A Method of Reducing Energy Consumption of Heating, Ventilation and Air Conditioning (HVAC) Equipment in a Multi-Room Building

Abstract

A method of reducing energy consumption of heating, ventilation and air conditioning (HVAC) equipment in a multi-room building. The energy consumption that would result from a particular candidate room allocation is used in order to determine how to allocate loads across the HVAC equipment. A candidate room allocation can be selected to optimise the efficiency of the HVAC equipment based on the proposed load that the HVAC equipment would be under should that candidate room allocation be selected.

Description

FIELD OF THE INVENTION

The present invention relates to method of reducing energy consumption of HVAC equipment in a multi-room building by allocating rooms based on their energy efficiency.

BACKGROUND TO THE INVENTION

Heating, ventilation and air conditioning (HVAC) power consumption contributes more than 40% of building energy consumption. Not only is this expensive, but it is damaging to the environment. HVAC energy consumption is widely cited to constitute to a large amount of building operating cost (Wang 2007).

Multi-room buildings such as hotels, meeting room and conference facilities, schools (classrooms) and flexible workspaces could benefit significantly if they were to prioritise allocating rooms based on the most energy efficient room available. Presently, room allocation is either done manually or is automatically selected based on the requirements such as room type, guest preference etc. It is typical to allocate rooms based on reducing the amount of empty time in the same room (i.e. gaps in usage).

One prior art room allocation method uses objective function optimisation to minimise the “room empty” nights between adjacent bookings (Li et al 2013).

Research into building HVAC energy optimisation has tended to focus on equipment operation scheduling or HVAC equipment control (Iddio et al 2020; Vu eta 2017, 2018). To date, there has been no research into the use of room allocation to optimise building HVAC energy consumption.

The present invention aims to optimise the energy consumption, instead of empty nights, by allocating rooms based on the most energy efficient room available.

SUMMARY OF THE INVENTION

Aspects of the present invention are set out in the appended independent claims. Certain variations of the invention are then set out in the appended dependent claims. Further aspects, variations and examples are presented in the detailed description below.

BRIEF DESCRIPTION OF THE DRAWINGS

For a better understanding of the invention and to show how the same may be carried into effect, there will now be described by way of example only, specific embodiments, methods and processes according to the present invention with reference to the accompanying drawings in which:

FIG. 1 shows a typical hotel room with equipment suitable for collecting data to perform the presently disclosed method. This includes a smart thermostat to collect the air conditioning or heating statistics such as power status (on/off), current room temperature, temperature set point (i.e. the desired temperature) and an energy meter to measure the HVAC equipment's power consumption.

FIG. 2 is a schematic showing an example of a typical commercial building setup where multiple HVAC compressors are deployed and each compressor supplies cooling/heating to multiple rooms inside the building.

FIG. 3 is a flow diagram outlining a method of determining an optimal room allocation to achieve minimal HVAC power consumption.

FIG. 4 is a diagram outlining a method of determining cooling (or heating) demand degree hour (CDDH).

FIG. 5 is a graph illustrating the relationship between number of occupied rooms and HVAC equipment's power consumption.

FIG. 6 is a graph illustrating a reduction in the number of compressors running when room allocation is performed in accordance with an embodiment of the present invention.

FIG. 7 is a graph illustrating the number of occupied rooms for each compressor in a multi-room building used to determine the next room to allocate.

DETAILED DESCRIPTION

When viewed from a first aspect, the present invention provides a method of reducing energy consumption of heating, ventilation and air conditioning (HVAC) equipment in a multi-room building, comprising the steps:

• • (i) obtaining energy usage profiles for a plurality of rooms and assigning each room an energy efficiency score, wherein the most energy efficient room is the room with the lowest energy usage and the least energy efficient room is the room with the highest energy usage,
  • (ii) determining which room to allocate by:
    • setting room occupancy status to occupied or unoccupied for each room for a given time point,
    • obtaining room reservation requirements for the given time point,
    • determining which room is most energy efficient and unoccupied at the given time point,
    • allocating a next room to be used according to the most energy efficient and unoccupied room.

Reducing energy consumption as employed herein refers to the process of reducing the amount of energy used for a given level of room occupancy. The baseline against which the reduction is measured is a prior art room allocation method in which rooms are either allocated manually or by utilising an “empty night” optimisation. The reduction is achieved by employing the present method wherein rooms are allocated based on which room is most energy efficient.

HVAC equipment as employed herein refers to heating equipment, ventilation equipment and cooling or air conditioning equipment. The term is intended to cover individual, standalone, pieces of equipment installed in each room and central HVAC equipment where conditioned air is delivered from a central equipment room to individual rooms via delivery ducts. Multiple networks may exist in a multi-room building.

Multi-room building as employed herein refers to any multi-room building, specifically where some rooms may be used and others not for prolonged periods of time (such as several hours, typically at least 24 hours). Examples of such buildings include, but are not limited to hotels, schools, conference facilities, meeting room facilities, flexible working/office buildings, apartment buildings, hospitals and retirement/nursing homes.

The multi-room building may selected from the group consisting: a school, a hotel, a meeting room facility, an office facility and a nursing home.

Energy usage profile as employed herein refers to the energy usage of a given room in the plurality of rooms or of the rooms on a specific HVAC network as a whole. The energy profile may vary over time, for example as seasons change, therefore the profile is not necessarily a fixed profile. The energy usage profile may be obtained by sensors located in each, or at least some of the, room(s) which monitor usage over a period of time.

Plurality of rooms as employed herein refers to more than one room. Typically, the plurality of rooms comprises rooms which have different HVAC equipment to others in the plurality of rooms and hence has more and less energy efficient rooms. However, it is possible for rooms using the same HVAC equipment (central HVAC equipment) to have different energy usage profiles. This may occur, for example, due to room size, differences in the primary HVAC equipment (i.e. the in-room equipment), orientation of the room (e.g. whether the room is facing the sun during the hottest part of the day) and differences in user comfort preferences, e.g. setpoint temperature. Because not all rooms in the plurality require heating or cooling to the same degree, there is scope to utilise the present method even where the plurality of rooms are controlled by a single HVAC network.

Energy efficient/efficiency as employed herein refers to the least or lower energy consumption.

Those skilled in the art will appreciate that, as outlined herein, the efficiency of HVAC equipment or an HVAC network can be understood to be the energy consumption required to service a particular heating and/or cooling demand. The efficiency of the HVAC equipment or network may, in general, be dependent on its percentage load.

