational simulation process includes a computational fluid dynamics (CFD) model and/or a conjugate heat transfer (CHT) model. After training the ANN, determining heat losses by the ANN corresponds to carrying out a respective CFD-CHT simulation. Using the ANN reduces the calculation effort and thus time and computational resources when the method for assessing the energy efficiency is executed, e,g. in terms of a web application.

Embodiments of the training method include generating at least 1.000 training data sets, preferably 2.000 training data sets, and even more preferably 10.000 training data sets. Embodiments require less than 1.000,2.000 or 5.000 training data sets.

In embodiments, the step of receiving tank data comprises: receiving computer aided design (CAD) data of the tank; and/or receiving material data indicative of material properties of components of the tank.

One may provide machine readable CAD data in a design database and/or a plurality of preselected materials for insulting, tube and/or wall materials of the tank in a material database.

In embodiments, the step of determining comprises: simulating at least one heat flow by the ANN; generating a digital twin of the tank having an associated tank identification code; and/or determining a heat flow within the tank and/or with an environment of the tank.

It is understood that the trained ANN may implement a model for predicting heat dissipa- tion/heat loss of a (e.g. warm water) tank being characterized by the tank data, based on the tank data and a predetermined measurement process. Hence, the ANN "simulates" heat flows. In embodiments, the tank data further comprises data indicative of functional properties of the tank, wherein said data indicative of functional properties of the tank comprises at least one of the group of: a number of heat exchangers, parameters characterizing a heat exchanger, a number of connections, a number of inlets, a number of outlets.

In embodiments, the tank data indicative of geometrical properties of the tank comprises at least one of the group of: a geometry, height, volume, width, diameter, form, inner wall thickness, outer wall thickness, inlet position, outlet position, a connection position.

In embodiments, the tank data indicative of material properties of the tank comprises at least one of the group of: a material of an insulation layer, material of an inner wall, a material of an outer wall.

The tank data may be provided and processed in a specific data format, e.g. in terms of Standard for the Exchange of Product model data (STEP), Initial Graphics Exchange Specification (IGES), Stereo Lithography (STL) or wave front. Other data formats are feasible.

In embodiments, the computer-implemented method further comprises at least one of the steps of: assigning a tank identification code to the tank; receiving input test method data indicative of the preset method for calculating the energy efficiency; providing a material database; providing a test standard database; identifying energy loss paths based on the determined heat flow; generating a database (VELDB) comprising data sets comprising, for a plurality of tanks, the tank identification code of the respective tank and the energy efficiency score value determined for the respective tank; generating an energy label (VELA) according to a test standard as a function of the determined generated energy efficiency score value (EESV); and/or outputting/displaying the tank identification code, a representative drawing of the tank, energy loss paths, the energy label and/or a time-dependent temperature, a heat loss as a function of a selected parameter comprised in the received tank data.

In embodiments, the tank data only comprises static data indicative of the structural properties of the warm water tank. E.g. the training data sets and input/received tank data is not time dependent or comprises process data indicative of an operation of the warm water tank. Rather, tank data is independent from the operation of the tank. This holds for the training data and the tank data when employing the method for assessing the energy efficiency.

In embodiments, the preset method for calculating the energy efficiency is defined according to Regulation (EU) No. 812/2013 (EU energy label).

Preferably, aspects and features of the presented methods and apparatuses are to be interpreted according to European Union Regulation No. 812/2013 of February 18, 2013 (EU energy label). However, in embodiments alternative definitions of tanks, EESVs, test standards for energy efficiency, etc. may be employed. The US Department of Energy provides a framework for assessing energy efficiency in 10 CFR 430, Subpart B, Appendix E (USEF). USEF may be used as a test standard for energy efficiency and/or a rating or alternative score value EESV.

Aspects of the method allow for efficiently planning, designing and manufacturing a tank having a desired energy efficiency without resorting to physical tests of prototypes or the use of thermal test chambers.

In embodiments, the tank data is indicative of a plurality of patches of insulation material, wherein the patches comprise different insulation materials. Hence, the method may take into account an isolation layer about a tank wall that is pieced together using different materials.

Preferably, the tank data and/or energy efficiency data, e.g. an anergy loss, are given and processed by the ANN in terms of SI units. The respective data is, in particular, represented by computer-processed using dimensionless variables representing physical observables in SI unit.

