WO2017064115A1

WO2017064115A1 - Energy buffer control and system

Info

Publication number: WO2017064115A1
Application number: PCT/EP2016/074457
Authority: WO
Inventors: Peter Coenen
Original assignee: Vito Nv
Priority date: 2015-10-12
Filing date: 2016-10-12
Publication date: 2017-04-20

Abstract

The present invention relates to methods, systems and devices such as controllers to control the use of energy such as electrical energy and thermal energy while taking into account storage losses, e.g. assuming electrical energy is used to charge a thermal buffer or a battery that is depleted as required by one or more users. As an example a cold store can be envisaged. For example, electrical power can be used to extract heat from a cold store and hence to lower the temperature inside. The cold store and its cargo or contents form a thermal buffer, which is charged with cooling energy produced from electric compressors. This buffer is depleted as warm cargo is added and/or due to losing cold energy to the environment or because heat energy is received from the environment.

Description

ENERGY BUFFER CONTROL AND SYSTEM

The present invention relates to methods, systems and devices to control the use of energy such as electrical energy and thermal energy while taking into account storage losses and other factors as well as software and a computer system to implement the methods.

Background

There are many thermal processes which are used privately of industrially which may consume electrical power to provide heating and/or cooling. The thermal energy can be stored in goods, building elements or dedicated materials such as water or phase change materials. Electrical power can be provided at different tariffs, e.g. a night tariff can be cheaper than a day tariff. These various options provide the possibility to buy electricity at moments when it is cheap or abundant (e.g. from renewables), convert it to thermal energy and store it in that form until it is needed.

For example, electricity can be bought on spot markets, e.g. a day ahead market. In this case the price for a 24 hour period is published say 10 hours ahead. Referring to Fig. 3, a power network 40 is shown whereby when operational expenses for renewable energy 44 are low, electricity prices in periods with a predicted abundance of renewable energy are low. When renewable energy becomes less available, electricity must be generated by more conventional means 42. The higher is the demand for conventional electricity, the higher is its cost. This means that when conventional generation is required, the most efficient plants are started first and with growing imbalance less efficient plants with higher C0₂ emissions are added. Thus electricity price as a technical control variable and environmental impact are related. Buying electricity cheap is in the first place a method to reduce environmental impact of human activity when that means renewable energy. The price versus time function can thus be viewed as a function expressing technical environmental impact versus time or technical availability of power with time. Thus price becomes a technical variable with which a system can be controlled.

High levels of renewables cannot be employed in current electrical power systems because of imbalances caused by renewable energy sources generating electricity at times when it is not needed. There has been a failure to integrate renewable energy sources into the electrical grid in such a way that imbalances are avoided. Methods for buffering thermal energy and thermal energy buffers are already known to the person skilled in the art. An example of a thermal energy buffer is for example a water heating unit of a domestic hot water system. Such a thermal energy buffer contains a thermal buffering medium, often water, contained in a tank and a controller controlling a heater of the thermal buffer. The heater can for example be an electrical heater provided at the bottom of the tank. When the temperature measured by the sensor drops below the minimum temperature, the controller activates the heater, for example until the maximum temperature is obtained. In such a configuration the minimum temperature represents the predetermined minimum amount of energy present in the thermal energy buffer. However, this method of operation is a simple response to a low temperature - no attempt is made to plan the operation of such a buffer.

Summary of the Invention The present invention relates to methods, systems and devices such as controllers and software and computer systems to control the use of energy such as electrical energy and/or thermal energy while taking into account energy storage losses, e.g. assuming electrical energy is used to power a device to charge a thermal buffer that is depleted through losses, e.g. self-depletion or as required by one or more users. As an example a cold store or a residential or industrial building heating system can be envisaged as embodiments of the present invention. The thermal buffer can contain a thermal buffer solid or liquid, and can include phase change materials. Generally, the thermal buffer will be insulated and such insulation heated or cooled to a temperature is included in the total heat capacity of the thermal buffer.

An aim of the present invention is to integrate renewable energy sources into an electrical grid (see Fig. 3) in such a way that imbalances are reduced or avoided. This can be achieved by taking into account the use of renewable energy in a predictive pathway to meet short term and/or longer term efficiencies. In accordance with embodiments of the present invention a method, a system, a controller and software is provided for a feed-forward predictive pathway and an operative pathway to be followed, the predictive pathway having predictive time steps and the operative pathway having operative time steps. The operative and the predictive time steps can have the same time intervals or different. The predictive pathway explores, in a feed-forward manner, at each predictive time step, a limited set of operative decisions for a thermal buffer charging device, the exploration predicting a state of charge of the thermal buffer for each next predictive time step using each one of a set of operative decisions to predict the next state of charge or value related to a state of charge of the thermal buffer. This procedure is repeated for each subsequent predictive time step thereby creating trajectories formed by branches of the predictive pathway until a time horizon is reached. The last step of the predictive pathway which results in reaching the time horizon, generates state of charge outcomes for the thermal buffer. One of the predictive outcomes is selected and an output signal (for example a digital or analogue signal), is derived from an operative decision chosen from the set of operative decisions of the initial i.e. first predictive time step of the selected trajectory or a later step such as a step corresponding to a current time. Due to how the methods of the present invention are carried out, the acceptable trajectory will include the selected outcome at the time horizon and the selected operative decision from the set of operative decisions of the first predictive time step and all intermediate calculations at intermediate predictive time steps. A valid trajectory will be continuous from an initial such as the current time to the time horizon.

For example, the state of charge outcome or outcomes (if there is more than one) of the predictive pathway at the time horizon is/are examined to see if any of them fall within an acceptable range, such as a state of charge of greater than 50%, 60%, 70% or 80%. If one or more outcomes is in an acceptable range, one of these outcomes is chosen as the optimised outcome. The operative decision of the first predictive step (e.g. initial or current step) can be chosen which is linked to this selected outcome in a valid trajectory. This operative decision is used in the first operative step of the operative pathway (e.g. initial or current time step), i.e. an output signal is generated that controls the buffer being charged according to this operative decision in this operative time step. If there is more than one valid trajectory that links an acceptable outcome at the time horizon and one of the operative decisions of the first operative step (e.g. initial or current time step), then one of the valid trajectories is chosen that meets an operation constraint. Then this operative decision is carried out, i.e. the amount of energy to charge the thermal buffer charging device is provided to the buffer as required by the selected operative decision for that operative time step.

If there is no outcome or no outcome that lies within an acceptable range, then the predictive pathway has failed and an operative decision is taken for the first operative time step (initial or current) based on other criteria, e.g. the operative decision of a previous operative time step is taken, no charging of the thermal buffer is performed, an average amount of charge is provided, the buffer is charged to a certain percentage state of charge such as 75 to 100%, a thermostat output is used to control the charging or similar.

As the calculation method travels forward through the predictive pathway exploring the branches and predicting the state of charge of the thermal buffer or a value related to the state of charge at predictive time steps, the availability of energy during each predictive time step of the predictive pathway can be input and a cost for the execution of each branch of the predictive tree structure can be determined and stored. Each predictive time step is a time step in the future from the initial or current time step, and at each of these future time steps an assessment can be made as to whether renewable energy or primary energy such as from fossil fuel fired power stations or nuclear power stations or a combination of these is used and the types used and their cost can be recorded. By this assessment renewable energy sources can be linked into an electrical grid in such a way that imbalances are reduced or avoided.

In embodiments according to the present invention of a thermal buffer such as a standalone or self-contained thermal buffer, e.g. a cold store or a residential or industrial building heating system, the thermal buffer has a source of energy, e.g. a source of energy which is directly usable as heat energy (example: a heat exchanger) or a source of energy that can be converted into heat energy (example: electricity and a resistive heater, a chiller, a compressor of a refrigerator). For example a refrigeration unit or chiller can convert electrical energy into heat energy, e.g. cooling. A geothermal plant can make use of heat energy directly from the Earth. Cooling can be delivered as ice blocks or frozen C0₂. The load on the energy source is the thermal buffer itself i.e. the cold store or the residential or industrial building heating system. Hence, the cold store and the residential or industrial building heating system are standalone or self-contained devices, neither the cold store nor the oven having a deliberate and dedicated external load such as would be the case for a distributed air conditioning system. For example, electrical power can be used to power a cooling device to extract heat from a cold store and hence to lower the temperature inside. The thermal buffer such as the cold store or the residential or industrial building heating system and its cargo or contents make up the thermal buffer, which is charged with heat energy such as cooling energy produced, for example from electric compressors, or is charged with heating energy, for example from resistance heaters or combustion of fossil fuels. This buffer when it is a cold store is depleted as warm cargo is added and/or heating energy from the environment enters the buffer and because heat energy is received from the environment. When this buffer is a residential or industrial building heating system, this system is depleted as cold materials enter the building and/or heat energy is lost to the environment.

In one aspect the present invention provides a method to control the use of energy while taking into account storage losses of energy stored in a thermal energy buffer, e.g. in a standalone or self-contained thermal buffer, that is depleted by self-depletion or as required by one or more users, the method comprising: controlling, by means of an output signal, (for example a digital or analogue signal), an energy source (such as an energy source powered by electrical power, or a heat energy source such as a heat exchanger or a geothermal plant) adapted to charge the thermal energy buffer;

calculating predicted iterations in a feed-forward manner through time along a predictive pathway in a plurality of prediction time steps from a current time instance up to a time horizon using a simulation model of the thermal energy buffer to predict a buffer state of charge or a value related to a buffer SoC at each prediction time step along a plurality of calculation trajectories which comprise branches, the buffer state of charge or the value related to the buffer state of charge being generated by a selection of a limited number of first operative decisions at the current time instance and by a selection of a limited number of second operative decisions at each subsequent time step up to the time horizon, executing an optimisation routine to reduce the number of branches at each or any predictive time step by pruning in accordance with a pruning selection criterion to obtain, at the time horizon, one or more outcomes which are buffer states of charge or values related to an SoC of the thermal buffer. If one or more outcomes is/are within an acceptable range then:

one of the outcomes of the predictive pathway is selected, and therefrom a selected first operative decision for a next operative time step is derived, the selected one of the first operative decisions being part of a calculated trajectory from the first time instance to the time horizon which includes an acceptable outcome, deriving a single output control signal from the selected first operative decision, and charging the thermal energy buffer in accordance with the single output signal.

The controller has an output signal that is used to control the energy flow, e.g. electricity flow required to charge the thermal energy buffer with cooling or heating energy.

The selection of a suitable trajectory can based on an overall cost or energy availability, or the trajectory that makes most use of renewable energy or least use of energy from burning fossil fuels or nuclear power stations, etc.

The step of calculating iterations through time in predictive time steps of the predictive pathway in accordance with a simulation model of the thermal energy buffer to predict a buffer state of charge or a value related to an SoC at each predictive time step along a plurality of calculation trajectories results in a tree structure with branches, each branch involving a calculation of future state of charge, or a value relating thereto, e.g. temperatures of a buffer medium. The method includes performing calculations along each branch of the tree structure and new branches being spawned at each iteration step of the predictive pathway and branches being pruned in accordance with a pruning mechanism.