An individual room can have a higher or lower energy consumption for many reasons including, for example, the outdoor temperature setpoint, the size of the room, etc.

Energy efficiency score as employed herein refers to how the HVAC equipment power consumption reacts to the in-room power status of the rooms it serves. This can be modelled using data gathered from in-room thermostat and energy meter(s) connected to the HVAC equipment.

As employed herein, in-room power status refers to whether the in-room HVAC controls are set to have the HVAC equipment on or off. Typically, this equipment takes the form of a thermostat, such as a smart thermostat.

A smart thermostat as employed herein refers to the operational control panel of heating and cooling equipment. Smart thermostats are generally networked and linked to the HVAC equipment. Typically, this will be wireless connectivity. Smart thermostats provide information on the status of the HVAC equipment e.g. on/off and typically also provide information on the desired temperature and existing temperature of the room. Some smart thermostats may provide additional data such as window-open detection, geocaching of the room's guest (i.e. whether they are currently in the room or not) etc.

In some embodiments the in-room power status also comprises a temperature setpoint. That is, the desired temperature in the room. Such as the temperature selected on a thermostat or smart thermostat.

In one embodiment wherein the energy usage profile is built using data from one or more sensors in the plurality of rooms.

In some embodiments the sensors are smart thermostats. Typically, the sensors are located in each room. In some embodiment each room is equivalent to a fan coil.

In one embodiment the method of any preceding claim wherein the energy usage profile is built using data on power consumption of the HVAC equipment over time.

Room occupancy status as employed herein refers to whether a room is occupied, or not, at a given time point. The room occupancy status determines whether or not a room is available to be allocated.

Room reservation requirements as employed herein refer to the number of rooms that are needed at a given time point, for example, tomorrow, at 10 am, next week. Room reservation requirements may also include a required duration of booking or may operate on a time block (such as per hour or per day or per week).

Typically, the given time point is the same time for both the room occupancy status and the room reservation requirements, although it is possible to map the given time points out into the future, based on the duration of room use required.

“Allocating the next room to be used” as employed herein refers to the step of determining which of the unoccupied rooms is the most energy efficient (e.g. has the best energy efficiency score) and assigning that room as the next room to be used. Should a guest request a specific room that is not the predetermined next room to be used, it is possible to overrule the next room to be used and the method will adapt to assign the next room to be used to the next guest.

Guest as employed herein refers to one or more users of the room. This can be the person reserving the room or a different user of the room.

Some multi-room buildings may have a plurality of rooms each served by local HVAC equipment, such as primary HVAC equipment (for example, a heater or air conditioning unit). It is possible to perform the present method on such buildings by acquiring energy usage profiles for each room in the same manner as those with central HVAC systems featuring one or more HVAC networks. The method is neutral as to the equipment being used.

HVAC network as employed herein refers to a number of rooms with a central HVAC system removed (physically distanced) from the rooms providing heating, ventilation and/or air conditioning to each of the rooms, via ducting. In-room controls can independently control the on/off status of the HVAC as well as setting the desired temperature (also referred to as the set point temperature, or temperature setpoint). For larger buildings, multiple HVAC networks may be present, each serving a number of rooms in the plurality of rooms.

An HVAC network may, as outlined above, include one or more air handling units and air may be cooled (or heated) centrally and distributed to rooms via ducts. However, it should be understood that HVAC networks may, additionally or alternatively, include arrangements in which the air cooling (or heating) function is distributed rather than centralised. For example, water and/or refrigerant may be used to cool the air without needing an air handling unit and/or ducts. It will be appreciated that combinations of different types of HVAC equipment may be used, as appropriate.

Thus, more generally, an HVAC network can be understood to be a number of rooms driven by common HVAC equipment. Regardless of what combination of specific HVAC equipment is used, in general there may be a number of compressors, and the efficiency the HVAC equipment may depend on its percentage load.

Thus, in some embodiments the HVAC equipment forms one or more HVAC networks. In some embodiments the HVAC networks each independently connect one or more rooms of the plurality of rooms.

In some embodiment the one or more HVAC networks each independently connect one or more rooms of the plurality of rooms.

Where rooms are served by central HVAC equipment in a HVAC network it is possible that certain networks are more efficient than others, that certain rooms served by a single network are more energy efficient and even that certain networks are more efficient at a specific capacity (for example, where less than half of the rooms on the network are currently occupied). All of these possibilities can be factored in to the energy usage profile for each room. The model f shown below, for example, will determine which equipment is the most efficient.

Some HVAC networks are more efficient than others for many reasons, including, but not limited to equipment type and age, number of sun facing rooms served, number of rooms being served by the network, number of windows in the rooms and physical location of the room within the building.

HVAC power data as employed herein refers to the amount of energy being consumed to heat, ventilate and/or cool a specific room in the plurality of rooms.

The method as disclosed herein comprises two stages. The first stage is to build or acquire an energy usage profile for each of the rooms in the plurality of rooms. The second stage is to model the room allocation requirements based on existing and upcoming usage and to allocate those rooms based on the information acquired in the first stage.

Although the solution to the problem may, prima facie, be simple i.e. allocate the most energy efficient room to the next guest, the dimensional space very rapidly expands in a real world situation where multiple rooms are available and bookings may have differing durations. The effect of this is that the optimal solution to the problem is extremely difficult to compute manually. Particularly given the number of times the calculations would need to be performed each day. For this reason, it is beneficial to convert the problem to be solved into a computer solvable format as described herein.

A simplified example of a real world problem for two rooms being occupied for two nights is detailed below in Example 2, which shows how the dimensional space and the number of possible solutions to the problem can very quickly become far too complex to be solved by the human brain.

In one embodiment the energy usage profile is made by modelling the relationship between the in-room power status and HVAC power data. Essentially, this refers to creating a mathematical model of the relationship.

In some embodiment the step of obtaining energy usage profiles includes the steps:

• • gathering in-room power status,
  • gathering HVAC power data, and
  • modelling the relationship between in-room power status and HVAC power data.

In one embodiment the modelling is:

• • Require: HVAC power consumption per time unit T=y
  • Require: Time unit T CDDH or in-room power status=X
  • Require: Number of rooms served by the compressor=n
    • Step 1: Scale X to x using the following formula,

$x=\frac{n\left(X-\min \left(X\right)\right)}{\max \left(X\right)-\min \left(X\right)}+1$

• • • Step 2: Fit model f: x→y=a×log(x)
  • Output: Model f

HVAC power consumption per time unit as employed herein refers to how much power it consumes over a given time period or “time unit”, for example, 1 hour or 1 day.

