Configuration and design of UPS redundant power supply system for IDC computer room

Abstract: One of the important indicators to measure the design and configuration level of an IDC computer room is not only whether its UPS redundant power supply system has high reliability, high anti-interference, high resistance to natural disasters and easy "manageability", but also whether it can obtain 100% high "availability" for IDC equipment and create an excellent power supply operating environment.

Keywords: Internet data center; multimedia data center; uninterruptible power supply redundancy configuration and design

1. Optional UPS redundancy with excellent "fault tolerance" function

Power supply system

As we all know, Internet Data Center (IDC) and Multimedia Data Center (MDC) are the control centers of high-speed Internet. Users have extremely high requirements for the "timeliness" of the remote processing, storage and transfer of information resources (data, voice and image information) they undertake. Even a "downtime" of only a few seconds will bring immeasurable losses to the safe operation of the entire Internet and the production and operation of users. In serious cases, it may even cause serious paralysis of social and economic life. Therefore, IDC must provide users with 365×24h continuous high-speed, safe and reliable information resource value-added services. To achieve this goal, from the design principle, the entire power supply system that is responsible for supplying power to the IDC room must adopt a redundant power supply solution with a high degree of "fault tolerance" to ensure that no matter when the mains power grid fails or when the inverter of a "double conversion, online UPS power supply" fails, or when performing daily maintenance/repair operations or when the fuse burns out/the circuit breaker switch "trips" for some reason, the interconnected devices should be powered by the inverter power of the "online UPS" and should not enter the state of being powered by the ordinary mains power supply/emergency standby generator set through the UPS AC bypass. This is because only the inverter power of the "online UPS" can provide the user's load with high-quality sinusoidal power with stable voltage, no frequency "sudden change", no interference and minimal waveform distortion. For "non-online UPS" including backup UPS and online interactive UPS, they mainly adjust the voltage of the input power supply, and there is no "substantial" improvement on the frequency fluctuation of the input voltage, various power supply interferences and voltage distortion. This means that there should be no single point "bottleneck" failure hidden danger in the entire power supply system. For this reason, a UPS redundant power supply system with a high degree of "fault tolerance" should be configured as much as possible. In other words, during the operation of such a UPS power supply system, even if some "components" accidentally "fail", the entire UPS power supply system must still be able to work normally.

According to the current technological development level of the UPS industry, it is advisable to select a UPS redundant power supply system with "dual bus input" and "dual bus output" power supply functions. It is composed of the following power supply systems.

1.1 "Dual bus input" redundant input power supply system

The dual bus input power supply system consists of mains power and standby generator set. Its basic configuration is: mains input power + standby generator set

Motor group + "auto-switching" control cabinet + input power distribution cabinet. The auto-switching control system constantly monitors the real-time operating status of various input power supplies and ensures that the most reliable power supply is always sent to the input of the UPS. For some important IDC computer rooms, their "emergency generator" power supply is actually a redundant generator power supply system composed of multiple diesel generators + generator parallel control cabinets. In order to ensure that the subsequent "N+1" type UPS redundant power supply system can operate absolutely safely and reliably, it is necessary to attach great importance to the "technical compatibility" and correct setting of "switching parameters" between various devices in the above redundant input power supply system. This is because if the selection and configuration of the equipment are unreasonable or the "switching parameters" are improperly set, it is possible to cause the subsequent UPS power supply system to "fail" or a momentary "power interruption" to the user's load. This can be illustrated by an example. 1) Due to the improper design of the input power supply system of a user's "1+1" type UPS parallel system, everything seemed to be "normal" during the years of operation. However, when the high voltage side of the user's input transformer has an "instant trip" fault for some reason, the battery pack in the UPS also has an "explosion and fire" fault;

2) Due to the design defects of the selected "1+1" parallel system (low price), a user's "1+1" UPS redundant parallel system will experience the following fault phenomenon when the input power supply is interrupted and then suddenly restored to normal power supply, or when someone on the same power grid is performing "mutual switching" between the mains power supply and the backup generator power supply:

——The UPS power supply system was in a fault state of providing "AC bypass power" to the user load for a long time; ——The inverter of one of the UPS entered the "automatic shutdown" state, which lasted for about 3 to 10 seconds;

——The entire "1+1" UPS parallel system experiences a "short shutdown" failure of about 2 to 6 seconds.

It is explained here that for -48V communication power supply, since they have adopted a dual-circuit three-phase AC power supply system and are equipped with a large-capacity -48V battery pack, the input end of this redundant communication power supply is generally directly connected to the output end of the redundant input power supply of the IDC, without connecting them to the "N+1" type UPS redundant parallel power supply system, thereby achieving the purpose of saving investment.

