Availability Design of UPS Power System

Preface

UPS power supply is a key device used in the industrial field to protect the load from power failure. There are two types of power failure protection for different load applications. One is ordinary computer equipment. When a power failure occurs, the UPS power supply needs to provide a backup power supply time of several minutes to more than ten minutes for the load. During this backup time, the load equipment will perform actions such as data storage to prevent data loss, and then the load will shut down. The load will still lose power after the UPS reaches the backup time, but this will not cause economic losses. The other is in data centers and industrial applications. The requirement for UPS is real uninterrupted power supply. The UPS system must provide continuous power supply 24 hours a day throughout the year. This article's discussion of reliability and availability is aimed at this situation.

The reliability of the power supply system can usually be expressed by MTBF (mean time between failures, or mean time between failures, expressed in hours). In addition, there is an easier-to-understand indicator AFR (annual failure rate). AFR and MTBF are inversely proportional, that is, AFR=8760/MTBF. Therefore, the longer the MTBF, the lower the annual failure rate.

For repairable systems, there is also an availability metric, which is defined as

A = MTBF / (MTBF + MTTR)

A is a percentage indicator, and MTTR is the mean time to repair a fault. If the system can recover very quickly when a fault occurs, then the system availability index is relatively high. For objects such as power grids, the use of availability indicators can more intuitively measure their reliability. For parallel redundant configurations that are often used in critical situations, availability indicators are more realistic than reliability indicators.

Reliability/availability indicators are statistical concepts. There is also a statistical correlation between the reliability/availability of a power system and the reliability/availability of each module that constitutes the system.

Assume that there are two power modules in the power system, and these two modules work in parallel, one of which is independent of the other, as shown in the figure below.

Then the relationship between the availability Asys of the system composed of these two modules and the availability A1 and A2 of each module is:

Asys = 1 – (1 – AFR1)×(1 – AFR2)

Another possibility is that the two modules in the system are connected in series, as shown in the figure below

Then the relationship between the availability Asys of the system composed of these two modules and the reliability A1 and A2 of each module is

Asys = A1×A2

Since the availability is definitely a value between 0 and 1, the overall availability of the two parallel modules is higher than their individual availability, and the availability of the two series modules is lower than their individual availability.

Reliability of UPS power supply

From the perspective of the design of a single UPS, the entire product can be divided into modules. The following figure is a typical UPS system structure diagram.

As can be seen from the figure, the dependency relationship between the various modules of the UPS is relatively complex, but the series and parallel relationships can still be divided as follows

The auxiliary power supply is connected in series with all other modules, so the availability of the auxiliary power supply directly limits the maximum availability level that the system can achieve;

The control module is also connected in series with other modules except the auxiliary power supply, so the availability of the control module will directly affect the overall availability design of the system;

For the load side, only the bypass module and the inverter module can be directly connected, and these two modules are connected in parallel;

The PFC/rectifier module is connected in parallel with the battery boost module, and then in series with the inverter module;

From the perspective of energy providers, the bypass power supply and the mains power supply are two independent power supplies, and the battery energy is provided by the mains through the charging module. If the charging module fails, the battery will have no energy storage, and in fact, it will not be able to achieve normal UPS functions. Therefore, the mains-charging module-battery are also connected in series. In this way, the availability series-parallel path diagram of the entire UPS system can be drawn

From this path relationship, we can see that there are 3 parallel paths in total, and each path is connected in series by several modules. Just as analyzed above, the availability of the auxiliary power supply and control module is connected in series on all paths, so if there are defects in the design of these two, the availability of the UPS cannot be very high. The battery loop has the largest number of modules in series and is also the path with the lowest availability.

To improve the availability of the system, we must first improve the availability of the critical path. From the path diagram, we can see that it is the control module and the auxiliary power supply. The auxiliary power supply is the key point of the entire UPS. If the auxiliary power supply does not work, the entire UPS will be paralyzed. There are many ways to improve the availability of the auxiliary power supply: one is to improve the design and improve the MTBF; one is to apply parallel redundant design to the auxiliary power supply to improve the availability; another is to use different auxiliary power supplies for the three availability paths of the UPS, which is equivalent to changing the original completely series path to parallel. These methods can be mixed in UPS design. Since the above three availability paths are in parallel, and the bypass path itself is the one with the highest availability, the most recommended design is to give priority to improving the availability of the bypass, and use a separate set of auxiliary power supplies for the bypass, and try to use a simple design for this power supply to have a high MTBF.

The control module is also a key point that affects all paths and must also have high availability. Referring to the processing method of the auxiliary power supply, a relatively independent bypass path can also be equipped with a separate control module, and high availability can be achieved by coordinating with other control functions. Similarly, the control module on the bypass should be as simple as possible to improve reliability. A recommended approach is that the bypass control module continuously detects the status of the UPS main control module, and automatically switches to bypass mode if the main control module is found. In addition, the availability of the main control module can also be improved through redundancy, such as using a dual MCU structure. When one MCU detects that another MCU has failed, it can take over the function of the other MCU, or take emergency measures such as switching to bypass to ensure that the load is not disconnected.