As employed herein, a time unit may be any unit of time but is likely to be an hour or a day or a week, or anywhere in between. This is to permit the modelling to function in varying ways for different building types. For example, a meeting room facility may be booked per hour but a hotel room is likely to be booked per day.

CDDH or cooling/heating demand degree hour (wherein cooling could equally be replaced with heating to give HDDH. Cooling and heating are interchangeable) refers to the demand on the HVAC system. By way of example, consider a situation in which the outside temperature for a one hour time unit is 30° C. and the setpoint temperature (the desired temperature) inside the room is 23° C., the difference (A) between the two is 7° C. This gives a CDDH of 7 hours. Should two rooms on a HVAC network have the same setpoint temperature of 23° C. this would be 14 hours CDDH, and so on. CDDH is a measure of the load on the HVAC system. By using hours, rather than days, the optimisation can be considerably more fine grained.

Room type factor as employed herein refers to the number of different room types. For example single room, twin room and double room would be three room types giving a b=3. Room types may refer to the size or category of room, for example a suite or a standard room in a hotel.

The model f can be determined by using historical data of HVAC equipment (e.g. compressor) and all rooms served by it, even if that is a 1:1 HVAC equipment to room ratio. Because hotel guests usually stay for a number of days, the data can be resampled daily which makes modelling easy. Due to the complexity of the non-linear constraints optimisation problem and the size of the variable (number of rooms), we used “log” function as it closely resembles the relationship between number of rooms and power consumption of the compressor and it is computationally easy. Exact mapping is not needed because the power consumption will depend on the weather and it is not possible to precisely predict the future, we can only predict the trend. It is possible to apply the modelling to longer or shorter periods of time (e.g. hour, day, week) based on the usage of the multi-room building.

Once the modelling of the HVAC power consumption to in-room power status has been performed, thereby identifying the most energy efficient rooms, it is necessary to then solve the problem of which room to allocate to minimise energy consumption.

In mathematical optimisation and decision theory, the goal is to minimise an objective function with subject to a set of constraints. An objective function is either a cost function that needs to be minimised or its negative (in specific domains, variously called a reward function, a profit function, a utility function, a fitness function, etc.), in which case it is to be maximised. A cost function is a function that maps an event or values of one or more variables onto a real number intuitively representing some “cost” associated with the event. Constraints describe rules that restrict the possible set of solutions, i.e. in the present case: “Do not allocate more rooms than are available in the building” or “Do not allocate rooms to guests that don't match the guest's preference”.

Constructing an objective/cost function with room occupancy status and room reservation requirements as employed herein refers to the process of converting the real world problem into one that can be solved by a machine (computer). For example, identifying the various steps needed to break down the problem and then converting the steps into mathematical problems. The cost function identifies the various factors at play in a given scenario and aims to quantify the cost of a given solution. In the present case, the cost of a specific room allocation. After we have mathematically formulated the cost function and the constraints, we can use standard mathematical optimisation and decision theory algorithms to optimise the problem and find the most energy efficient solution.

The cost function created for the present invention predicts the energy consumption of all rented rooms at a given allocation, i.e. it answers the question: “How much energy will I consume if I allocate my rooms in a certain way”. To be able to perform this prediction the cost function utilises the recorded past data to estimate the load on the HVAC network under the supplied room allocation scenario. It then uses the models created for each HVAC network to predict the resulting energy consumption. This is then summed up over all HVAC networks in the property to give the total energy cost of a given room allocation. HVAC network may be considered to include individual HVAC equipment serving individual rooms within the building.

Minimising the objective/cost function to minimise HVAC energy consumption as employed herein refers to calculating the optimal room allocation solution that will consume the lowest possible amount of energy by the HVAC equipment.

In some embodiment the step of determining which room to allocate includes the steps:

• • constructing an objective function with room occupancy status and room reservation requirements, and
  • minimising the objective function to minimise HVAC energy.

As outlined elsewhere herein, for example with regard to certain non-limiting examples described later, the objective function may be constructed from the load profiles, and minimised subject to the current occupancy and the incoming new reservations. Thus, in some embodiments, the objective function is constructed based on the energy usage profile for each of the plurality of rooms, or the compressor efficiency profile for each compressor, as appropriate. In some embodiments, the method comprises minimising the objective function to minimise HVAC energy subject to the room occupancy status and room reservation requirements.

In one embodiment the optimisation is:

• • Require: Current rented rooms (occupied) and blocked rooms (i.e. unavailable for some other reason)
  • served by each compressor Z_i (i=1 . . . n)
  • Require: Model of each compressor f_i
  • Require: Incoming guests information G_k
    • Step 1: Form cost function: J=F(Z)=Σ_i=1ⁿf_i(Z_i)
    • Step 2: where Z=Z+X×G
    • Step 3: Form constraint C₁: Z≤Z_max
    • Step 4: Form constraint C₂: 1_n×X=1_p (1_n is a n by 1 vector of which elements are 1)
    • Step 5: Minimize J with respect to X subject to constraints C₁ and C₂
  • Output: Optimal Allocation X

A worked example is shown in example 2, below.

In some embodiments the method can be carried out manually, however, it is more efficient to program the method into a computer where complex mathematics are involved and to obtain the best results. This is due to the complexity of the maths. Advantageously, by making the problem machine solvable it can be quickly and repeatedly carried out.

It is possible to determine the room efficiency once and then use that going forward to influence which rooms are allocated. This method is encompassed by the present invention. However, obtaining continued data on the present room energy usage and energy efficiency of the HVAC equipment and, in some cases, applying machine learning to that modelling, provides a more sophisticated embodiment of the invention. Furthermore, as mentioned above, the calculations involved in allocating the rooms in a complex multi-room building for a prolonged period of time are highly complex and better suited to computer implementation.

Thus, in one embodiment the method is a computer implemented method.

In some embodiments the computer implemented invention utilises machine learning to refine the energy usage profile by gathering data on one or more parameters selected from: time of room reservation requirements, orientation of room, weather forecast, predicted deterioration in HVAC equipment efficiency due to age etc.

Time of room reservation requirements as employed herein may relate to time of day or time of year, for example. This is because cooling or heating requirements vary throughout the day and likewise from season to season.

