Abstract:The proper operation of the UPS system in data centers is crucial for their management and operation. PUE, pPUE, CLF/PLF, and RER are universal metrics for evaluating data center energy efficiency. The design of the UPS system and the choice of operation modes play a significant role in these efficiency metrics. The quality of local utility power supply greatly impacts the UPS system's operation, and flexible UPS operation modes can greatly optimize data center energy consumption.
Keywords:UPS; Energy Efficiency Metrics; Equipment Selection; Power Quality; Energy Consumption Optimization
I. Introduction
National People's Congress delegate and President of the China Academy of Information and Communications Technology, Liu Du, said during an interview with People's Daily Finance that "new infrastructure" is the foundation for the innovative development of the digital economy, a key force for cultivating new momentum, and a driving force for the transformation of development models. "New infrastructure" will inject new momentum into economic development. From the system diagram of new infrastructure, we can see that the construction of data centers occupies a significant position; with the rapid economic development of our country, the construction of data centers is also experiencing explosive growth, and the increase in energy consumption has attracted more and more attention. According to statistics from relevant professional agencies, the annual energy consumption of data centers in our country has already exceeded 1.5% of the total social energy consumption. Professional data center technicians need to attach great importance to the increase in data center energy consumption, fully understand the energy efficiency indicators of data centers, and provide optimized and reasonable energy conservation and emission reduction plans for different types of data centers. The UPS system, as the power supply and guarantee system of the critical equipment in data centers, plays a crucial role in data center energy consumption through its system construction and operation methods.
Figure 1: New Infrastructure System Diagram
Data Center Energy Efficiency Indicators Overview
Cost reduction and efficiency improvement are crucial topics in economic development. The ever-increasing energy consumption of data centers is also a significant issue for new infrastructure. The construction and operation of UPS systems in data centers play a vital role in reducing energy consumption. Before discussing the construction and operation methods of UPS systems, understanding the evaluation system of energy efficiency indicators for data centers helps us build a superior operational system. For different types of data centers, the difference lies in their applications, but they all share basic consistent energy efficiency indicators. Currently, the following four types of energy efficiency indicators are commonly used in data centers:
- PUE, or Power Usage Effectiveness, is a comprehensive indicator widely used internationally to measure the energy efficiency of data center infrastructure. The calculation formula is: PUE = PTotal / PIT. PTotal refers to the total energy consumption of all types of equipment within the data center, while PIT is the total energy consumption of IT equipment. The energy consumption of UPS systems accounts for a significant portion of the infrastructure, and it is clear that optimizing the operation of UPS systems can enhance the data center's PUE.
- pPUE: pPUE is an advanced discussion within the PUE concept, focusing on the evaluation and analysis of the energy efficiency of specific areas or equipment within data centers. When using the pPUE metric for assessing data center energy efficiency, a crucial step is to properly partition the data center.
- CLF/PLF: The CLF/PLF stands for the ratio of Cooling/Power Load to IT Equipment Energy Consumption. From a management perspective, these ratios are more accurate in reflecting data center energy efficiency than PUE. In this evaluation system, the size of PLF is closely related to the UPS system.
- RER: The utilization of renewable energy is an important issue for data centers in the future. RER is an index used to measure the use of renewable energy in data centers. An increase in the RER index can reduce carbon emissions from data centers. The main renewable energies currently include solar, wind, hydro, biomass, geothermal, and ocean energy, among others.
Section 3: Construction of Data Center UPS System and Optimization of System Operation Energy Consumption
Data centers, as one of the key directions of the new infrastructure, often face challenges along with opportunities. The new infrastructure also places higher demands on the development of data centers, such as energy consumption. The energy-saving and optimized operation of the UPS system is a critical aspect we need to focus on. Constructing and optimizing the operation of the data center UPS system requires thorough consideration of factors such as UPS system selection, power supply requirements of IT equipment, UPS operation modes, and ensuring power quality. Below is a brief analysis:
- Rational UPS System Design
The proper construction of UPS systems is crucial for data centers. Currently, the market offers both traditional high-frequency and industrial-frequency UPS units, as well as modular UPS units. The power distribution technology of high-voltage DC supply and distribution systems replacing conventional UPS supply and distribution systems is gradually being accepted by some data centers, but high-frequency or industrial-frequency UPS supply and distribution systems remain the mainstream. The reasonable construction of the system has a significant impact on the energy consumption system of data centers; the following is a brief analysis:
- Industrial Frequency UPS Power Supply Solution
More than 50% of data centers currently use frequency conversion UPS for power supply, primarily due to the mature and reliable可控硅 phase control rectification technology it employs, along with the "1+1" dual bus power supply scheme in power distribution. Its advantages include: resolving single-point failures, good fault tolerance, high reliability, and simple control. However, for data centers where costs are increasingly high, factors such as large installation area requirements, high load-bearing capacity for installation sites, low power factor, and high operating costs also制约 its further development.