As indicated above electricity can be bought on spot markets, e.g. from a day ahead market. In this case the price for a 24 hour period is published say 10 hours ahead. A second element required to optimize the buying time is the prediction of the charge in the buffer. For example, this prediction can translate into the prediction of the thermal buffer state of charge such as the cold store or residential or industrial building heating system state of charge or a value related to SoC such as a temperature in the future or at a plurality of times in the future. A system, method or controller or software and computer systems according to embodiments of the present invention can have an input related to a measurement of a thermal capacity or one or more values related to such thermal capacities, e.g. actual cold store temperatures or residential or industrial building heating system temperatures. More than one temperature can be measured, e.g. the temperature of a heat energy buffer fluid such as water or a phase change material may be measured at different heights in the buffer. The one or more temperatures determine the state of charge of the buffer or a value related to the SoC. A system, method or controller according to embodiments of the present invention has a single output to control the electricity flow, e.g. for transforming the electricity flow into a charging of the buffer with thermal energy, e.g. inherent to at least some of the embodiments of this invention is a phase change.

The system, method or controller or software and computer systems according to embodiments of the present invention is adapted to find a good and often an optimal solution, e.g. a cheap or the cheapest solution or one that has a good or the best environmental impact or one that combines renewable energy with energy from fossil fuel power stations or nuclear power stations to reduce imbalances. This can be done by using an optimisation technique with dynamic programming (DP) as one example of an optimisation technique. The optimisation technique used may be based on a Markov chain in which the predictive process of the predictive pathway goes through a sequence of random variables with a Markov property defining serial dependence only between adjacent periods linking time points in the development of the process of the predictive pathway. Thus the system, method or model or a controller or software and computer systems according to embodiments of the present invention can follow a chain of linked events along the predictive pathway forming trajectories, where what happens next depends only on the current state of the system. Hence a system, method or controller or software and computer system according to embodiments of the present invention can be adapted to iterate through time (e.g. in steps of a convenient length such as perhaps 1 hour) along the predictive pathway and to examine for every time step the result in terms of buffer state of charge or a value related to the SoC (e.g. the SoC is related to the temperature of a buffer medium in a thermal energy buffer) for some or all possible operative decisions for the output (e.g. charging device such as a compressor is OFF, charging device operation such as a compressor operation is at any value between 0 and 100%, e.g. 70%, 100%, etc.). This procedure follows a large number of calculation trajectories in the predictive pathway and can lead to a very large number of possible combinations. For example, when the time horizon is 34 hours and three operative decisions or output states are possible at each time step, and each operative decision or output state results in a new branch and a new trajectory this results in 17 x 10¹⁵ combinations. In case of fine control it may be required to use a higher time resolution (shorter periods) and more output possibilities (e.g. greater than three) so the number of possible combinations or branches may be significantly higher, in fact prohibitively higher.

A system, method or controller or software and computer systems according to embodiments of the present invention can start a procedure of the predictive pathway as late as possible in order to have the latest information about the charge in the buffer and early enough in order to finish before a new cyclic event such as a new electricity price applies. Hence a high resolution or a short calculation time would be preferred. Accordingly, a system, method or controller or software and computer systems according to embodiments of the present invention reduces the number of branches to be examined by means of a selection procedure of pruning mechanisms based on a pruning criterion. Four different criteria are described and any one of these or any combination of these can be used with any or all of the embodiments of the present invention.

A system, method or controller or software and computer systems according to embodiments of the present invention can be adapted to add a branch and bound (as a first pruning mechanism) in addition to an optimisation method such as DP. If the optimisation calculation such as a DP calculation is started using the measured temperature of the buffer medium as a value related to the SoC, predicting from each possible operative decision results in a new temperature prediction of the buffer medium or prediction of a value related to the SoC at the next time step. The predicted temperature or value related to the SoC, can be used in a pruning selection criterion. For example, once this predicted SoC or value related to the SoC exceeds a minimum or maximum value the procedure can be assumed to have reached the end of a realistic technical option and this branch can be terminated. For example, no more calculations need be performed on this branch and any previously achieved results can be discarded freeing up memory. Instead of temperature any other value related to the State of Charge of the thermal energy buffer may be used, or if it is available the state of charge itself.

A second pruning mechanism requires, for example, a preliminary simulation of an arbitrary control strategy, e.g. a thermostat strategy so as to generate a benchmark for comparison with any of the pruning methods of the present invention. Cost can be seen as a valve, energy at a low price means abundant energy is available and the cost valve is wide open whereas a high price of energy means a scarcity of energy and a valve setting with a small opening. Hence simulation models for use with the present invention can make use as cost modelled technically as a valve. The cost of this arbitrary control strategy which is related to the state of a cost valve and to the technical availability of an energy source is used as a limit, e.g. the maximum cost of any branch in the optimisation tree such as the DP tree, i.e. a minimum setting of a cost valve. This is used in a similar way to a maximum or minimum temperature or SoC in the first pruning mechanism as a pruning selection criterion. More expensive branches (those with the cost valve opened least) are discarded. For example, no more calculations are performed on such a branch and any previously achieved results are discarded. This pruning mechanism is especially helpful in the later iterations when the number of combinations becomes potentially extremely high and is therefore a need for a very efficient means of pruning.

Due to the large number of possible combinations, it is inevitable that an identical or similar SoC (or value related to the SoC) may be reached at a given time (i.e. iteration within the predictive pathway) but these SoC's (or values related to the SoC) can have different costs, i.e. with a cost valve at different openings which is related to different technical availabilities of an energy source. A similar SoC (or value related to the SoC) may be determined by a difference between two SoC's (or two values related to the SoC's) being less than a certain threshold value. The cost is used as a pruning selection criterion. As a third pruning operation, in such a situation, the or any more expensive branch i.e. one with a lower cost valve opening or a lower technical availability of an energy source is discarded as soon as this occurs. Finally, an identical or similar cost, i.e. identical or similar opening of a cost valve, which is related to the technical availability of an energy source may lead to different SoCs (or different values related to the SoC's) at a given time. A similar cost may be determined by a difference between two costs being below a threshold value which is related to differences in openings of cost valves or a difference between different technical availabilities of an energy source separated by less than a threshold value. The value of the SoC (or different values related to the SoC's) is/are used as a pruning selection criterion. As a fourth pruning operation, in such a situation, a branch yielding the or any lower SoC is discarded. Under certain circumstances the fourth pruning operation can be to discard a branch yielding a higher SoC. This can occur when increasing the SoC would be done by purchasing energy at an excessively high price.

The combination of some or all of these pruning mechanisms can reduce the calculation time of the optimisation algorithm such as DP algorithm by a factor of 1000 or significantly more. This allows a system, method or controller according to embodiments of the present invention to be used in real time applications with on-line calculation.

The SoC (the SoC is related to the temperature of buffer medium in the thermal energy buffer) of the thermal buffer is function of the losses and the charging, e.g. the use of compressor power to cool a buffer medium, or a heater to heat an oven. As the optimisation method e.g. such as DP investigates the future, it is possible to predict a SoC (or value related to the SoC) at a later time. A simulation or model of the thermal buffer is required as part of the algorithm. This simulation or model can be any convenient function like constant losses or charge (linear temperature changes), RC models (exponential changes or sum of ditto), or lumped circuit models or any other convenient function or combination thereof. The selection of a simulation or model is not expected to be a limitation on the present invention.

The model parameters can be calibrated either manually or automatically, e.g. if a system identification (SI) algorithm is used. The SI does not need to run together with the optimisation such as DP but may be run as part of a different task, e.g. once per day. So even complex processing will not influence the performance of the optimisation such as DP. If the buffer simulation or model functions are complex then the SI may be performed using a parameter optimising method such as Levenberg-Marquart, Gauss- Newton algorithm (GNA) or the method of gradient descent to find the optimal parameter fit. Depending on the model used, it may be required to use different functions for buffer charging and discharging. In this case either an additional input is required to determine the current state (charging device such as compressor is ON or OFF) or this information has to be derived from the evolution of SoC (e.g. temperature change).

Brief description of the drawings

Fig. 1 shows a system according to an embodiment of the present invention.

Fig. 2 shows a system according to another embodiment of the present invention.

Fig. 3 shows a power network with which embodiments of the present invention can be used.

Fig. 4 shows buffer according to another embodiment of the present invention.

Fig. 5 shows measured temperature points as a thermal buffer is depleted.

Fig. 6 shows a tree structure of a predictive pathway according to an embodiment of the present invention.

Definitions "Technical availability" of a source of energy refers to the ease with which such energy is available for use. For example the cost of an energy may be low which is indicative of a surplus of energy hence energy is easily available. If the cost is high this means that a surplus is low or non-existent and this may require the use of more expensive short term energy production systems which will drive up the cost. Low and high costs can also be related to environmental factors. When there is a surplus of energy from renewable sources the technical availability is high. However, if the surplus is low or non-existent, less environmentally friendly sources of power have to be used and the technical availability of energy is low. "Cost" as used in this application relates to a control variable used to decide how a thermal energy buffer is to be operated. Cost does not necessarily relate to what is actually charged to a user. Cost therefore acts like a valve - a high cost closes the valve because there is a low usage and hence it reduces the flow of energy. A low cost motivates use and therefor opens a valve, i.e. allows more flow of energy. Thus in this application reference is made to a "cost valve" which is controlled by the low cost opening it up and high cost closing it down.

A "predictive pathway" as used in this application relates to a process of a systematic enumeration and selection of a limited number of operative decisions over a period of time. It involves a process of calculating at a sequence of time steps, a predicted state of charge or a value related to a state of charge of a thermal buffer or a battery based on a simulation or model of the thermal buffer or battery. The predictive process of the predictive pathway can go through a sequence of random variables with a Markov property defining serial dependence only between adjacent periods linking time points in the development of the process of the predictive pathway. The predictive process follows a chain of linked events along the predictive pathway, wherein what happens in the next predictive time step depends only on the current state of the system. The predictive process iterates through time (e.g. in steps of a convenient length such as perhaps 1 hour) along the predictive pathway and examines for every time step the result in terms of buffer state of charge or a value related to the SoC (e.g. SoC is related to the temperature of a buffer medium in a thermal energy buffer) for some or all possible operative decisions (e.g. charger device such as a compressor is OFF, charger device operations such as compressor operation is at any value between 0 and 100%, e.g. 70%, 100%, etc.). This procedure follows a number of calculation trajectories in the predictive pathway.

The "operative pathway" as used in this application relates to driving a device for charging the thermal buffer in time steps. The amount of energy used to alter the state of charge of the thermal buffer is determined by outcomes of the predictive pathway. The times steps of the operative pathway do not need to be the same length of time as the time steps of the predictive pathway.

A "value related to a state of charge" is a known concept in which, for example the state of charge of a buffer is estimated or modelled by a number of temperatures such as two temperatures measured at the top and bottom of a thermal buffer. The temperature is used as a measure of the energy stored in the buffer medium. More accurate values can be obtained for the SoC of the thermal buffer by measuring the temperature of layers of a stratified thermal buffer and using the temperature and thermal capacities of all the layers in the calculation of State of Charge. For a battery it is often difficult to measure state of charge directly although the voltage of some types of battery can give some idea of the SoC. For batteries a better way is usually to monitor charging and discharging current and voltage and estimate the charge from these values. "Branch and bound" as used in this application relates to a systematic enumeration of a set of candidate operative decisions at time points between a current time and the time horizon. The set of candidate operative decisions at a plurality of time points can be thought of as forming a rooted tree. The method explores branches of this tree, which exploration involves using individual members of the set of operative decisions at predictive time steps to predict states of charge. In "Branch and bound" a state of charge of a spawned branch is checked at time steps against upper and lower estimated bounds, and is discarded if it does not lie in an acceptable range.