Sensors positioned in the rooms can provide useful information to the energy usage profile, measuring energy consumption over time, for example every 15 minutes or more. Sensors in rooms may measure, for example, air conditioner related activities and energy consumption over time such as heating or a/c setting and changes in setting (set point temp), current temp etc.

Once the problem has been put into a machine solvable format as described herein, the solution to the problem can be performed in a number of different ways known in the mathematical art. Including, but not limited to: AMPL and APOPT.

In one embodiment the allocating the next room to be used step is carried out using mixed integer programming.

Mixed integer programming as employed herein refers to a type of mathematical optimisation in which some or all of the variables are restricted to being integers, for example, which compressor to use (compressor 1, 2 etc.).

The first aspect of the invention extends to a computer programmed to carry out the method of the first aspect of the present disclosure.

The first aspect of the invention also extends to a non-transitory computer-readable medium comprising instructions which, when executed by a processor, cause the processor to carry out the method of the first aspect of the present disclosure.

The first aspect of the invention also extends to a computer software product comprising instructions which, when executed by a processor, cause the processor to carry out the method of the first aspect of the present disclosure.

In a set of potentially overlapping embodiments, the energy profile for one or more rooms may comprise a predetermined energy profile. It will be understood that a ‘predetermined’ energy profile is one which is not dependent on any sensors. A predetermined energy profile may be a static profile for that room, or may be a profile that is selected from a plurality of energy profiles for that room, each of which corresponds to different conditions for that room (e.g. the season). Thus a multi-room building may include at least some rooms for which the energy profile is static and/or at least some rooms which do not include any sensors. Predetermined energy profile(s) for such a room or rooms may be set during installation or configuration of the system.

As outlined hereinabove, where rooms are served by central HVAC equipment in a HVAC network it is possible that certain networks are more efficient than others, that certain rooms served by a single network are more energy efficient and even that certain networks are more efficient at a specific capacity. In some embodiments, the energy usage profile for each room comprises information regarding the efficiency of the HVAC network connected to said room at a specific capacity. It has been appreciated that the efficiency of a compressor can vary significantly depending on its load. As such, it may be preferable to allocate a room serviced by a particular compressor for optimal loading. For example, if there are two rooms available for allocation where one is serviced by a first compressor that is already in use (i.e. it is already servicing another room on the same HVAC network) and another is serviced by a second compressor not currently in use, it may be most efficient that the next room to be allocated be one on the first, already in use compressor rather than enabling the second compressor.

As such, analysis may be performed on different possible configurations of rooms that may be allocated that are unoccupied within the room reservation requirements. These possible configurations are referred to herein as ‘candidate room allocations’. Each candidate room allocation provides a possible combination of rooms and thus the compressors required to provide heating and/or cooling to the rooms being considered for allocation. Each candidate room allocation will have an associated impact on the capacity at which one or more of the compressors will operate at, should that candidate room allocation be selected.

Thus it will be appreciated that where the rooms of the multi-room building are served by multiple HVAC networks, the efficiency of each network can be understood to be the energy needed for that network to serve the heating and/or cooling demand as required of it. Embodiments of the present invention may reduce energy consumption by improving efficiency through smart allocation of the expected compressor load across the HVAC networks, rather than analysing the energy consumption of specific rooms on a room-by-room basis. The energy consumption of a given candidate allocation (i.e. specific scenario) corresponds to the efficiency across all HVAC networks involved in fulfilling that candidate allocation.

Thus, when viewed from a second aspect, the present invention provides a method of reducing energy consumption of heating, ventilation and air conditioning (HVAC) equipment in a multi-room building, wherein the HVAC equipment comprises a plurality of compressors, wherein each compressor supplies cooling and/or heating to one or more rooms in said multi-room building, the method comprising the steps:

• • (i) obtaining a compressor efficiency profile for each compressor, wherein the compressor efficiency profile for each compressor comprises information regarding the efficiency of said compressor when operated at a specific capacity;
  • (ii) determining which room and respective compressor to allocate by:
    • setting room occupancy status to occupied or unoccupied for each room for a given time point,
    • obtaining room reservation requirements for the given time point,
    • obtaining a plurality of candidate room allocations based on the room occupancy status for each room and the room reservation requirements for the given time point,
    • analysing the plurality of candidate room allocations to determine a resultant energy consumption for the plurality of compressors for each of said candidate room allocations,
    • determining which candidate room allocation is most energy efficient at the given time point,
    • allocating a next room and respective compressor to be used according to the most energy efficient candidate room allocation.

The second aspect of the invention extends to a computer programmed to carry out the method of the second aspect of the present disclosure.

The second aspect of the invention also extends to a non-transitory computer-readable medium comprising instructions which, when executed by a processor, cause the processor to carry out the method of the second aspect of the present disclosure.

The second aspect of the invention also extends to a computer software product comprising instructions which, when executed by a processor, cause the processor to carry out the method of the second aspect of the present disclosure.

It will be appreciated that the optional features outlined hereinabove in respect of various embodiments of the first aspect of the invention apply equally to the second aspect of the invention as appropriate.

In some embodiments of either of the foregoing aspects, the method further comprises receiving a request for allocation of one or more rooms of the multi-room building. The request may be used to generate, determine, or otherwise obtain the room reservation requirements for the given time point.

When viewed from a third aspect, the present invention provides a method of distributing load across a plurality of compressors in a heating, ventilation and air conditioning (HVAC) system for a multi-room building in which each compressor supplies cooling and/or heating to one or more rooms in said multi-room building, the method comprising the steps:

• • (i) obtaining a compressor efficiency profile for each compressor, wherein the compressor efficiency profile for each compressor comprises information regarding the efficiency of said compressor when operated at a specific capacity;
  • (ii) determining which of a plurality of candidate room allocations to select for a given time point by:
    • analysing the plurality of candidate room allocations to determine a resultant energy consumption for the plurality of compressors when operated at a resultant load for each of said candidate room allocations,
    • determining which candidate room allocation is most energy efficient at the given time point,
    • allocating a next room and respective compressor to be used according to the most energy efficient candidate room allocation.

The third aspect of the invention extends to a computer programmed to carry out the method of the third aspect of the present disclosure.

The third aspect of the invention also extends to a non-transitory computer-readable medium comprising instructions which, when executed by a processor, cause the processor to carry out the method of the third aspect of the present disclosure.

The third aspect of the invention also extends to a computer software product comprising instructions which, when executed by a processor, cause the processor to carry out the method of the third aspect of the present disclosure.