It is understood that the aspects and properties of warm water may stand for a general fluid being held in the tank. Hence, further aspects of this disclosure may relate to assessing the energy efficiency of keeping a fluid in a desired temperature range and isolated from a surrounding. One may consider a fridge or freezer a tank/container and the inner air the fluid.

According to another aspect, an apparatus is provided, wherein the apparatus comprises processing means implemented to carry out the method steps as disclosed above or below with respect to embodiments of the method for assessing an energy efficiency of a tank. The apparatus may be a server or processing device accessible through a web interface. According to a further aspect, a computer-readable medium storing computer program instructions, wherein the computer program instructions, when executed by a processing device, cause the processing device according to aspects or embodiments disclosed above or below with respect to specific examples, cause the processing device to perform operations comprising the method as disclosed herein. The computer-readable medium is, in particular, a non- transitory computer-readable medium.

In embodiments, a computer-program or computer-program product comprises a program code for executing the above-described methods and functions by a computerized device when run on at least one computerized device, in particular when run on a personal computer. A computer program product, such as a computer program means, may be embodied as a memory card, USB stick, CD-ROM, DVD or as a file which may be downloaded from a server in a network. For example, such a file may be provided by transferring the file comprising the computer program product from a wireless communication network. In embodiments, the method is implemented as a web app.

Further possible implementations or alternative solutions of the invention also encompass combinations - that are not explicitly mentioned herein - of features described above or below in regard to the embodiments. The person skilled in the art may also add individual or isolated aspects and features to the most basic form of the invention.

Further embodiments, features and advantages of the present invention will become apparent from the subsequent description and dependent claims, taken in conjunction with the accompanying drawings, in which:

Fig. 1 shows a flow chart of method steps involved in embodiments of the method for assessing an energy efficiency of a tank;

Fig. 2 shows an illustrative embodiment of a hot water storage tank;

Fig. 3 shows an example for an energy score value in terms of an energy efficiency label;

Fig. 4 shows method steps for receiving tank data;

Fig. 5 shows method steps for determining a heat loss;

Fig. 6 shows method steps for outputting an energy efficiency score value;

Fig. 7 shows an example for a displayed heat loss in relation to energy efficiency score values;

Fig. 8 shows a flow chart of method steps involved in embodiments of a method for training an ANN; and Fig. 9 shows a flow chart of method steps involved in embodiments of a method for generating training data for an ANN.

In the Figures, like reference numerals designate like or functionally equivalent elements, unless otherwise indicated.

Next, embodiments and aspects of methods for assessing, predicting or estimating energy efficiencies , in particular in terms of heat losses, of tanks are presented. The method is, for example implemented in terms of a web application facilitating the development and production of tanks, in particular water storage tanks.

Fig. 1 shows an embodiment of the method in terms of method steps. The method predicts an energy efficiency score as a function of tank data. In a first step S1, tank data is received. Tank data characterizes the tank to be assessed from a design perspective and includes, in this embodiment, characteristic parameters indicative of a geometry or form of the tank, functional features, such as a number of connections, e.g. inlets and outlets, and material properties, e.g. of the insulation material used. The entirety of tank data received may be referred to as tank parameters.

In Fig. 2, an embodiment of a water storage tank 1 is shown to illustrate tank parameters. The tank 1 may have a cylindrical form with a height h as geometric tank parameters. The tank 1 includes one inlet 2 and one outlet 3. The number of those connections and the location can be tank parameters. Further, a heating element 4 or heat exchanger 4 is provided in the interior of the tank 1.

Referring to Fig. 1 , in step S2, a heat loss of the tank 1 is determined. After having received the tank data, a trained artificial neural network (ANN) classifies or predicts a heat loss as a function of the tank data. The ANN is implemented to determine the heat loss of the tank 1 according to a preset test standard, e.g. as specified in Regulation No. 812/2013 of February 18, 2013 for the Ell energy label. The heat loss of the tank is translated into an energy efficiency score value (EESV) in step S3. When generating the EESV one may refer to a database TSDB containing specifications of various test standards, so that the EESV corresponding to the heat loss can be calculated and output in step S4. For example, the Ell energy label defines ratings A through F, wherein A refers to a highest energy efficiency and F/G to a lowest.

The determined energy loss and EESV are obtained from tank data without carrying out physical measurements in a controlled environment as required by the chosen test standard in the real world. Manufacturing a prototype tank according to the tank data is not required during a development process because of the method of Fig. 1. Hence, a "virtual" EESV is generated by the method prior to prototyping and during a design stage. This allows amending the design at an early stage of the design and development process for a tank. By changing tank parameters and repeating the method one may design a tank having desired tank parameters and being eligible for a specific energy label.

