A computer room includes rack mounted computing devices and multiple air handling systems for cooling the computer room. A model based cooling control system is implemented on resources of a provider network that includes the rack mounted computing devices. The model based cooling control system use
A computer room includes rack mounted computing devices and multiple air handling systems for cooling the computer room. A model based cooling control system is implemented on resources of a provider network that includes the rack mounted computing devices. The model based cooling control system uses a model of the computer room to iteratively perform calculations to determine a set of control parameters for cooling the computer room. The set of control parameters are selected to reduce or optimize an amount of energy and/or water used to cool the computer room while satisfying one or more conditions for cooling the computer room such as a maximum allowable temperature of the room and/or a maximum allowable temperature of the rack mounted computing devices in the computer room.
대표청구항▼
1. A data center comprising: rack mounted computing devices in a room of the data center;sensors configured to sense conditions related to the rack mounted computing devices; anda cooling system, comprising: air handling devices; anda cooling control system, implemented on at least one of the rack m
1. A data center comprising: rack mounted computing devices in a room of the data center;sensors configured to sense conditions related to the rack mounted computing devices; anda cooling system, comprising: air handling devices; anda cooling control system, implemented on at least one of the rack mounted computing devices, wherein the cooling control system is configured to, for each present time interval of a plurality of time intervals: receive condition information from the plurality of sensors;iterate through sets of control parameters for controlling the air handling devices, wherein for each iteration, the cooling control system is configured to determine, based on a transient model, predicted room conditions for a given set of control parameters and predicted energy usage or water usage associated with the given set of control parameters;select one of the sets of control parameters for controlling the plurality of air handling devices that results in less energy usage or less water usage than other ones of the sets of control parameters, wherein the predicted room conditions for the selected set of control parameters do not exceed a constraint related to cooling the rack mounted computing devices; andgenerate control outputs for the air handling devices such that the air handling devices operate in accordance with the selected set of control parameters for a subsequent time interval,wherein the cooling control system is further configured to:monitor an amount of time required to complete said receive, said iterate, and said select; andin response to the monitored amount of time exceed a threshold, request that an additional one of the rack mounted computing devices be allocated for implementation of the cooling control system. 2. The data center of claim 1, wherein the cooling control system is configured to release at least one of the rack mounted computing devices allocated to implement the cooling control system in response to the monitored amount of time falling below another threshold. 3. The data center of claim 1, wherein the transient model is a transient reduced order computational fluid dynamics (CFD) model of the room and the air handling devices. 4. The data center of claim 1, wherein the plurality of air handling devices are configured to operate in multiple cooling modes, wherein the multiple cooling modes comprise two or more of: a free-cooling mode, a direct evaporative cooling mode, or a mechanical cooling mode. 5. The data center of claim 4, wherein to iterate through the plurality of sets of control parameters for controlling the plurality of air handling devices, the cooling control system is configured to iterate through the multiple cooling modes; and wherein to select the one of the sets of control parameters for controlling the air handling devices that results in less energy or less water usage usage than other ones of the sets of control parameters, the cooling control system is configured to select a cooling mode for controlling the air handling devices that results in less energy usage or less water usage than other ones of the multiple cooling modes. 6. A method comprising: implementing a cooling control system on at least one of multiple rack mounted computing devices in a data center;receiving, by the cooling control system, condition information from sensors that monitor conditions in the rack mounted computing devices;iterating, by the cooling control system, through sets of control parameters for controlling air handling devices that provide cooling for the rack mounted computing devices, wherein said iterating comprises, for each iteration: determining, based on a transient model, predicted room conditions for a given set of control parameters and predicted energy usage or water usage associated with the given set of control parameters;selecting, by the cooling control system, one of the sets of control parameters for controlling the plurality of air handling devices that results in less energy usage or less water usage than other ones of the sets of control parameters, wherein the predicted room conditions for the selected set of control parameters do not exceed a constraint;generating, by the cooling control system, control outputs for controlling the air handling devices such that the air handling devices operate in accordance with the selected set of control parameters for a subsequent time interval;monitoring an amount of time required to complete said receiving, said iterating, and said selecting; andrequesting, in response to the monitored amount of time exceeding a threshold, an additional rack mounted computing device be allocated for implementation of the cooling control system. 