Method for controlling a direct evaporative cooling unit and direct evaporative cooling system

Abstract

A method for controlling a direct evaporative cooling unit includes predicting water consumption and energy consumption of the direct evaporative cooling unit based on possible supply air temperatures of the direct evaporative cooling unit, selecting a target supply air temperature based on optimizing an objective function, the objective function including water consumption and energy consumption, and controlling the direct evaporative cooling unit according to the target supply air temperature.

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims the benefit of and priority to U.S. Provisional Patent Application No. 63 / 336,240, filed April 28, 2022, U.S. Provisional Patent Application No. 63 / 403,016, filed September 1, 2022, and U.S. Provisional Patent Application No. 63 / 403,018, filed September 1, 2022, the disclosures of all of which are incorporated herein by reference. [Background technology]

[0002] The present disclosure relates generally to direct evaporative cooling, e.g., for data centers. A data center may be a building, facility, etc. that includes computing hardware (e.g., servers, computer processing units, hard drives, etc.). The computing hardware typically generates heat during operation due to, e.g., electrical resistance within such computing hardware. Furthermore, the computing hardware may need to be kept within a proper temperature range in order to function properly. Thus, a need exists to remove heat from the data center (i.e., to cool the data center).

[0003] One approach for cooling data centers can be direct evaporative cooling. Direct evaporative cooling uses the evaporation of water to create a cooling effect that can be used to affect the temperature of the data center. However, direct evaporative cooling consumes both water and energy, both or either of which may be scarce and / or valuable resources in certain geographic areas. Thus, techniques that can reduce the resource consumption of direct evaporative cooling units for data centers and / or detect failures in direct evaporative cooling units for data centers would be valuable. Summary of the Invention

[0004] One implementation of the present disclosure is a method for controlling a direct evaporative cooling unit, the method including: predicting water consumption and energy consumption of the direct evaporative cooling unit based on possible supply air temperatures (or other control variables) of the direct evaporative cooling unit, selecting a target supply air temperature (or target for other control variables) based on optimizing an objective function, the objective function including water consumption and energy consumption, and controlling the direct evaporative cooling unit according to the target supply air temperature (or target for other control variables).

[0005] In some embodiments, the water consumption includes at least one of water evaporated during operation of the direct evaporative cooling unit or water drained from a water tank of the direct evaporative cooling unit. In some embodiments, predicting the water consumption is based on a predicted change in the amount of airflow over the evaporative medium of the direct evaporative cooling unit and the humidity of the airflow over the evaporative medium. In some embodiments, the energy consumption includes the energy consumption of a fan of the direct evaporative cooling unit.

[0006] In some embodiments, a supply air temperature of the direct evaporative cooling unit is associated with a bypass damper position as a function of an outdoor air temperature. Anticipating energy consumption may be based on the bypass damper position. Controlling the direct evaporative cooling unit according to the target supply air temperature may include controlling the bypass damper to a position determined based on the target supply air temperature and the outdoor air temperature.

[0007] In some embodiments, predicting the water consumption and energy consumption of a direct evaporative cooling unit based on the possible supply air temperatures of the direct evaporative cooling unit includes utilizing a plurality of curves representing the relationship between pressure difference and flow rate for a plurality of bypass damper positions.

[0008] Another implementation of the present disclosure is a method for controlling a group of direct evaporative cooling units serving a data center, the method including: selecting a target supply air temperature for the group of direct evaporative cooling units by performing a first optimization of objectives based on projected water and energy consumption of the group of direct evaporative cooling units, allocating the projected water and energy consumption to the group of direct evaporative cooling units by performing a second optimization constrained by the target supply air temperature, and controlling the group of direct evaporative cooling units according to a result of the allocating.

[0009] Another implementation of the present disclosure is a method for fault detection for direct evaporative cooling of a data center, the direct evaporative cooling affecting humidity and temperature of the data center. The method includes generating a prediction for humidity of the data center and a prediction for a temperature of the data center using a model, and determining a fault condition in response to an actual measurement for humidity deviating from the prediction for humidity while tracking the prediction for temperature, or determining a fault condition in response to an actual measurement for humidity deviating from the prediction for temperature while tracking the prediction for humidity.

[0010] In some embodiments, the method includes altering control of direct evaporative cooling of the data center to resolve the fault condition. The model may use inputs including a bypass profile and an outdoor air temperature value. In some embodiments, the model is further based on a design efficiency of the direct evaporative cooling unit. In some embodiments, the method includes determining a degradation in the design efficiency based on a comparison of the predicted value and the actual value.

[0011] Another implementation of the present disclosure is a method for fault detection for a fleet of direct evaporative cooling units, the method including detecting one or more of the direct evaporative cooling units as outliers by performing a peer analysis on the performance of the direct evaporative cooling units, detecting a fault by examining the outlier, determining a cause of the fault using an analysis method, and initiating a recommended action to resolve the cause of the fault.

[0012] In some embodiments, the performance of the direct evaporative cooling units is quantified as an efficiency of the direct evaporative cooling units. In some embodiments, detecting one or more of the direct evaporative cooling units as outliers by performing a peer analysis includes using a generalized extreme studentized deviation. In some embodiments, determining the cause of the failure using the analysis method also includes performing a model-based calculation. In some embodiments, the recommended action includes one or more of mechanical maintenance, cleaning, replacing a sensor, replacing an evaporative media, or replacing an air filter.

[0013] Another implementation of the present disclosure is a method for controlling a group of direct evaporative cooling units serving a data center, the method including: selecting a target supply air temperature for the group of direct evaporative cooling units by performing a first optimization of objectives based on projected water and energy consumption of the group of direct evaporative cooling units, allocating the projected water and energy consumption to the group of direct evaporative cooling units by performing a second optimization constrained by the target supply air temperature, and controlling the group of direct evaporative cooling units according to a result of the allocating.

[0014] The method includes selecting a target supply air temperature by performing a first optimization that includes predicting predicted water and energy consumption for a plurality of possible supply air temperatures. The water consumption can include an amount of water evaporated during operation of the direct evaporative cooling units. In some embodiments, the predicted water consumption is predicted based on a predicted change in the amount of air flow over the evaporative medium of the direct evaporative cooling units and the humidity of the air flow over the evaporative medium. In some embodiments, the method includes predicting the predicted energy consumption by determining a fan speed based on a pressure differential and a volumetric flow rate associated with the target supply air temperature.

[0015] In some embodiments, the method includes controlling the direct evaporative cooling units according to the results of the dispensing, including causing a movement of a bypass damper of the direct evaporative cooling units. In some embodiments, the method includes predicting water and energy consumption based on possible supply air temperatures by utilizing a plurality of curves representing a relationship between pressure difference and flow rate for a plurality of bypass damper positions.

[0016] Another implementation of the present disclosure is a direct evaporative cooling system including a controller programmed to predict water and energy consumption of the direct evaporative cooling system based on a selected supply air temperature, determine a target supply air temperature by adjusting the selected supply air temperature to improve a value of an objective function, the objective function including the predicted water and energy consumption of the direct evaporative cooling system, and control the direct evaporative cooling system according to the target supply air temperature.

[0017] In some embodiments, the water consumption includes the amount of water evaporated during operation of the direct evaporative cooling system. In some embodiments, the controller is programmed to forecast the water consumption based on predicted changes in air flow rate over the evaporative medium of the direct evaporative cooling unit and humidity of the air flow across the evaporative medium.

[0018] In some embodiments, the controller is programmed to predict water and energy consumption using multiple curves representing the relationship between pressure differential and flow rate for multiple bypass damper positions.

[0019] In some embodiments, the direct evaporative cooling unit includes a bypass damper. The controller can be configured to control the direct evaporative cooling system according to a target supply air temperature by causing a change in the position of the bypass damper. The controller can also be programmed to predict energy consumption by determining a fan speed based on a pressure differential and a volumetric flow rate associated with a selected supply air temperature.

[0020] One implementation of the disclosure is a method for fault detection for direct evaporative cooling of a data center. The direct evaporative cooling affects humidity and temperature of the data center. The method includes generating a prediction for humidity of the data center and a prediction for temperature of the data center using a model, and determining a fault condition in response to an actual measurement for humidity deviating from the prediction for humidity while tracking the prediction for temperature, or determining a fault condition in response to an actual measurement for humidity deviating from the prediction for temperature while tracking the prediction for humidity.

[0021] In some embodiments, the method includes altering control of direct evaporative cooling of the data center to resolve the fault condition. The model may use inputs including a bypass profile and an outdoor air temperature value. In some embodiments, the model is further based on a design efficiency of the direct evaporative cooling unit. In some embodiments, the method includes determining a degradation in the design efficiency based on a comparison of the predicted value and the actual value.

