Optimizing chiller energy efficiency in a multiple data center set-up
A method includes: integrating, by the computing device, a combination of user identified parameters in a template-based framework with historical operations data of a plurality of data centers in a multiple data center set-up to produce a feature set of parameters; clustering, by the computing device, the exclusive feature set of parameters into an optimal number of groups, each of which comprise similar physical attributes of chillers associated with any of the plurality of data centers; generating, by the computing device, analytical models for data center infrastructure component energy efficiency optimization of the multiple data center set-up based on the optimal number of groups; and providing recommendations from the analytical models as to which parameters are to be adjusted to have a more efficient energy utilization of the data center infrastructure components.
1. A method, comprising:
integrating, by a computing device, a combination of user identified parameters in a template-based framework with historical operations data of a plurality of data centers in a multiple data center set-up to produce a feature set of parameters;
clustering, by the computing device, the feature set of parameters into an optimal number of groups, each of which comprise similar physical attributes of data center infrastructure components associated with any of the plurality of data centers;
generating, by the computing device, analytical models for data center infrastructure component energy efficiency optimization of the multiple data center set-up based on the optimal number of groups; and
providing recommendations from the analytical models as to which parameters are to be adjusted to have a more efficient energy utilization of the data center infrastructure components.
2. The method of claim 1 , wherein the data center infrastructure component comprises one of a chiller, a water pump, a computer room air handler, and a cooling tower.
3. The method of claim 1 , further comprising dynamically generating feature variables by incorporating the user identified parameters and data center design parameters from the template-based framework.
4. The method of claim 3 , further comprising identifying a precise combination of parameters of the user identified parameters with a corresponding relevant weightage.
5. The method of claim 4 , wherein the precise combination of parameters with its corresponding relevant weightage are integrated with the historical operations data to generate the feature set of parameters to cluster into the optimal number of groups.
6. The method of claim 1 , further comprising:
determining that a class coherence score of the clustered optimal number of groups is not within an acceptable limit;
notifying the user that the class coherence score is not within the acceptable limit; and
suggesting corrective actions to be taken such that the class coherence score will be placed within the acceptable limit.
7. The method of claim 1 , further comprising determining that a class coherence score of the clustered optimal number of groups is within an acceptable limit and training a machine learning based clustering algorithm to minimize loss function over a prediction range.
8. The method of claim 1 , further comprising ranking each group of the optimal number of groups and generating a data center infrastructure component attribute score using a weighted function of the user identified parameters.
9. The method of claim 8 , wherein the weighted function using an additional hyper-parameter regulates influence of the user inputs on the data center infrastructure component energy efficiency modeling process.
10. The method of claim 9 , further comprising combining the data center infrastructure component attribute score with the historical operations data to dynamically generate feature variables using transformation functions that are then used to generate the optimal number of groups.
11. The method of claim 1 , wherein the user identified parameters are provided within a configurable template.
12. The method of claim 1 , further comprising classifying a new data center infrastructure component into one of the optimal number of groups that has similar physical attributes to the new data center infrastructure component.
13. A computer program product comprising one or more computer readable storage media having program instructions collectively stored on the one or more computer readable storage media, the program instructions executable to:
obtain data center infrastructure components historical operations data of data centers in a multiple data center set-up;
compute a data center infrastructure component attribute score obtained of from a list of attributes in a template framework and which is associated with data centers in a multiple data center set-up; and
create data center infrastructure component data buckets using a clustering algorithm, wherein the data center infrastructure component data buckets are an optimal number of groupings of data center infrastructure components each of which include similar attributes as defined in templates of the template framework.
14. The computer program product of claim 13 , further comprising assigning priorities to attributes in the list of attributes, which control impact of template inputs and the data center infrastructure component data buckets.
15. The computer program product of claim 13 , further comprising determining a class coherence score of each of the data center infrastructure component data buckets.
16. The computer program product of claim 15 , wherein the class coherence score signifies whether the data center infrastructure component data buckets are valid data center infrastructure component data buckets.
17. The computer program product of claim 13 , further comprising using seasonality and ambient temperature conditions for each data center to compute the data center infrastructure component data generalization data.
18. The computer program product of claim 13 , further comprising classifying a new data center infrastructure component into an existing one of the data center infrastructure component data buckets based on a minimum distance from a data center infrastructure component data bucket centroid.
19. The computer program product of claim 13 , further comprising weighting one or more attributes in the list of configurable attributes.
20. A system comprising:
a processor, a computer readable memory, one or more computer readable storage media, and program instructions collectively stored on the one or more computer readable storage media, the program instructions executable to:
integrate a combination of user identified parameters within a template-based framework with historical operations data of a plurality of data centers in a multiple data center set-up to produce a feature set of parameters;
cluster the feature set of parameters into an optimal number of groups, each of which comprise similar physical attributes of data center infrastructure components associated with any of the plurality of data centers; and
generate analytical models for data center infrastructure component energy efficiency optimization of the multiple data center set-up based on the optimal number of groups.