A "branch" is a time step that belongs to a sequence of predictions of state of charge derived from operative decisions and a simulation or model of the thermal buffer starting from the starting point, i.e. a current time, and ending at the time point currently being investigated. Once a branch is completed, from start to time horizon, and identified as an acceptable or an optimal one, an operative decision from a predictive step such as the first predictive step can be executed in the operative pathway.

As known from control theory a "trajectory" is a time-ordered set of states of a dynamical system. In accordance with embodiments of the present invention the time-ordered set of states is in the form of a tree with branches. A valid trajectory is one which has branches running from the current time over a series of time steps to the event horizon whereby each branch meets the requirements of methods of the present invention, i.e. has not been pruned.

"Pruning" involves discarding whole branches and stopping trajectories. It is not necessary to complete the calculation of a branch from start to horizon before pruning it. Pruning in early stages helps reduce calculation time.

"Horizon" is the end time of the time interval over which the optimization, i.e. the predictive pathway, occurs. In accordance with embodiments of the present invention there is no fixed state for the buffer at the horizon. Any state outcome that can be reached through a branch that was not pruned, is considered a valid one. Conventional schemes start from a known state at the time horizon and work back.

A "decision" or "operative decision" also called an "output state" (but not a calculated state of charge) relates to the determination of how much the thermal buffer is to be charged over a time step, for example an amount of energy that is to be used to charge the thermal buffer in the time step. The thermal buffer may be charged with heat energy directly from any suitable heat source such as a heat exchanger or from a geothermal plant, or can be charged by a device powered by an energy source, such as an electric heater, a combustion heater, a refrigeration unit, a chiller, comprising for example a compressor. A battery charger can be used to charge a battery. A decision is a choice of control action for the device which charges the buffer in the next time step interval, for example compressor to be set to on, off, at any value between 0 and 100%, e.g. 70%.... For every time step of the predictive pathway all possible decisions are tested.

A "load" is what depletes a thermal buffer. In embodiments of the present invention "load" refers to the losses that occur in the buffer rather than external use of the heat energy from the buffer in another device or system. The load and losses are synonymous for most of the embodiments of the present invention, the load being the buffer itself.

In a "feed-forward" system or method as used in the present invention, control variable adjustments are not error-based. Instead they are based on knowledge about the process in the form of a mathematical model of the process and knowledge about or measurements of the process disturbances. In embodiments of the present invention the measurements are of SoC or a value related to the SoC such as temperature of a buffer medium in a thermal buffer.

Description of the preferred embodiments The present invention relates to methods, systems and devices such as controllers to control the use of energy such as electrical energy and thermal energy while taking into account storage losses, e.g. assuming that heat energy is used directly or electrical energy is transformed to heat energy to charge a thermal buffer that is depleted by losses to the environment or as required by one or more users. As an example a cold store can be envisaged. For example, electrical power can be used to power a device that extracts heat from a cold store and hence to lower the temperature inside. The cold store and its cargo or contents form a thermal buffer, which is charged with cooling energy obtained directly from a heat exchanger for example, or produced by transforming electrical energy to thermal energy by electric compressors. This buffer is depleted as warm cargo is added and/or due to losing cold energy to the environment or because heat energy is received from the environment.

With reference to Fig. 6 in one aspect of the present invention, which can be applied to all embodiments of the present invention, a method is provided to control the use of energy while taking into account losses of energy stored in a thermal energy buffer that is depleted as required by losses or by one or more users, the method comprising: controlling, by means of an output signal 62, heat energy or other forms of energy such as electrical power used to charge the thermal energy buffer or battery;

calculating iterations 52 through time in predictive time steps PTO, PT1 PTH from a current or initial time PTO to a time horizon PTH in accordance with a simulation model of the thermal energy buffer or battery to predict a buffer state of charge (or a value relating to the buffer SoC) at each predictive time step along a plurality of calculation trajectories PT1 to PT13 each of which comprises branches, the buffer state of charge (or a value relating to the buffer SoC) being generated by selection of a limited number of possible operative decisions 01-03 at each predictive time step for charging the buffer, executing an optimisation routine to reduce the number of branches at any or some or each predictive time step by pruning in accordance with a pruning selection criterion (for example PT4, PT5, PT8 are pruned and terminated) to obtain one or more branches at the time horizon (PT1, PT2, PT3, PT6, PT7, PT9, PT10, PT11, PT12, PT13) and selecting one of these trajectories (PT7 shown as thick black line) and determining from this trajectory a single output signal 62, and

charging the thermal energy buffer or battery in accordance with the output signal 62.

In Fig. 6 the output signal 62 is shown as the operative decision 02 (selected from a valid trajectory PT7 between PTO and PTH as an example) of the first or initial predictive time step PT1 which is used by the controller 3 to control the operation of the buffer 10 for the first operative time step OD1. To achieve this, the calculations of the predictive pathway need to be completed early in the operative time step OD1 which is possible with embodiments of the methods of the present invention due to the reduction of computational complexity by means of pruning mechanisms. The output signal 62 can be emitted from a signal output unit 64 for delivery to the controller 3. The signal output unit 64 may be a relay for example, which has an input from the predictive pathway (or has the same signal as an internal signal when the controller 3 calculates the iterations of the predictive pathway) and a binary output - the controller 3 controls the buffer charging device to be either on or off. If there are many potential operative decisions which could be selected the signal output unit 64 can receive an analog value from the selected trajectory and convert this into a digital value between 0 and 1 and the controller 3 drives the buffer charging device at a value between zero and full load.

The method of calculating iterations through time in predictive time steps TO, Tl .... TH in accordance with a simulation model of the thermal energy buffer to predict a buffer state of charge (or a value relating to the buffer SoC) at each predictive time step along a plurality of calculation trajectories results in a tree structure with branches as shown schematically in Fig. 6, each branch involving a calculation of a future state of charge, (or a value relating to the SoC, e.g. one or more temperatures of a buffer medium). This tree structure has branches and new branches are spawned at each iteration step and are being discarded based on a pruning mechanism.

As the calculation method travels forward exploring the branches and predicting the state of charge of the thermal buffer or a value related to the state of charge generated by operating the thermal buffer in accordance with each or any of a set of operative decisions of how to charge the thermal buffer at real times in the future, the availability of energy during each predictive time step of the predictive pathway can be input and a cost for the execution of each branch of the predictive tree structure can be determined and stored. Each predictive time step is a time step in the future from the current or initial time step, and at each of these future time steps an assessment can be made as to whether renewable energy or primary energy such as from fossil fuel fired power stations or nuclear power stations or a combination of these is used. By this assessment renewable energy sources can be linked into an electrical grid in such a way that imbalances are reduced or avoided.

Embodiments The above method may be applied to any of the following or similar thermal buffers or electric batteries each of which is an embodiment of the present invention, of which some are described in more detail below:

A commercial or residential water boiler (for example Fig. 1 or Fig. 4 or a boiler 49 in building 48 of Fig. 3) operates as a standalone or self-contained thermal buffer most of the time as hot water is usually only used sporadically. A commercial oven or other industrial process involving heating, e.g. an oven operated more or less continuously typically uses energy in the first 30 mins to one hour to heat up the materials of the oven after which the oven loses energy mainly through the insulation of the walls. Thus such an oven can be a standalone or self-contained thermal buffer and it is usually relatively easy to make a model or simulation for such an oven when operated continuously.

A residential house, especially with underfloor heating operates like a standalone or self- contained thermal buffer. Suitable thermal storage systems, whether below ground, stratified warm water or chilled water thermal storage systems, above ground warm water or chilled water thermal system and many other thermal storage systems are described in "Sustainable Thermal Storage Systems", Lucas B Hyman at al, McGraw Hill 2011, ISBN 978-0-07-175297-8.

A cold store for storing perishable items for example, operates as a thermal buffer (for example see Fig. 2). Such a cold store can operate as a standalone or self-contained thermal buffer.

Electrical energy storage, e.g. in lead carbon batteries can be used with solar farms as an electrical energy buffer. For such a solar farm, thousands of batteries can be needed to provide power during the night when the solar panels are not delivering power. During the day the batteries can be charged by the solar panels. Solar panel output is not always very regular and load requirements can vary which mean that the batteries might not reach 100% charge during the day (typically in a seven hour charging cycle). Under those circumstances the batteries can be charged from the electric grid during the night or at times when electricity is cheap, e.g. at midday, nights or at weekends or holidays. Buffer embodiment Figure 1 shows a system 1 according to an embodiment of the present invention for performing a method according to an embodiment of the present invention and thereto comprises a thermal energy buffer 10 with an optional inner storage space 28 and a controller 3 provided to perform a method according to the invention. The inner storage space 28 can include cargo to be processed or kept at a certain temperature. Some buffers such as a boiler in residential or industrial premises store water so that the buffer medium and the cargo are the same object and a separate inner storage 28 is not required.

The thermal energy buffer 10 and the controller 3 can be incorporated in a single device, or the controller 3 may be remote from the thermal energy buffer. For example, the controller 3 and the thermal energy buffer 10 can also be physically different devices, for example when several thermal energy buffers 10 are connected to a single controller 3, allowing reduction of the number of controllers 3 necessary.

The thermal energy buffer 10 contains a thermal buffering medium 2 which preferably is a liquid thermal buffering medium but can be a solid buffering medium and can be a phase change material. The thermal buffering medium 2 can be any medium known to the person skilled in the art which allows to store thermal energy in it, but preferably is water as water is known to have good thermal storage properties, is safe and is widely available. To operate at temperatures below 0°C a salt such as sodium chloride can be added to the water. Moreover, but not limited thereto, in such case the thermal energy buffer 10 can also be used to store thermal energy in a warm buffering medium 2 such as warm water. The warm water is in thermal contact with the optional container 28 to keep objects warm or to heat them as in an oven. Hence the buffer may have a heating unit 18a for charging the thermal buffer with heat energy which can be under control of the controller 3 with an optional manual override. The heating unit 18a can have an electrical heater 4 which can be under control of the controller 3 with an optional manual override. An electrical heater 4 is however not critical for the invention and the thermal energy buffer 10 can also be used in combination with a heat pump such that heat recovered by the heat pump can be used to charge the thermal energy buffer 10, or a source of heat energy can be used, e.g. from a district heating system, from a geothermal plant, excess heat from a power station, or can be from a heat exchanger, etc. Embodiments of the present invention can include several different means for charging the thermal buffer 10 with heat energy. The heater 4 shown in figure 1 is an electric heater and is situated at the bottom of a tank inside the thermal energy buffer 10. Such a configuration is however not critical for the invention. It is for example possible to provide a heater 4 which is not electric but which, for example, can be a combustion heater that uses wood, gas, petrol, diesel fuel, etc. Also, the position of the heater 4 is not critical for the invention and can be at the bottom, near the middle, near the top, etc. However, by providing the heater 4 near the bottom it has been found that natural heat convection of the thermal buffering medium 2 when heated by the heater 4 allows that the thermal buffering medium 2 is heated homogeneously.

Alternatively, the thermal energy buffer 10 can include a cooling unit or refrigeration unit 18b for charging the thermal buffer 10 with thermally cold energy which can be under control of the controller 3 with an optional manual override. Hence the thermal energy buffer 10 can be a cold store. The optional inner store room or container 28 is in thermal contact with the cold water.

Other buffering media are included within the scope of the present invention such as for example gels having good thermal storage properties, or phase change materials.

The controller 3 is adapted to control the energy charging unit 18a and/or 18b. Loss of thermal energy from the thermal energy buffer 10, for example from the optional store or container 28, can occur due to convection or conduction, e.g. through the walls of the thermal buffer 10 and/or through opening doors or access openings or other forms of heat loss.