[0022] Another implementation of the present disclosure is a method for fault detection for a fleet of direct evaporative cooling units, the method including detecting one or more of the direct evaporative cooling units as outliers by performing a peer analysis on the performance of the direct evaporative cooling units, detecting a fault by examining the outlier, determining a cause of the fault using an analysis method, and initiating a recommended action to resolve the cause of the fault.

[0023] In some embodiments, the performance of the direct evaporative cooling units is quantified as an efficiency of the direct evaporative cooling units. In some embodiments, detecting one or more of the direct evaporative cooling units as outliers by performing a peer analysis includes using a generalized extreme studentized deviation. In some embodiments, determining the cause of the failure using the analysis method also includes performing a model-based calculation. In some embodiments, the recommended action includes one or more of mechanical maintenance, cleaning, replacing a sensor, replacing an evaporative media, or replacing an air filter.

[0024] Another implementation of the present disclosure is a method for controlling a direct evaporative cooling unit, the method including: predicting water consumption and energy consumption of the direct evaporative cooling unit based on possible supply air temperatures of the direct evaporative cooling unit, selecting a target supply air temperature based on optimizing an objective function, the objective function including the water consumption and the energy consumption, and controlling the direct evaporative cooling unit according to the target supply air temperature.

[0025] In some embodiments, the water consumption includes at least one of water evaporated during operation of the direct evaporative cooling unit or water drained from a water tank of the direct evaporative cooling unit. In some embodiments, predicting the water consumption is based on a predicted change in the amount of airflow over the evaporative medium of the direct evaporative cooling unit and the humidity of the airflow over the evaporative medium. In some embodiments, the energy consumption includes the energy consumption of a fan of the direct evaporative cooling unit.

[0026] In some embodiments, a supply air temperature of the direct evaporative cooling unit is associated with a bypass damper position as a function of an outdoor air temperature. Anticipating energy consumption may be based on the bypass damper position. Controlling the direct evaporative cooling unit according to the target supply air temperature may include controlling the bypass damper to a position determined based on the target supply air temperature and the outdoor air temperature.

[0027] In some embodiments, predicting the water consumption and energy consumption of a direct evaporative cooling unit based on the possible supply air temperatures of the direct evaporative cooling unit includes utilizing a plurality of curves representing the relationship between pressure difference and flow rate for a plurality of bypass damper positions.

[0028] Another implementation of the present disclosure is a method for controlling a group of direct evaporative cooling units serving a data center, the method including: selecting a target supply air temperature for the group of direct evaporative cooling units by performing a first optimization of objectives based on projected water and energy consumption of the group of direct evaporative cooling units, allocating the projected water and energy consumption to the group of direct evaporative cooling units by performing a second optimization constrained by the target supply air temperature, and controlling the group of direct evaporative cooling units according to a result of the allocating.

[0029] Another implementation of the present disclosure is a method for fault detection for direct evaporative cooling of a data center, the direct evaporative cooling affecting humidity and temperature of the data center. The method includes generating a prediction for humidity of the data center and a prediction for temperature of the data center using a model, and triggering a sensor fault in response to a measurement for humidity deviating from the prediction for humidity while the measurement for temperature tracks the prediction for temperature, or in response to a measurement for humidity deviating from the prediction for temperature while tracking the prediction for humidity.

[0030] In some embodiments, the method also includes performing an action to resolve the sensor fault in response to triggering the sensor fault. Performing the action to resolve the sensor fault may include moving the sensor, cleaning the sensor, or replacing the sensor. Performing the action to resolve the fault may include altering control of direct evaporative cooling of the data center to adapt to the fault condition.

[0031] In some embodiments, the model uses inputs including a bypass profile and an outdoor air temperature value. The model may be further based on a design efficiency of the direct evaporative cooling unit. In some embodiments, the method includes determining a reduction in the design efficiency based on a comparison of the predicted value and the actual value.

[0032] Another implementation of the present disclosure is a method for sensor fault detection for a first sensor measuring a first condition and a second sensor measuring a second condition. The device is operable to act on the first condition and the second condition. The method includes: generating a predicted value for the first condition and a corresponding predicted value for the second condition using a physics-based model that defines a physics-based relationship between the first condition and the second condition; triggering a sensor fault in response to determining that the measurement value for the first condition deviates from the predicted value for the first condition and that the measurement value for the second condition tracks the predicted value for the second condition; and ceasing to trigger the sensor fault in response to determining that the measurement value for the first condition deviates from the predicted value for the first condition and that the measurement value for the second condition deviates from the predicted value for the second condition.

[0033] In some embodiments, the method also includes performing an action to resolve the sensor fault in response to indicating the sensor fault. Performing an action to resolve the sensor fault can include moving the first sensor or the second sensor, cleaning the first sensor or the second sensor, and / or replacing the first sensor or the second sensor. In some embodiments, the method includes altering control of direct evaporative cooling of the data center to adapt to the fault condition.

[0034] In some embodiments, the physics-based model uses a bypass position of the direct evaporative cooling unit as an input. The physics-based model may be based on the efficiency of the direct evaporative cooling unit. In some embodiments, a first sensor measures the supply air humidity of the direct evaporative cooling unit and a second sensor measures the supply air temperature of the direct evaporative cooling unit.

[0035] In some embodiments, a method includes controlling a direct evaporative cooling unit by determining a target supply air temperature that minimizes an objective function that accounts for water consumption of the direct evaporative cooling unit.

[0036] Another implementation of the present disclosure is a method for sensor failure detection for a first sensor measuring a first indoor air condition and a second sensor measuring a second indoor air condition, the method including determining a predicted relationship between the first indoor air condition and the second indoor air condition based on thermodynamic principles, evaluating a first measurement of the first indoor air condition from the first sensor and a second measurement of the second indoor air condition from the second sensor to determine whether the first and second measurements satisfy the predicted relationship between the first indoor air condition and the second indoor air condition, and triggering a failure of the first sensor or the second sensor in response to determining that the first measurement from the first sensor and the second measurement from the second sensor do not satisfy the predicted relationship between the first indoor air condition and the second indoor air condition.

[0037] In some embodiments, the first sensor and the second sensor are associated with a first unit of the plurality of equipment units. The method may also include attributing the fault to the first sensor or the second sensor by performing a peer analysis on the plurality of equipment units. Performing the peer analysis may include determining whether a measurement from the first sensor or a measurement from the second sensor is an outlier. In some embodiments, performing the peer analysis includes using a generalized extreme studentized deviation.

[0038] In some embodiments, the method includes, in response to the failure, performing one or more of cleaning, moving, or replacing the first sensor or the second sensor.

[0039] The present disclosure will become more fully understood from the following detailed description taken in conjunction with the accompanying drawings, in which like reference numerals refer to like elements and in which: [Brief description of the drawings]

[0040] [Figure 1] FIG. 1 is a block diagram of a data center according to some embodiments. [Diagram 2] FIG. 1 is a diagram of a direct evaporative cooling (DEC) unit, according to some embodiments. [Diagram 3] 1 is a flowchart of a process for controlling a DEC unit according to some embodiments. [Figure 4] 1 is a graph illustrating a control approach for a DEC unit according to some embodiments. [Diagram 5] 1 is a set of flowcharts providing a process for controlling a DEC unit, according to some embodiments. [Figure 6] 1 is a flowchart of a process for controlling one or more DEC units, according to some embodiments. [Figure 7A] 7 is a set of graphs illustrating data from examples of the process of FIG. 6, in accordance with some embodiments. [Figure 7B] 7 is a graph illustrating data from an example for the process of FIG. 6, according to some embodiments. [Figure 8] 7 is a set of graphs illustrating data from examples of the process of FIG. 6, in accordance with some embodiments. [Figure 9] 1 is a flowchart of a process for model-based fault detection for a DEC unit, according to some embodiments. [Figure 10A] 10 is a set of graphs relating to an exemplary execution of the process of FIG. 9 in accordance with some embodiments. [Figure 10B] 10 is a set of graphs relating to an exemplary execution of the process of FIG. 9 in accordance with some embodiments. [Figure 11A] 1 is a flowchart of a process for modeling the operation of a DEC unit according to some embodiments. [Figure 11B] 11B is a graph illustrating steps of modeling the process of FIG. 11A according to some embodiments. [Figure 11C] 11B is a graph illustrating steps of modeling the process of FIG. 11A according to some embodiments. [Figure 11D] 11B is a graph illustrating steps of modeling the process of FIG. 11A according to some embodiments. [Figure 12] 11B is a flowchart of a process for controlling a DEC unit, for example, using the model from the process of FIG. 11A, according to some embodiments. [Figure 13] 13 is a graph relating to control of a DEC unit according to some embodiments. [Figure 14] 1 is a set of graphs relating to control of a DEC unit according to some embodiments. [Figure 15] 1 is a set of graphs relating to control of a DEC unit according to some embodiments. [Figure 16] 1 is a set of graphs relating to control of a DEC unit according to some embodiments. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0041] Generally, with reference to the drawings, the teachings may be applied in various direct evaporative cooling systems, for example, using computer room air conditioning systems, as described in U.S. Patent Nos. 9,635,786 (issued April 25, 2017) and 9,521,783 (issued December 13, 2016) and / or U.S. Application Nos. 17 / 482,181, 18 / 071,336, or 18 / 071,327, the entire disclosures of which are incorporated herein by reference. A direct evaporative cooling (DEC) unit (DEC system) uses a pump that circulates a fluid through an evaporative medium, a tank, and a pump. A supply fan operates to blow air across the evaporative medium, creating evaporation of the fluid supplied from the tank. In some embodiments, the pump operates to push enough fluid into the tank to pull salts or other impurities from the fluid back into the tank (rather than leaving such salts in the evaporative medium as the fluid evaporates). Although salt is described as a primary example of a substance that may be present in the fluid, it is contemplated that the fluid may contain any number or type of impurities that may increase in concentration as the fluid evaporates (e.g., dissolved salts or minerals, particulate matter, other fluids that do not evaporate within the evaporative medium, etc.) Although salt is described throughout this disclosure for ease of explanation, the same or similar control strategies may be used to treat other impurities in the fluid.

