Peak shifting and synchronization of compressor duty cycles for optimal energy efficiency

Abstract

A system for performing energy optimization of compressors used in refrigeration and HVAC (Heating Ventilation and Cooling) through the synchronization of or adjustment of run cycles so as to minimize overlapping energy demands and to reduce peak energy usage in a given measurement window resulting in lowered cap tag and demand based or transmission bill charges.

Description

FIELD OF THE INVENTION

[0001] The present disclosure relates to systems and methods for the management and energy optimization of compressors used in refrigeration and HVAC (Heating Ventilation and Cooling).BACKGROUND OF THE INVENTION

[0002] With the rising costs of energy, and ongoing struggle to contain or limit environmental impacts, initiatives to reduce energy consumption have become increasingly important and sought after. Due to the nature of the energy supply system, peak loads are an important factor in energy pricing and an emphasis on reducing these peaks can result in a substantial reduction in energy costs as the demand charges are extrapolated over multiple months of billing and can account for up to 30% of a commercial monthly charge. In addition to cost savings, lowering peak usage and limiting peak stacking contributes to lowering the amount of grid capacity a utility is required to supply.

[0003] In looking at the appliances that are large energy consumers in an average home or commercial establishment, compressors make up a substantial portion of a facilities energy use. What's more, compressors generally work with a predictable and repeatable pattern of starting and stopping making them ideal candidates for coordination and synchronization.

[0004] Take the example of a refrigeration system, or an HVAC (heating ventilation and conditioning) system. The compressor comes on to cool the facility or the unit, once cool the compressor turns off and stays off until the unit of facility warms up to a preset temperature differential. Once the room or the refrigerator warms up again to the given threshold, the compressor comes on again repeating the cycle. When left undisturbed, the unit will repeat this pattern almost like clockwork. When doors are opened or food is put it, the cycle changes, but returns to the typical pattern a short time later.

[0005] When a compressor initially starts, there is a marked spike in energy use followed by a steady use which continues while the compressor is running. Once the cooling cycle has finished the compressor's energy use falls to zero or near zero.

[0006] Commercial establishments such as QSRs (Quick serve restaurants) have multiple compressor systems for devices such as their HVAC, refrigerator, walk-in freezer(s), icemakers, drink coolers, frosty and ice creme makers, to name just a few. Such facilities are ideal candidates for energy reduction from compressor energy use optimization.

[0007] When multiple compressors are used, having the peaks start at the same time creates peak-stacking and can lead to a larger energy usage spike which can substantially add to the demand charges for a given facility. Consider 5 compressors starting all at once and the peak energy use of the facility vs. having each of the compressors start separately or in a staggered fashion.

[0008] Even if the compressors start at a slight offset, when running concurrently the aggregation of load can increase peak usage at the site. In an ideal situation, the compressors would be adjusted to run in timeslots when the other compressors are off, taking turns to run one at a time or minimizing the number that are running concurrently.

[0009] The estimation and prediction of the run cycles becomes complex when one considers outside influences including outdoor temperature variations, internal temperature increase from cooking facilities, external factors such as door opening, occupancy, or putting warm items in a freezer or refrigerator as well as equipment age and maintenance. This quickly makes the problem a dynamic one and one that can benefit from machine learning.SUMMARY OF THE INVENTION

[0010] Disclosed herein is synchronization system for shifting and coordinating the cycle times of compressors in order to minimize overlapping energy demands to reduce peak energy usage resulting in lowered cap tag charges for individual sites. Further applications across multiple sites for community and grid level optimization are also proposed.

[0011] The system and method disclosed herein further accurately predicts and estimates the cycle times of multiple compressors and adjusts their run cycles to align in such a way that peak energy usage would be reduced. The present disclosure provides for machine learning algorithms for prediction and to anticipate changes in the patterns and adjust these accordingly within a given set of thresholds based on configurable constraints determined from the application of the compressor.

[0012] Accordingly, it is desired to provide a system and method that manages the run cycles of compressors in a multi-compressor environment organizing and shifting and shaping energy use in such a way as to minimize simultaneous run cycles and minimize peak energy usage.

[0013] It is further desired to provide a system and method for predicting and mapping energy usage events related to the compressors which, with the aid of machine learning, will predict the compressor run cycles dynamically as they adjust to outside influences. The system alters the run cycle timing dynamically and alters its peak shifting algorithms as needed. Machine learning based on historical measurements and metering are also applied to further improve upon the model over time.

[0014] One simple example involves a two-compressor deployment. A house with two HVAC systems, one for its upstairs bedrooms and one for its downstairs living area. Both systems are run by independent compressors, both run independently with no direct correlation in that each has its own thermostat that controls a particular zone with a setpoint, which is the desired temperature of the space that unit controls. Running independently, it is a common occurrence that the compressors run simultaneously and the energy use at a given time includes the energy use of both compressors. These two values are then added to all other systems that are running to determine an instantaneous peak energy value.

[0015] Duty cycles on the HVAC units vary, but for simplicity of our example, let's assume a duty cycle of 50%, meaning the compressor has a cooling cycle time which is equal to the rest cycle time. A duty cycle of 60% would reference the HVAC unit and compressor running 60% of the time. To further illustrate the 50% duty cycle, we can assume each cycle lasts 15minutes. Therefore, the compressors will alternate on for 15 minutes and off for 15 minutes.