Fig. 3 shows a representation of an Ell energy label VELA. For example, in step S4, a virtual energy label VELA is generated and presented on a display.

Fig. 4 depicts in more detail a user interaction with a software application, e.g. running as a web app, implementing the method. Step S1 requires a user, who intends to develop a water storage tank, to enter tank data. First, in step S11 geometric tank data is input. Geometric tank data may be input as computer aided design (CAD) data characterizing the topology and dimensions of the tank. For example, the user inputs geometric tank data indicative of a tank having a flat bottom, a flat head, a specific height and diameter and wall thicknesses in step S11.

Next, the user selects materials for the tank walls and insulation layers from a choice of materials in step S12. A material database MDB contains a choice of materials for walls, insulation layers or the like.

Finally, the user may select a test standard for evaluating the energy efficiency in step S13. The chosen test standard specifies the measurements and calculation method for the energy efficiency. In this example, an EU energy label is considered. However, one may contemplate of other test standards, e.g. according to Australian, US or UK regulations. A test standard database TSDB contains specifications according to various regulations.

Steps S11 , S12 and S13 can be supported by a graphical user interface.

Fig. 5 depicts in more detail the aspect of determining the heat loss of the "virtual" tank corresponding to the input tank data. In step S21 a digital twin of the tank to be assessed is generated. The digital twin of the tank has a unique associated tank identification code or digital twin ID (DTID).

Next, a computational simulation of heat flows in the tank and between the tank and its environment is carried out in step S22. The computational simulation involves a trained ANN and input from the TSBD and MDB. The trained ANN is chosen as a function of the user selection regarding the test standard in step S13. The simulated heat flows are obtained in step S23. The simulated heat flows would occur in a test procedure of the tank according to the input tank parameters and the chosen test standard.

The heat loss simulation by the ANN allows for a further analysis in step S24. By changing input tank parameters the user may identify correlations between tank parameters and specific heat flows. For example, heat flows and heat loss may be influenced by the choice of the wall and/or insulation material. Changing the intended material can lead to reduced heat flows and thus an improved energy efficiency in terms of the EESV.

Steps S41 and S42 depicted in Fig. 6 refer to outputting the determined EESV. The digital twin identification (DTID), the tank parameters of the digital twin and the determined EESV are stored in a virtual energy label database VELDB in step S41. In step S42 a VELA is produced that may be printed and placed on a tank to be manufactured according to the design data referring to the DTID.

Outputting the EESV may involve displaying viewgraphs as shown in Fig. 7. Fig. 7 shows heat losses (y-axis) as a function of a tank volume (x-axis), wherein the tank volume is a representative tank parameter. The colored regions refer to Ell energy ratings A+ (green, lowest sections) through G (red, highest section). Between the colored regions curves separating the ratings from one another can be identified. At about 250 I tank volume a heat loss of 25 W is marked with a dot SR. The dot SR indicates the simulation result. Heat loss is determined by the trained ANN and can be analyzed in connection with various tank parameters. The presented method thus facilitates a design and production process for storage tanks.

Next, training methods for the involved ANN are depicted. Investigations of the applicant have shown that a feed forward ANN with a normalization input stage, a first layer having five ReLu (rectified linear unit) neurons, a second layer of five ReLu neurons and an output layer indicating a heat loss accurately predicts heat losses in tanks. This ANN can be written as [n,5,5, 1], "n" stands for a normalization layer, and the integers between the commas designate the width of a respective neuron layer.

For example, an [n,5,5, 1] ANN receives, variables in SI units, indicative of i) a material, ii) a thickness of the insulative layer [m], iii) a number of connections, iv) a tank diameter in [m], and v) a tank height in [m]. The five input parameters i) - v) are processes and yield, in SI units, an energy loss in [W] and a corresponding EU energy label score. Fig. 8 shows a flow chart of method steps for training an [n,5,5, 1 ] ANN. Generally, in step S5 training data is retrieved. A respective training data set comprises a training input data set (tank parameters) and a training output data set (a heat loss value). In step S6 the dimension of the ANN to be generated and trained is set, e.g. [n,5,5, 1 ], and the training proposal is defined, e.g. simulating a specific heat flow path. In step S7 a suitable training algorithm is executed. The training algorithm involves calculating a loss function that is indicative of an accuracy of the trained ANN. If the loss function becomes essentially constant during training and does not change anymore (within a predetermined tolerance) training may be terminated. Then, the ANN configuration, e.g. in terms of the internal weight functions of the neurons, is stored in step S8. As a result a trained ANN for a specific simulation aspect is made available.