[0042] The tank is initially filled and as evaporation occurs the liquid level drops to provide cooling. As the fluid evaporates, the salt concentration increases. When the salt reaches a threshold level in the tank (e.g., measured by a sensor), the tank is drained. Also, at regular intervals, the tank is fully flushed (e.g., for cleaning and sanitation purposes). In some embodiments herein, the controller is programmed to implement a water usage control process, e.g., optimizing water usage. One goal of the control process is to manage the increase in salt concentration in the tank to adjust the duration that salt accumulates to a threshold limit and a fixed interface for flushing the tank, thereby reducing the overall number of tank flushes (e.g., by preventing the salt concentration from reaching the threshold until a pre-set flush time). Implementing such control can include using a penalty function or constraint based on such conditions to make an on / off decision for the DEC system.

[0043] In some embodiments, the control circuitry may also be programmed to provide fan power optimization. Fans on servers / computers / CPUs etc. in a data room may control the exhaust temperature coming out of the CPU (measured by a sensor) so that the fan speeds up with increasing supply temperature or increasing CPU power usage. As CPU fans speed up, a vacuum builds up behind them, so the supply fans of a cooling (e.g. DEC) system have control of the static pressure in front of the CPU (e.g. feedback control by a pressure sensor). The supply fans are indirectly controlled by the CPU fans.

[0044] Therefore, since the supply air temperature to the CPU affects the required fan speed and has a cubic relationship with the power for the fans to flow, at some point it may be more efficient to turn on evaporative cooling to reduce the supply temperature than to run all the fans harder based on the cost of water, power, OAT, and fan model. The bypass damper may be controlled to make such a transition (e.g., from free cooling and / or outdoor air ventilation to cooling the air using the DEC system). At some point, 100% of the supply air will pass through the evaporative media, at which point the control circuitry may again increase the fan speed. An optimization process or other control decision process may be performed to make equipment control decisions (fan speed, DEC on / off or other setpoints, damper position, etc.) that can be made to improve overall efficiency.

[0045] Additionally, the control circuitry can be programmed to execute a control process to reduce the overall water consumption by the DEC system and the fan power consumption (and in some embodiments the power consumption of other components of the DEC system, such as pumps, etc.) by coordinating the operation of all such systems in an energy efficient manner. Coordinating the water consumption and the fan power consumption can include building a model and performing operations to select the supply air temperature provided by the DEC system over time, which results in the minimization of an objective function that considers both the water consumption and the fan power consumption. The supply air temperature value can then be used to control the DEC system, for example, by controlling an actuator to affect the bypass damper position.

[0046] In some embodiments herein, the controller enables detection of a fault in the DEC unit by comparing predicted temperature and humidity values ​​determined by the model with measured values. Based on a particular discrepancy, a fault may be detected. In some embodiments, the fault may be automatically corrected, for example, by adding an offset to the sensor reading in response to detecting that a faulty sensor is providing values ​​that are offset from the actual measured values.

[0047] Such features improve operation of the DEC unit by, for example, reducing resource usage in cooling the data center, improving cooling of the data center. As used herein, the controller(s), control circuitry, etc. may in various embodiments be provided in a cloud-based system, an on-site or off-site server, in a CPU / server / etc. cooled by the equipment being controlled, locally (e.g., on an edge device) by the equipment, etc., and / or combinations thereof.

[0048] The features herein may be used in addition to, in conjunction with, or otherwise complement the features described in U.S. patent application Ser. No. 16 / 579,686, filed Sep. 23, 2019, the entire disclosure of which is incorporated herein by reference.

[0049] Referring now to FIG. 1, a diagram of a data center 100 is shown, according to some embodiments. The data center 100 includes a plurality of server racks, shown as server racks 102a, 102b, 102c, 102d, 102e, and 102f, arranged parallel to one another. The server racks are positioned with a cold space 104 providing a cold aisle around or between the server racks 102a. Hot aisles 106a, 106b, and 106c are positioned between pairs of server racks (with a hot aisle 106a between server racks 102a and 102b, a hot aisle 106b between server racks 102c and 102d, and a hot aisle 106c between server racks 102e and 102f). The hot aisles 106a, 106b, and 106c are sealed off from the cooling space 104 except for air that may flow through the server racks 102a-102f. An exhaust fan 110 operates to draw hot air from the hot aisles 106a, 106b, 106c. A number of direct evaporative cooling units 108a, 108b, 108c, 108d, 108e, 108f are included to provide air (e.g., chilled air) to the cold space 104.

[0050] The server racks 102a-102f hold various servers, processors, hard drives, routers, and other computing hardware that generate heat during operation (e.g., due to internal electrical resistance). The server racks 102a-102f may include fans that draw relatively cool air from the cold space 104, across the computing hardware, and into the hot aisles 106a, 106b, 106c. The airflow may also be driven by a pressure differential across the server racks 102a-102f, created, for example, by the operation of the exhaust fan 110. Thus, air reaching the hot aisles 106a, 106b, 106c is heated by the server racks such that the hot aisles 106a, 106b, 106c have an air temperature that is greater than the air temperature of the cooling space 104. The direct evaporative cooling units 108a, 108b, 108c, 108d, 108e, 108f operate to provide air to the cooling space 104, including air that is cooled by direct evaporative cooling. The server racks 102a-102f are thereby cooled, for example, such that the server racks 102a-102f are maintained within a target temperature range.

[0051] 2, a block diagram of a DEC unit 108 serving a server rack 102 is shown, according to some embodiments. In some embodiments, the DEC unit 108 may be any one of the DEC units 108a-108f of FIG. 1, and the server rack 102 may be any one of the server racks 102a-102f of FIG. 1. FIG. 1 shows the server rack 102 separating a cold space 104 and a hot space 106, which in some embodiments may be one of the hot aisles 106a-106c of FIG. 1. In other embodiments, the DEC unit 108 serves a single server rack 102 in a data center (e.g., a modular computing room) having a cold space 104 and a hot space 106 separated by the server rack 102. As shown, the server rack 102 may include multiple CPU fans 200a, 200b, 200c operable to push air from the cold space 104 to the hot space 106, which may be an exhaust. In some embodiments, fans may additionally or alternatively be included at an exhaust port 202 in the hot space 106 to push air from the hot space 106 outside the data room (i.e., the outside environment).

[0052] The DEC unit 108 operates to provide air to the cold space 104. Air is delivered to the cold space 104 by the DEC unit 104, referred to as supply air. As illustrated, the DEC unit 108 is configured to provide a supply air temperature T 供給 (which can be measured by temperature sensor 204) and supply air humidity (which can be measured by humidity sensor 206) into the cold space 104. s The supply air is a combination of air flow through face channels 208 in which evaporative media 210 is positioned and bypass channels 212 that are open to the air flow and allow air to bypass evaporative media 210. The supply air flow rate w s is the surface channel 208 (w f ) and the flow rate through the bypass channel 212 (w b) (i.e., w s =w f +w b ).