[0016] In such a scenario, it is easy to envision how one could possibly align the compressor timing such that each compressor can be synchronized to run when the other is off thus alternating rest and run cycle times across the units. In such a case, the overlapping of peaks is eliminated, and the energy use would be aligned on a constant level with compressor 2 coming on when compressor 1 shuts off and vice versa.

[0017] Of course, temperature variances will affect the duty cycle, as will outside influences such as the opening of doors or windows, or even the maintenance and age of equipment and the changing of filters. It may be required that for a short time the peaks overlap but overall, the sharp spike of both compressors starting at once is eliminated and the overlap can be minimized.

[0018] Peak demand charges are typically calculated in 5-minute intervals that are then averaged over a 15-minute period. In such a case, even having the system minimize one-or two of the 5-minute measurement intervals within the measurement period will result in savings.

[0019] It is thus desired to provide a system and method that can adjust the timing and shape of compressor run time cycles so as to minimize the overlap of these cycles thus reducing the overall peak energy usage for site(s).

[0020] It is still further desired to provide a method by which such a system can predict and estimate the requires shape and start times of such run cycles including the application of machine learning from historical data and predictions made from a variety of sensors and outside feeds to improve upon its prediction capabilities.

[0021] It is still further desired to provide a system and method by which a set of outside constraints can be applied to the system as parameters to which adjustments can be made. For example, food safety and comfort settings may apply to how hot or cold the environment can get as well as compressor health in allowing a minimum time between cycles is enforced.

[0022] A system is provided that communicates with compressor management systems such as smart thermostats, refrigeration controllers, intelligent RTUs, or other compressor controllers, including direct control over compressor controls.

[0023] It is further desired to provide a system that uses outside data collected from a system of sensors including temperature and humidity sensors as well as outside feeds and information such as weather reports, external temperature sensors and temperature forecasts.

[0024] In one configuration, a system of predicting the timing cycles of compressors in refrigeration and HVAC applications is provided which can map energy usage cycles into the future and where they may overlap. Such a system applies capabilities to adjust subsequent cycles in a way as to optimize the overlap of energy use at any given time, primarily focused on 5-minute intervals which are used to calculate utility peak usage when averaged over a 15-minute period.

[0025] In another configuration, a system is provided that sends commands to an RTU or heat-pump system to alter setpoints and temperature differentials to shift the resultant energy usage cycles and adjust the duration of such cycles.

[0026] In yet another configuration, a system is provided that send commands to a refrigeration controller system to alter setpoints and temperature differentials to again shift the resultant energy usage cycles and adjust the duration of such cycles.

[0027] In yet another configuration, a system is provided that sends commands to a smart thermostat to alter setpoints and temperature differentials to again shift the resultant energy usage cycles and adjust the duration of such cycles.

[0028] In yet another configuration, a system is provided that sends commands to a controller system which may include direct control over the thermostat turning it on and off based on input from thermostats and other sensors in or in proximity of the unit.

[0029] In yet another configuration, a system is provided that sends commands to a controller device that may utilize the benefit of other inputs to control the run cycles of a given piece of equipment. For example, a sensor known to activate the unit, such as a door open sensor or a proximity sensor may be used to trigger the starting or stopping of a given compressor. Regardless of how the compressor is started or stopped, the temperature is measured to ensure it stays within the established norms, constraints, and desired settings all the while adjusting the cycles with small micro-adjustments to fit within an optimal window.

[0030] In certain aspects the system comprises software executing on a computer which software is connected via a network to a plurality of control devices, each control device associated with a temperature regulation device and configured to control one or more compressors of the temperature regulation device. The software is configured to receive a temperature reading indicative of a temperature of a space controlled by the temperature regulation device, the software further configured to receive data indicative of energy use of each compressor. Said software further connected to a storage which contains historical data indicative of an amount of time to achieve a set temperature in the space, said software determining an order of operation energy use for one or more compressors for each of the energy control devices, the order of operation determined based on said software determining a minimum peak energy usage which are predicted to meet one or more temperature constraints for each of the spaces. Said software measuring actual time to temperature for each of the spaces based on the order of operation and said software saving data associated with actual time to temperature in the storage such that a next determination of time to temperature is based in part on the data associated with actual time to temperature.

[0031] Therefore one or more of foregoing and other objects of the invention are achieved by providing a system for controlled cycling of compressors. The system includes software executing on a computer which software is connected via a network to at least one control device configured to control a plurality of compressors based on temperature readings. The software is configured to receive temperature readings indicative of a temperature of each of a plurality of spaces, each space associated with one of the plurality of compressors. The software is further configured to receive data indicative of energy use of each compressor. The software is further connected to a storage which contains historical data indicative of operation patterns of each compressor associated with data indicative of one or more conditions associated with such operation patterns. The software determines a predicated usage pattern for each of the plurality of compressors over a period of time based on data associated with a predicted condition for said period of time, said predicted usage pattern indicative of at least one cycle of each compressor being on and off and time periods associated with the compressor being on and off and a power usage during said time period associated with the compressor being on. The software determines a modified cycle for each compressor such that at least a first group of one or more of the plurality of compressors is off or reduced in usage while at least a second group of one or more of the plurality of compressors is on or increased in usage relative to the predicted usage pattern. The software generates control instructions for the at least one control device to modify a temperature setting and / or a temperature band for each space in order to control each compressor according to the modified cycle.