In some embodiments of the third aspect, the method further comprises:

• • obtaining a plurality of candidate room allocations for the given time point,

In some embodiments of the third aspect, the method further comprises:

• • setting room occupancy status to occupied or unoccupied for each room for a given time point, and
  • obtaining room reservation requirements for the given time point,
  • wherein the step of obtaining a plurality of candidate room allocations is based on the room occupancy status for each room and the room reservation requirements for the given time point.

In some embodiments of the third aspect, the step of determining which candidate room allocation is most energy efficient at the given time point includes the steps:

• • constructing an objective function with room occupancy status and room reservation requirements, and
  • minimising the objective function to minimise HVAC energy.

It will be appreciated that the optional features outlined hereinabove in respect of various embodiments of the first and/or second aspect(s) of the invention apply equally to the third aspect of the invention as appropriate.

This includes, but is not limited to, the optional features described previously in respect of the step of “determining which room to allocate” in embodiments of the first aspect of the invention; the step of “determining which room and respective compressor to allocate” in embodiments of the second aspect of the invention; and/or the step of “determining which of a plurality of candidate room allocations to select for a given time point” in embodiments of the third aspect of the invention, where these optional features are equally applicable to the corresponding steps in the embodiments of the other aforementioned aspects of the invention.

In the context of this specification “comprising” is to be interpreted as “including”.

Aspects of the invention comprising certain elements are also intended to extend to alternative embodiments “consisting” or “consisting essentially” of the relevant elements.

Where technically appropriate, embodiments of the invention may be combined.

Embodiments are described herein as comprising certain features/elements. The disclosure also extends to separate embodiments consisting or consisting essentially of said features/elements.

Technical references such as patents and applications are incorporated herein by reference.

Any embodiments specifically and explicitly recited herein may form the basis of a disclaimer either alone or in combination with one or more further embodiments.

EXAMPLES

Mathematical Solution to the Problem

The problem of how to optimise room allocation to save HVAC energy consumption in a central HVAC building can be formulated as so:

Given a property of n compressors/HVAC equipment which has q number of room types (e.g. deluxe room, suite etc), there are currently Z rooms occupied and in m days (or hours) there are p rooms that will be occupied, the question is how to allocate those p rooms to minimise the power consumption of n compressors in the coming m days/hours. The objective function can be expressed as:

J=F(Z)=Σ_i=1ⁿf_i(Z_i)

where Z=Z+X×G.

The calculation of X×G is the tensor contraction that is X^nxp×G^pxqxm=(XG)^nxqxm and the element (XG)_ijl with i=1 . . . n, j=1 . . . q, l=1 . . . m is calculated as:

$\begin{array}{cc}{\left(\mathrm{XG}\right)}_{\mathrm{ijl}}=\sum _{k=1}^p{X}_{\mathrm{ik}}\times G_{\mathrm{kjl}}& \left(1\right)\end{array}$

The objective is to find X that minimizes/.

The first constraint is that at any day in the coming m days/hours, the number of rented rooms of each room type of each compressor cannot exceed the maximum number of that room type of that compressor, that is:

Z≤Z_max  (2)

The second constraint is that each room can only assigned to one and only one compressor that is:

Σ_i=1ⁿX_ik=1 for k=1 . . . p and X_ik can only be 0 or 1,

The above equation can be expressed in vector form as:

1_n×X=1_p  (3)

Minimising (1) with constraints (2) and (3) can be done using some optimization suites, e.g. Python Gekko AMPL.

Determining f

The mapping/modelling of f can be determined by using historical data of compressor i and all the keys under it. Because rooms are usually occupied for a positive number of days/hours, the data can be resampled as daily/hourly which makes modelling easier.

Because of the constraints of the linear programming, only certain kinds of function can be used. Here we used log function as it closely resembles the relationship between keys and power consumption of the compressor. We also do not need the exact mapping as the power consumption will depend on the weather and we cannot precisely predict the future, we can only predict the trend.

Application

m days can be chosen arbitrarily from the guest booking record; however, a large m and G will make solving (1) extremely long. The more realistic scenario is that optimal room allocation happens when a guest checks in. For this case, m will be the duration of the room usage. G will be reduced to a matrix (instead of a 3D tensor). Minimising (1) can be solved quickly (near real time) to give the best room for the guest.

Simulation of Energy Saving Made Using an Embodiment of the Presently Disclosed Method

Hotel A, is 154 guest room hotel in which the rooms' HVAC is supplied by 21 Daikin VRV compressors (i.e. 21 HVAC networks). Each of the 21 compressors supply cooling to several of the hotel's rooms. Each compressor is equipped with an energy monitor device to record its power consumption and each room (including individual rooms in a suite) is equipped with smart sensors to collect data and control thermostats.

A simulation was conducted for the first two weeks of February 2020 (1 Feb. 2020 to 15 Feb. 2020). On the first day of February 2020, there were 96 guest rooms occupied in the hotel (62% occupancy) and there were 96 guest rooms reserved in the hotel the next two weeks. The simulation is conducted with smart allocation (SA) method of those 96 guest rooms compared to the actual allocation that occurred in reality. Actual allocation was performed manually by the staff at the hotel, in the way it would normally be done. The smart allocation method works by prioritising allocation of guests to the guest rooms using the more efficient compressors first (i.e. on the most efficient HVAC network).

Table 1 shows SA method saves 17% of energy consumption. This proves the effectiveness of the smart allocation method.

TABLE 1
Actual power used	Smart allocation	Saving	Savings
(kWh)	power used (kWh)	(kWh)	(%)
30853	25535	5318	17

Example 2—Simplified Worked Example of which Room to Allocate

A simple hotel with 4 rooms, namely room 11 (twin), room 12 (double), room 21 (twin), room 22 (double) and 2 HVAC networks (compressors):

• • Compressor 1 (VRV1) supplies rooms 11 and 12
  • Compressor 2 (VRV2) supplies cooling to rooms 21 and 22

Currently there are guests staying in rooms 11 (twin) and 22 (double) and both are going to check-out tomorrow.

Today one guest will check-in for a double room (selected from 12 or 22) for one day and tomorrow another guest will check-in to a twin room (selected from 11 or 21) for one day. The constraints described below handle the fact that room 22 is not available for today's double room guest.

What is the best allocation given the models and the information?