The training data used for training the ANN can be obtained through a computational simulation of ("virtual") tanks and their heat losses, wherein each tank is characterized by set of tank parameters (tank data). The computational simulation deploys CFD and CHT models, for example. Carrying out CFD/CHT computations requires large computational resource so that a real-time computation of heat losses, e.g. replacing steps S2, S22 is not feasible.

Referring to Fig. 9, steps S21, S22', S23, S24 and S3 are repeatedly carried out, and each loop produces one training data set based on different tank data. Steps S21, S23, S23, S24 and S3 correspond to the method steps described above. However, step S22 involves the deployment of CFD and CHT models in order to numerically simulate the heat flows, heat flow paths and to calculate an energy efficiency. In embodiments, 2,000 training data sets are generated and stored in the training data base TBD. For example, a one-dimensional model is deployed for generating the training data.

It is understood that in the above-depicted embodiment, the training input data sets include, in a suitable data format, the warm water tank data as explained above. The tank data characterizes or defines a real water tank or serves as a model of a warm water tank. The corresponding training output data set for a water tank is a ‘standing loss’ (S) in terms of the heating power dissipated from the hot water storage tank at given water and ambient temperatures, expressed in W. For a heater, the training output data set is a ‘water heating energy efficiency’ (r|_wh) in terms of the ratio between the useful energy provided by a water heater or a package of water heater and solar device and the energy required for its generation, expressed in %. The standing loss and/or the energy efficiency is preferably obtained according to the measurement and calculation methods of Regulation No. 812/2013 of February 18, 2013 (Ell energy label), in particular Art. 2, Art. 5 and/or Annex VII and Annex VIII. The same holds for the standing loss/heat loss and/or the heating energy efficiency determined by the trained ANN. The ANN implements the before-mentioned energy efficiency calculations/measurements through its prior training.

In summary, in an embodiment, a computer-implementable method for assessing or predicting an energy efficiency of a warm water tank is disclosed, wherein an energy efficiency rating can be determined in real time. Assessing the energy efficiency during a design stage prior to manufacturing the tank renders the tank development process more effective thereby saving energy and resources. The designed tank according to tank parameters characterizing the tank may be rated and labeled with a VELA.

Reference signs:

1 tank

2 inlet

3 outlet

4 heat exchanger h height

TDB Training database

TSDB Test standard database

MDB material database

SR simulation result

51 Receiving tank data

511 User inputs tank topology and dimensions/CAD

512 User inputs material data from database

513 User inputs desired test rating standard

52 Determining a heat loss

521 Generating a digital twin

522 Simulating tank by ANN

523 Calculating heat flow

524 Analyzing heat flow paths

53 Generating an EESV

54 Outputting the EESV

541 Storing digital twin ID and EESV in a virtual energy label database (VELDB)

542 Generating virtual energy efficiency label (VELA) S50 ANN training

55 Loading plurality of training input data and associated training output data

56 Setting ANN dimension and training proposal

57 Training of ANN S8 Storing trained ANN configuration

Claims

1 . A computer-implemented method for assessing an energy efficiency of a warm water tank (1), said warm water tank being suitable for holding warm water, comprising:

Receiving (S1) tank data indicative of geometrical and material properties of the warm water tank;

Determining (S2), by a trained artificial neural network (ANN) and based on the received tank data, a heat loss of the warm water tank; generating (S3) an energy efficiency score value (EESV) based on a preset method for calculating the energy efficiency according to a test standard for energy efficiency and as a function of the determined heat loss; and outputting (S4) the EESV.

2. The computer-implemented method of claim 1 , wherein the warm water tank is implemented to store water at temperatures between 4°C and 100°C.

3. The computer-implemented method of claim 1 or 2, wherein the warm water tank is a water heater comprising a heating element, the water heater having a rated heat output of < 70 k (20 kW)W.

4. The computer-implemented method of any one of claims 1 - 3, wherein the warm water tank has a storage volume of < 1000 liters .

5. The computer-implemented method of any one of claims 1 - 4, wherein the trained ANN implements a model of the warm water tank in a predetermined environment and predicts a heat loss rate of warm water in the warm water tank to the predetermined environment.

6. The computer-implemented method of any one of claims 1 - 5, wherein the trained ANN is implemented to predict the heat loss rate and/or the energy efficiency in real-time.