1.2 High reliability "N+1" type UPS redundant power supply system

In order to make the Internet data center have the "all-weather" operation characteristics of 365×24h, the reliability requirements for the UPS products that provide power to it are very stringent. This is because for a UPS product with a reliability of only 99.999%, it may cause the Internet to be down for up to 316s in a year. Even if the reliability of the UPS product is increased to 99.999999%, the downtime that may be caused in a year is still 320ms. For the IDC computer room, if a downtime failure of up to 320ms really occurs, it will cause great losses. This is because the instantaneous power interruption time allowed by most computers is 10 to 18ms. Otherwise, the user's network control operating system or running software will be damaged. Therefore, if the IDC computer room really has the ability to provide 365×24h continuous and uninterrupted operation characteristics, it is definitely not something that can be achieved by the UPS single machine that the current UPS industry can provide. So far, we can only produce UPS products with lower and lower failure rates. However, we cannot produce UPS products with a "zero failure rate". Under current technical conditions, the use of "N+1" UPS redundant parallel power supply system is the best power supply solution to eliminate single-point "bottleneck" failures. It is a power supply system that ensures that the inverter output voltage of each UPS unit is at the same amplitude, frequency and phase (the "circulation" between various UPS units is equal to zero), and puts "N+1" UPS units with the same output power into parallel output state to operate.

In order to make the UPS parallel power supply system have the necessary "fault tolerance" function, the maximum load of the user should not exceed the total output power of N UPS units. When the UPS parallel system works normally, the "N+1" UPS units will evenly share the load current. When a UPS fails, the failed UPS will automatically go offline by performing the "selective tripping" operation. At this time, the remaining N UPS will continue to provide users with high-quality inverter power.

A large number of operational practices have shown that as the number of UPS units in a UPS redundant system increases, it will not only cause the reliability of the entire UPS redundant parallel power supply system to gradually decrease, but also cause the "output power margin" of the entire UPS redundant parallel system to gradually decrease (which means: the UPS parallel system's ability to resist output overload is also gradually decreasing). Therefore, from the perspective of application technology, users should try to choose the most reliable "1+1" or "2+1" UPS redundant power supply system. To illustrate this issue, please refer to Table 1.

Table 1 Reliability of a certain type of UPS multi-machine redundant direct parallel power supply system Parallel scheme 1 + 12 + 13 + 14 + 15 + 16 + 1
System output power "margin" / % 100 50 33 25 20 16
Ratio of MTBF between system and single machine 5.5 4.1 2.9 2.1 1.3 0.98
From Table 1, it can be seen that for the "1 + 1" type parallel system, its MTBF is 5.5 times that of a single machine. It can be seen that the use of redundant parallel power supply scheme can indeed significantly improve the reliability of the entire power supply system. However, excessively increasing the number of single machines will cause a "huge sacrifice" in the reliability of the parallel system, and the reliability of the "6 + 1" type parallel system is even lower than that of a single machine.

Configuration and design of UPS redundant power supply system for IDC computer room (1)

Figure 2 Dual bus output UPS redundancy with load synchronization controllers LBS and STS

Power supply system (UPS power supply products of Liebert)

Figure 1 Economical dual-bus input, dual-bus output UPS redundant power supply system

Table 2 Comparison of the prices of Internet value-added services of “bandwidth” Network rate/bit/s≤64k64～128k128～256k256～384k384～512k512～768k768～1024k1024k～2M
Relative cost of network usage 11.31.82.43.34.56.17
Network rate/bit/s2M～4M4M～8M8M～10M10M～20M20M～34M34M～100M100M～200M
Relative cost of network usage 91619305090150

1.3 "Dual Bus Output" UPS Redundant Output Power Distribution System

In order to eliminate the "single point bottleneck" fault hidden danger that may occur between the output end of the UPS parallel system and the user end, it is necessary to configure the UPS dual bus output power distribution system. Its basic configuration is a UPS output power supply system composed of an "N+1" type UPS redundant power supply system (preferably a "1+1" or "2+1" type parallel power supply solution) + an output power distribution cabinet + a load automatic transfer switch (LTS). For some places with extremely high requirements, a power supply system with extremely high "fault tolerance" function composed of a load synchronization controller (LBS) + two sets of "N+1" type UPS redundant power supply systems should also be configured. Given that (30-50)% of the servers and disk arrays currently used in IDC computer rooms are products that adopt a "dual power input power supply" system, for these devices, the power supplies from two sets of "N+1" type parallel systems can be directly connected to the two input ends of this "dual power input device". For critical loads that use a "single power input power supply" mode, the power supplies from the two "N+1" parallel systems are first connected to the two input ends of the "load automatic transfer switch" (commonly STS-type static switches and SS-type fast transfer switches), and then the user's critical equipment is connected to the output end of the "load automatic transfer switch". Two typical dual-bus input, dual-bus output UPS redundant power supply systems are shown in Figures 1 and 2.