[0032] Other objects are achieved by providing a method for controlled cycling of compressors comprising one or more of the steps of: providing software executing on a computer which software is connected via a network to at least one control device configured to control a plurality of compressors based on temperature readings; receiving temperature readings indicative of a temperature of each of a plurality of spaces, each space associated with one of the plurality of compressors; receiving data indicative of energy use of each compressor; obtaining historical data indicative of operation patterns of each compressor associated with data indicative of one or more conditions associated with such operation patterns; determining a predicated usage pattern for each of the plurality of compressors over a period of time based on data associated with a predicted condition for said period of time, said predicted usage pattern indicative of at least one cycle of each compressor being on and off and time periods associated with the compressor being on and off and a power usage during said time period associated with the compressor being on; determining a modified cycle for each compressor such that at least a first group of one or more of the plurality of compressors is off or reduced in usage while at least a second group of one or more of the plurality of compressors is on or increased in usage relative to the predicted usage pattern; and generating and transmitting control instructions with said software for the at least one control device to modify a temperature setting and / or a temperature band for each space in order to control each compressor according to the modified cycle.

[0033] In certain aspects the software monitors the temperature readings compared to one or more thresholds and further modifies the modified cycle based one or more of the temperature readings falling outside of at least one of the one or more thresholds such that the software generates additional control instructions to control one or more of the plurality of compressors in response to the monitored temperature readings.

[0034] In certain aspects the control instructions are configured to cause at least one of the one or more compressors to cycle on and off more frequently during the period of time. In other aspects the control instructions are configured to cause at least one of the one or more compressors to remain on during an on cycle for a longer time period during the period of time. In further aspects the control instructions are configured to cause at least one of the one or more compressors to remain on during an off cycle for a longer time period during the period of time. In yet other aspects the modified cycle changes a start time for at least one of the plurality of compressors during the period of time relative to the predicted usage pattern. In still other aspects the predicted condition is a predicted weather and the one or more conditions associated with such operation patterns are one or more weather conditions. In yet other aspects the predicted condition is a predicted business volume the one or more conditions associated with such operation patterns are one or more business volume conditions. In still other aspects the temperature band is a range of temperatures set relative to the temperature setting. In yet other aspects the control instructions are further generated or modified during the time period based on sensor data from sensors associated with one or more of the plurality of spaces which sensor data is measured during the time period and which sensors are sensors other than the temperature sensor which measures the temperature readings. In still other aspects the predicted condition is a predicted frequency and / or duration of opening of one or more doors associated with one or more of the plurality of spaces and the software monitors actual conditions associated with openings and / or duration of openings of the one or more doors during the time period and adjusts the control instructions based on a comparison of the actual conditions to the predicted conditions.

[0035] While the focus in the descriptions and the examples used herein relate to compressors and in particular HVAC and refrigeration, it should not be construed as limiting in any way. Similar mechanisms to coordinate energy usage and adjust peaks may well be applicable to any number of equipment types and could benefit from a similar system.

[0036] Other aspects and features will become apparent from consideration of the following description taken in conjunction with the accompanying drawings.BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The drawings illustrate example embodiments in which:

[0038] FIG. 1 provides a functional flow drawing of an example system.

[0039] FIG. 2 shows additional details of FIG. 1.

[0040] FIG. 3A provides an overview of types of commands implemented by the system.

[0041] FIG. 3B provides a functional flow diagram of the control implementation of the present system.

[0042] FIG. 4 shows example compressor cycles and related settings.

[0043] FIG. 5 shows an example of overlapping compressor cycles.

[0044] FIG. 6 shows an example of shifted peaks and operations of compressors.

[0045] FIG. 7 shows unoptimized overlapping peaks from 3 compressors.

[0046] FIG. 8 shows optimized peaks from 3 compressors.

[0047] FIG. 9 provides an example software logic implementation.

[0048] FIG. 10 shows how scheduling of compressors is created and modified.DETAILED DESCRIPTION

[0049] The reference numbers in specific figures refer to elements in those figures.

[0050] Example embodiments as described herein reduce the peak energy usage in multi-compressors systems.

[0051] Turning to the drawings, FIG. 1 is a high-level overview of system. A computer 10 is connected to a storage system 30 and runs software that includes a control application 20 and a prediction system 40. The communicate through a communications system 50 over a network 2 to send control instructions 4 to various compressor control systems 60 described in the subsequent drawings.

[0052] Turning now to FIG. 2 we see the various components that make up the system. A prediction system 40 is connected to billing data 32, prediction data 34, and historical data 36. The prediction system 40 also leverages external feeds 41 such as weather reports. A measurement and metering system 45 is used to gather and store the historical data 36 and a machine learning system 47 adjusts the prediction data dynamically based on actuals and historical measurements.

[0053] A control application 20 is driven by the outputs of the prediction system and makes adjustments to the compressor systems 60 through a communication system 50. External sensors 42 may also provide additional inputs to the control application 20. Metering data 43 is also used to validate and capture actual energy use of the various compressors when running and in particular when there are multiple phases and multiple modes to better predict the energy usage curves. In addition, constraint thresholds 44 are used to limit the variances that can be made to existing cycles. There are further described in the subsequent figures but can be summarized as simply thresholds that cannot be exceeded in terms of temperatures and cycle times related to safety, be that compressor health or food safety, or legal requirements for worked environments.