[0053] The DEC unit 102 includes a supply fan 214 operable to push air into the surface channel 208 and the bypass channel 212 with one or more dampers included to direct the airflow from the supply fan 214 into the surface channel 208, the bypass channel 212, or some combination thereof. As shown, the DEC unit 102 includes an actuator 216 operable to reposition the bypass damper 218 and the surface damper 220. The bypass damper 218 and the surface damper 220 may mechanically interact and may be referred to herein as a single damper (e.g., a bypass damper), for example, such that the damper(s) direct air completely through the bypass channel 212 at a maximum damper position, completely through the surface channel 208 at a minimum damper position, and partially through both the bypass channel 212 and the surface channel 208 at different rates throughout a range of damper positions between the minimum and maximum positions. In some embodiments, the bypass damper 218 and the face damper 220 are independently controllable and can be set to any position (e.g., fully open, fully closed, 20% open, 40% open, 75% open, etc.) independent of one another. The actuator 216 is thereby controllable to direct different amounts of airflow provided by the supply fan 214 through the face channel 208 and the bypass channel 212 at different times, for example, to implement a control strategy, as described below. The power consumed by the supply fan 214 to provide a certain amount of airflow may depend on the damper position due to the different resistance to airflow in the face channel 208 compared to the bypass channel 212.

[0054] The airflow through the surface channel 208 passes through the evaporative medium 210 where water evaporates from the evaporative medium 210 into the airflow. The DEC unit 102 includes a water tank 222 configured to hold water and a water pump 224 configured to pump water from the water tank 222 to the evaporative medium. In FIG. 1, a volume of water (pumped water mass m p ) is pumped into the evaporation medium 210, and the water (evaporated water mass m e ) evaporates through the surface channel 208 into the airflow, and the remaining water (return water mass m e ) returns to the tank 222. The tank 222 can be controlled to periodically drain the tank 222 (i.e., periodically open to allow a certain amount of water to flow out, a drained water mass m d t) which may be drained on a set schedule to ensure that the water in the tank is sanitary (e.g., preventing algae or bacterial growth, etc.). The tank 222 also contains a utility water mass m u 1. Water is received from a utility 228 or other source, with a quantity of water received from a utility shown in FIG.

[0055] As water evaporates from the evaporative medium 210, the evaporated water leaves behind dissolved salts in the water. e may be flushed to return the salt to the tank. Over time, the concentration of dissolved salts in the tank will increase due to some of the water evaporating, and may eventually become so high that the water is no longer suitable for use in the evaporative medium 210. The tank 222 may then be drained and refilled to provide fresh water for use in evaporative cooling.

[0056] The DEC unit 108 is shown to include a controller 230. The controller 230 may include circuitry configured (e.g., programmed) to perform operations described herein in connection with controlling the DEC unit 108, and in some embodiments, controlling the CPU fans 200a-200c, and / or the operation of the computing hardware of the server rack 102. The controller 230 may also, or alternatively, provide fault detection to the DEC unit 108. In some embodiments, the controller 230 includes one or more processors and a non-transitory computer-readable medium that stores program instructions that, when executed by the one or more processors, cause the one or more processors to perform the operations attributed to the controller 230 herein. The controller 230 may, in various embodiments, be provided on the computing hardware of the server rack 102, provided remotely from the DEC unit 108, and / or some combination thereof, included locally as part of the DEC unit 108 (e.g., packaged, coupled with, etc., the DEC unit 108). 2, the controller 230 may be in communication with the supply fan 214, the actuator 216, the water pump 224, the drain 226, the temperature sensor 204, the humidity sensor 206, and other sensors (shown as pressure sensor 232 measuring the pressure differential across the server rack 102, and temperature sensor 234 positioned in the hot space 106 measuring the exhaust air temperature, as well as temperature sensor 236 positioned to measure the outdoor air temperature of the outdoor air entering the DEC unit 108 and the supply fan 214). Although several examples of sensors are shown in FIG. 2, it is contemplated that any number or type of sensors may be present in the DEC unit 108, the server rack 102, the cold space 104, the hot space 106, downstream or upstream of the DEC unit 108, in the tank 222, or located anywhere else in the system, in various embodiments. The controller 230 may execute the various processes shown in the figures and described below, including using various equations, algorithms, etc. described herein.

[0057] In the example of Figures 1-2, the cost of cooling a data center can be attributed to three main components: electricity to power the fans, electricity to power the water pump(s), and the cost of water. The CPU fans 200a-200c as well as the supply fan 214 (and / or exhaust fan 110) can have their speeds modulated, and thus the fan speed s ファン Based on the affinity law, power P ファン vs flow w ファン should have a relationship of P ファン =as ファン 3 w ファン =as ファン

[0058] The fans 200a-200c, 214 may be controlled by the controller 230 based on the exhaust temperature of the air leaving the server rack 102 (e.g., as measured by sensor 234) such that the total flow desired by the CPU fans is:

number

number

[0059] Tank 222(c タンク The concentration of the salt in may follow a differential equation, where c u is the salt concentration in the water as received from the utility 228.

number

[0060] Before the concentration limit is reached, in some embodiments, water is not drained from the tank 222 and the concentration is integrated as new water from the utility 228 replaces that lost to evaporation.

number

number

number

[0061] Water is therefore drained in proportion to the amount of evaporation, the rate depending on the concentration just as the number of tank changes depends on the concentration before the drain was opened. The total amount of water evaporated is governed by the difference between the supply temperature and the outdoor air temperature.

number

[0062] In some embodiments, controller 230 is programmed to determine whether, at a given time, to increase cooling by increasing fan speed or by using water in evaporative cooling. For example, at any particular moment, controller 230 may determine fan energy assuming the air is at the outdoor temperature (e.g., measured by sensor 236) and may determine fan energy assuming a temperature achievable by evaporative cooling (e.g., from the design value of the equipment). Controller 230 may use logic that indicates that the point at which it makes sense to turn on evaporative cooling occurs when the marginal cost increase from fan power is equal to the marginal cost increase from the water and electricity used by the evaporative cooling system.

number

[0063] This tipping point occurs under conditions that are highly dependent on electricity and water price structures, as well as outdoor air conditions, which may be taken into account by the controller 230. Adjustments may be made by the controller 230 to allow such logic to take into account the inability to run evaporative cooling at very low loads. Assuming a trade-off here, the controller 230 may initially run the supply fan 314 at an increasing speed, and then as the load gets higher, evaporative cooling (i.e., use of water and surface channels 208) is turned on to reduce the supply air temperature. When it is no longer possible to reach the desired supply air temperature using evaporative cooling at the current fan speed, the controller 230 may increase the fan speed, thereby providing more outside air as well as increasing the amount of ventilation performed. In some embodiments, the controller 230 uses neural networks, reinforcement learning, or other forms of machine learning to determine when to turn on evaporative cooling versus running the fan faster. Furthermore, when evaporative cooling is used, extremum-seeking control may be used to find the supply air temperature that minimizes the total cost without incurring any modeling errors.

[0064] 3, a flow chart of a process 300 for determining whether to use evaporative cooling is shown, according to some embodiments. Process 300 may be performed by controller 230, for example.

[0065] At step 302, a weather forecast is obtained and a cooling load is predicted. The weather forecast may be accessed by the controller 230, for example, via the Internet, from a third party weather service (e.g., a government weather service). The weather forecast may indicate, for example, outdoor air temperature and / or outdoor air humidity for the next hour, the next day, etc. The cooling load may be predicted, for example, based on the weather forecast and / or based on predictions of operation of the computing equipment of the server rack 102.

[0066] At step 304, a penalty function for exceeding a desired temperature is generated. The desired temperature may be a maximum allowable temperature for the data center (e.g., based on building operator rules, etc.). The penalty function may be formulated as follows:

number

number

number

[0067] If the tank 222 is not yet filled, the penalty on the time period of interest (perhaps equal to the time before the next required tank flush) is compared to a threshold. A small penalty below the threshold means that the tank 222 does not need to be filled, while a large penalty indicates that the tank 222 needs to be prepared for evaporative cooling by filling it. Thus, in step 308, a determination is made whether the tank 222 is empty and the penalty is greater than the threshold. If so (step 308: yes), then in step 310, the tank 222 is filled and prepared for evaporative cooling. If not (step 308: no), then the process 300 skips step 310 and proceeds to step 312.

[0068] In step 312, it is determined whether the tank 222 is full and the supply air temperature T 供給 is the temperature set point T 設定点A determination is made whether evaporative cooling is greater than . If yes (i.e., the tank 222 is ready for evaporative cooling and needs to cool the supply air to reach the set point), evaporative cooling is enabled in step 314. In step 314, the actuator 216 can operate the dampers 218, 220 to direct air through the evaporative medium 210 and the water pump 224 can operate to supply water to the evaporative medium 210 such that the airflow from the supply fan 214 is cooled by direct evaporative cooling in the surface channels 208.

[0069] If the answer is "no" in step 312 (i.e., the tank is not full or 供給 is the temperature set point T 設定点 If t is less than or equal to t , direct evaporative cooling is not enabled and the process proceeds to step 316. Step 316 begins when process 300 resets t , which indicates the amount of time (e.g., minutes) between iterations of process 300. s t s The value of may be selected, for example, by a user.

[0070] Thereby, process 300 of FIG. 3 can determine whether to use evaporative cooling in an advanced manner that may facilitate water conservation.