2. Optional with high anti-interference and excellent EMC characteristics

UPS redundant power supply system

The use of this power supply system is to create an excellent power supply operating environment for the information network to obtain 100% high "availability" (the "data throughput" and "data transmission rate" of the information network equipment in the IDC and MDC computer rooms can reach the nominal working values ​​of these equipment). A large number of operating practices have shown that whether the electromagnetic compatibility (EMC) characteristics of the UPS product itself are excellent and whether the various "interferences" that may appear in the UPS power supply system can be reduced as much as possible are one of the important factors that determine whether the "bit error rate" of Internet equipment can be reduced and whether the "actual data transmission rate" of the network can be improved. If the above problems are not handled properly, the Internet that could have run at high speed will always be in a "derating" state of low speed and small data throughput (it means that the actual "availability" of the Internet is extremely low), resulting in a huge waste of information resources. To illustrate this problem, please see Table 2.

As can be seen from Table 2, if the anti-interference performance of the power supply system is poor, thus forcing Internet devices to enter a low-speed "derating" state, it will not only cause a large number of network users to suffer huge "invisible economic losses" due to reduced "work efficiency", but will also lead to a significant decline in the profits of "telecom and network operators".

Examples can be given to illustrate the impact that power “interference” may have on the “bit error rate” and “data transmission speed” of the Internet.

CIRRUSLOGIC design center was established in Shenzhen

Cirrus Logic Corporation (NASDAQ: CRUS) has opened a new electronic design and applications center in Shenzhen, expanding its global leadership in digital entertainment electronics such as audio/visual systems, personal video recorders and, especially, China's rapidly growing DVD market.

The design center will support Cirrus Logic's large number of consumer electronics customers in China, strengthen support for certain key time-frequency customers in the region, and expand Cirrus' business into the fastest growing Chinese electronics market in the world.

In the process of running its local area network, it was accidentally discovered that if the inverter of the UPS was turned off and the UPS was powered by the AC bypass channel, the "data packet transmission rate" of the local area network was higher than the "data packet transmission rate" when the UPS inverter was used for power supply. The reason for this "abnormal" phenomenon is that there is serious "modulation interference" in the output of this UPS. Therefore, in today's Internet era, when examining the quality of a UPS, it is not limited to whether it can ensure uninterrupted inverter power supply to users. This is because, even when we choose this type of UPS product, on the surface, both the UPS itself and the Internet devices seem to be running "normally", and they do not cause any "downtime" failures on the Internet. However, since the Internet devices at this time are in an "inefficient operation" state of "low data transmission rate" and "small data throughput", that is, under this condition, the technical potential of the Internet devices cannot be fully utilized, and the consequence is a sharp decline in the actual "availability" of the Internet.

3. Optional "compatibility" with IDC's centralized monitoring system

Good UPS redundant power supply system

In order to meet the information network's needs for unmanned or undermanned management and remote centralized monitoring of IDC and MDC computer rooms, thereby improving the "manageability" of the UPS power supply system, the Internet data center should not only be equipped with a network management system for real-time monitoring of all its IT equipment, but also establish a centralized monitoring system for "non-IT equipment" to monitor and analyze the operation of equipment such as air conditioners, power distribution cabinets, battery packs, generator sets, water leakage alarms, safety systems, and fire protection systems in real time. To this end, the selected UPS redundant power supply system should be configured with the following communication interfaces:

1) Configure RS232/RS485 interface, Modem or SNMP adapter on UPS;

2) "Relay dry contact" type output communication interface used to display the working status/alarm information of the UPS;

3) "Relay dry contact" type input communication interface for "user-defined" input signals (access control, smoke, temperature, humidity and other alarm signals);

4) Configure necessary data collectors for various “non-IT equipment”;

5) Configure the corresponding centralized monitoring and management software package or communication protocol.

By using the above-mentioned input/output communication interface and the corresponding power management software or the user's existing network management centralized monitoring system, the so-called intelligent IDC and MDC computer room centralized monitoring system can be formed. In such a network management system, the main control functions that can be realized are:

1) Check the real-time operating parameters of the UPS that can be observed on the LED/LCD display of the UPS (for example, input/output voltage, current, frequency, active power/apparent power, power factor, and battery charge/discharge voltage/current, etc.);

2) "Operational Chronicles" of various "computer room environment control equipment" used by the "Internet Data Center" (information such as faults/alarms/operations performed by users, etc., arranged in chronological order since the various equipment was turned on);

3) In case of a fault, the network broadcast alarm (pop-up alarm window), automatic dialing of the telephone or mobile phone, automatic paging or sending of E-mail, etc. are executed to notify the on-duty personnel to go to the site to troubleshoot or repair in time;

4) Perform a programmable battery capacity "self-test" operation on the UPS backup battery pack. If the "actual capacity (Ah)" of the battery is found to be low, a pre-alarm signal of "battery needs to be replaced" should be automatically issued;

5) The "user-defined" alarm signals (such as temperature/humidity, access control, fire alarm signals, etc.) are incorporated into the unified centralized monitoring system of the IDC room or intelligent building through the UPS input communication interface;

6) Personnel designated by the competent authorities of IDC and MDC shall “readjust”/set the operating parameters of various corresponding equipment or perform remote fault analysis/diagnosis operations in a hierarchical manner according to different authorization levels.