• • Hotel has two compressors→n=2
  • Two guests are going to stay→p=2
  • There are only two room types (twin and double)→q=2
  • We consider only two days (today and tomorrow)→m=2
  • Step 1—Form Z_i∈N^q×m=2×2:
    • Compressor 1: has one guest staying for today in the twin room 11, hence:

$Z_1=\left[\begin{array}{cc}1& 0\\ 0& 0\end{array}\right]$

• • • Note: Twin-room is the first row, double room is second row, today is the first column and tomorrow is the second column. In summary, compressor 1 currently services one room (11) but tomorrow will have no rooms to service.
    • Compressor 2: has one guest staying in the double room 22 for today, hence:

$Z_2=\left[\begin{array}{cc}0& 0\\ 1& 0\end{array}\right]$

• • • In summary, compressor 2 has one room to service today (room 22) but none tomorrow.
    • Matrix Z is the matrix of Z₁ and Z₂ that is:

$Z=\left[\begin{array}{c}{Z}_1\\ Z_2\end{array}\right]$

• • • Matrix Z is a three-dimensional matrix (or tensor) not the stacking of two matrices. Z has the dimension of n×q×m=2×2×2 (2 compressors, 2 room types, 2 days)
    • Because of the high dimension data, it is easier to visualise it as a list: the elements of the tensor, that is:
      • z₁₁₁=1; z₁₁₂=0; z₁₂₁=0; z₁₂₂=0; z₂₁₁=0; z₂₁₂=0; z₂₂₁=1; z₂₂₂=0
    • To put into table form this can be shown as:

	TABLE 2
	Compressor	Room type	Day	Z
	1	1	1	1
	1	1	2	0
	1	2	1	0
	1	2	2	0
	2	1	1	0
	2	1	2	0
	2	2	1	1
	2	2	2	0

• • Step 2—Form the G matrix:
    • New Guest 1: check-in today for a double room, check-out tomorrow

$G_1=\left[\begin{array}{cc}0& 0\\ 1& 0\end{array}\right]$

• • • Note: The first row is again for twin rooms, the second for double. The first column is for today and the second is for tomorrow.
    • New Guest 2: check-in tomorrow for a twin-room

$G_2=\left[\begin{array}{cc}0& 1\\ 0& 0\end{array}\right]$

• • • Matrix G is a 3-dimensional tensor that combines G₁ and G₂. That is
      • g₁₁₁=0; g₁₁₂=0; g₁₂₁=1; g₁₂₂=0; g₂₁₁=0; g₂₁₂=1; g₂₂₁=0; g₂₂₂=0
    • To put into table form this can be shown as:

	TABLE 3
	Guest	Room type	Day	G
	1	1	1	0
	1	1	2	0
	1	2	1	1
	1	2	2	0
	2	1	1	0
	2	1	2	1
	2	2	1	0
	2	2	2	0

• • Step 3—Form the decision variable for each guest:
    • Guest 1: X₁=[x₁₁ x₂₁]
    • Guest 2: X₂=[x₁₂ X₂₂]
    • Matrix X is the combination of the two vectors above:

$X=\left[\begin{array}{cc}{x}_{11}& x_{12}\\ x_{21}& x_{22}\end{array}\right]$

• • • Each column corresponds to each guest.
  • Step 4—Form the XG matrix which is the calculation of X×G (matrix multiplies with tensor). This the tensor contraction that is X^n×p×G^p×q×m=(XG)^n×q×m and the element (XG)_ijl with i=1 . . . n, j=1 . . . q, l=1 . . . m is calculated as

${\left(\mathrm{XG}\right)}_{\mathrm{ijl}}=\sum _{k=1}^p{X}_{\mathrm{ik}}\times G_{\mathrm{kjl}}$

• • • we can write it down in detail:

xg₁₁₁=x₁₁g₁₁₁+x₁₂g₂₁₁=0

xg₁₁₂=x₁₁g₁₁₂+x₁₂g₂₁₂=x₁₂

xg₁₂₁=x₁₁g₁₂₁+x₁₂g₂₂₁=x₁₁

xg₁₂₂=x₁₁g₁₂₂+x₁₂g₂₂₂=0

xg₂₁₁=x₂₁g₁₁₁+x₂₂g₂₁₁=0

xg₂₁₂=x₂₁g₁₁₂+x₂₂g₂₁₂=x₂₂

xg₂₂₁=x₂₁g₁₂₁+x₁₂g₂₂₁=x₂₁

xg₂₂₂=x₂₁g₁₂₂+x₂₂g₂₂₂=0

• • Step 5—Form the Z matrix:

Z=Z+XG

z₁₁₁=1,z₁₁₂=x₁₂,z₁₂₁=x₁₁,z₁₂₂=0,z₂₁₁=0,z₂₁₂=x₂₂,z₂₂₁=1+x₂₁,z₂₂₂=0

• • Z is a combination of the current occupied rooms and all possible allocations of upcoming bookings.
  • Step 6—Form the constraint Z_max
    • Z_max is defined such as for each compressor, for each room type, for every day, the number of rented rooms should not exceed the maximum number of rooms for that room type for that compressor. In this case, room 22 is already occupied and therefore the guest checking in today who requires a double room can only be allocated room 12.
    • In this case, the Z_max has the same size of Z with all the elements are 1 (because each compressor only has one twin room and one double room). The constraint is:

Z≤Z_max

• • Step 7—Form the constraint on decision variable X
    • This constraint says that one guest can only stay at one compressor (because they can only stay in one room, that is:

Σ_i=1ⁿx_ik=1 for k=1 . . . p and x_ik can only be 0 or 1.

• • • For this case, the constraints on X are:

x₁₁+x₁₂=1

x₂₁+x₂₂=1

• • Step 8—Form the objective/cost function:
    • Get the function log (modelling) coefficient for two compressors [a₁ a₂]->this is from the model-fitting. f=a×log (z)
    • Get the room-type coefficient [b₁ b₂]: this is weighting between room, e.g. executive room consumes more power than single room-> in this case it is [1 1] because twin room and double room are somewhat the same but it can be weighted differently.
    • The cost function is:

$J=F\left(\stackrel{\_}{Z}\right)=\sum _{i=1}^n{f}_i\left({\stackrel{\_}{Z}}_i\right)$ $\mathrm{Where}:$ $f_i\left({\stackrel{\_}{Z}}_i\right)=\sum _{t=1}^m{a}_i\log \left(\sum _{j=1}^q{b}_j{\stackrel{\_}{z}}_{\mathrm{ijt}}+1\right)$

• • • Noter: +1 is added to avoid the case of log (0) which is undefined.
    • In this case:

f₁(Z₁)=a₁ log(b₁z₁₁₁+b₂z₁₂₁+1)+a₁ log(b₁z₁₁₂+b₂z₁₂₂+1)=a₁ log(2+x₁₁)+a₂ log(x₁₂+1)

f₂(Z₂)=a₂ log(2+x₂₁)+a₂ log(x₂₂+1)

• • • The total cost function:

J=f₁(Z₁)+f₂(Z₂)=a₁ log(2+x₁₁)+a₂ log(x₁₂+1)+a₂ log(2+x₂₁)+a₂ log(x₂₂+1)

• • Step 9—Solving cost function and interpret the result:

Depends on the values of a₁ and a₂ the result can be:

$X=\left[\begin{array}{cc}1& 1\\ 0& 0\end{array}\right]$

This means both guests are going to be allocated to the compressor 1->means guest 1 to room 12 and guest 2 to room 11.