[0071] Referring now to FIG. 4, a graphical representation illustrating a control approach that may be implemented by controller 230, according to some embodiments, is shown. FIG. 4 shows a graph 400 illustrating operation of supply fan 214 and evaporative cooling (e.g., by running pump 224) over a range of cooling needs (demand, load). Graph 400 illustrates that the fan initially operates without evaporative cooling in a first zone 402. Then, in a second zone 404, evaporative cooling is gradually turned on at a constant fan speed (by controlling surface damper 220 and bypass damper 218 to gradually direct more air through evaporative media 210). As cooling demand increases in a third zone 406, both the evaporative cooling rate and fan speed are increased together by directing all airflow across evaporative media 210 while increasing fan speed.

[0072] 5, a pair of flow charts are shown illustrating a first portion 501 and a second portion 502 of a process 500, according to some embodiments. The process 500 may be performed by the controller 230 in some embodiments. In some embodiments, the first portion 501 is performed before the second portion 502. The first portion 501 may run offline and the second portion 502 may run online to control the DEC unit 108.

[0073] In step 504, a fan model is trained. Training the fan model involves training the fan equation P ファン =as ファン 3 and w ファン =as ファン Training the fan model may include identifying a parameter a in the leakage and w s =w cpu +w 漏れ (P s). Training the fan model may be performed using measurements of fan power consumption along with training data representing other variables (e.g., temperature, damper position, airflow measurements, pressure measurements, etc.). Various other equations for fan operation are provided below and may be used in step 504.

[0074] At step 506, a load forecasting model is trained. The load forecasting model may be configured to forecast the amount of heat (i.e.,

number

[0075] The model resulting from the first portion 501 of the process 500 may be formulated as follows:

number

number

[0076] In step 508, a weather forecast is obtained (e.g., from a weather service) and a cooling load is predicted. The cooling load may be predicted using the load forecast model trained in step 506. Step 508 may, for example, be performed using the load forecast model trained in step 506 to forecast the cooling load based on the external air temperature T OAand heat from computing devices

number

[0077] In step 510, a penalty function is generated to penalize temperatures of the server rack 102 that are above a desired temperature, i.e., above a desired maximum temperature of the server rack 102. The penalty function may be, for example, c ペナルティ =r ペナルティ (T 排気 -T sp ) 2 It can be given by:

[0078] In step 512, the cost plus penalty is optimized in response to draining the tank 222 at least every N days. Step 512 can include considering water usage as follows, where water used to dilute salt is noted separately from water used for evaporation using an additional subscript d:

number

number

number

number

number

[0079] In step 514, the DEC unit 108 is controlled using the supply temperature setpoint and evaporative cooling enable decision to achieve the DEC operation determined to minimize the objective function, thereby operating as the culmination of the preceding steps of process 500. In step 516, the second portion 502 of process 500 can be repeated, for example, on a periodic schedule (e.g., every 15 minutes), allowing process 500 to account for changes in load or weather forecasts.

[0080] In some embodiments, step 514 includes controlling the computing device to act on the heat generated thereby, i.e.,

number

[0081] 6, a flow chart of a process 600 for controlling a DEC is shown, according to some embodiments. Process 600 may be performed by controller 230, in some embodiments. In some embodiments, process 600 may be used in combination with process 500. Process 600, in various embodiments, may provide advantageous operation of DEC unit 102, such as in FIG. 2, and / or multiple DEC units 102a-102f in FIG. 1.

[0082] The process 600 may be based on (eg, use, include algorithmic steps derived therefrom, etc.) the following equation related to the efficiency of a DEC unit (eg, DEC unit 108):

number

[0083] In step 602, heat generated from a CPU (e.g., a server in a server rack 102, etc.) is estimated from historical temperature data. The historical temperature data can include hot aisle temperature (exhaust temperature), cold aisle temperature, and supply temperature. Bypass values ​​(damper positions) can also be used. As an example of a data center including multiple aisles and DEC units (e.g., similar to FIG. 1), FIG. 7A shows a first graph 700 of supply temperatures from multiple DEC units over a period of time (e.g., several days), a second graph 702 of hot aisle temperatures over the same period or a portion thereof, a third graph 704 of cold space (cold aisle) temperatures over the same period, and a fourth graph 706 showing bypass positions over the same period according to a set of experimental data. In some embodiments, step 602 is performed by taking an average over multiple DEC units, multiple aisles, etc. In other embodiments, separate estimations are performed for multiple models.

[0084] Based on the temperature measurements and the bypass position, step 602 may include estimating the heat generated by the CPU.

number

number

number

[0085] In step 604, the cold aisle temperature is predicted from the DEC bypass profile (bypass position over time), the outdoor air temperature, and the DEC design value. In some embodiments, step 604 can be considered as validating the DEC model used by process 600 and involves the use of such a model, i.e.:

number

[0086] By setting an efficiency value from the design state (e.g., 92%) and using an average bypass profile from historical data (e.g., the average position of each bypass damper at each time step from the data shown in graph 706 of FIG. 7A), the cold side (aisle) temperature can be predicted in step 604. FIG. 8 shows an exemplary first graph 800 of bypass positions used in an exemplary implementation of step 604, and a second graph 802 including a line plotting predicted / estimated cold aisle temperatures over time, as derived in step 604. FIG. 8 further includes a third graph 804 illustrating that such calculated cold aisle values ​​track actual cold aisle values, thereby validating the approach used by process 600.

[0087] In step 606, the hot aisle temperature is predicted. The hot aisle temperature may be predicted, for example, by using the cold aisle temperature prediction from step 604 and the CPU heat estimation / prediction from step 602. Step 606 includes: 高温 for

number

[0088] In step 608, the amount of water mass evaporated is predicted. Step 608 may assume that the same efficiency value (i.e., efficiency) holds for humidity (e.g., 92%) such that the portion of the air passing through the evaporative media 210 increases its humidity by 92% (or other selected efficiency percentage) halfway between the outdoor air humidity and 100% humidity. Because the amount of water evaporated is equal to the amount of water required to increase the humidity of the air by a known amount, step 608 may use that information to perform a water mass balance and calculate the total amount of water evaporated as a function of the outdoor air humidity and 100% humidity.

[0089] In step 610, one or more DEC units are controlled using a forecast, for example, a temperature forecast and / or a water mass evaporation forecast. In some embodiments, the water mass forecast is used in a manner similar to that described with reference to, for example, FIGS. 3-4. For example, step 610 can include:

number

number

number

number

number

[0090] Step 610 may also additionally or alternatively include optimally controlling fan speed and / or optimally controlling bypass position to optimally control water-related components. Various such features are described elsewhere herein and may be used in step 610 in some embodiments.

[0091] 9, a flow chart of a process 900 for model-based fault detection for a direct evaporative cooling unit is shown, according to some embodiments. In some embodiments, the controller 230 is programmed to execute the process 900. The process 900 may, in various embodiments, use other process modeling, equations, etc. described herein.

[0092] In step 902, the supply air temperature and supply air humidity of the DEC unit are estimated (or predicted) based on the bypass profile and the outdoor air temperature. Estimating the supply air temperature and supply air humidity may include, for example, determining predicted values ​​of those variables based on the actual bypass position and the actual outdoor air temperature for each time step over a period of time. Estimating the supply air temperature and supply air humidity may be done using models described elsewhere herein (e.g., with respect to process 600). Like the models used in process 600, the estimates (or predictions) may be based on ideal design efficiencies or other design values ​​from the manufacturer.

[0093] The estimate determined in step 902 is illustrated in Figures 10A-B. Figure 10A shows a set of graphs for a first DEC unit (DEC A, e.g., shown as DEC 102a in Figure 1) including a first graph 1000 comparing measured humidity ratio to predicted (or estimated) humidity ratio, a second graph 1002 showing measured and predicted (or estimated) supply relative humidity, a third graph 1004 showing measured and predicted (or estimated) supply air temperature, and a fourth graph 1006 showing bypass position. FIG. 10B shows a set of graphs for a second DEC unit (DEC B, e.g., shown as DEC 102b in FIG. 1 ) including a first graph 1050 comparing the measured humidity ratio to the predicted (or estimated) humidity ratio, a second graph 1052 showing the measured and predicted (or estimated) supply relative humidity, a third graph 1054 showing the measured and predicted (or estimated) supply air temperature, and a fourth graph 1056 showing the bypass position.

[0094] In step 904, the predicted supply air temperature is compared to the measured supply air temperature. Comparing the predicted supply air temperature to the measured supply air temperature may include determining a difference (e.g., gap) between the predicted supply air temperature and the measured supply air temperature (e.g., at a given time, summed or integrated over a period of time, etc.). In some embodiments, a statistical metric of the difference between the predicted supply air temperature and the measured supply air temperature is evaluated (e.g., variance of the difference, average of the difference, standard deviation of the variance, etc.). Comparing the predicted supply air temperature to the measured supply air temperature may include determining whether such difference, sum of the difference, statistical metric, etc., exceeds a corresponding threshold. If so, the comparison may be deemed to indicate a discrepancy between the predicted supply air temperature and the measured supply air temperature.