[0054] The compressor systems 60 may communicate with individual units in the system in various ways including Smart RTU / Heat pump control 62, refrigeration controllers 66, smart thermostats 64, and / or compressor controls 68. These methods may vary by manufacturer and by the systems employed and may be provided via an API (application programming interface). As shown, the prediction system can use a variety of information from storages including billing data 32, prediction data 34 and historical data 36 in order to predict the run cycles of each of the several compressors at a particular facility. The external sensors 42 and metering data 43 shown communicates with the control application 20, but it is understood that this data is accessible either directly or indirectly by the prediction system 40. The initial prediction looks at the expected use of each compressor for a period of time. This prediction is based on external feeds 41 such as weather predictions along with stored data indicative of patterns of usage. These patterns of usage may be indicative of expected usages or needs based on the time of day, day of the week, the particular day and other indicators of business needs. These baseline patterns for each of the compressors would indicate a particular time period, usually an hour or several hours where the compressors are expected to turn on and off, running through one or several duty cycles. This cycling would be to maintain various temperature settings in a variety of spaces associated with the facility. This prediction for each compressor may assume all compressors start their first “on” or active cycle at the same time which would be similar to the combined curve 730 shown in FIG. 7. While one example of a duty cycle previously reference herein is described in terms of percentage of overall run time, the duty cycle can be much more specific in terms of frequency of cycling, patterns, run times of each cycle, overall usage and other indicators of the pattern of usage of the compressor.

[0055] With the prediction determined, next the control application 20 will generate commands for several of the compressors. For example, FIG. 3a depicts variety of commands that may be issued on controller systems. The control application 20 may issue a setpoint adjustment to delay a compressor run cycle 300. This in effect delays the turning on of the compressor shifting the run cycle to the right on a time curve.

[0056] The control application may also issue a setpoint adjustment down command to accelerate the cycle 310. This in effect accelerates the turning on of the compressor shifting the run cycle to the left on a time curve.

[0057] The control application may also issue a setpoint temperature differential up command to lengthen the cycle 320 by adjusting the temperature band. This in effect makes the run cycle longer by extending the temperature range which must be achieved from the setpoint to the temperature differential setting. In essence increasing the duration of the run cycle. For example, a setpoint of 70 deg F may come with a temperature band or differential in the thermostat of +1.5 degrees for cooling such that once the room reaches 71.5 deg F., the thermostat commands operation of cooling. A lower temperature band of 1.5 degrees (or a different value) may be set such that the cooling stays on until 68.5 degrees is reached. In the cooling context, the larger this band is on the upper end, the later the thermostat will call for cooling and the later the compressor will turn on. The smaller then value of the lower band, the compressor will run for less time. As a result, in scenarios where adjusting the setpoint is not possible, but adjusting the bands is, the effect of adjusting the setpoint may be achieved by the thresholding or temperature band adjustments.

[0058] Thus, the control application may also issue a setpoint temperature differential down command to shorten the cycle 330 by adjusting the temperature band. This in effect makes the run cycle shorter by extending the reducing the temperature range which must be achieved from the setpoint to the temperature differential setting. In essence decreasing the duration of the run cycle.

[0059] The control application may also issue Adjust Multi-phase cooling (increase / reduce usage during cycle) 340. This in effect makes changes to the energy use during the cycle by using more or less energy through the multi-phase values provided. For example, in a two identical compressor system running two phases, phase 1 alone would use 1 / 2 the energy that phase 2 would use where both identical compressors are running. In a three-phase system where phase 1 may have a variable speed motor running the first compressor allowing it to run a 1 / 2 speed, the energy use may be only 1 / 4 of the maximum use, depending on the efficiency of the compressor and the settings on the variable speed setting. Metering data can be used along with external sensors and billing data to learn and adapt using the machine learning components of the system.

[0060] Thus, the control application 20 determines an adjusted usage pattern which creates a modified cycle for the compressors. This modified cycle may initially be based on prediction information only and not real time data feeds. Thus, the modified cycle may be determined in advance of, for example a 3 hour period, how the predicted patterns of usage should be adjusted in order to deliver comparable cooling to the various spaces associated with the appliances, but doing so in a way that has the compressors cycling on at different times to avoid peak stacking. In order to determine how the predicted patterns can be modified, the system may look at the overall predicted energy usage for each compressor during the period of time in question. In order to deliver the same average temperatures in the various spaces, it may initially be assumed that this overall usage over the time period should be the same, just the cycling on and off may be shorter or longer in order to result in the same energy use over time per compressor, just done at different times. There may however be additional energy needs to deliver adequate cooling due to more frequent startups. However, at the same time, if more frequent startups from one compressor result in more overall energy usage to maintain the same average temperature, other compressors which may run for longer but less frequently as a result of the modification may see an overall decrease in usage over the time period.

[0061] Since the system has real time temperature readings in the various spaces from the temperature sensors in those spaces, the results of the cycle adjustments compared to overall energy usage to maintain the desired temperature average can be tracked and the system will continue to learn how modifications to the duty cycle impact performance and usage since the historical data 36 continues to update as impacts from modifications to the compressor cycling are implemented.