The thermal energy buffer 10 shown in figure 1 comprises an inlet 5 and an outlet 6. Optionally, the inlet 5 can be positioned such that the thermal buffering medium 2 enters the thermal energy buffer at the bottom and optionally, the outlet 6 can be positioned such that the thermal buffering medium 2 exits the thermal energy buffer 10 at the top. This has as a consequence that heated thermal buffering medium 2, which rises to the top due to convection, becomes near to the outlet 6. As the heater 4 preferably is located near the bottom, cold thermal buffering medium 2 entering near the bottom through the inlet 5, is heated by the heater 4 and afterwards rises to the top where the outlet 6 is located. Such a configuration has been found to further improve the homogeneous heating of the thermal buffering medium 2. The thermal buffer 10 is preferably a standalone or self-contained buffer and hence has no external load (the thermal buffer is itself the load), such that no buffering medium 2 is extracted from the thermal buffer and none is added to it except to compensate for any losses by conduction, convection or evaporation for example. For example, if a heat exchanger 30 is used to charge the thermal buffer 10 as shown in Fig. 4, then the buffering medium 2 can be circulated, e.g. pumped through the heat exchanger 30 which is supplied with heating or cooling fluid from an energy source 38. For this operation the thermal energy buffer 10 shown in Fig. 4 comprises an outlet 36 and an inlet 35. Optionally, the outlet 36 can be positioned such that the thermal buffering medium 2 leaves the thermal buffer at the bottom (rather cold) and optionally, the inlet 35 is positioned such that the thermal buffering medium 2 enters the thermal buffer 10 at the top after being heated by the heat exchanger 30. This has as a consequence that cold thermal buffering medium 2, which falls to the bottom, exits through outlet 36. Such a configuration has been found to further improve the homogeneous heating of the thermal buffering medium 2 which is in thermal contact with the optional store or container 28. As shown in Fig. 4, the controller 3 can include or can be a computer 8 which is part of or is connected to the controller 3, over for example a computer network 7. The computer network 7 can for example be a LAN or the internet and can be a physical wire or, for example, wireless network such as for example a WiFi or Bluetooth network. The controller 3, for example, is provided with a server application, for example a web server application, allowing the computer 8 to log in to the website to control the cold store 10'.

The exact configuration of the inlet 5; 35 and the outlet 6; 36 is not critical for the invention. Although they are shown here as pipes entering and leaving the thermal energy buffer 10 at the bottom or at the top, this is not critical for the invention. Preferably, the inlet 5; 35 and the outlet 6; 36 are configured such that the thermal energy buffer 10, (e.g. preferably the tank provided in it), is substantially always, preferably always, filled with thermal buffering medium. For example, the outlet 6; 36 and the inlet 5; 35 can be controlled by controllable valves VI, V2 which can be under control of the controller 3. The level of filling of the thermal energy buffer 10 is preferably obtained (e.g. by controlling valves VI and V2) by configuring the inlet 5; 35 and the outlet 6; 36 such that when thermal buffering medium 2 is drawn from the thermal energy buffer 10 through the outlet 6; 36, new thermal buffering medium 2 is led into the thermal energy buffer through the inlet 5; 35 until the thermal energy buffer 2 is, or preferably its tank, is filled again with thermal buffering medium 2 such that the tank remains substantially filled, preferably filled. The volume of the thermal buffering medium 2, and accordingly the tank of the thermal energy buffer in which it is contained, is preferably subdivided in at least one part 21 and suitably in a number of parts 21, 22, 23, 24, 25, 26, 27. Preferably, at least two parts are provided, more preferably even more such as for example at least three, four, five, six, seven eight, etc. The number of parts is not limited and can be determined by the person skilled in the art. As can be seen in figure 1 or Fig. 4, the parts 21, 22, 23, 24, 25, 26, 27 subdividing the volume of the thermal energy medium are provided on top of each other along an upright direction forming a stack of parts 21, 22, 23, 24, 25, 26, 27.

As can be seen in figure 1 or figure 4, the different parts 21, 22, 23, 24, 25, 26, 27 of the volume of the thermal buffering medium 2 together form the total thermal buffering medium 2 present in the thermal energy buffer 10 and the thermal energy buffer 10 comprises a number of respective one or more temperature sensors 11, 12, 13, 14, 15, 16, 17 for each part 21 , 22, 23, 24, 25, 26, 27 for sensing a temperature of the thermal buffering medium 2 contained in the corresponding part 21 , 22, 23, 24, 25, 26, 27. In combination with the preferred stack of parts 21 , 22, 23, 24, 25, 26, 27, it has been found that such a configuration allows an efficient way of sensing the temperature profile of the thermal buffering medium 2 as the temperature varies substantially only in height direction. As for the parts, the number of temperature sensors is not limited and can be determined by the person skilled in the art.

Although the parts 21, 22, 23, 24, 25, 26, 27 are indicated as such in figure 1 or figure 4, it is to be understood that the parts 21, 22, 23, 24, 25, 26, 27 only imaginarily subdivide the volume of thermal buffering medium 2 and not physically.

Preferably, the sensors 11 , 12, 13, 14, 15, 16, 17 are placed along the thermal energy buffer 10 such that the position of each of these sensors corresponds to the position of each of the corresponding parts 21, 22, 23, 24, 25, 26, 27 subdividing the total volume of the thermal buffering medium 2. Thereto, preferably the temperature sensors are equidistantly distributed along the height of the thermal energy buffer 10, or along the height of the tank 18 comprised by the thermal energy buffer 10 and containing the thermal buffering medium 2. The sensors 11 to 17 are in communication with the controller 3.

Optionally, thermal energy can be removed from or circulated through the thermal energy buffer 10 by extracting the buffering medium 2, e.g. the buffering medium 2 may be pumped out of the thermal energy buffer 10 or may be drained by gravity. The extracted charged buffering medium 2 may be used directly (e.g. in an air conditioning system or in an industrial process) and then discarded or used in some other way but it is preferred if the buffer is self-contained. Alternatively, the buffering medium 2 may be passed through a heat exchanger 30 for extraction of energy or addition of energy before the buffering medium 2 is returned to the thermal energy buffer 10, 10' for re-charging with warm or cold energy. Hence, the heat exchanger 30 can be provided with heating or cooling fluid from an energy source 38.

The controller 3 for buffering thermal energy of a thermal energy buffer is preferably adapted to perform a method according to the present invention or implement a system in accordance with the present invention. The controller can be implemented as a microcontroller for example and may include a processor such as a microprocessor or an FPGA and one or more memories. The processor can be adapted to execute any of the software required to execute any embodiment of the present invention. The controller 3 can include or can be a computer 8 which is part of or is connected to the controller 3, over for example a computer network 7. The computer network 7 can for example be a LAN or the internet and can be a physical wire or, for example, wireless network such as for example a WiFi or Bluetooth network. The controller 3, for example, is provided with a server application, for example a web server application, allowing the computer 8 to log in to the website to control the thermal energy buffer 10.

The controller 3 can calculate a value representing the amount of total thermal energy present in the thermal energy buffer by multiplying the temperature measured by each sensor 11 , 12, 13, 14, 15, 16, or 17 corresponding to respectively part 21 , 22, 23, 24, 25, 26, 27 of the thermal buffering medium 2 with the volume of the corresponding part of thermal buffering medium 2 such as to obtain a value representing the partial thermal energy contained in the corresponding part of the thermal buffering medium 2 and adding the resulting partial thermal energy values to each other to obtain the total thermal energy in the thermal energy buffer 10. This calculation can also provide a State of Charge (SoC) of the thermal energy buffer 10 by the controller 3 calculating the ratio of the total thermal energy present to the maximum possible thermal energy stored in the thermal energy buffer. The present invention provides a system 1 and devices such as the controller 3 to control the use of energy such as electrical energy and thermal energy while taking into account storage losses, e.g. assuming electrical energy is used to charge the thermal energy buffer 10 that is depleted as required by one or more users. As an example a cold store can be envisaged. For example, electrical power can be used to extract heat from a cold store and hence to lower the temperature inside. The cold store and its cargo or contents (buffering medium 2) form the thermal energy buffer 10, which is charged with cooling energy produced from electric compressors, or a heat exchanger, for example. This thermal energy buffer 10 is depleted as warm cargo is added and/or due to losing cold energy to the environment, optionally due to buffering medium being extracted from the thermal energy buffer 10 or because heat energy is received from the environment.

As indicated above electricity can be bought on spot markets, e.g. from a day ahead market. In this case the price for a 24 hour period is published say 10 hours ahead. To optimise the use of electrical energy it is an aspect of embodiments of the present invention that the controller 3 is preferably adapted to predict the state of charge in the thermal energy buffer 10 over a time period long enough that significant changes in the price of electricity occur and hence an optimisation can be carried out. For example, this prediction can translate into the prediction of the cold store temperature or state of charge in the future or at a plurality of times in the future. This time period can be 10 hours, 24 hours or one or more days, for example.

As explained above system 1, or controller 3 according to embodiments of the present invention executes a method according to an embodiment of the present invention and has an input related to a measurement of one or more actual temperatures of the buffer such as cold store temperatures, e.g. as measured by one or more temperature sensors 11, 12, 13, 14, 15, 16, 17. Hence, more than one temperature can be measured, e.g. the temperature of a heat energy buffer fluid such as water or a phase change material may be measured at different heights in the thermal energy buffer 10. The one or more temperatures determine the state of charge of the buffer. The State of Charge is one parameter that is preferably used by the controller 3 to control the energy flow such as electricity flow i.e. that flow required to charge the thermal energy buffer 10. The controller 3 has an output signal that is used to control the electricity flow i.e. required to charge the thermal energy buffer 10 with cooling or heating energy.

The system 1 or controller 3 according to embodiments of the present invention executes a method according to an embodiment of the present invention and hence is preferably adapted to find a good and often an optimal solution, e.g. a cheap or the cheapest solution or one that has a good or the best environmental impact or one the combines renewable energy with energy from fossil fuel power stations or nuclear power to reduce imbalances. This can be done by using an optimisation technique and dynamic programming (DP) is one example of an optimisation technique. The optimisation technique used may be based on a Markov chain in which the process goes through a sequence of random variables with a Markov property defining serial dependence only between adjacent periods linking time points in the development of the process. Thus, the system 1, or the controller 3 may include a system model and is adapted to follow a chain of linked events, where what happens next depends only on the current state of the system. Hence the system 1, or the controller 3 according to embodiments of the present invention is adapted to execute a method according to an embodiment of the present invention and hence is preferably adapted to iterate through time (e.g. in steps of a convenient length such as perhaps 1 hour) and to examine for every time step the result in terms of buffer state of charge (SoC - or a value related to the SoC, e.g. the temperature of the buffer medium 2) for all possible operative decisions for the output signal. This output signal may control one or more devices such as turning a charging device OFF such as a compressor OFF (to stop charging of the thermal energy buffer 10) or turning a charging device ON such as a compressor ON to charge the thermal energy buffer 10, operating a charging device such as a compressor at any value between 0 and 100%, e.g. 70%, 100% load, turning a pump ON or opening a controllable valve (e.g. to extract medium 2) or turning a pump OFF or closing a controllable valve. This procedure potentially leads to a very large number of possible combinations. For example, when the time horizon is 34 hours and three operative decisions or output states are possible at each time step, this results in 17 x 10¹⁵ combinations. In case of fine control it may be required to use a higher time resolution (shorter periods) and more output possibilities (e.g. greater than three) so the number of possible combinations or branches may be significantly higher, in fact prohibitively higher.