[0095] In step 906, the predicted supply air humidity is compared to the measured supply air humidity. Comparing the predicted supply air humidity to the measured supply air humidity may include determining a difference (e.g., gap) between the predicted supply air humidity and the measured supply air humidity (e.g., at a given time, summed or integrated over a period of time, etc.). In some embodiments, a statistical metric of the difference between the predicted supply air humidity and the measured supply air humidity is evaluated (e.g., variance of the difference, mean of the difference, standard deviation of the variance, etc.). Comparing the predicted supply air humidity to the measured supply air humidity may include determining whether such difference, sum of the difference, statistical metric, etc., exceeds a corresponding threshold. If so, the comparison may be considered to indicate a discrepancy between the predicted supply air humidity and the measured supply air humidity.

[0096] In some embodiments, the process 900 is performed for a group of DEC units. In such embodiments, the process 900 (e.g., comparing steps 904 and 906) can include performing a peer analysis to compare the performance of the DEC units and detect outliers. For example, comparing the predicted and measured supply air temperatures in step 904 can include using a peer analysis to detect outlier supply air temperatures (predicted or measured) across a group of DEC units. As another example, comparing the predicted and measured supply air humidity in step 906 can include using a peer analysis to detect outlier supply air humidity (predicted or measured) across a group of DEC units. As another example, the supply air temperature and humidity can be used to quantify the efficiency of the DEC units, which can be compared to each other to detect outliers. The detection of outliers can be performed using a generalized extreme Studentized deviance analysis.

[0097] In step 908, in response to the discrepancy determined in step 904 or 906, it is determined that a fault has occurred. Various faults are possible. For example, if step 904 determines that the supply air temperature is not inconsistent, while step 906 determines that the supply air humidity is inconsistent, step 908 may include determining that a humidity sensor (e.g., humidity sensor 206) measuring the supply air humidity has a fault. FIG. 10B illustrates an example of such a scenario, where the third graph 1054 shows the actual supply temperature and the predicted supply temperature closely following each other, while gaps consistently appear between the predicted and measured supply relative humidity in the second graph 1052, and between the predicted and measured supply relative humidity ratios in the first graph 1050. Because the predicted supply air temperature is still met, step 908 may infer that the error causing the discrepancy between the humidities is a humidity sensor failure.

[0098] As another example, if step 906 determines that no inconsistency has occurred in the supply air humidity but step 904 determines that an inconsistency has occurred in the supply air temperature, step 908 may include determining that a temperature sensor (e.g., humidity sensor 206) that measures the supply air temperature has failed.

[0099] As another example, if step 904 determines that the measured supply air temperature is consistently higher than predicted (anticipated, estimated) and step 906 determines that the supply air humidity is consistently lower than predicted (but the supply air humidity and supply air temperature maintain the predicted relationship), step 908 may include determining that the DEC unit is operating at less than predicted efficiency or that some other control or equipment failure has occurred. DEC unit efficiency may be lost due to deterioration of the evaporative media 210, mechanical problems with the dampers 218 / 220, problems with the supply fan 214, etc. Thus, in such a scenario, a loss of efficiency has occurred and a failure may be determined in step 908 indicating that the DEC unit could benefit from maintenance.

[0100] As another example, a fault may be detected if an outlier (inconsistency) is found for one of the DEC units of the population (relative to other units in the population) in steps 904 and / or 906. Process 900 may include performing analytical methods and / or model-based calculations to examine the outlier(s) to determine the cause of the fault and / or to determine recommendations for resolving the fault (e.g., maintenance steps, automated control actions, etc.).

[0101] In step 910, in response to detecting a fault in step 908, operation of the DEC unit may act to resolve or compensate for the fault. For example, in an example where the fault indicates that a sensor is faulty and providing measurements that deviate from actual or predicted values ​​by a certain amount, the control logic of the DEC unit may be updated to add said certain amount of offset to measurements from the corresponding sensor before use in the control, thereby compensating for the detected fault. As another example, step 908 may include running the DEC unit through a diagnostic or self-repair routine (e.g., flushing the evaporative media 210 with extra water to clear salt buildup that may cause degradation). As another example, step 908 may include ordering and performing maintenance tasks for the DEC unit, such as mechanical maintenance, cleaning, sensor replacement, evaporative media replacement, and / or air filter replacement. For example, one or more sensors (e.g., temperature sensor, humidity sensor) may be cleaned, moved (e.g., to a location predicted to provide a more reliable or useful measurement), replaced (e.g., an existing sensor is discarded and a new, replacement sensor is installed), or otherwise maintained or adjusted to resolve the sensor fault. Triggering a fault (e.g., triggering a sensor fault) as in steps 908-910 may thereby cause the fault to be resolved.

[0102] 11A, a flowchart of a process 1100 for creating a model that may be used in controlling one or more DEC units is shown, according to some embodiments. In some embodiments, the controller 230 is programmed to execute the process 1100. The model generated in the process 1100 may be used in various control processes described herein, such as the process 1200 of FIG. 12, described below.

[0103] In step 1102, a fan model relating fan speed and volumetric flow to pressure difference is fitted. Step 1102 can use a fan pressure rise equation where the fan pressure rise is a function of the volumetric flow rate and the fan speed ratio, for example:

number

[0104] Step 1102 may include, for example, fitting coefficients (coefficient 1, coefficient 2, coefficient 3) using a nonlinear least squares approach based on actual fan data (e.g., measurements). Speed ​​refers to fan speed (e.g., revolutions per minute) and flow refers to volumetric flow rate (e.g., cubic meters per hour). ΔP is the equipment design delta pressure (measured in inches of water column, e.g., from product literature). 流れ ) is also the equipment design value (e.g., from product literature, measured in cubic meters per hour). FIG. 11B shows the volumetric flow rate (i.e., flow) on the horizontal axis and the

number

[0105] In step 1104, the flow coefficients are fitted to a number of bypass damper positions. Step 1104 is performed using the function

number

[0106] Step 1104 may include, for example, fitting a curve (e.g., in increments of 0.1) for each of a set of bypass positions, for example, 10 or 11 bypass positions ranging from bypass=0 to bypass=1, based on actual data (e.g., measurements). An example of such a curve is shown in graph 1160 of FIG. 11C, which is used to determine the value of the flow coefficient Cv at each bypass position evaluated.

[0107] In step 1106, a mapping between bypass damper position and flow coefficient is created. For example, the discrete values ​​of flow coefficient Cv at multiple damper positions from step 1104 may then be used in an interpolation to determine a continuous function that determines the flow coefficient Cv for any bypass value, such as plotted in graph 1170 of FIG. 11D. Step 1106 thereby provides a mapping between bypass damper position and flow coefficient.

[0108] In step 1108, models are created for volumetric flow rate, CPU heat, and supply air temperature. The models used in step 1108 are, for example, p is the heat transfer coefficient, function Q CPU =Flow*C p (T 高温 -T 供給 ) based on the required volumetric flow (flow) of CPU heat (Q CPU ) and supply air temperature (T供給 ) The relationship in step 1108 can be a regression model based on site data that maps to T 高温 (Hot aisle temperature) constraints, limits, goals, etc., e.g., Q CPU To adjust for different levels of flow and / or T 供給 By increasing or decreasing 高温 The objective of the present invention may be to keep the temperature substantially constant.

[0109] In step 1110, a model is created for supply air temperature and water consumption. The model created in step 1110 may be, for example, a water mass balance based on the design efficiency value of the DEC unit. The model created in step 1110 may be similar to the water evaporation modeling discussed above, for example, based on the following set of rules:

number

number

[0110] 12, a process 1200 for optimizing control of a DEC unit is shown, according to some embodiments. In some embodiments, the controller 230 is programmed to execute the process 1200. The process 1200, in some embodiments, may use the model created in the process 1100.

[0111] In step 1202, a value for the supply air temperature is selected. In the first instance of step 1202, the selected value for the supply air temperature starts the optimization. The selected value may, in some embodiments, be the value used for control in the preceding time step.

[0112] In step 1204, the volumetric flow rate and bypass damper position are determined based on the supply air temperature. The bypass position is determined, for example, by T 供給 =Bypass*T oa + (1-Bypass) * DEC Cooling T, where DEC Cooling T = Dry Bulb T-Efficiency * (Dry Bulb T-Wet Bulb T), where efficiency is a given value (e.g., 92%) from the DEC unit product documentation. Such a formula may be used with the selected supply air temperature from step 1202 and the outdoor air temperature (e.g., from a weather service, from a sensor) to determine the bypass position (i.e., value for bypass).

