System and methods for simulation-based optimization of data center cooling equipment

1. A method for optimizing cooling power consumption in a data center including a plurality of racks and at least one cooling unit, the method comprising:

receiving information regarding the data center, including a layout of the plurality of racks and the at least one cooling unit within the data center;

determining an initial power consumption value for the data center by implementing an energy model;

predicting at least one temperature value and at least one airflow rate value by implementing a numerical simulation model;

defining constraints associated with operation of the at least one cooling unit, including maximum and minimum airflow rate values for the at least one cooling unit, maximum and minimum supply temperature values for the at least one cooling unit, and a maximum inlet temperature for each rack of the plurality of racks;

determining at least one optimal cooling set point value for the at least one cooling unit using an optimization solver based on the initial power consumption value and the at least one predicted temperature value and the at least one predicted airflow rate value, the optimization solver minimizing a power consumption value of a power consumption function while maintaining the constraints associated with operation of the at least one cooling unit; and

controlling operation of the at least one cooling unit using the at least one optimal cooling set point value.

2. The method of claim 1 , wherein the optimization solver iteratively determines the at least one optimized cooling set point until a stopping condition is met by iteratively varying one or more inputs received by at least one of the energy model and the numerical simulation model.

3. The method of claim 2 , further comprising reducing a number of iterations by averaging at least one input received by at least one of the energy model and the numerical simulation model.

4. The method of claim 2 , further comprising:

dividing a representation of the data center into a computational grid comprising a plurality of grid cells;

determining an inlet temperature value for at least one grid cell of the plurality of grid cells, wherein the inlet temperature value is associated with an inlet temperature of at least one rack of the plurality of racks.

5. The method of claim 4 , further comprising:

determining a lowest value for an inlet temperature value for the plurality of racks in each iteration;

averaging the lowest value associated with each iteration over the number of iterations to determine an average minimum inlet temperature for the plurality of racks;

determining a highest value for an inlet temperature for the plurality of racks in each iteration; and

averaging the highest value associated with each iteration over the number of iterations to determine an average maximum inlet temperature for the plurality of racks,

wherein the constraints include the average minimum inlet temperature for the plurality of racks and the constraint associated with the maximum inlet temperature for each rack is replaced by the average maximum inlet temperature for the plurality of racks.

6. The method of claim 5 , further comprising calculating an airflow velocity value and a temperature value for at least one grid cell, wherein the inlet temperature value is based on the airflow velocity value and the temperature value.

7. The method of claim 6 , wherein the airflow velocity value and the temperature value are calculated using a potential flow method.

8. The method of claim 2 , wherein the received information includes at least one real-time measurement value including at least one of rack power, rack airflow, cooler airflow, supply temperature of a cooler, and an outside temperature value, and the one or more inputs includes the at least one real-time measurement value.

9. The method of claim 2 , further comprising statistically tuning at least one of the energy model and the numerical simulation model.

10. The method of claim 1 , wherein implementing the numerical simulation model includes generating a regression model, the regression model based at least in part on at least one operating parameter of the data center, including a total load of the plurality of racks, an air ratio, and a supply temperature of the at least one cooling unit.

11. The method of claim 10 , wherein the regression model is generated using a least square regression technique.

12. A system for optimizing cooling power consumption in a data center including a plurality of racks and at least one cooling unit, the system comprising:

at least one input configured to receive information regarding the data center, including a layout of the plurality of racks and the at least one cooling unit within the data center;

a programmable device in communication with the at least one input, the programmable device comprising:

a memory for storing the received information;

at least one processor coupled to the memory and configured to:

determine an initial power consumption value for the data center by implementing an energy model;

predict at least one temperature value and at least one airflow rate value by implementing a numerical simulation model;

define constraints associated with operation of the at least one cooling unit, including maximum and minimum airflow rate values for the at least one cooling unit, maximum and minimum supply temperature values for the at least one cooling unit, and a maximum inlet temperature for each rack of the plurality of racks;

determine at least one optimal cooling set point value for the at least one cooling unit using an optimization solver based on the initial power consumption value and the at least one predicted temperature value and the at least one predicted airflow rate value, the optimization solver minimizing a power consumption value of a power consumption function while maintaining the constraints associated with operation of the at least one cooling unit; and

control operation of the at least one cooling unit using the at least one optimal cooling set point value.

13. The system of claim 12 , wherein the at least one processor is configured to use the optimization solver by iteratively determining the at least one optimized cooling set point until a stopping condition is met by iteratively varying one or more inputs used by at least one of the energy model and the numerical simulation model.

14. The system of claim 13 , wherein the at least one processor is configured to:

divide a representation of the data center into a computational grid comprising a plurality of grid cells;

determine an inlet temperature value for at least one grid cell of the plurality of grid cells, wherein the inlet temperature value is associated with an inlet temperature of at least one rack of the plurality of racks.

15. The system of claim 14 , wherein the at least one processor is configured to:

determine a lowest value for an inlet temperature value for the plurality of racks in each iteration;

average the lowest value associated with each iteration over the number of iterations to determine an average minimum inlet temperature for the plurality of racks;

determine a highest value for an inlet temperature for the plurality of racks in each iteration; and

average the highest value associated with each iteration over the number of iterations to determine an average maximum inlet temperature for the plurality of racks,

wherein the constraints include the average minimum inlet temperature for the plurality of racks and the constraint associated with the maximum inlet temperature for each rack is replaced by the average maximum inlet temperature for the plurality of racks.

16. The system of claim 13 , wherein the received information includes at least one real-time measurement value including at least one of rack power, rack airflow, cooler airflow, supply temperature of a cooler, and an outside temperature value, and the one or more inputs includes the at least one real-time measurement value.

17. The system of claim 13 , wherein the at least one processor is configured to statistically tune at least one of the energy model and the numerical simulation model.

18. The system of claim 13 , wherein the at least one processor is configured to implement the numerical simulation model by generating a regression model, the regression model based at least in part on at least one operating parameter of the data center, including a total load of the plurality of racks, an air ratio, and a supply temperature of the at least one cooling unit.