The time horizon can be selected such that the future date of the horizon makes a compromise between short term and long term advantages. There is no requirement for the time horizon to be the same for each determination of the outcomes. For example let us assume that the time horizon is set at 96 time steps of 15 minutes = 24 hours. This can be advantageous as there is often a daily cycle of use in thermal buffers of industrial applicability. Hence it is useful to select the time horizon which respects cyclic behaviour. The controller according to embodiments of the present invention can be adapted to buy electricity on the day ahead market. Prices are published at say 16:00 for each day from 00:00 to 24:00. At that point in time, e.g. if the current time is 16:00, the prices for today and tomorrow are known so the horizon can be set to 8hours + 24hours or 32 time 4 predictive time steps of 15 minutes = 128 time steps. Just before the prices were published the time horizon can be set to 8hours. Similar considerations can be made based on other constraints which can affect the thermal buffer such as the weather forecast, in particular the forecasting of outside temperature. A time horizon may be determined based on several parameters and there is no requirement for it to be constant. For example, electricity prices can fluctuate with ½ day periods : it is usually expensive in the morning and in the evening and less expensive at midday. A time horizon can be selected that allows to predict beyond the next expensive phase. 8 hours horizon at 9:00 would not be enough because electricity will be expensive at 9+8 = 17:00 but it is acceptable at 16:00 because prices will be low at 16+8= 24:00. Optionally a long or maximum time horizon can be selected so that unexpected low or negative prices (e.g. due to a lot of wind at night) and can be included within the predictive optimization. In such a case the maximum time horizon may be set by the calculation time for all the predictive time steps up to the event horizon which must all be calculated well within each operative time step. The system 1, or controller 3 according to embodiments of the present invention algorithm adapted to execute a method according to an embodiment of the present invention can start a control procedure as late as possible in order to have the latest information about the charge in the thermal energy buffer 10 and early enough in order to finish before a new electricity price applies. Hence a high resolution or a short calculation time is preferred.

The system 1, or controller 3 according to embodiments of the present invention adapted to execute a method according to an embodiment of the present invention is adapted to examine at each predictive time step a number of predetermined possible alternatives for the controlling operative decision for the operative pathway. The system 1 or controller 3 according to embodiments of the present invention adapted to execute a method according to an embodiment of the present invention preferably applies a method to reduce the number of branches to be examined at each time step.

The system 1 or controller 3 according to embodiments of the present invention adapted to execute a method according to an embodiment of the present invention is preferably adapted to execute a branch and bound method in addition to an optimisation method such as DP, as a first pruning mechanism. If the optimisation calculation such as a DP calculation is started using a value related to the state of charge, such as one or more measured temperatures of the medium 2 in the thermal energy buffer 10, each possible decision results in a new prediction of these one or more temperatures at a future time, e.g. after lapse of the next time step. Once these one or more predicted temperatures exceed minimum or maximum values the procedure is assumed to have reached the end of a realistic technical option and this branch and this trajectory is terminated. For example, no more calculations are performed on it and any previously achieved results are discarded. The system 1 or controller 3 according to embodiments of the present invention adapted to execute a method according to an embodiment of the present invention is preferably adapted to perform a second pruning mechanism based on a preliminary simulation of an arbitrary control strategy, e.g. a thermostat strategy to generate a benchmark for comparisons. The cost of this strategy, i.e. a setting of a cost valve or an availability of an energy source, is used to identify the maximum cost of any branch in the optimisation tree or DP tree, and is used in the method according to the present invention in a similar way to the maximum or minimum temperature in the first pruning. More expensive branches are discarded. This pruning mechanism is especially helpful in the later iterations when the number of combinations becomes extremely high and is therefore a very efficient means of pruning.

Due to the high number of possible combinations, it is inevitable that an identical or similar SoC may be reached at a given time (iteration) but at a different cost, i.e. a different setting of a cost valve which relates to a different technical availability of an energy source. A similar SoC may be determined by a difference between two SoC's being less than a certain threshold value. The system 1 or controller 3 according to embodiments of the present invention adapted to execute a method according to an embodiment of the present invention is preferably adapted to perform a third pruning operation, i.e. in such a situation the or any more expensive branch is discarded as soon as this occurs, i.e. one having a lower opening of a cost valve or availability of an energy source.

Finally, an identical or similar cost (i.e. identical or similar opening of a cost valve or a technical availability of an energy source) may lead to different SoCs at a given time. A similar cost may be determined by a difference between two costs being below a threshold value. The system 1 or controller 3 according to embodiments of the present invention adapted to execute a method according to an embodiment of the present invention is preferably adapted to perform a fourth pruning operation, i.e. in such a situation a branch yielding the or any lower SoC is discarded.

The SoC (temperature) of the thermal energy buffer 10 is a function of the load on it (any losses) and the charging operation, e.g. the use of compressor power to provide cooling energy or operation of a heater to provide hot thermal energy or a heat exchanger to provide hot or cold thermal energy. As the optimisation method e.g. DP investigates the future it is possible to predict a SoC. A simulation or model of the thermal energy buffer 10 is required as part of the algorithm that is to execute in the controller 3. This simulation or model can be any convenient function like constant losses or charge (linear temperature changes), RC models (exponential changes or sum of ditto), lumped ciruits or any other convenient function or combination thereof.

The model parameters can be calibrated either manually or automatically if a system identification (SI) algorithm is used. The SI does not need to run together with the optimisation such as DP but may be run as part of a different task, e.g. once per day. So even complex processing will not influence the performance of the optimisation such as DP. If the buffer simulation or model functions are complex than the SI may be performed using Levenberg-Marquart (or Gauss-Newton algorithm (GNA) or the method of gradient descent) to find the optimal parameter fit. Depending on the model used, it may be required to use different functions for buffer charging and discharging. In this case either an additional input is required to determine the current state (e.g. compressor is ON or OFF) or this information must be derived from the evolution of the SoC (e.g. evolution of a value related to the SoC such as the temperature changes of the buffer medium 2).

Once the predictive calculations are performed and after pruning, there will be/are one or more trajectories containing a sequence of potential operative decisions of different branches that result in different states of the buffer at different time steps and which are present at the time horizon. All or any of these states comply to imposed limits (due to the pruning mechanism that has been applied) so the branches of a valid trajectory can be executed. If there is no branch surviving at the time horizon then the predictive pathway has failed. An alternative method of deciding on the operative decision for the current time is then used, e.g. based on a thermostat, maintaining the operative decision from a previous operative time step, etc. If there is more than one branch surviving at the horizon, one branch is selected according to selection criteria or a selection criterion.

As the calculation method travels forward exploring the branches and predicting the state of charge of the thermal buffer or a value related to the state of charge as caused by operating the thermal buffer in accordance with each or any of a set of operative decisions of how to charge the thermal buffer, the availability of energy during each predictive time step of the predictive pathway can be input and a cost for the execution of each branch of the predictive tree structure can be determined There are various ways that an acceptable or optimal trajectory can be selected: a) Because energy used to charge the buffer is not or need not be equally available during the intended time interval of the whole predictive process (availability of a certain type of energy source such as wind power fluctuates over time having an effect on cost) different branches will require more or less available energy (e.g. will cost more or less). The trajectory (which is a sum of all its branches) that uses the energy which is most available (e.g. the cheapest branch) can be regarded as an acceptable or optimal branch. The first (starting) operative decision of this trajectory is the proper control action 46 (operative decision - see Fig. 3) to be taken for the current operative time step. b) The trajectory that uses the energy which is most acceptable from an environmental point of view (least environmental impact) can be selected. The first (starting) operative decision of this trajectory at the current time in the predictive pathway is then the proper control action 46 (operative decision) to be taken for the current operative time step. c) . Each predictive time step is a time step in the future from the current or initial time step, and at each of these future time steps an assessment can be made as to whether renewable energy or primary energy such as from fossil fuel fired power stations or nuclear power stations or a combination of these is used. By this assessment renewable energy sources can be linked into an electrical grid in such a way that imbalances are reduced or avoided and a valid trajectory can be selected which meets a good compromise of renewable energy and energy from fossil fuel power stations or nuclear power stations. c) The trajectory that uses the energy which is most acceptable from other points of view (e.g. commercially, politically more acceptable, such as a branch not using nuclear power) can be selected. The first (starting) operative decision which was used in the predictive pathway at the current time is then the proper control action 46 (operative decision) to be taken for the current operative time step. Following the end of the predictive pathway and the selection of a branch as the outcome of the predictive pathway, a flag and a variable can be set: the flag signals that the optimization algorithm has found a valid solution for the current operative time step. The outcome of the predictive pathway is one of the first (starting) operative decisions of a selected branch and an output signal is generated to execute the selected outcome. The variable is set to the optimal decision e.g. as a binary or an analog 0..10V setpoint for the device providing the energy to the buffer, e.g. a compressor inverter. If no outcome is available from the predictive pathway then the output signal can be generated to activate a relay to select a back-up control signal e.g. from a thermostat.

A selected and executed operative decision will expire at the next time step. One option is that the second operative decision of the optimal branch may be executed. This is advantageous when the optimisation calculations are computationally expensive. Thus the operative pathway may allow the decisions for a number of operative time steps ( e.g. up to five) to be taken from one calculated predictive pathway. However, this need be not optimal, for example:

Consider two time intervals, the first starts at tO and ends at tl and the second starts at tl and ends at t2. Assume that operative decisions DO-1 and Dl-2 are the optimal way to go from state SO to SI and from SI to S2. At tO DO-1 is executed. This decision expires at tl. Decision Dl-2 is based on the calculated state SI. However by the time tl is reached a measurement of SI would be more accurate than the predicted SI. So instead of using Dl-2, the predictive pathway and the optimization can restart at tl. The states previously known as SI, S2, now become SO, SI so again only the first decision needs to be executed. This method is much more accurate than conventional methods as it takes into account any deviations between predicted states and real states as soon as possible.

Note that calculations preferably should be made in real time because all the parameters in the model description formulas may change: temperatures and time constants. It can be impractical to store all results for all decisions on all states for all combinations of inputs.

Embodiment of a cold store

As indicted above and algorithm is described that assumes heat energy can be used directly or other forms of energy such as electricity can be transformed to heat energy and used to charge a buffer that is depleted as required. An example of such a thermal buffer is a cold store 10' . Electricity can be used to extract heat from a cold store (Fig. 2), e.g. through a cooling unit 18b optionally under the control of a controller 3. The controller 3 for controlling the cold store 10' is preferably adapted to perform a method according to the present invention or implement a system in accordance with the present invention. The controller 3 can be implemented as a microcontroller for example and may include a processor such as a microprocessor or an FPGA and one or more memories. The processor can be adapted to execute any of the software required to execute any embodiment of the present invention.

The controller 3 can include or can be a computer 8 which is part of or is connected to the controller 3, over for example a computer network 7. The computer network 7 can for example be a LAN or the internet and can be a physical wire or, for example, wireless network such as for example a WiFi or Bluetooth network. The controller 3, for example, is provided with a server application, for example a web server application, allowing the computer 8 to log in to the website to control the cold store 10'.