For UPS, the battery is the key to ensure that the UPS can continue to maintain power supply when the mains or bypass power outage occurs. However, the battery has the most series links, which is also the link with the weakest availability. Generally, the battery specification will state that the charging current should not exceed 0.15CC, which means that it will take several times longer to recharge the battery after the UPS is fully discharged. In this sense, its availability is generally below 20%. However, since the battery does not work continuously, as long as the mains power is restored before the battery is discharged and there is no power outage during the recharging process, the load will still not be affected. From this perspective, the battery availability is still very high in the case of a short power outage.

Let's re-examine the reliability of the battery circuit. There is also a charger module between the battery and the mains. If the charger is damaged, the battery cannot be recharged after it is fully discharged, resulting in a power outage on the load during the next mains power outage. However, the charger only works when the battery needs to be charged. Therefore, if the status of the charger can be monitored in time and an alarm is issued in time when an abnormality is found in the charger, the problems caused by the charger failure can be avoided, thereby improving the availability of the entire UPS. The same means are available for batteries. After being used many times, the battery will also face the problem of capacity reduction and failure. However, if the battery failure can be detected through battery status monitoring and replaced in time, the availability of the UPS can also be effectively improved.

Reliability of UPS systems

Since UPS is not a separate application system, but is combined with other environmental factors, these external factors must also be taken into consideration. As mentioned earlier, the backup time of UPS batteries is limited. If the power outage lasts for a long time, causing the battery to be discharged, the load will still be powered off. Therefore, UPS availability will be affected by the probability of long-term power outages.

In order to solve this bottleneck, a backup power supply with complementary characteristics to the battery can be added to the UPS system: it does not need to react quickly when the mains power is cut off, but it can continue to provide power under long-term power outage conditions. The fuel generator set is the most suitable choice. Therefore, an automatic switching device can be added to the UPS system configuration to switch to the generator set after the mains power is cut off. This can greatly improve the availability of the UPS system under long-term power outage conditions. In this way, the availability path of the UPS system becomes

Although an additional ATS for switching between AC power and generator is connected in series in the availability path, which increases the probability of failure of the monotonic path, it is still worth it compared to the availability problems caused by long-term power outages.

Another branch of UPS application is the DC UPS system which is currently emerging. The idea of ​​the DC system is to improve efficiency, reduce the conversion links in the power system, and convert the power distribution part from the original AC to DC. The application structure of an ideal DC UPS system server application from mains power to 12V terminal is shown in the figure below.

It can be seen that the ideal DC UPS system can improve efficiency by replacing the inverter link of the UPS in the AC system and the PFC link in the server power supply with an isolated DC/DC link. However, in the DC UPS system, since the battery voltage has a relatively large range of variation, in order to obtain a more optimized efficiency curve, it is also possible to use a two-stage structure in the subsequent server power supply. That is, through a simple conversion, the input range of the server power supply isolation DC/DC conversion stage is reduced to obtain better energy-saving effects. The structure at this time is shown in the figure below.

In this DC UPS system, there is no bypass circuit in AC UPS, only a mains to battery circuit, which also serves as a charger. Therefore, from the perspective of the availability and reliability of a single UPS, there are only two DC UPS reliability links, one of which is the two-stage conversion plus the auxiliary power supply and control board, and the other is the battery, as shown in the figure below.

Compared with AC UPS, DC UPS power supply lacks the bypass circuit of AC UPS, and lacks a circuit to improve availability. However, the battery directly supplies power to the load, and the availability is higher than that of AC UPS. Therefore, the DC power supply system has both advantages and disadvantages in terms of availability. But on the other hand, the DC system is easier to parallel than AC UPS, so the availability can be increased by increasing the number of parallel units.

Availability of power distribution systems

For general UPS system applications, there are two common configuration methods. One is dual-machine hot backup, as shown in the figure below.

In this configuration, the two UPSs work in complete parallel. Based on the above availability principle, the second configuration will have higher availability than the first.

This reflects a clear difference between availability and reliability. For two parallel redundant UPSs, since the number of devices is doubled, the probability of failure will also increase, so the MTBF of the entire system will decrease statistically. However, since one of the UPSs will still work after the failure, as long as the failed UPS can be repaired quickly, the load will still be effectively protected, and the availability will be improved. From the perspective of load, it is more meaningful to evaluate the availability of the system than reliability.

In the definition of availability, the shorter the time it takes for the power system to recover, the better the availability. Therefore, designing the power system as a modular and easily replaceable structure can greatly reduce maintenance time, thereby significantly improving availability.

For computer room applications, the concept of dual bus is widely used. For critical server loads, two sets of power input are generally provided. Correspondingly, two sets of independent power buses can also be used in the power distribution part. Combined with the UPS itself supporting dual bus input, many combinations can actually be constructed. After comparing different methods, a typical structure that is recommended is shown in the figure below.

Here, two sets of independent AC power are supplied to two UPS systems, and then each UPS system is used as a bus, which can give full play to the advantages of high availability of AC dual bus, UPS internal dual bus and load dual bus.

in conclusion

This article analyzes the internal design of UPS, the availability of UPS system and power distribution system, and provides ideas for improving the availability of UPS power system. The analysis results show that the use of bypass power supply independent of the mains power supply, adding multi-CPU monitoring, adding battery monitoring and other measures in UPS can significantly improve the availability of UPS. On the other hand, at the system level, choosing a modular structure, shortening the maintenance and replacement time, and using more parallel structures can also significantly improve the availability.