[0113] The volumetric flow rate may be determined in step 1204 using a model relating volumetric flow rate, CPU heat, and supply air temperature, such as the model output from and described with reference to step 1108 of process 1100. The volumetric flow rate may be based, for example, on expected CPU heat (e.g., load forecast). Step 1204 thereby outputs the volumetric flow rate and the bypass damper position.

[0114] In step 1206, a flow coefficient Cv is selected based on the bypass damper position. Step 1206 may be performed using a function output from step 1106 of process 1100 (e.g., the function illustrated in graph 1170 of FIG. 11D) to determine the value of Cv.

[0115] In step 1208, a pressure differential is determined based on the flow coefficient (Cv) from step 1206 and the volumetric flow rate (Flow) from step 1204. The pressure differential is determined by the function discussed with reference to step 1104 of process 1100.

number

number

[0116] In step 1210, the fan speed is determined based on the pressure differential ΔP from step 1208 and the volumetric flow rate from step 1204. Step 1210 uses the fan model from step 1102, e.g.

number

[0117] In step 1214, a water mass balance is used to calculate the amount of water evaporated and / or drained based in part on the supply air temperature. Step 1214 may consider all water consumed by the DEC over the time horizon. Step 1214 may use, for example, the model created in step 1110 of process 1100.

[0118] In step 1216, the value of an objective function that takes into account both energy and water consumption is calculated based on the outputs of steps 1212 and 1214. The objective function may have the following form:

number

[0119] Step 1216 provides a value of the cost function, assuming the supply air temperature selected in step 1202. In step 1218, the supply air temperature is adjusted with the goal of reducing (or otherwise improving depending on the goal / preference) the value of the objective function, and the process returns to step 1202, with the adjusted value being used as the supply air temperature input to initialize steps 1204-1216 of process 1200. Steps 1202-1218 can be repeated such that values ​​of the objective function are calculated for different supply air temperatures, e.g., using gradient descent or other optimization methods, to select the supply air temperature in 1218 that is predicted to drive the objective function to an optimal value. Thus, after iterating through steps 1202-1218, a supply air temperature is determined that improves (e.g., minimizes, or, if formulated differently, maximizes or approaches the goal) the value of the objective function.

[0120] In step 1220, the DEC unit is controlled according to the supply air temperature that best improves (e.g., minimizes) the value of the objective function. Step 1220 may include, for example, controlling the supply fan 214 and actuator 216. The process 1200 may be repeated periodically, for example, every minute, every 15 minutes, etc., to provide optimal control of the DEC unit over time. The DEC unit is thereby controlled in a manner that considers the trade-off between consuming more water to reduce the power consumption of the fan and consuming more power at the fan to reduce water consumption. The benefits of such an approach are demonstrated in the experimental results shown in Figures 13-16 and discussed below.

[0121] In some embodiments, process 1200 is performed on a group of DEC units. The supply air temperature output from the optimization may be used as an overall control target for the group of DEC units. Controlling the group of DEC units may further include allocating load across the group of DEC units by distributing water consumption and fan power consumption among the individual DEC units, for example, by running a second optimization constrained to provide the same overall supply air temperature output from process 1200. For example, it may be more efficient to provide a larger amount of evaporative cooling with one DEC unit while the other DEC units bypass evaporative cooling entirely, compared to providing an equal amount of evaporative cooling from all DEC units, while both scenarios achieve the same overall supply air temperature. Thus, controlling the group of DEC units may include running process 1200 to determine a target overall supply air temperature, and then using that target overall supply air temperature as a constraint for the second optimization to determine an allocation of load across the group of DEC units that optimally achieves the target overall supply air temperature.

[0122] 13, a graph 1300 is shown in accordance with some embodiments and experimental results. The graph 1300 compares supply air temperature values ​​determined by optimizing only fan power usage, optimizing only water usage, or optimizing fan power and water consumption together over a period of time (as in process 1200). As illustrated, optimizing the fan only provides a lower supply air temperature because optimizing the fan relies on increased use of direct evaporative cooling (and therefore more water consumption) to reduce the fan speed required to provide sufficient cooling. Alternatively, optimizing water usage only provides a higher supply air temperature, thereby requiring more fan speed to provide higher flow rates while reducing the use of evaporative cooling (and corresponding water consumption). The graph 1300 further shows that optimizing the fan and water consumption together provides a supply air temperature between those provided by the other two approaches, representing an intermediate approach that provides a balance between water consumption savings and fan power savings.

[0123] 14-16, a set of graphs according to some embodiments and experimental results are shown, which further illustrate the advantages of integrated optimization of water consumption and fan power, as in process 1200. FIG. 14 shows a set of graphs 1400 illustrating fan and water costs in a scenario where only fan power is optimized (i.e., water consumption is not considered in the control). FIG. 15 shows a set of graphs 1500 illustrating fan and water costs in a scenario where only hydraulic power is optimized (i.e., fan power is not considered in the control). Graph 1500 shows that, relative to the fan-focused approach of graph 1400, water costs are significantly reduced (e.g., 87% in one period), but fan power consumption increases (which may result in an increase in overall costs). FIG. 16 shows a set of graphs 1600 illustrating fan and water costs in a scenario where both water cost and fan power cost are considered in the control optimization (e.g., by process 1200). Graph 1600 illustrates that overall costs are reduced relative to the other examples (e.g., by more than 20% during the evaporation period) and water consumption remains significantly reduced relative to the example of Figure 14 (e.g., by 47% during the evaporation period). The results of Figures 14-16 thereby attest to the effectiveness of the various features described herein.

[0124] In various embodiments, various other control strategies may be implemented. For example, in some embodiments, the optimization (e.g., as in process 1200) is formulated as a multi-objective optimization that adjusts water and electricity costs to construct a Pareto front. An operating point may then be selected based on the Pareto front, for example, by examining how much water savings are achieved for various levels of electricity savings. As another example, for a data center served by multiple DEC units as in FIG. 1, fan energy may be saved by opening all dampers wide open to dry half of the DEC units any time less than half the cooling is needed. If more than half the cooling is required to be optimal, additionally, comparing whether turning off half the units operates at a lower cost and keeps temperatures within acceptable limits may become the new optimum. As another example, a high-level / low-level optimization architecture may be provided in which loads can be allocated across multiple DEC units, e.g., to determine the optimal way to provide functionality for any combination of flows and temperatures from N DEC units in a high-level optimization, and then during run-time, low-level optimization can be used to determine the optimal operating points for each DEC unit with the decisions / allocations made in the high-level optimization.

[0125] The hardware and data processing components used to implement the various processes, operations, example logic, logic blocks, modules and circuits described in connection with the embodiments disclosed herein may be implemented or performed using general purpose single or multi-chip processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor may be a microprocessor, or any conventional processor, controller, microcontroller, or state machine. A processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors in cooperation with a DSP core, or any other such configuration. In some embodiments, certain processes and methods may be performed by circuitry specific to a given function. The memory (e.g., memory, memory unit, storage device) may include one or more devices (e.g., RAM, ROM, flash memory, hard disk storage, etc.) for storing data and / or computer code for completing or facilitating the various processes, layers, and modules described in this disclosure. The memory may be or include volatile or non-volatile memory, and may include database components, object code components, script components, or any other type of information structure for supporting the various activities and information structures described in this disclosure. According to an exemplary embodiment, the memory is communicatively coupled to the processor via a processing circuit and includes computer code for executing (e.g., by the processing circuit or processor) one or more processes described herein.

[0126] The present disclosure contemplates methods, systems, and program products on any machine-readable medium for accomplishing various operations. The embodiments of the present disclosure may be implemented using existing computer processors, or by dedicated computer processors for suitable systems incorporated for this or another purpose, or by hardwired systems. An embodiment within the scope of the present disclosure includes a program product including a machine-readable medium for carrying or having stored thereon machine-executable instructions or data structures. Such machine-readable media may be any available medium accessible by a general-purpose or special-purpose computer or other machine with a processor. By way of example, such machine-readable media may include RAM, ROM, EPROM, EEPROM, or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of machine-executable instructions or data structures and that can be accessed by a general-purpose or special-purpose computer or other machine with a processor. Combinations of the above are also included within the scope of machine-readable media. Machine-executable instructions include, for example, instructions and data that cause a general-purpose computer, a special-purpose computer, or a special-purpose processing machine to perform a certain function or group of functions.

[0127] Although the figures and description may show a particular order of method steps, unless otherwise specified above, the order of such steps may differ from that shown and described. Also, unless otherwise specified above, two or more steps may be performed simultaneously or partially simultaneously. Such variations may depend, for example, on the software and hardware systems selected, as well as designer choice. All such variations are within the scope of this disclosure. Similarly, software implementations of the described methods may be realized using standard programming techniques with rule-based logic and other logic to accomplish the various connection, processing, comparison, and decision steps.