A cold store 10' comprises thermal insulation to keep objects - "cargo" - therein cool without a large leakage of heat into the store. The large amount of insulation and the cargo once cold provide a thermal energy buffer within the cold store 10' . The cold store 10' and its cargo are representative of a thermal energy buffer as described above and comments related to the buffer 10 of the first embodiment apply to this embodiment. Cold store 10' is charged with cold energy produced from, for example a cooling unit 18b having for example electric compressors, and the cold store is depleted as warm cargo is added and also cold energy is lost to the environment through the walls and doors of the cold store 10' .

The thermal buffer 10' is preferably a standalone or self-contained buffer and hence has no external load (the thermal buffer is itself the load), such that no heat energy is extracted from the thermal buffer and none is added to it except to compensate for any losses. For example, if a heat exchanger 30 is used to charge the thermal buffer as shown in Fig. 2, then the buffering medium 2 can be circulated, e.g. pumped through the heat exchanger 30 which can be provided with heating or cooling liquid from energy source 38. For this operation the thermal energy buffer 10' shown in Fig. 2 comprises an outlet 36 and an inlet 35. Optionally, the outlet 36 can be positioned such that the thermal buffering medium 2 leaves the thermal buffer at the top and optionally, the inlet 35 is positioned such that the thermal buffering medium 2 enters the thermal buffer 10 at the bottom after being cooled by the heat exchanger 30. This has as a consequence that warm thermal buffering medium 2, which rises to the top, exits through outlet 36. Such a configuration has been found to further improve the homogeneous heating of the thermal buffering medium 2 and the optional container or store 28 which is in thermal contact with the buffering medium.

The cold store 10' is operated as described for methods of calculating the predictive pathway and obtaining an output signal to drive the cold store during time steps of the operative pathway as explained below. General operation applicable to all embodiments and in particular the cold store

The operation of a controller 3 or a system 10 or 10' or a method according to embodiments of the present invention can be based on a demand and supply as follows. Accordingly in any of the embodiments of the present invention the system or controller or the method selects a control action (46 see Fig. 3) e.g. in the range { 0,1 } for every time step t=l T (i.e. within the time range of t = 1 with integer steps up to T) T being the time horizon. The result of a (46) is an energy demand dt [J], (J being a unit of energy such as Joule) which is a function depending on physical parameters of the thermal energy buffer. For example for a cold store 10' these parameters may include parameters such as the outside and inside temperature (T_out and Tin respectively).

This demand needs to be met by a supply, e.g. fossil fuel plants such as gas fired plants, nuclear power stations (42) or renewable energy (44). However, covering 1 J of demand at building level requires primary energy (42) and results in a health H [NO_x] (the health aspect is represented by an amount of nitrogen oxides, for example) and environmental impact E [CO2] (the environmental impact is represented by an amount of carbon dioxide in the atmosphere, for example), the amount of which depends on the balance between demand and supply at system level p_t.

When expressed as an objective this results in:

Ot = H(g(Ut,Tout,Tin, ),pt) + E(g(Ut,Tout,Tin, ),/?t ) The system or controller or the method according to embodiments of the present invention can optionally make use of the objective of the control strategy to provide the control values that minimize the time integrated objective: c* = H( g (ut, T<mt, Tin, ), pt) + E(g (ut, Tout, Tin, ), pt)

It is not expected that the present invention is limited by mathematical methods of solving such an equation.

An element required to optimize the environmental impact is the prediction of the charge in the buffer or a buffer charge versus time function. The system or controller or the method according to embodiments of the present invention use is made of a buffer and load model to predict its state of charge or a value relating to the state of charge.

In the case of electric storage in a battery, a state of charge (SoC) algorithm can be used as known from the prior art. It usually involves coulomb counting and accounting for losses as self-discharge or power to auxiliary devices such as the battery management system. These algorithms need to be complemented by an electric load versus time prediction. These may be identical to load patterns known from experience. For example, in the system or controller or the method according to embodiments of the present invention for the cold store 10' as a thermal energy buffer, the state of charge can translate to the prediction of a value related to the SoC of the cold store such as the cold store temperature. The cold store buffer 10' is depleted by losses. The losses are due to heat transfer to the environment and activity in the cold store. They depend on more or less time invariant parameters as the cold store and cargo heat capacity and on time variant parameters as outside air temperature and activity in the cold store. For the system or controller or the method according to embodiments of the present invention, the model can be described as

T_in(t+At) = Tbulk(t+At) - (Tbulk(t+At) - Tin(t)) x e '

Tbuik(t+At) = Tout - (Tout - Tbuik(t)) x e ^Atk2

The buffer is charged by means of an electrically powered heat pump according to :

Tin(t) = Tevap(O) + (Tevap - T_in(0))^■ e^"^³

The equation is evaluated by iteration where :

At = iteration step

t = at time t

t + At = at next iteration step

Tin = cold store internal temperature representing the state of charge

Tbuik = bulk temperature, it is never measured and is not constant, a new value is

calculated on every iteration as shown in equation 2.

τι = time constant for the first phase of the heating (loss) curve where the colder air heats to the bulk temperature

τ2 = time constant for the second phase of the heating (loss) curve where the cold store and everything in it heats up to the ambient air

τ3 = time constant for the cooling phase where the cold store and everything in it cools to the evaporator temperature

Tout = outside ambient temperature

Tevap = practical inside air end temperature for cooling

The third expression is valid throughout the cooling process so it includes the losses during cooling. The ambient air plays a considerable role in the losses. This variable is optionally included in the equations making the model more accurate. However it requires a prediction of the outside air temperature to be available.

The system or controller or the method according to embodiments of the present invention involving a buffer 10 such as the cold store 10', can have an input to measure the actual buffer e.g. cold store temperature (in general: a value to determine the state of charge of the thermal energy buffer) and an output to control the energy flow such as electricity flow (in general: charging the buffer). The measured temperature is Tin(0). There may be a plurality of temperature sensors and each sensor can be associated with a layer or part of the buffer such as the cold store 10'. Tin(0) can then be calculated as an average value taking into account all of the sensor outputs.

It is not expected that the present invention is limited by mathematical methods of obtaining a simulation or system model of the state of charge of a thermal energy buffer. For example, the system or controller or the method according to embodiments of the present invention can use an optimisation method such as dynamic programming (DP) to find an economical solution. It iterates through time (e.g. in steps of 1 hour) and examines for every time step the result in terms of buffer state of charge (SoC) or a value which relates to the SoC which can be calculated, for example as a temperature of the cold store, for all possible operative decisions for the output (e.g. charging device such as compressor OFF, charging device such as compressor operated at any value between 0 and 100%, e.g. 70%, compressor 100% of full load). At the same time the costs for executing the branch can be estimated and stored based on the availability of energy at that time. This leads potentially to a very large number of possible combinations: when the time horizon is 34 hours and three operative decisions or output states are possible, this results in 17xl0¹⁵ combinations. It may be required to use higher time resolution and more output possibilities so the number of possible combinations or branches may even be significantly higher. The system or controller or the method according to embodiments of the present invention should preferably start off as late as possible in order to have the latest information about the charge in the buffer but early enough in order to finish before a new electricity price applies. A short calculation time is therefore required. The number of branches to be examined is preferably reduced, sometimes drastically reduced. This is accomplished by applying one or more pruning mechanisms. A first such pruning mechanism is adding a branch and bound applied on top of the optimization method such as DP. If an optimisation calculations such as a DP calculation is started using the measured temperature of the cold store (as a value related to the SoC of the store), each possible operative decision taken over the next predictive time step results in a new temperature prediction. Once this predicted value related to the SoC of the store such as the predicted temperature exceeds minimum or maximum values this branch is terminated, no more calculations are performed on it and previous results are discarded.

A second pruning mechanism requires a preliminary simulation of an arbitrary control strategy, e.g. a thermostat strategy to generate a benchmark for comparisons. The cost of this strategy is used as the maximum cost of any branch in the optimization tree, such as the DP tree, similar to a maximum or minimum temperature. More expensive branches are discarded. This pruning mechanism is especially helpful in the later iterations when the number of combinations becomes extremely high and is therefore is a very efficient means of pruning.

Due to the high number of possible combinations, it is inevitable that an identical SoC or identical value related to the SoC of the store may be reached at a given time (iteration) but at a different cost. The more expensive branch is discarded as soon as this occurs.

Finally, an identical cost may lead to different SoCs or values relating to the SoC at a given time. The branch yielding the lower SoC or the value relating to the lower SoC can be discarded, for example.

The combination of these pruning mechanisms can reduce the calculation time of the DP algorithm by a factor of 1000 or significantly more. This allows the algorithm to be used in real time applications with on-line calculation.

The SoC (e.g. temperature) of the buffer or a value related to the SoC is a function of the load on the buffer (e.g. losses) and the charging (e.g. compressor power). As the optimization method such as the DP algorithm is looking into the future it needs to be possible to predict the SoC or the value related to the SoC in the future. The system or controller or the method according to embodiments of the present invention is adapted to use a simulation of the buffer. This simulation can be any convenient function like constant losses or charge (linear temperature changes), RC models (exponential changes or sum of ditto), lumped circuits or any other convenient function or combination thereof.

The model proposed above is a practical simplified model. More accurate predictions can potentially be obtained using more complex models. In the system or controller or the method according to embodiments of the present invention, the model parameters can be calibrated either manually or automatically if a system identification (SI) algorithm is used. The SI algorithm does not need to run together with the optimization method such as the DP algorithm but may be run as part of a different task, e.g. once per day. So even complex processing will not influence the performance of the optimization method such as DP. If the buffer simulation functions are complex than the SI may be performed using Levenberg-Marquart (or Gauss-Newton algorithm (GNA) or the method of gradient descent) to find the optimal parameter fit. Depending on the model used, it may be required to use different model functions for buffer charging and discharging. In this case either an additional input is required to determine the current state (charging device such as a compressor is ON or OFF) or this information must be derived from the evolution of SoC (or evolution of a value related to SoC such as a temperature change of the buffer medium).

As the number of inputs to a model increases, the calculation time may increase but this will result in almost no increase in overall time because the number of simulations does not increase, just the complexity of them and these can be calculated off-line.

Examples of other models are :

Higher order models can be used such as a second order heating or cooling model. The formulas for cooling or heating will be similar to those for losses described above so a fourth time constant would need to be added.

Another example of model takes into account convective and conductive losses and heat transfer from the cargo. This model requires an equation to describe these three cases of heat transfer, and it would need to take into account thermal mass of inside air and cargo and the buffer wall thermal resistance such as a cold store wall thermal resistance. This model can be extended to include the mass of the buffer building such as the mass of the cold store building. For example, the model can include a concrete floor as an additional thermal mass and/or loss path.

An aspect of embodiments of the present invention is the determination of the inputs of the buffer model equations. It is usually not practical to determine these parameters manually as a lot of labour will be involved. So an automated approach is preferred such as system identification or SI.

In the case of the proposed model, the SI needs to fit three time constants and a value representing the SoC such as a bulk temperature if this is not available through measurement. Measurement data is collected and equations are selected which apply (for example, is the charging device such as a compressor on or off ?) and tries to fit the proposed equations to these data. This can be done with a least squares approach, for example, for exponential functions that are fitted to the measured data. One such fitting method is known as Levenberg-Marquart. An example of a fit ("x" points) on data measured ("+" points) during approximately three hours while the compressor is running and shown in Fig. 5. The fit of measured points to calculated has been made as is described above. Tin(0) is the first measured data point, Tau3 and Tevap are fit results. In Fig. 5 the X axis is time [s], and the Y axis is inside air temp [°C].

For more complex models, more parameters need to be fitted and more equations are involved and the SI becomes more complex.