[0062] FIG. 3b depicts various methods of issuing commands for the control of compressors. The control application 20 can receive data from external sensors 42 and external metering data 43 in order to validate and control the compressors. The programming established by the prediction system 40 along with these inputs provides the parameters by which commands are issued to the various controller systems. A set of constraints 44 are provided as input to the system establishing thresholds that must be adhered to. These are related to safety and comfort and may be established by norms, legal requirements, safety requirements or user definable settings.

[0063] For example, food safety may dictate that the refrigerator may not be above 40 degrees for more than 5 minutes. Cal / OSHA have established norms for heat illness prevention in indoor spaces that require heat index monitoring imposing thresholds. In some scenarios, set temperatures should be adjusted in 1° F. (0.5° C.) increments, and failure to follow this caution may result in equipment damage.

[0064] Other constraints include operating parameters for compressors such as ensuring there is a minimum off period before cycling a compressor to prevent damage flood back and allow pressures to equalize and overheating. These may vary by manufacturer and are programable in the system with defaults of 5 minutes.

[0065] In essence, these be summarized as simply thresholds that cannot be exceeded in terms of temperatures and cycle times. As an example, when turning off a compressor, the compressor should not be restarted for 2 minutes at a minimum. This is for the lifespan and operating health of the compressor. When dealing with freezers or refrigerators, norms for food safely and freshness can be established and incorporated to ensure that temperatures never exceed these levels compromising food quality or freshness. With HVAC system, norms have been established in some jurisdictions that mandate minimal or maximal temperature and humidity levels for the comfort and safely of workers. In some cases, business settings whereby end users can opt-in to a varying degree of settings can be established. For example, those willing to allow adjustments up to 2 degrees may receive a certain level of savings or compensation whereby others that have opted in to allow up to 4 degrees of variation may obtain even more savings or compensation.

[0066] Thus, the modified cycle that is based on historical and weather metric predictions may be further adjusted and the controls further adjusted in real time based on real time sensor feeds such as occupancy sensors, information from the POS (point of sale) system, door sensors. Each of these external sensors 42 may be related to a particular space and thus associated with one or more of the compressors. As but one example a front door sensor, while generally applicable to the overall space in which several other compressors sit, is mainly applicable to the room / space HVAC / cooling unit since the opening of that door on a more regular basis than predicted can cause changes in the duty cycle to maintain the desired temperature. The prediction and then the modified cycle can have built in expectations from these types of external sensors, for example that the front door is expected to open and shut a certain number of times and be open a certain amount of time within the relevant time period. If the door opens more frequently, this can indicate that the HVAC unit and its associated compressor is expected to need to run longer in order to maintain the desired temperature. A similar logic pattern can be applied to other spaces, for example a refrigerator or freezer door, its door sensor, how long the door is open and how often can be built in to the prediction and the modified cycle's expectations, again if the door opens more often (or less often), these external sensor feeds 42 can be used by the software to further refine the control instructions that modify how the compressors run.

[0067] FIG. 4 shows a typical energy use curve 400 which raises and lowers in a square waveform pattern indicating the times when the compressor is running and using energy (raised curve) and when the compressor is stopped (lowered curve) and not using energy. The duty cycle energy use 405 is indicated where the energy use curve 400 is raised. For example, a large compressor using a lot of energy would be a higher curve whereas as smaller compressor would be a lower value. Thus, the graph shows the amount of instantaneous power being used over time.

[0068] When looking at the energy use curve 400 and once again imagining it as a graph over time, the x-axis or horizontal axis would be mapped as time. The rest cycle time 410 is depicted when the y axis power measurement is at zero or lowest. The length of the rest cycle time 410 is the amount or time the compressor is off.

[0069] Similarly when looking at the cooling cycle time 420 we can again see that the energy use curve 400 is now at its highest along the y axis or vertical axis representing that the energy use is at its highest. This depicts the compressor being on and using energy and the length along the horizontal or x axis of the cooling cycle time 420 is the amount of time the compressor is running.

[0070] The dotted line depicts a temperature change curve 450 and is overlayed on the graph to show the various temperature settings and how they are related to the power utilization. This is a direct correlation in that when the energy use is at its lowest, such as during the rest cycle time 410. The temperature change curve 450 varies between the temperature differential 440 and the setpoint 430. In essence, when the temperature reaches the setpoint 430, the compressor is turned off and it begins the rest cycle 410. As the temperature increases over time and reaches the temperature differential 440, the compressor then starts again to begin the cooling process.

[0071] FIG. 5 shows the peak with overlap of cooling cycles. As shown, the energy use curve of compressor 1500 has its peak 1550.

[0072] Similarly, energy use curve of compressor 2510 has its peak 2560. When we now overlay the two energy use curves 520, we see the resultant peak over time is double what each of the peaks represented as combined peak 570.

[0073] In this illustration, both compressors operate at the same time so the power use is the sum of the power used by both compressors, thus resulting in a cycle with a large peak followed by a prolonged period of no use. While the overall use may average out the same, this does not account for the peak usage being charged at higher rates and thus costing more money.

[0074] FIG. 6 shows how peak shifting by changing temperature settings on compressor controls in a refrigeration unit can be achieved. In the initial energy model 600 the usual energy usage curve is shown with a dotted line again depicting the temperature change. In this case, the dotted line is annotated with setting showing how when the energy use is zero the temperature is 37 which is the setpoint value. Then the dotted line starts to slowly climb up since the compressor is not running and the space in the refrigerator is warming up due to normal heat transfer. When the Temperature reaches 38 (temperature differential of 1 deg F.), this coincides with the compressor again being turned on and energy being used. This cycle is repeated.