7. The computer-implemented method of any one of claims 1 - 6, wherein the ANN is a feed forward ANN comprising: an input layer having input nodes for receiving input data; an output layer having output nodes for outputting output data; and intermediate layers each having a plurality of nodes; and wherein the tank data is the input data, and heat flows being indicative of the heat loss, the heat loss and/or the EESV is the output data.

8. The computer-implemented method of claim 7, wherein nodes of the pluralities of nodes have a rectified linear activation function.

9. The computer-implemented method of claim 7 or 8, wherein the input layer is a normalization layer.

10. The computer-implemented method of any one of claims 7 - 9, wherein training (S7) the ANN comprises:

Providing a set of training input data and an associated set of training output data;

Receiving the set of training input data;

Generating output data by the ANN;

Calculating a loss function value based on a comparison between generated output data and the associated training output data; and adapting the weights of the nodes as a function of the generated output data and/or the loss function value.

11. The computer-implemented method of claim 10, further comprising: generating training data by a computational simulation process implemented to compute a heat loss and/or heat flows based on a set of tank data and a selected test standard for energy efficiency, wherein a set of training input data is tank data and the associated set of training output data is the computed heat loss and/or heat flows.

12. The computer-implemented method of claim 11 , wherein the computational simulation process includes a computational fluid dynamics (CFD) model and/or a conjugate heat transfer (CHT) model.

13. The computer-implemented method of any one of claims 1 - 12, wherein the step of receiving tank data comprises: receiving computer aided design (CAD) data of the warm water tank; and/or receiving material data indicative of material properties of components of the warm water tank.

14. The computer-implemented method of any one of claims 1 - 13, wherein the step of determining comprises: simulating at least one heat flow by the ANN; generating a digital twin of the warm water tank having an associated tank identification code; and/or determining a heat flow within the warm water tank and/or with an environment of the warm water tank.

15. The computer-implemented method of any one of claims 1 - 14, wherein the tank data further comprises data indicative of functional properties of the warm water tank, wherein said data indicative of functional properties of the warm water tank comprises at least one of the group of: a number of heat exchangers, parameters characterizing a heat exchanger, a number of connections, a number of inlets, a number of outlets.

16. The computer-implemented method of any one of claims 1 - 15, wherein the tank data indicative of geometrical properties of the warm water tank comprises at least one of the group of: a geometry, height, volume, width, diameter, form, inner wall thickness, outer wall thickness, inlet position, outlet position, a connection position.

17. The computer-implemented method of any one of claims 1 - 16, wherein the tank data indicative of material properties of the warm water tank comprises at least one of the group of: a material of an insulation layer, material of an inner wall, a material of an outer wall.

18. The computer-implemented method of any one of claims 1 - 17, wherein the tank data only comprises static data indicative of the structural properties of the warm water tank.

19. The computer-implemented method of any one of claims 1 - 18, wherein the tank data does not comprise process data indicative of an operation of the warm water tank.

20. The computer-implemented method of any one of claims 1 - 19, further comprising: assigning a tank identification code to the warm water tank; receiving input test method data indicative of the preset method for calculating the energy efficiency; providing a material database; providing a test standard database; identifying energy loss paths based on the determined heat flow; generating a database (VELDB) comprising data sets comprising, for a plurality of warm water tanks, the tank identification code of the respective warm water tank and the energy efficiency score value determined for the respective warm water tank; generating an energy efficiency label (VELA) according to a test standard as a function of the determined generated energy efficiency score value (EESV); and/or outputting/displaying the tank identification code, a representative drawing of the warm water tank, energy loss paths, the energy label and/or a time-dependent temperature, a heat loss as a function of a selected parameter comprised in the received tank data.

21. The computer-implemented method of any one of claims 1 - 20, wherein the preset method for calculating the energy efficiency is defined according to Regulation (Ell) No. 812/2013 (Ell energy label).

22. The computer-implemented method of any one of claims 1 - 21 , wherein the steps of receiving (S1) tank data, and outputting (S4) the EESV are implemented as a web interface for a user.

23. The computer-implemented method of claim 22, wherein the steps of determining (S2) the heat loss and generating (S3) the energy efficiency score value (EESV) are carried out by a processing device communicatively coupled to the web interface.

24. A medium storing computer program instructions, wherein the computer program instructions, when executed by a processing device, cause the processing device to perform operations comprising the method according to any one of claims 1 - 23.