If the behaviour of a buffer such as a cold store differs significantly over time, i.e. the time constants depend on time, this may be taken into account e.g. by identifying time constants for day night/ weekday/ weekend/holiday operation. In most cases, with a time horizon more than one day/night cycle, the equation that allows to calculate a next state based on a previous state and a decision will be different depending on the start time of the optimization and the number of iterations performed. To achieve this a set of look-up tables can be used, one for each set of time constants.

Further embodiment One or more aspects of the system, controller or method according to embodiments of the present invention can be implemented in digital electronic circuitry, or in computer hardware, firmware, software, or in combinations of them. The processing system may include a storage subsystem that has at least one disk drive and/or CD-ROM drive and/or DVD drive. In some implementations a user interface subsystem may be provided for a user to manually input information or adjust the operation. More elements such as network connections, interfaces to various devices, and so forth, may be included in some embodiments. The various elements of the processing system may be coupled in various ways, including via a bus subsystem e.g. a single bus, but will be understood to those in the art to include a system of at least one bus. The memory of the memory subsystem may at some time hold part or all of a set of instructions that when executed on the processing system implement the steps of the method embodiments described herein. The present invention also includes a computer program product which provides the functionality of any of the methods according to embodiments of the present invention when executed on a computing device. Such computer program product can be tangibly embodied in a carrier medium carrying machine-readable code for execution by a programmable processor. The present invention thus relates to a carrier medium carrying a computer program product that, when executed on computing means, provides instructions for executing any of the methods as described above. The term "carrier medium" refers to any medium that participates in providing instructions to a processor for execution. Such a medium may take many forms, including but not limited to, nonvolatile media, and transmission media. Non-volatile media includes, for example, optical or magnetic disks, such as a storage device which is part of mass storage. Common forms of computer readable media include, a CD-ROM, a DVD, a flexible disk or floppy disk, a tape, a memory chip or cartridge or any other medium from which a computer can read. Various forms of computer readable media may be involved in carrying one or more sequences of one or more instructions to a processor for execution. The computer program product can also be transmitted via a carrier wave in a network, such as a LAN, a WAN or the Internet. Transmission media can take the form of acoustic or light waves, such as those generated during radio wave and infrared data communications, Transmission media include coaxial cables, copper wire and fibre optics, including the wires that comprise a bus within a computer. Accordingly, the present invention also includes a software product which when executed on a suitable computing device carries out any of the methods of the present invention. Suitable software can be obtained by programming in a suitable high level language such as C and compiling on a suitable compiler for the target computer processor. Target computer processor can be (for example but not limited to): the general purpose processor (CPU) in a computer system, a graphical processor (such as a GPU) of a computer system, a general purpose processor present in a display system, a graphical processor (such as a GPU) present in a display system, an embedded processor present in a display system, a processor present in a panel such as a LCD panel or OLED panel or plasma panel, a processor present in the driver system of a liquid crystal display panel.

Accordingly the present invention provides a computer program product to control the use of energy while taking into account storage losses of energy stored in a thermal energy buffer that is depleted as required by one or more users.

The computer program product comprising code segments that are executable on a processing engine is adapted to provide controlling, by means of an output signal, power used to extract energy from the thermal energy buffer or a battery and/or used to charge the thermal energy buffer or battery. The computer program product comprising code segments that are executable on a processing engine is adapted to provide the step of calculating iterations through time in predictive time steps in accordance with a simulation model of the thermal energy buffer or battery to predict a buffer state of charge or a value related to the buffer state of charge at each predictive step along a plurality of calculation trajectories each of which comprise branches, the buffer state of charge or the value related to the buffer state of charge being generated by a selection of a limited number of possible operative decisions at each predictive time step for the charging of the buffer or battery. The computer program product comprising code segments that are executable on a processing engine is adapted to provide the step of executing an optimisation routine to allow reduction of the number of branches at any or each predictive step by pruning in accordance with a pruning mechanism selection criterion. This procedure is continued up to the time horizon. If there is at least one valid trajectory at the time horizon this can be used to obtain a single output signal. The computer program product comprising code segments that are executable on a processing engine is adapted to provide a step of extracting energy from the thermal energy buffer and/or charging the thermal energy buffer in accordance with the output signal.

The computer program product comprising code segments that are executable on a processing engine is adapted to provide a step of generating a tree structure by calculating iterations through time in steps in accordance with the simulation model of the thermal energy buffer to predict a buffer state of charge or a value related to the buffer state of charge at each predictive time step along a plurality of calculation trajectories results in the tree structure with the method performing calculations along each branch and new branches being spawned at each iteration step or pruned, each branch involving calculations of future states of charge.

The computer program product comprising code segments that are executable on a processing engine is adapted to provide a step of the prediction of the charge in the state of charge of the thermal energy buffer or batter or a value related to the state of charge such as a prediction of a temperature of the thermal energy buffer in the future or at a plurality of times in the future.

The computer program product comprising code segments that are executable on a processing engine is adapted to provide a step of allowing measurement of one or more actual temperatures of the thermal energy buffer and for using these as an input of a state of charge.

The computer program product comprising code segments that are executable on a processing engine is adapted to provide a step of allowing measurement of temperatures of a heat energy buffer fluid such as water or a phase change material at different heights in the thermal energy buffer.

The computer program product comprising code segments that are executable on a processing engine is adapted to provide a step of, for every time step, the result in terms of buffer state of charge or value related to the state of charge for some or all possible operative decisions for the output.

The computer program product comprising code segments that are executable on a processing engine is adapted to allow a step of selection of the operative decisions from operations of the device which charges the buffer such as compressor OFF, compressor operation at any value between 0 and 100%, e.g. 70%, compressor operation at 100%. The computer program product comprising code segments that are executable on a processing engine is adapted calculate forward exploring the branches and predicting the state of charge of the thermal buffer or battery or a value related to the state of charge as caused by operating the thermal buffer or battery in accordance with each or any of a set of operative decisions of how to charge the thermal buffer or battery, and to determine or to receive the input of the availability of energy during each predictive time step of the predictive pathway and determining a cost for the execution of each branch of the predictive tree structure.

The computer program product comprising code segments that are executable on a processing engine is adapted to decide at a future time step whether renewable energy or primary energy such as from fossil fuel fired power stations or nuclear power stations or a combination of these is to be used for each branch independently. The source or sources and cost or costs is/are recorded and stored. The computer program product comprising code segments that are executable on a processing engine is adapted to link use of renewable energy sources into an electrical grid in such a way that imbalances are reduced or avoided.

The computer program product comprising code segments that are executable on a processing engine is adapted to provide a step, when performing calculations for a certain branch, of a first pruning carried out when the buffer state of charge or value related to the state of charge exceeds a minimum or a maximum value by terminating this branch.

The computer program product comprising code segments that are executable on a processing engine is adapted to provide that no more calculations are performed on the terminated branch and any previously achieved results are discarded.

The computer program product comprising code segments that are executable on a processing engine is adapted to provide a second pruning mechanism which is when a preliminary simulation of a technical availability of an energy source, for example that the energy source has a larger availability, is used as the second pruning selection criterion. The computer program product comprising code segments that are executable on a processing engine is adapted to provide when identical or similar states of charge or values related to a state of charge are reached at a certain time in an iteration the state of charge or value related to a state of charge which has the largest technical availability of an energy source is selected as a third pruning criterion.

The computer program product comprising code segments that are executable on a processing engine is adapted to provide that when an identical or similar availability is reached for two different states of charge or values related to states of charge then the lower state of charge or value related to a state of charge is discarded as a fourth pruning criterion.

Concluding remarks

In one aspect the present invention provides a method to control the use of energy while taking into account storage losses of energy stored in a thermal energy buffer that is depleted as required by one or more users, the method comprising: controlling, by means of an output signal, power used to extract energy from the thermal energy buffer and/or used to charge the thermal energy buffer;

calculating iterations through time in steps in accordance with a simulation model of the thermal energy buffer to predict a buffer state of charge or a value related to the buffer state of charge at each step along a plurality of calculation trajectories each of which comprise branches, the buffer state of charge or the value related to the buffer state of charge being generated by a selection of a limited number of possible operative decisions at each step for charging of the buffer,

executing an optimisation routine to reduce the number of branches at each step by pruning in accordance with a pruning mechanism selection criterion to obtain a single output signal, and

extracting energy from the thermal energy buffer and/or charging the thermal energy buffer in accordance with the output signal.

The method described above may further comprise steps of: generating a tree structure as the step of calculating iterations through time in steps in accordance with the simulation model of the thermal energy buffer to predict a buffer state of charge or a value related to the buffer state of charge at each step along a plurality of calculation trajectories resulting in the tree structure with the method performing calculations along each branch and new branches being spawned at each iteration step, each branch involving calculations of future states of charge or values relating to the state of charge.

The prediction of the state of charge in the thermal energy buffer or value related to the state of charge, mentioned above, can be a prediction of a temperature of the thermal energy buffer in the future or at a plurality of times in the future.

Any of the method steps described above may also include measurement of one or more actual temperatures of the thermal energy buffer, for example, temperatures of a thermal energy buffer fluid such as water or a phase change material can be measured at different heights in the thermal energy buffer.

One of the main steps of the method includes calculating for every time step the buffer state of charge or value related to the state of charge for all possible decisions for the output. These decisions can be selected, for example from the charging device such as a compressor being OFF, the charging device operation such as compressor operation being at any value between 0 and 100%, e.g. 70%, compressor operation at 100%.

Any of the method steps described above can also include one or more pruning steps based on one or more pruning selection criteria. For example, when calculating within a certain branch, a first pruning mechanism is carried out, wherein the first pruning mechanism comprises terminating that branch when the buffer state of charge or value related to the state of charge exceeds a minimum or a maximum value. In such a case one option is that no more calculations are performed on the terminated branch and any previously achieved results are discarded. Alternatively, or additionally a second pruning mechanism can be a preliminary simulation of a technical availability of an energy source which can be used as a second pruning selection criterion.

Alternatively, or additionally, when identical or similar states of charge or values related to the state of charge are calculated at a certain time in an iteration, the state of charge or the value related to the state of charge which is associated with the largest technical availability of an energy source is selected in accordance with a third pruning selection criterion, whereby a state of charge or the value related to a state of charge associated with a lower technical availability of the energy source can be discarded.

Alternatively, or additionally, when identical or similar technical availabilities are reached for different states of charge or values related to states of charge then the lower state of charge or value related to a state of charge is discarded in accordance with a fourth pruning selection criterion.

Another aspect of the present invention is the provision of a system with control of the use of energy while taking into account storage losses of energy stored in a thermal energy buffer that is depleted as required by one or more users, the system comprising: means for controlling, by means of an output signal, power used to extract energy from the thermal energy buffer and/or used to charge the thermal energy buffer; means for calculating iterations through time in steps in accordance with a simulation model of the thermal energy buffer to predict a buffer state of charge or a value related to the buffer state of charge at each step along a plurality of calculation trajectories each of which comprise branches, the buffer state of charge or the value related to the buffer state of charge being generated by a selection of a limited number of possible operative decisions at each step to charge the buffer,

means for executing an optimisation routine to reduce the number of branches at each step by pruning in accordance with a pruning mechanism selection criterion to obtain a single output signal, and

means for extracting energy from the thermal energy buffer and/or charging the thermal energy buffer in accordance with the output signal. Means can be provided in the system for generating a tree structure by calculating iterations through time in steps in accordance with the simulation model of the thermal energy buffer to predict a buffer state of charge or a value related to the buffer state of charge at each step along a plurality of calculation trajectories resulting in the tree structure with the method performing calculations along each branch and new branches being spawned at each iteration step, each branch involving calculations of future states of charge or values related to future states of charge. The prediction of the charge in the thermal energy buffer can be a prediction of a temperature of the thermal energy buffer in the future or at a plurality of times in the future. Means for measurement of one or more actual temperatures of the thermal energy buffer can be provided. The means for measurement can be adapted to measure temperatures of a thermal energy buffer fluid such as water or a phase change material at different heights in the thermal energy buffer.

