Effect of Cooling Systems on the Energy Efficiency of Data Centers: Machine Learning Optimisation

This paper proposes a control method to improve the energy efficiency and performance of cooling fans used for cooling. In Industry 4.0, a large number of digital data are used, and a large number of data centers are created to handle these data. These data centers consist of information technology (IT) equipment, power systems, and cooling systems. The cooling system is essential to prevent failure and malfunction of the IT equipment, which consumes a considerable amount of energy. This paper proposes a method to reduce the energy used in such cooling systems and to improve the temperature control performance. This paper proposes an fuzzy proportional integral(FPI) controller that controls the input value of the proportional integral(PI) controller by the fuzzy controller according to the operation state, a VFPI (Variable Fuzzy Proportional Integral) controller that adjusts the gain value of the fuzzy controller, and a variable fuzzy proportion integration-variable limit (VFPI-VL) controller that adjusts the limit value of the fuzzy controller’s output value. These controllers control the fan applied to the cooling system and compare the energy consumed and temperature control performance. When the PI controller consumes 100% of the power consumed, the FPI is 50.5%, the VFPI controller is 44.3%, and the VFPI-VL is 32.6%. The power consumption is greatly reduced. In addition, the VFPI-VL controller is the lowest in temperature variation, which improves the energy efficiency and performance of the cooling system using a fan. The methods presented in this paper can not only be applied to fans for cooling, but also to variable speed systems for various purposes and improvement of performance and efficiency can be expected.

Download Full-text

Analyzing the Cooling Efficiency of the Thermal Environment in Data Centers

International Journal of Grid and High Performance Computing ◽

10.4018/ijghpc.2014100102 ◽

2014 ◽

Vol 6 (4) ◽

pp. 15-27

Author(s):

Yongmei Xu ◽

Jingru Zhang ◽

Yuhui Deng ◽

Lan Du ◽

Rong Jiao

Keyword(s):

Energy Efficiency ◽

Air Temperature ◽

Thermal Load ◽

Air Conditioning ◽

High Speed ◽

Data Centers ◽

Thermal Environment ◽

External Environment ◽

Cooling Systems ◽

Standard Data

Given the explosive growth of data, scalability and fault tolerance have become a fundamental challenge for data center network structures. Temperature in data centers significantly affects the failure ratio of high-speed network devices. Various types of air distribution schemes influence the temperature of network equipment differently, and the cooling cost in data centers dominates the overall energy cost. On the basis of the energy efficiency of cooling systems, this study analyzes and compares the thermal load distribution in the enclosure of standard and non-standard data centers by considering the effects of the external environment. Analysis results demonstrate that the external environment significantly affects the thermal load of non-standard data centers. By leveraging on the air temperature outside data centers and on the inlet/outlet of IT equipment, the air temperature and return air temperature of air conditioning are calculated when performing hot and cold aisle containment. The calculations indicate that sealing an appropriate aisle (hot or cold aisle) can significantly reduce the energy consumption of cooling systems in terms of the external air temperature outside data centers. Furthermore, if the air temperature outside data centers is higher than the temperature at the inlet of IT equipment, sealing the cold aisle outperforms sealing the hot aisle. By contrast, the aisle to be sealed depends on the energy efficiency ratio of the air conditioning.

Download Full-text

Cooperatively Improving Data Center Energy Efficiency Based on Multi-Agent Deep Reinforcement Learning

Energies ◽

10.3390/en14082071 ◽

2021 ◽

Vol 14 (8) ◽

pp. 2071

Author(s):

Ce Chi ◽

Kaixuan Ji ◽

Penglei Song ◽

Avinab Marahatta ◽

Shikui Zhang ◽

...

Keyword(s):

Energy Efficiency ◽

Reinforcement Learning ◽

Resource Utilization ◽

Data Center ◽

Data Centers ◽

Optimization Method ◽

Comparison Method ◽

Cooling Systems ◽

Model Free ◽

Multi Agent

The problem of high power consumption in data centers is becoming more and more prominent. In order to improve the energy efficiency of data centers, cooperatively optimizing the energy of IT systems and cooling systems has become an effective way. In this paper, a model-free deep reinforcement learning (DRL)-based joint optimization method MAD3C is developed to overcome the high-dimensional state and action space problems of the data center energy optimization. A hybrid AC-DDPG cooperative multi-agent framework is devised for the improvement of the cooperation between the IT and cooling systems for further energy efficiency improvement. In the framework, a scheduling baseline comparison method is presented to enhance the stability of the framework. Meanwhile, an adaptive score is designed for the architecture in consideration of multi-dimensional resources and resource utilization improvement. Experiments show that our proposed approach can effectively reduce energy for data centers through the cooperative optimization while guaranteeing training stability and improving resource utilization.

Download Full-text

Development of an energy evaluation and design tool for dedicated cooling systems of data centers: Sensing data center cooling energy efficiency

Energy and Buildings ◽

10.1016/j.enbuild.2015.03.040 ◽

2015 ◽

Vol 96 ◽

pp. 357-372 ◽

Cited By ~ 7

Author(s):

Jinkyun Cho ◽

Joonyoung Yang ◽

Changkeun Lee ◽

Jinyoung Lee

Keyword(s):

Energy Efficiency ◽

Data Center ◽

Data Centers ◽

Design Tool ◽

Cooling Systems ◽

Sensing Data ◽

Energy Evaluation ◽

Data Center Cooling

Download Full-text

Energy Efficiency in Data Centers: Recommendations for Government/Industry Coordination

10.2172/1218528 ◽

2008 ◽

Author(s):

none,

Keyword(s):

Energy Efficiency ◽

Data Centers

Download Full-text

Statistical and machine learning models for optimizing energy in parallel applications

The International Journal of High Performance Computing Applications ◽

10.1177/1094342019842915 ◽

2019 ◽

Vol 33 (6) ◽

pp. 1079-1097 ◽

Cited By ~ 2

Author(s):

Mark Endrei ◽

Chao Jin ◽

Minh Ngoc Dinh ◽

David Abramson ◽

Heidi Poxon ◽

...

Keyword(s):

Machine Learning ◽

Energy Efficiency ◽

High Performance ◽

Large Scale ◽

Energy Use ◽

Parallel Applications ◽

Learning Models ◽

Trade Off ◽

Time Required ◽

Machine Learning Models

Rising power costs and constraints are driving a growing focus on the energy efficiency of high performance computing systems. The unique characteristics of a particular system and workload and their effect on performance and energy efficiency are typically difficult for application users to assess and to control. Settings for optimum performance and energy efficiency can also diverge, so we need to identify trade-off options that guide a suitable balance between energy use and performance. We present statistical and machine learning models that only require a small number of runs to make accurate Pareto-optimal trade-off predictions using parameters that users can control. We study model training and validation using several parallel kernels and more complex workloads, including Algebraic Multigrid (AMG), Large-scale Atomic Molecular Massively Parallel Simulator, and Livermore Unstructured Lagrangian Explicit Shock Hydrodynamics. We demonstrate that we can train the models using as few as 12 runs, with prediction error of less than 10%. Our AMG results identify trade-off options that provide up to 45% improvement in energy efficiency for around 10% performance loss. We reduce the sample measurement time required for AMG by 90%, from 13 h to 74 min.

Download Full-text