[0075] Next, a peak shift energy model 1610 is shown whereby the above energy curve has been shifted over by 15 minutes. To achieve this, we again look at the energy usage curve and see once again that at the setpoint temperature of 37 the energy use drops to zero as the compressor is turned off at the setpoint. Now however, the temperature differential has been changed to 2 degrees, i.e. 39 is when the compressor will turn on instead of the previous 38. Therefore the temperature rise to 39 takes more time, thus shifting over the start of the compressor run cycle again at the temperature differential of 39. The length of the cooling cycle is also extended and can be seen as being longer on the x-axis or horizontal time axis on the curve. This is because the compressor is now working longer to cool the temperature by two degrees (39-37) instead of the previous one degree (38-37). Next, we see the temperature differential is again moved to 38, so the same curve as in the initial energy model 600 is resumed in terms of length of cooling and resting cycles. In essence, the movement of the temperature differential by one degree for a single cycle has resulted in a shift in time of the cooling cycles. This is one of the mechanisms used to move a compressor cycle over in time to fit into a slot that the prediction system has outlined in order to minimize overlapping usage and overall peak usage.

[0076] Next, the shifted energy model 2620 shows the temperature differential is now three degrees to 40 instead of 39 (or the normal cycle of 38). With an even larger temperature differential of 3 degrees, the shifted energy model 2 delays the starting of the compressor cooling cycle further as the temperature gradually goes from the setpoint of 37 when the compressor last turned off, to when the temperature reaches the temperature differential of 40. Further, the compressor also runs longer as it must cool from a higher temperature to reach a 3-degree differential temperature. Assuming that the settings are now left as they are with a setpoint of 37 and a temperature differential of 40. In such a case, the curve of energy use was shifted to the right by the initial change of adding the additional degrees to the temperature differential. The shape of the energy usage curve is also reshaped by having the longer cooling cycles and rest cycles due to the larger temperature differential. This shift is not a one-time event like the shifted energy model 1 but is both shifted and reshaped. The compressor runs with a new cadence, unlike the initial energy model 1 with longer cooling cycles and resting cycles.

[0077] It should be noted that the illustration uses one degree, but fractions of degrees can be used to more granularly move over a setting. Further, metering data showing the operation of a compressor coupled with simply modifying the thermostat to trigger an on or off cycle can be done to more immediately control the operation of the compressor.

[0078] It should also be noted that while the illustration adjusted the temperature differential, the same result could have been obtained by adjusting the setpoint. Moving the setpoint up one degree would stop the compressor from cooling earlier as the temperature T+1 is reached faster than temperature T. Similarly moving the setpoint down one degree during a cooling cycle would cause the compressor to run longer as it must now cool to a lower temperature.

[0079] These methods of adjusting the settings in a smart thermostat are also meant simply to be illustrative and by no means limiting.

[0080] As another example of more direct controls let's take a system where metering is in effect and the control system is able to determine whether a compressor is running or not. Let's assume that in such a case, an internal temperature thermometer is also readable, and the control system is able to both monitor the internal temperature and know when the compressor is running. Here, the system could exercise control over the compressor in a more direct fashion by simply setting a low enough setpoint when it wants to turn the compressor on, and a high enough setpoint when it wants to turn the compressor off. A set of constraints is also implemented into the system to ensure that no food safety, workplace regulations, or compressor operating safety are compromised. It should be simply intended to depict only one way that direct control of the compressor being turned on or off can be achieved.

[0081] FIG. 7 we see an example of overlapping peaks from three compressors. Compressor 1700 runs with a cooling cycle 701 and a resting cycle 702 with a peak energy use 703. Compressor 2710 runs with a cooling cycle 711 and a resting cycle 712 with a peak energy use 713. Compressor 3720 runs with a cooling cycle 721 and a resting cycle 722 with a peak energy use 723.

[0082] The combined energy usage curve 730 shows how peak 733 is the combined peaks 703713 and 723 when in a given point of time all three of the compressors are on at a given time.

[0083] FIGS. 5 and 7 may equally be considered to represent the baseline or predicted energy usage for the individual components and the combined system in the typical scenario where everything turns on all at once at a particular time. Without the time shifting of the start times or end times or duration or combinations thereof the building or space in question could operate in a way that causes fully additive peaks. Yet, this prediction of the usage needs for each compressor and the timing thereof enables the software to compute and implement the adjustments described herein to reduce peak usage loads.

[0084] FIG. 8 shows the same three compressors of FIG. 7 but with the energy usage curves shifted and shaped into a more optimized model. Compressor 1700 has been shifted to have its cooling cycle start at a desired time. Compressor 2710 has been shifted to have its compressor start and the opposite time as compressor 1700 and its energy cycle has been adapted to lengthen slightly the cooling cycle and resting cycles to align with compressor 1 so that there is a perfect cadence of one and off for these two units. Compressor 3 has also been shifted to start at the same time as compressor 1. When we now look at the combined energy usage curve 800 we see that the new peak 833 only encompasses the combination of peaks 703 and 723 or peaks 713 and peaks 723 as there is no occurrence of peaks 713 and 703 that occur at the same time. Thus the peak usage has shifted lower, but the same overall energy usage remains the same.