In the system, the means for calculating can be adapted, for every time step to calculate a result in terms of buffer state of charge or a value related to the state of charge for all possible decisions for the output. The decisions can be selected from the charging device such as a compressor being OFF, he charging device operation such as the compressor operation being at any value between 0 and 100%, e.g. 70%, compressor operation at 100%, for example.

In such a system, means can be provided for a first pruning of a branch when the buffer state of charge or the value related to the state of charge exceeds a minimum or a maximum value, by terminating this branch. In such a case no more calculations need to be performed on the terminated branch and any previously achieved results can be discarded. In the system either alternatively or additionally, means can be provided for a second pruning mechanism whereby a preliminary simulation of a technical availability of an energy source can be used as the second pruning selection criterion.

Additionally, or alternatively, when identical or similar states of charge or values related to the state of charge are reached at a certain time in an iteration, the state of charge or the value related to the state of charge which has the largest technical availability of an energy source can be selected as a third pruning criterion. Additionally or alternatively, when an identical or similar availability is reached for two different states of charge or values related to states of charge then the lower state of charge or value related to a state of charge is discarded as a fourth pruning criterion.

A controller for control of the use of energy while taking into account storage losses of energy stored in a thermal energy buffer that is depleted as required by one or more users, the controller comprising: means for controlling, by means of an output signal, power used to extract energy from the thermal energy buffer and/or used to charge the thermal energy buffer; means for calculating iterations through time in steps in accordance with a simulation model of the thermal energy buffer to predict a buffer state of charge or a value related to the buffer state of charge at each step along a plurality of calculation trajectories each of which comprise branches, the buffer state of charge or the value related to the buffer state of charge being generated by a selection of a limited number of possible operative decisions at each step for charging of the buffer, means for executing an optimisation routine to reduce the number of branches at each step by pruning in accordance with a pruning mechanism selection criterion to obtain a single output signal, and means for extracting energy from the thermal energy buffer and/or charging the thermal energy buffer in accordance with the output signal.

The controller can include means for generating a tree structure by calculating iterations through time in steps in accordance with the simulation model of the thermal energy buffer to predict a buffer state of charge or a value related to the buffer state of charge at each step along a plurality of calculation trajectories resulting in the tree structure further including performing calculations along each branch and new branches being spawned at each iteration step, each branch involving a calculation of a future state of charge or a value related to the buffer state of charge.

The prediction of the charge in the thermal energy buffer can be a prediction of a temperature of the thermal energy buffer in the future or at a plurality of times in the future. Means for measurement of one or more actual temperatures of the thermal energy buffer can be provided. The means for measurement can be adapted to measure temperatures of a heat energy buffer fluid such as water or a phase change material at different heights in the thermal energy buffer. The means for calculating can be adapted, for every time step, to calculate the result in terms of buffer state of charge or a value related to the state of charge for all possible decisions for the output. The decisions can be selected from the charging device such as a compressor being OFF, the charging device operation such as the compressor operation being at any value between 0 and 100%, e.g. 70%, compressor operation at 100%.

For a branch, means can be provided for a first pruning when the buffer state of charge or the value related to the state of charge, exceeds a minimum or a maximum value by terminating this branch. In such a case no more calculations need to be performed on the terminated branch and any previously achieved results can be discarded.

In the controller, means can be provided, either additionally or alternatively, for a second pruning mechanism wherein a preliminary simulation of a technical availability of an energy source can be used as the second pruning selection criterion. Additionally or alternatively, when an identical or similar state of charge or value related to a state of charge is reached at a certain time in an iteration, the state of charge or the value related to the state of charge which has the largest technical availability of an energy source is selected as a third pruning criterion. Additionally or alternatively, when an identical or similar availability is reached for two different states of charge or values related to states of charge then the lower state of charge or value related to a state of charge is discarded as a fourth pruning criterion.

In another aspect a computer program product is provided for carrying out any of the methods of the present invention. A storage medium such as an optical disk (CD-ROM or DVD-ROM), magnetic tape, magnetic disk such as a hard drive, a solid state memory device such as a USB memory such as a flash memory, a random access memory can be used to store the computer program product. The storage medium can be a non-transient storage medium.

Claims

1. A method to control the use of energy while taking into account storage losses of energy stored in a thermal energy buffer or a battery that is depleted by self-depletion or as required by one or more users, the method comprising:

controlling, by means of an output signal, power used to charge the thermal energy buffer or the battery;

calculating iterations through time in steps in accordance with a simulation model of the thermal energy buffer or battery to predict a buffer state of charge or a value related to the buffer state of charge at each step along a plurality of calculation trajectories each of which comprises branches from an initial time step to a time horizon, the prediction of the buffer state of charge or a value related to the buffer state of charge being generated by a selection of a limited number of possible operative decisions for charging of the thermal buffer or the battery at each step,

executing an optimisation routine to reduce the number of branches at any or each step by pruning in accordance with a pruning mechanism selection criterion and selecting a trajectory which reaches the time horizon from which a single output signal is derived, and

charging the thermal energy buffer in accordance with the output signal.

2. The method of claim 1 further comprising generating a tree structure comprising branches, the method performing calculations along each branch and new branches being spawned at each iteration step, each branch involving a calculation of a future state of charge or a value relating to a future state of charge.

3. The method of claim 2 wherein the prediction of the state of charge or the value related to the state of charge of the thermal energy buffer is a prediction of a temperature of the thermal energy buffer in the future or at a plurality of times in the future.

4. The method of any previous claim, wherein for every time step the result in terms of a buffer state of charge or a value related to the state of charge is calculated for all operative decisions for charging the thermal buffer.

5. The method of claim 4, wherein the operative decisions are selected from no charge, or operation of a charging device at any value between 0 and 100%.

6. The method according to any previous claim further comprising, for a branch, a first pruning mechanism is carried out, wherein the first pruning mechanism comprises terminating the branch when the buffer state of charge or the value related to the state of charge exceeds a minimum or a maximum value.

7. The method of claim 6 wherein no more calculations are performed on the terminated branch and any previously achieved calculation results are discarded.

8. The method of any previous claim further comprising a second pruning mechanism, the second pruning mechanism being a preliminary simulation of a technical availability of an energy source, the technical availability of the energy source being used as the second pruning selection criterion.

9. The method of any previous claim wherein if identical or similar states of charge or values related to a state of charge are reached at a certain time in an iteration, the state of charge or the value related to the state of charge which has the largest technical availability of an energy source is selected this being a third pruning selection criterion.

10. The method of any previous claim wherein if identical or similar availabilities of energy are reached for different states of charge or values related to states of charge then the lower or lowest state of charge or value related to a state of charge is discarded this being a fourth pruning selection criterion.

11. The method of any of the previous claims wherein charging the thermal energy buffer in accordance with the output signal is from one or more energy sources of an electric grid whereby renewable energy sources are selected in such a way that imbalances are reduced or avoided.

12. The method of any of the previous claims wherein selection of the trajectory is the trajectory that uses energy which is most available for charging the buffer or battery.

13. The method of any of the claims 1 to 11, wherein selection of the trajectory is the trajectory which uses energy which has a low environmental point of view.

14. The method of any of the claims 11 to 13, wherein the selection of a trajectory is the trajectory which avoids or reduces imbalances uses renewable energy or energy from fossil fuel fired power stations or nuclear power stations or a combination of these.

15. A system with control of the use of energy while taking into account storage losses of energy stored in a thermal energy buffer or a battery that is depleted by self-depletion or as required by one or more users, the system comprising:

means for controlling, by means of an output signal, power used to charge the thermal energy buffer or battery;

means for calculating iterations through time in steps in accordance with a simulation model of the thermal energy buffer or battery to predict a buffer state of charge or a value related to the buffer state of charge at each step along a plurality of calculation trajectories each of which comprise branches, the buffer state of charge or the value related to the buffer state of charge being generated by a selection of a limited number of operative decisions at each step for charging the buffer or battery,

means for charging the thermal energy buffer or the battery in accordance with the output signal.

16. The system of claim 15 further comprising means for generating a tree structure from calculating iterations through time in steps in accordance with the simulation model of the thermal energy buffer to predict a buffer state of charge or a value related to the buffer state of charge at each step along a plurality of calculation trajectories, resulting in the tree structure with the means for generating a tree structure performing calculations along each branch and new branches being spawned at each iteration step, each branch involving calculations of future states of charge or a value related to a state of charge.

17. The system according to claim 15 or 16 wherein for a branch, means are provided for a first pruning when the buffer state of charge exceeds a minimum or a maximum value by terminating this branch.

18. The system according to any of the claims 15 to 17, wherein for a branch, means are provided for a second pruning mechanism and wherein a preliminary simulation of a technical availability of an energy source is used as the second pruning selection criterion.

19. The system according to any of claims 15 to 18 wherein when identical or similar states of charge or values related to the state of charge are reached at a certain time in an iteration, the state of charge or the value related to the state of charge which has the largest technical availability of an energy source is selected as a third pruning criterion.

20. The system according to any of the claims 15 to 19 wherein when identical or similar availabilities are reached for different states of charge or values related to states of charge then the lower state of charge or value related to a state of charge is discarded as a fourth pruning criterion.

21. A controller for controlling use of energy while taking into account storage losses of energy stored in a thermal energy buffer or a battery that is depleted by self-depletion or as required by one or more users, the controller comprising:

means for executing an optimisation routine to reduce the number of branches at each step by pruning in accordance with a pruning mechanism selection criterion to obtain a single output signal, and means for controlling charging the thermal energy buffer or the battery in accordance with the output signal.

22. The controller of claim 21 further comprising means for generating a tree structure from calculating iterations through time in steps in accordance with the simulation model of the thermal energy buffer to predict a buffer state of charge or a value related to the buffer state of charge at each step along a plurality of calculation trajectories, resulting in the tree structure with the means for generating a tree structure performing calculations along each branch and new branches being spawned at each iteration step, each branch involving calculations of future states of charge or a value related to a state of charge.

23. The controller according to claim 21 or 22 wherein for a branch, means are provided for a first pruning when the buffer state of charge exceeds a minimum or a maximum value by terminating this branch.

24. The controller according to any of the claims 21 to 23, wherein for a branch, means are provided for a second pruning mechanism and wherein a preliminary simulation of a technical availability of an energy source is used as the second pruning selection criterion.

25. The controller according to any of claims 21 to 24 wherein when identical or similar states of charge or values related to the state of charge are reached at a certain time in an iteration, the state of charge or the value related to the state of charge which has the largest technical availability of an energy source is selected as a third pruning criterion.

26. The controller according to any of the claims 21 to 25 wherein when identical or similar availabilities are reached for different states of charge or values related to states of charge then the lower state of charge or value related to a state of charge is discarded as a fourth pruning criterion.

27. A computer program product which when executed on a computer system executes any of the methods of claims 1 to 14 or a non-transient signal storage medium storing a computer program product which when executed on a computer system executes any of the methods of claims 1 to 14.