Claims

1. 1. A method of controlling a direct evaporative cooling unit, comprising: forecasting water consumption and energy consumption of the direct evaporative cooling unit based on control variables of the direct evaporative cooling unit; selecting targets for the control variables based on optimizing an objective function, the objective function including the water consumption and the energy consumption; and controlling the direct evaporative cooling unit in accordance with the target for the control variable.

2. The method of claim 1 , wherein the water consumption includes at least one of water evaporated during operation of the direct evaporative cooling unit or water drained from a water tank of the direct evaporative cooling unit.

3. 2. The method of claim 1, wherein forecasting the water consumption is based on a predicted change in the amount of airflow over the evaporative medium of the direct evaporative cooling unit and the humidity of the airflow across the evaporative medium, and the controlled variable is supply air temperature, supply air humidity, or bypass damper position.

4. The method of claim 1 , wherein the energy consumption comprises the energy consumption of a fan of the direct evaporative cooling unit.

5. The method of claim 1 , wherein the control variable of the direct evaporative cooling unit is related to a bypass damper position as a function of an outside air temperature, and the forecasting of energy consumption is based on the bypass damper position.

6. 2. The method of claim 1 , wherein the control variable is a supply air temperature, the target is a target supply air temperature, and controlling the direct evaporative cooling unit includes controlling a bypass damper to a position determined based on the target supply air temperature and an outdoor air temperature.

7. 2. The method of claim 1, wherein predicting the water consumption and the energy consumption of the direct evaporative cooling unit based on the control variables of the direct evaporative cooling unit includes utilizing a plurality of curves representing a relationship between pressure difference and flow rate for a plurality of bypass damper positions.

8. 1. A method of controlling a fleet of direct evaporative cooling units serving a data center, comprising: selecting a target supply air temperature for the group of direct evaporative cooling units by performing a first optimization of an objective based on projected water and energy consumption of the group of direct evaporative cooling units; Distributing the predicted water and energy consumption among the direct evaporative cooling units by performing a second optimization constrained by the target supply air temperature; and and controlling the direct evaporative cooling units according to a result of the distributing.

9. 9. The method of claim 8, wherein selecting the target supply air temperature by performing the first optimization comprises projecting the projected water and energy consumption for a plurality of possible supply air temperatures.

10. The method of claim 8 , wherein the water consumption comprises an amount of water evaporated during operation of the group of direct evaporative cooling units.

11. 10. The method of claim 8, comprising forecasting a predicted water consumption based on a predicted change in humidity of the airflow over the evaporative medium of the group of direct evaporative cooling units and the airflow over the evaporative medium.

12. 10. The method of claim 8, comprising predicting projected energy consumption by determining a fan speed based on a pressure differential and a volumetric flow rate associated with the target supply air temperature.

13. 9. The method of claim 8, wherein allocating the predicted water and energy consumption among the group of direct evaporative cooling units includes determining different control actions for different units of the group of direct evaporative cooling units, and controlling the group of direct evaporative cooling units according to the results of the allocating includes operating the different units according to different control decisions.

14. 9. The method of claim 8, comprising forecasting the water and energy consumption of the direct evaporative cooling units based on possible supply air temperatures by utilizing a plurality of curves representing the relationship between pressure differential and flow rate for a plurality of bypass damper positions.

15. 1. A direct evaporative cooling system comprising: forecasting water and energy consumption of the direct evaporative cooling system based on a selected supply air temperature; determining a target supply air temperature by adjusting the selected supply air temperature such that a value of an objective function is improved, the objective function including the predicted water and energy consumption of the direct evaporative cooling system; and controlling the direct evaporative cooling system according to the target supply air temperature.

16. 16. The direct evaporative cooling system of claim 15, wherein the water consumption comprises an amount of water evaporated during operation of the direct evaporative cooling system.

17. 16. The direct evaporative cooling system of claim 15, wherein the controller is programmed to predict the water consumption based on a volume of airflow over an evaporative medium of a direct evaporative cooling unit and a predicted change in humidity of the airflow across the evaporative medium.

18. 16. The direct evaporative cooling system of claim 15, wherein the controller is programmed to predict the water consumption and the energy consumption using a plurality of curves representing the relationship between pressure differential and flow rate for a plurality of bypass damper positions.

19. the direct evaporative cooling unit comprising a bypass damper; The direct evaporative cooling system of claim 15 , wherein the controller is configured to control the direct evaporative cooling system according to the target supply air temperature by causing a change in a position of the bypass damper.

20. 16. The direct evaporative cooling system of claim 15, wherein the controller is programmed to predict the energy consumption by determining a fan speed based on a pressure differential and a volumetric flow rate associated with the selected supply air temperature.

21. 1. A method for fault detection related to direct evaporative cooling of a data center, the direct evaporative cooling affecting humidity and temperature of the data center, the method comprising: using a model to generate a prediction for the humidity of the data center and a prediction for the temperature of the data center; the measurement related to the humidity deviating from the predicted value for the humidity while the measurement related to the temperature tracks the predicted value for the temperature; or triggering a sensor failure in response to the measurement value for the temperature deviating from the predicted value for the temperature while the measurement value for the humidity tracks the predicted value for the humidity.

22. The method of claim 21 , comprising, in response to triggering the sensor fault, performing an action to resolve the sensor fault.

23. 23. The method of claim 22, wherein the performing an action to resolve the sensor fault comprises moving a sensor, cleaning the sensor, or replacing the sensor.

24. 22. The method of claim 21, comprising altering control of the direct evaporative cooling of the data center to accommodate a fault condition.

25. The method of claim 21 , wherein the model uses inputs including a bypass profile and at least one of an outdoor air temperature or outdoor air humidity value.

26. The method of claim 21 , wherein the model is further based on a design efficiency of a direct evaporative cooling unit.

27. 27. The method of claim 26, further comprising determining the design inefficiency degradation based on a comparison of the predicted values ​​and actual values.

28. 1. A method for sensor fault detection for a first sensor measuring a first condition and a second sensor measuring a second condition, wherein an instrument is operable to act on the first condition and the second condition, the method comprising: generating a predicted value for the first state and a corresponding predicted value for the second state using a physics-based model that defines a physics-based relationship between the first state and the second state; triggering a sensor fault in response to determining that a measurement for the first condition deviates from the predicted value for the first condition and a measurement for the second condition tracks the predicted value for the second condition; ceasing to trigger the sensor fault in response to determining that a measurement value for the first condition deviates from the predicted value for the first condition and that a measurement value for the second condition deviates from the predicted value for the second condition.

29. 30. The method of claim 28, comprising, in response to indicating the sensor fault, performing an action to resolve the sensor fault.

30. 30. The method of claim 29, wherein performing the action to resolve the sensor fault comprises moving the first sensor or the second sensor, cleaning the first sensor or the second sensor, or replacing the first sensor or the second sensor.

31. 30. The method of claim 28, comprising altering control of direct evaporative cooling of the data center to accommodate the fault condition.

32. 30. The method of claim 28, wherein the physics-based model uses as inputs a bypass position of a direct evaporative cooling unit, an outdoor air temperature, or an outdoor air humidity.

33. 30. The method of claim 28, wherein the physics-based model is based on the efficiency of a direct evaporative cooling unit.

34. 30. The method of claim 28, wherein the first sensor measures a supply air humidity of a direct evaporative cooling unit and the second sensor measures a supply air temperature of a direct evaporative cooling unit.

35. 30. The method of claim 28, further comprising triggering a control and / or equipment fault when ceasing to trigger the sensor fault.

36. 1. A method for sensor fault detection for a first sensor measuring a first indoor air condition and a second sensor measuring a second indoor air condition, the method comprising: determining a predicted relationship between the first indoor air condition and the second indoor air condition based on thermodynamic principles; evaluating a first measurement of the first indoor air condition from the first sensor and a second measurement of the second indoor air condition from the second sensor to determine whether the first and second measurements satisfy the predicted relationship between the first indoor air condition and the second indoor air condition; and in response to determining that the first measurement from the first sensor and the second measurement from the second sensor do not satisfy the predicted relationship between the first indoor air condition and the second indoor air condition, triggering a fault in the first sensor or the second sensor.

37. 37. The method of claim 36, wherein the first sensor and the second sensor are associated with a first unit of a plurality of equipment units, the method further comprising attributing the fault to the first sensor or the second sensor by performing a peer analysis on the plurality of equipment units.

38. 37. The method of claim 36, wherein performing the peer analysis includes determining whether the measurement from the first sensor or the measurement from the second sensor is an outlier.

39. 37. The method of claim 36, wherein performing the peer analysis comprises using generalized extreme studentized deviance.

40. 37. The method of claim 36, further comprising, in response to the failure, performing one or more of cleaning, moving, or replacing the first sensor or the second sensor.

Images