- High-Frequency Machine UPS Power Supply Solution
Historically, the concern with high-frequency UPS systems was the technical bottleneck of high-power IGBT modules. With the advancement in the manufacturing process of high-power IGBT modules, high-frequency UPS systems using IGBT rectification have gradually been accepted by a growing number of data center technicians. The technical change in high-frequency machines that data center technicians recognize is the elimination of the need for an output isolation transformer, as their power distribution systems adopt a "1+1" dual-bus power supply scheme like that of conventional frequency UPS systems. Compared to conventional frequency UPS systems, these advantages, such as smaller footprint, lower load-bearing requirements, single-point fault resolution, high fault tolerance, high power factor, and lower operating costs, are significant. Of course, high-frequency UPS power supply solutions still have many drawbacks, such as more harmonic distortion and higher "zero ground" voltage at the load end. Due to some imported server manufacturers' excessive requirements for "zero ground" voltage less than 1V, non-specialist technicians often tend to choose conventional frequency machines when building UPS systems for data centers with a high number of computing nodes. However, from the construction choices of over 80% of newly built data centers, the adoption of high-frequency machine systems is an inevitable trend for the future power supply systems of data centers.
- Constructing a Reasonable Capacity Selection for UPS Systems
Building a reasonably sized UPS system is crucial for data centers. The construction of a UPS system involves numerous factors, with efficiency and cost being two key considerations. From a technical standpoint, we focus on efficiency. To grasp the construction of a UPS system, two key numbers are essential: 50% and ±15%. Understanding the significance of 50% lies in the fact that the system operates at a higher efficiency when the load is above 50%; below this threshold, efficiency sharply declines. ±15% refers to the efficiency at the point where the user selects the efficiency-output power relationship curve and the DC voltage variation is ±15%. Technical professionals who fully comprehend these percentages can construct a UPS system with optimal capacity. From a cost perspective, higher efficiency can lead to cost savings, and selecting the appropriate capacity can also be more optimized.
(II) UPS System Operation Mode Selection
For a data center, the operation of the UPS system is determined by the needs of IT equipment such as computing, storage, and switching within the center. The total power of the IT equipment determines the capacity of the UPS system, while the power input loop requirements of the IT equipment determine the composition of the UPS system. Below, we discuss the optimization of energy consumption in data center UPS systems from various factors.
1. UPS standalone operation
For small-scale data centers or those with investment limitations, many opt for standalone operation. Energy-saving optimization for the UPS system in such data centers is straightforward; equipment selection is crucial. Choosing high-efficiency, high-frequency machines or modular UPS units can fundamentally improve the data center's PUE. If the equipment is already designed for industrial frequency machines, the operation mode should be selected reasonably based on the quality of the municipal power supply and the importance of the equipment. If the quality of the municipal power supply is reliable, the economic operation mode available in most UPS systems can be chosen.
Figure 2: UPS System Single Machine Energy-Saving Operation Mode
2. UPS dual system parallel operation
Figure 3: UPS Interactive Hot Standby (ATS) Operation
As data center scale expands or the importance of IT equipment increases, the selection and operation of data center UPS systems must transition to parallel operation. Of course, there are various parallel operation methods, such as master-slave hot standby, interactive hot standby, interactive hot standby (ATS), and redundant (N+X) parallel operation. Most data centers opt for interactive hot standby (ATS) or redundant (N+X) parallel operation. For those choosing the interactive hot standby (ATS) system, the most direct way to optimize energy consumption efficiency is to use high-efficiency high-frequency units, which can ensure system reliability while also improving and reducing the energy consumption of the units during individual unit rotation operation.
UPS operates multiple systems in parallel
Figure 4: UPS Redundancy (N+X) Operation
As data center scale continues to expand or as IT equipment demands increased power redundancy, the size of the UPS system construction in data centers is growing significantly. The number of issues to address and constraints to consider is also increasing. For instance, many highly critical data centers now require high power redundancy for IT equipment, gradually moving from dual power to 4, 6, or even 8 power sources. For such IT equipment, a single parallel UPS system is no longer sufficient. We need to design multiple N+X redundancy systems like those shown in Figure 3 to meet their needs. For the widespread use of multi-power equipment in data centers, most centers actually employ an A/B power supply mode, using two or more N+X redundancy systems to ensure IT equipment power supply. This also presents an opportunity to reduce energy consumption in data centers. In areas with reliable municipal power quality, we can alternate between energy-saving modes or common bypass modes for A/B UPS systems, or even directly use high-reliability municipal power. This approach can ensure the reliability of the data center while also reducing its energy consumption.
Ankorri Energy Consumption and Statistical Analysis (Energy Management) Solution
Introduction to Energy Efficiency Management Solutions
Establish an efficient energy consumption monitoring and management system to measure the energy consumption data of various energy-consuming equipment in buildings in real-time, and to statistically analyze the collected data. The system can reasonably determine the economic indicators and performance evaluation indicators for energy consumption in different areas of buildings, identify energy usage patterns and energy waste, and enhance the awareness of personnel in proactive energy conservation.
Establish the basic framework for a smart energy management system for data centers, conducting real-time monitoring of each energy-consuming link.
Carbon Emission Data-ization: The system enables per capita energy consumption analysis within buildings (including water, electricity, and energy), achieving data-driven low-carbon office operations.