[0085] The prediction engine looks ahead one or two cycles and also incorporates input from various external sensors and events. In one example, the prediction engine, along with the machine learning component measures the outside temperature as being 1100 degrees F. and extrapolates from previous data that the cooling cycle for the HVAC unit will be increase by 2 minutes when compared to earlier cycles when the outside temperature was 90. We will assume for simplicities sake that nothing else changes with any of the other compressors. The system them adjusts the temperature cycle and predicted curve re-runs the optimization algorithm.

[0086] Turning to FIG. 9 example logic shows how the baseline of overlapping compressors is calculated and managed through the various systems. Once started 900 the system gathers historical, billing and simulation data 910 as well as external variables 920 that may affect cooling cycles. The predicted cycle of compressor timing is made 930 for the foreseeable future, one hour is chosen for depiction but it may be any time value that spans one or more cycles allowing the system to exert control. The compressors are adjusted 940 by shaping and shifting the various curves to come up with an optimized pattern of energy use and compressor timing that adhered to the ideal model. Thresholds and constraints are monitored 950 first and foremost to ensure that no safety protocols or other constraints are being approached or compromised. If there are concerns, the readjustments 980 is made and a new model is created. Next metering data is monitored to validate and reconcile the timing and ensure compressors are running when they are expected to be running. If not, again readjustments 980 are made. Further, temperature sensors and other sensors are monitored and measured 970 to ensure that the system is performing as expected. if not, again readjustments 980 are made. The system repeats continually adjusting and tweaking the settings in an ever-changing environment. The use of machine learning for the prediction and for the combination and real time calculation make the real time adjustments possible.

[0087] FIG. 10 shows the program generation and resulting changes as the system operates. An ordered queue of duty cycle times 1030 is sent to the control module to issue controls to the compressors 1060. Normally this queue 103 is made up of predicted cycle times 1010 that are generated from an expected duty cycle 1000 that is generated from the machine learning components of the system that predict the requirements of running the compressors. When unexpected events occur 1080 or constraints 1070 thresholds are approached, a queue reprioritization 1050 system adjust or skips 1040 cycles changing the programming 1030 dynamically.

[0088] The system is made up of the multiple sub-systems and components. A prediction system 40 which estimates when the compressors will come on and for how long using outside sensor data, historical data, and prediction data gleaned from machine learning and knowledge of the system and the operation. This knowledge comes in part from the visibility to the current cycles, the internal temperatures and other external sensors. As an example, when we know it's 90 degrees outside, we know from past experience that it takes 30 mins to cool the place to 74 from 80 when the shift starts. We also know, that if we were to pre-cool to 70 one hour earlier, we could more quickly arrive at the desired temperature.

[0089] The second component of the system is the control application 20. Here, in the most rudimentary form the system is able to control the setpoints on the thermostat and adjust these to move the cooling cycles and shape the cooling cycles. For example, increasing the setpoint will result in shortening the current cycle as the compressor will turn of sooner when the higher temperature is reached. Lowering the setpoint will extend the current cycle, and increasing the temperature differential will increase the gap between cooling cycle times.

[0090] A baseline or running program is generated and followed by the control application for all compressors present in the system. The predicted data comes from the application of machine learning to predicted data as well as knowledge of external variables and sensor data. The program is then adjusted dynamically in near real time to achieve the optimal results.

[0091] Ventilation systems, while not outlined in the examples, also play a role in the efficiency of HVAC systems and the exhausting of the conditioned air on premises. It is contemplated that, while direct control of these is not proposed, they work as external factors, and the system reacts to the resultant temperature changes. These may also be inputs to the machine learning and program generation when it comes to predicting cooling cycles.

[0092] The algorithms used for calculating and adjusting the setpoints and cooling cycles and rest cycles also takes into account the energy used by each appliance. In the simplest sense, like stacking blocks of various sizes, the system will do it's best to minimize the overall energy usage at any given time where each compressor may utilize more or less energy.

[0093] When managing multiple sites in a community or a given sector on the grid, the coordination of peaks can also benefit the utility by reducing overall peak demand on the system. While there is currently limited potential to obtain direct savings for the end client from such multi-site coordination, it is contemplated that this capability will arouse the interest of utilities or virtual utilities as a valuable methodology for overall peak reduction and energy reduction. If nothing else, it can provide the substantial benefit of reducing overall energy use and optimizing the grids'ability to supply energy.

[0094] The control and management of multiple entities and the coordination capability of resources across entities can apply the same algorithms and the same machine learning capabilities to benefit the utility in terms of overall system peak demand reduction. With an ability to manage and adjust loads by load shaping and shifting, an entity controlling a multitude of end points can provide this type of aggregate load reduction and shaping.

[0095] The examples depicted are shown as simple ones for illustration purposes. When looking at an active restaurant where the refrigerator door is opened and closed frequently, ice is being made based on demand of lunch rushes and volume of activity and external temperature and business volume is ever changing.

[0096] While the disclosure is susceptible to various modifications, and alternative forms, specific examples thereof have been shown in the drawings and are herein described in detail. It should be understood however that the disclosure is not limited to the particular forms or methods or embodiments disclosed.