Compressor Allocation

As outlined previously, the approach described can be used for distributing load across the various compressors in a multi-room building (e.g. a hotel). It will be appreciated by those skilled in the art that the efficiency of a compressor is at least partially dependent on the operational load or capacity of that compressor. Depending on the number of rooms that the compressor is actively serving (i.e. how many rooms for which it is currently supplying heating and/or cooling and the degree of heating and/or cooling needed), the efficiency of the compressor can vary, in some cases significantly. A compressor that is operating but not at full capacity is sometimes referred to as being under ‘part-load conditions’, and its efficiency under such conditions is referred to as the compressor's ‘part-load efficiency’.

The part-load efficiency of a compressor can be included in an efficiency profile for that compressor. This profile includes information regarding the efficiency of the compressor across its operational capacity range.

When allocation of a new room in the multi-room building is required, e.g. for a new room booking in a hotel, a system operating in accordance with an embodiment of the present invention can choose the room based on which compressor would be the most energy efficient choice, i.e. to allocate a room served by the most efficient compressor.

For each possible room allocation (which may be for allocating a single room or multiple rooms) or ‘candidate room allocation’, analysis can be carried out to determine what the total energy efficiency will be for that candidate. For example, one candidate room allocation may allocate half the rooms to one compressor, and half to another; while another candidate room allocation may allocate all of the rooms to one compressor.

In general, it is usually more efficient to allocate a new room to a compressor that is already operating under part-load conditions than to turn on an idle compressor (i.e. one that is not actively providing heating and/or cooling to any room at the current time). It may be most efficient to allocate to the compressor running closest to full capacity, however this is not always the case and there may be cases in which it is more efficient to allocate to some other compressor.

Factoring in the total number of rooms each compressor would ultimately serve were that candidate followed, the efficiency profile for those compressors can be used to determine the total energy efficiency associated with that candidate. Rather than kWh vs load (CDDH), a predetermined power profile may be used which models efficiency vs percentage load. For a given candidate room allocation, the load that allocation will generate if followed can be estimated. This may assume a standard load of 1 unit, but this may vary, e.g. in response to prior knowledge regarding the guest such as their nationality, or the reserved room category.

In one implementation, the system could iterate through all candidates to make a determination as to the most efficient candidate, and then the rooms and associated compressors can be allocated accordingly. However, this ‘brute force’ approach may be computationally complex, particularly for large multi-room installations with many rooms and many HVAC networks. Thus it is generally preferred to use mathematical optimisation techniques, such as the technique described previously.

In order to find the best distribution of the load to the HVAC networks subject to current load distribution that allows the whole system of networks to satisfy the load in the most efficient way. To achieve this, a mathematical optimisation can be performed according to the following procedure:

• • a) Start with a random allocation and consult the objective function to calculate the resulting energy consumption of the HVAC networks.
  • b) Perform a gradient descent (i.e. find the direction of biggest reduction in energy consumption) and modify the allocation in that direction.
  • c) Repeat until the improvement each step is less than a specified threshold.

The efficiency profile that is used for each of the compressors may, as outlined previously, be learned or updated over time based on sensor data. Sensor measurements may be used to model the relationship between the load associated with the rooms and the energy consumption of the compressor under that load. Alternatively, the system can operate without the use of sensors, in which the efficiency profile is instead based on a predetermined model of that relationship, e.g. a model that reflects that a higher load percentage can be serviced more efficiently than a low load percentage. Such a predetermined model need not quantify this in absolute terms.

An exemplary case study of the impact an embodiment of the present invention can have on the number of compressors running is illustrated in the graph of FIG. 6. This system for allocating rooms is referred to as ‘Smart Room Allocation’ (SRA). The number of daily running compressors provides an indication as to the performance of the SRA, where a reduction in the number of compressors running may generally lead to a reduction in energy consumption for the multi-room building.

In this case study, the SRA system is applied to a Covid-19 quarantine hotel with 21 VRV compressor configured to supply cooling to 154 rooms in the hotel. Each compressor has a particular label: CU-01 A; CU-01 B; CU-01 C; and CU-02 to CU-19.

The SRA system provides recommendations to the front desk every day by assuming there will be 20 guests checking in each day. The front desk uses this information to allocate incoming guests to a room. It should be noted, however, that the recommendation is not necessarily always followed by the front desk, for example to accommodate the individual situations or needs of a guest.

The case study is conducted for a duration of two months, during which time the occupancy of the hotel ranges between 25% to 65%.

FIG. 6 shows a comparison of the number of daily running compressors with and without the SAR system. The ‘without SAR’ scenario is simulated by randomly assigning reservations each day of the case study duration. As can be seen from the graph, the ‘without SAR’ scenario leads to the use of at least one additional compressor at almost all times, and in several instances leads to significantly more compressors being in use that need not have been.

It should be noted that the number of running compressors slowly reduces at the beginning of the case study due to the current staying guests.

At the middle of FIG. 6, there is an uptrend as indicated by the trend arrow 30. This uptrend occurs due to the front-desk not following the recommendation from the SAR during this period. In fact, the front-desk only followed 50% of the recommendations due to various reasons. Had the front-desk followed the SAR recommendations, it is foreseeable that further energy savings could have been made.

In this particular example, the use of the SAR approach described herein may lead to a reduction of 13.3% in the number of running compressors. It would be expected that this would lead to a reduction in the energy consumption of the HVAC system.

FIG. 7 is a graph illustrating the number of occupied rooms for each compressor in a multi-room building used to determine the next room to allocate. In particular, FIG. 7 shows the number of compressors allocated on 29 Apr. 2021 in the above example.