③ Regional Energy Efficiency Ratio: Achieves comparison of energy consumption within building units, facilitating energy consumption assessment.
④ Concurrent Energy Efficiency Ratio: Achieves energy consumption comparison across the same year, period, and region, facilitating energy-saving data analysis.
Energy Consumption Evaluation Management: Analyze the per unit area energy consumption and per capita energy consumption indicators based on the standard constraint values, standard values, and guiding values of the energy consumption quota standards.
⑥ Energy Consumption Competition Ranking: Compare energy consumption across different functional areas, achieve energy consumption rankings, and enhance staff awareness of energy conservation.
Our energy consumption data utilizes comprehensive analysis, statistics, printing, and query functions, and allows for the selection of different report formats based on the needs of the energy consumption monitoring and management system, providing a reliable basis for energy operation and management departments.
Energy consumption data collection is available for real-time query, and statistical analysis is conducted based on the collected data. It monitors abnormal energy usage and alarms for faults in intelligent energy meters, enhancing the system's informatization and automation levels.
2. Energy Management System Hardware Configuration
|
Application Scenarios |
Model |
Image |
Enhanced Protection Features |
|
Energy Consumption Management Cloud Platform |
AcrelCloud-5000 |
|
Utilizing technologies such as ubiquitous IoT, cloud computing, big data, mobile communications, and intelligent sensing, the platform offers services including energy data collection, statistical analysis, energy efficiency analysis, energy consumption alerts, and equipment management. The platform is widely applicable across various industries. |
|
Intelligent Gateway |
Anet Series Network Management |
|
Utilizing an embedded hardware computer platform equipped with multiple downstream communication interfaces and one or more upstream network interfaces, it serves as a bridge between the collection terminal and the platform system in an information collection system. It is capable of collecting and summarizing data from devices such as water meters, gas meters, electricity meters, and microcomputer protection terminals according to different collection protocols, and uses the corresponding protocols to transmit data from the field equipment to the platform system. |
|
High-voltage critical circuits or low-voltage incoming line cabinets |
APM810 |
|
Equipped with full electrical measurement, power statistics, power quality analysis, and network communication functions, this series of instruments is mainly used for comprehensive monitoring and diagnosis of power grid supply quality and power management. The instruments feature a modular design, allowing customers to easily add switch input/output, analog input/output, SD card recording, and Ethernet communication by simply inserting the corresponding modules into the back. |
|
APM520 |
|
Three-phase full electrical quantity measurement, 2-63 harmonic orders, unbalance degree, maximum demand, payment support, limit alarm, SOE, 4-20mA output. |
|
|
Low-voltage junction box |
AEM96 |
|
The three-phase multi-functional energy meter integrates the measurement of three-phase power parameters, energy metering, and evaluation management. It provides statistics on energy data for the past 24 hours, 31 days, and the previous 12 months. It features 63-point harmonic detection and total harmonic content measurement, and comes with switch input and relay output to enable "remote signaling" and "remote control" functions. It also has alarm output and can be widely used in various control systems, SCADA systems, and energy management systems. |
|
Power cabinet |
ACR120EL |
|
Measure all common electrical parameters, such as three-phase current, voltage, active and reactive power, electricity consumption, harmonics, etc., and features comprehensive communication networking capabilities, making it highly suitable for real-time power monitoring systems. |
|
DTSD1352 |
|
The DIN35mm rail-mounted installation structure is compact, capable of measuring electrical energy and other electrical parameters, and supports clock and rate period parameter settings. It boasts high precision, excellent reliability, and performance indicators that comply with the national standards GB/T17215-2002 and GB/T17883-1999, as well as the power industry standard DL/T614-2007 for electricity meters. Additionally, it features an energy pulse output function and can exchange data with a supervisory computer via an RS485 communication interface. |
|
|
AEW100 |
|
Three-phase full electric quantity measurement, residual current, 2-63rd harmonics, support charge rate, magnitude, cable temperature, optional 2G/4G communication. |
V. Conclusion
Ensuring the safety, efficiency, and reliability of our data centers is our top priority for all staff members. It's also our duty to guarantee that they operate in an environmentally friendly and low-carbon manner. Proper selection and construction of the UPS system, such as choosing the right capacity, opting for high-efficiency single-machine operation, blending grid power with UPS power, and directly supplying IT equipment with increasingly high-quality grid power, all require data center designers and operators to promote and accumulate more experience.
[Reference]
Ren Yan. New Infrastructure Injects New Momentum into Economic Development[J]. People's Daily Finance Channel, 2020, 5.
[2] Zhou Heliang. Handbook of Electrical Engineers. China Electric Power Press, 2010, 4: 223-230.
Li Changchun. Discussion on the Power Quality of Key Equipment in Data Centers[J]. Information System Engineering, 2017, 12.
Li Changchun, Zhang Qian, Qiao Weilin, Discussion on Energy Consumption Optimization of Data Center UPS Systems
[5] Ankorri Enterprise Microgrid Design & Application Handbook, 2022.5 Edition