Claims

1. A system for controlled cycling of compressors comprising:software executing on a computer which software is connected via a network to at least one control device configured to control a plurality of compressors based on temperature readings;the software is configured to receive temperature readings indicative of a temperature of each of a plurality of spaces, each space associated with one of the plurality of compressors;the software further configured to receive data indicative of energy use of each compressor;said software further connected to a storage which contains historical data indicative of operation patterns of each compressor associated with data indicative of one or more conditions associated with such operation patterns;said software determining a predicated usage pattern for each of the plurality of compressors over a period of time based on data associated with a predicted condition for said period of time, said predicted usage pattern indicative of at least one cycle of each compressor being on and off and time periods associated with the compressor being on and off and a power usage during said time period associated with the compressor being on;said software determining a modified cycle for each compressor such that at least a first group of one or more of the plurality of compressors is off or reduced in usage while at least a second group of one or more of the plurality of compressors is on or increased in usage relative to the predicted usage pattern;said software generating control instructions for the at least one control device to modify a temperature setting and / or a temperature band for each space in order to control each compressor according to the modified cycle.

2. The system of claim 1 further comprising: said software monitoring the temperature readings compared to one or more thresholds and further modifying the modified cycle based one or more of the temperature readings falling outside of at least one of the one or more thresholds such that the software generates additional control instructions to control one or more of the plurality of compressors in response to the monitored temperature readings.

3. The system of claim 1 wherein said control instructions are configured to cause at least one of the one or more compressors to cycle on and off more frequently during the period of time.

4. The system of claim 1 wherein said control instructions are configured to cause at least one of the one or more compressors to remain on during an on cycle for a longer time period during the period of time.

5. The system of claim 1 wherein said control instructions are configured to cause at least one of the one or more compressors to remain on during an off cycle for a longer time period during the period of time.

6. The system of claim 1 wherein the modified cycle changes a start time for at least one of the plurality of compressors during the period of time relative to the predicted usage pattern.

7. The system of claim 1 wherein the predicted condition is a predicted weather and the one or more conditions associated with such operation patterns are one or more weather conditions.

8. The system of claim 1 wherein the control instructions are further generated or modified during the time period based on sensor data from sensors associated with one or more of the plurality of spaces which sensor data is measured during the time period and which sensors are sensors other than the temperature sensor which measures the temperature readings.

9. The system of claim 1 wherein the temperature band is a range of temperatures set relative to the temperature setting.

10. The system of claim 1 wherein the predicted condition is a predicted frequency and / or duration of opening of one or more doors associated with one or more of the plurality of spaces and the software monitors actual conditions associated with openings and / or duration of openings of the one or more doors during the time period and adjusts the control instructions based on a comparison of the actual conditions to the predicted conditions.

11. A method for controlled cycling of compressors comprising:providing software executing on a computer which software is connected via a network to at least one control device configured to control a plurality of compressors based on temperature readings;receiving temperature readings indicative of a temperature of each of a plurality of spaces, each space associated with one of the plurality of compressors;receiving data indicative of energy use of each compressor;obtaining historical data indicative of operation patterns of each compressor associated with data indicative of one or more conditions associated with such operation patterns;determining a predicated usage pattern for each of the plurality of compressors over a period of time based on data associated with a predicted condition for said period of time, said predicted usage pattern indicative of at least one cycle of each compressor being on and off and time periods associated with the compressor being on and off and a power usage during said time period associated with the compressor being on;determining a modified cycle for each compressor such that at least a first group of one or more of the plurality of compressors is off or reduced in usage while at least a second group of one or more of the plurality of compressors is on or increased in usage relative to the predicted usage pattern;generating and transmitting control instructions with said software for the at least one control device to modify a temperature setting and / or a temperature band for each space in order to control each compressor according to the modified cycle.

12. The method of claim 11 further comprising: said software monitoring the temperature readings compared to one or more thresholds and further modifying the modified cycle based one or more of the temperature readings falling outside of at least one of the one or more thresholds such that the software generates additional control instructions to control one or more of the plurality of compressors in response to the monitored temperature readings.

13. The method of claim 11 wherein said control instructions are configured to cause at least one of the one or more compressors to cycle on and off more frequently during the period of time.

14. The method of claim 11 wherein said control instructions are configured to cause at least one of the one or more compressors to remain on during an on cycle for a longer time period during the period of time.

15. The method of claim 11 wherein said control instructions are configured to cause at least one of the one or more compressors to remain on during an off cycle for a longer time period during the period of time.

16. The method of claim 11 wherein the modified cycle changes a start time for at least one of the plurality of compressors during the period of time relative to the predicted usage pattern.

17. The method of claim 11 wherein the predicted condition is a predicted weather and the one or more conditions associated with such operation patterns are one or more weather conditions.

18. The method of claim 11 wherein the control instructions are further generated or modified during the time period based on sensor data from sensors associated with one or more of the plurality of spaces which sensor data is measured during the time period and which sensors are sensors other than the temperature sensor which measures the temperature readings.

19. The method of claim 11 wherein the temperature band is a range of temperatures set relative to the temperature setting.

20. The method of claim 1 wherein the predicted condition is a predicted frequency and / or duration of opening of one or more doors associated with one or more of the plurality of spaces and the software monitors actual conditions associated with openings and / or duration of openings of the one or more doors during the time period and adjusts the control instructions based on a comparison of the actual conditions to the predicted conditions.

Images