7. The method of claim 6, wherein the transient model is based, at least in part, on a computational fluid dynamics (CFD) model of the room. 8. The method of claim 6 further comprising decreasing a quantity of the rack mounted computing devices implementing the cooling control system in response to the monitored amount of time falling below another threshold. 9. The method of claim 6, wherein the threshold is an amount of time equal to or less than a minute. 10. The method of claim 6, wherein said determining, based on the transient model, predicted room conditions for the given set of control parameters and predicted energy usage or water usage associated with the given set of control parameters comprises: determining, for future time intervals, based on the transient model, predicted room conditions and predicted energy usage or water usage for the given set of control parameters; andwherein said selecting the one of the sets of control parameters for controlling the plurality of air handling devices that results in less energy usage or less water usage than other ones of the sets of control parameters comprises:selecting a set of control parameters that results in an overall energy or water usage over the future time intervals that is less than an overall energy or water usage over the future time intervals for other ones of the sets of control parameters. 11. The method of claim 10, wherein said determining, for the future time intervals the predicted room conditions and the predicted energy or water usage for the given set of control parameters comprises: calculating predicted room conditions and predicted energy or water usage for one of the time intervals on one of the computing devices; andcalculating, concurrently with said calculating for the one of the time intervals, predicted room conditions and predicted energy or water usage for another time interval of the time intervals on another of the computing devices. 12. The method of claim 6, wherein the set of control parameters comprises two cooling modes, wherein the two cooling modes comprise two of: a free-cooling mode, an evaporative cooling mode, or a mechanical cooling mode; and wherein said selecting the one of the set of control parameters for controlling the air handling devices comprises selecting one of the two cooling modes. 13. The method of claim 12, wherein the set of control parameters comprises two or more of: an air handler speed, a rate of water injection to an evaporative cooler used in the evaporative cooling mode, a ratio of recirculated air to fresh air, a flow rate of chilled water used in the mechanical cooling mode, or a temperature of chilled water used in the mechanical cooling mode. 14. A non-transitory computer readable storage medium storing program instructions, wherein the program instructions, when executed, are configured to: implement a cooling controller using a rack mounted computing device of a plurality of rack mounted computing devices in a data center;wherein the cooling controller is configured to: receive condition information from sensors that monitor conditions in the rack mounted computing devices;iterate through sets of control parameters for controlling air handling devices that provide cooling for the rack mounted computing devices, wherein for each iteration predicted room conditions for a given set of control parameters and predicted energy or water usage associated with the given set of control parameters are determined based on a transient model;select one of the sets of control parameters for controlling the air handling devices that results in less energy or water usage than other ones of the sets of control parameters, wherein the predicted room conditions for the selected set of control parameters do not exceed a constraintmonitor a time required to complete said receive, said iterate, and said select; andin response to the monitored time exceeding a threshold, request that an additional rack mounted computing device be allocated for implementation of the cooling controller. 15. The non-transitory computer readable storage medium of claim 14, wherein the transient model is a reduced order computational fluid dynamics (CFD) model of the room and the air handling devices. 16. The non-transitory computer readable storage medium of claim 14, wherein the cooling controller is configured to: in response to the monitored time falling below another threshold, request that a rack mounted computing device allocated for implementation of the cooling controller be released. 17. The non-transitory computer readable storage medium of claim 14, wherein the threshold is less than a minute. 18. The non-transitory computer readable storage medium of claim 14, wherein the sensors comprise virtual sensors configured to generate condition signals indicating conditions in the rack mounted computing devices based on respective relationships between other measurements related to the rack mounted computing devices and the condition signals generated by the virtual sensors. 19. The non-transitory computer readable storage medium of claim 18, wherein the condition information from the sensors comprises respective inlet temperatures at respective ones of the rack mounted computing devices from the virtual sensors. 20. The non-transitory computer readable storage medium of claim 18, wherein the condition information from the sensors comprises a measure of recirculation in one of the racks in which the rack mounted computing devices are mounted, wherein the measure of recirculation is generated by one of the virtual sensors.
연구과제 타임라인
LOADING...
LOADING...
LOADING...
LOADING...
LOADING...
이 특허에 인용된 특허 (6)
Brunell, Brent Jerome; Mathews, Jr., Harry Kirk; Kumar, Aditya, Adaptive model-based control systems and methods for controlling a gas turbine.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.