As can be seen, on that on 29 Apr. 2021, 8 out of the 21 compressors did not have any room rented meaning those compressors can be turned off as shown in FIG. 2. Based on the compressor efficiency profiles (or ‘load profiles’) of each compressor, the SRA generated room recommendations for Apr. 30, 2021 check-ins. In this particular example, the next rooms are allocated according to the closest-to-full-load compressor first, e.g. CU-09, CU-03, CU-08. This avoids turning on idle compressors which would generally lead to a waste in energy. It will of course be appreciated that, depending on the compressor efficiency profiles, it may be more efficient to allocate a room served by a compressor that is not necessarily the one operating closest to its full load capacity, e.g. if some of the compressors have different part-load efficiency profiles.

Thus it will be appreciated that embodiments of the present invention may provide an arrangement which can be used to determine the impact on energy consumption that a particular allocation of rooms in a multi-room building will have, e.g. across multiple HVAC networks or compressors. Rather than by assessing the energy efficiency of each room individually, the efficiency of the allocation across the various HVAC networks for a given candidate room allocation (that is, a particular scenario of rooms being allocated for a particular time period) can be used to determine the optimal room—and therefore compressor—allocation.

While specific embodiments of the present invention have been described in detail, it will be appreciated by those skilled in the art that the embodiments described in detail are not limiting on the scope of the claimed invention.

CITATIONS

• Iddio, E., Wang, L., Thomas, Y., McMorrow, G., & Denzer, A. (2020). Energy efficient operation and modeling for greenhouses: A literature review. Renewable and Sustainable Energy Reviews, 117 (January 2019), 109480. https://doi.org/10.1016/j.rser.2019.109480
• Li, Y., Wang, B., & Caudillo-Fuentes, L. A. (2013). Modeling a hotel room assignment problem. Journal of Revenue and Pricing Management, 12(2), 120-127. https://doi.org/10.1057/rpm.2011.37 Vu, H. D., Chai, K. S., Keating, B., Tursynbek, N., Xu, B., Yang, K., Yang, X., & Zhang, Z. (2017). Data Driven Chiller Plant Energy Optimization with Domain Knowledge. Proceedings of the 2017 ACM on Conference on Information and Knowledge Management—CIKM '17, 1309-1317. https://doi.org/10.1145/3132847.3132860
• Vu, H. D., Keating, B. D., & Alleyne, A. G. (2018). Energy Optimization of an In-Service Building Chiller Plant via Extremum Seeking Control. International High Performance Buildings Conference, 3271-3281.

Claims

1-19. (canceled)

20. A method of distributing load across a plurality of compressors in a heating, ventilation and air conditioning (HVAC) system for a multi-room building in which each compressor supplies cooling and/or heating to one or more rooms in said multi-room building, the method comprising the steps:

(i) obtaining a compressor efficiency profile for each compressor, wherein the compressor efficiency profile for each compressor comprises information regarding the efficiency of said compressor when operated at a specific capacity;

(ii) determining which of a plurality of candidate room allocations to select for at a given time point by: analysing the plurality of candidate room allocations to determine a resultant energy consumption for the plurality of compressors when operated at a resultant load for each of said candidate room allocations, determining which candidate room allocation is most energy efficient at the given time point, allocating a next room and respective compressor to be used according to the most energy efficient candidate room allocation.

21. The method of claim 20 wherein the step of obtaining compressor efficiency profile for each compressor includes the steps:

gathering in-room power status,

gathering HVAC power data, and

modelling the relationship between in-room power status and HVAC power data.

22. The method according to claim 20 wherein the step of determining which candidate room allocation is most energy efficient at the given time point includes the steps:

constructing an objective function based on the compressor efficiency profile for each compressor, and

minimising the objective function to minimise HVAC energy subject to the room occupancy status and room reservation requirements.

23. The method of claim 20 wherein the allocating the next room to be used step is carried out using mixed integer programming.

24. The method of claim 20 wherein the compressor efficiency profile is built using data from one or more sensors in the plurality of rooms.

25. The method of claim 20 wherein the compressor efficiency profile is built using data on power consumption of the HVAC equipment over time.

26. The method of claim 24 wherein the sensors are smart thermostats.

27. The method of claim 24 wherein the sensors are located in each room.

28. The method of claim 27 wherein each room is equivalent to a fan coil.

29. The method of claim 20 wherein the multi-room building is selected from the group consisting: a school, a hotel, a meeting room facility, an office facility and a nursing home.

30. The method of claim 20 wherein the method is a computer implemented method.

31. The method according to claim 30 wherein the computer implemented method utilises machine learning to refine the compressor efficiency profiles by gathering data on one or more parameters selected from: time of future room reservation requirements, orientation of room, weather forecast, age of the HVAC equipment, weather forecast and setpoint temperature.

32. A computer programmed to carry out a method of distributing load across a plurality of compressors in a heating, ventilation and air conditioning (HVAC) system for a multi-room building in which each compressor supplies cooling and/or heating to one or more rooms in said multi-room building, the method comprising the steps:

(i) obtaining a compressor efficiency profile for each compressor, wherein the compressor efficiency profile for each compressor comprises information regarding the efficiency of said compressor when operated at a specific capacity;

(ii) determining which of a plurality of candidate room allocations to select for at a given time point by: analysing the plurality of candidate room allocations to determine a resultant energy consumption for the plurality of compressors when operated at a resultant load for each of said candidate room allocations, determining which candidate room allocation is most energy efficient at the given time point,

allocating a next room and respective compressor to be used according to the most energy efficient candidate room allocation.

33. A non-transitory computer-readable medium comprising instructions which, when executed by a processor, cause the processor to carry out a method of distributing load across a plurality of compressors in a heating, ventilation and air conditioning (HVAC) system for a multi-room building in which each compressor supplies cooling and/or heating to one or more rooms in said multi-room building, the method comprising the steps:

(i) obtaining a compressor efficiency profile for each compressor, wherein the compressor efficiency profile for each compressor comprises information regarding the efficiency of said compressor when operated at a specific capacity;

(ii) determining which of a plurality of candidate room allocations to select for at a given time point by: analysing the plurality of candidate room allocations to determine a resultant energy consumption for the plurality of compressors when operated at a resultant load for each of said candidate room allocations, determining which candidate room allocation is most energy efficient at the given time point,

allocating a next room and respective compressor to be used according to the most energy efficient candidate room allocation.

34. (canceled)