In recent period, the author heard about a novel saying through communication with several customers within industry. It goes that modularized UPS of a certain manufacturer has a fabulous function, i.e. single power module can be used as a complete UPS and it is also listed to be a highlighted selling point. It seems that marketing also requires to be kept in pace with the times. Is it the previous modularized UPS of "decentralized bypass"? The technology which has remained unchanged for over a decade is currently transformed to be a new selling point, which can be authentically said that "knowledge is power".

Industry peers who are familiar with development of modularized UPS should know that system framework of modularized UPS has two different technical routes from the outset: decentralized bypass and centralized bypass. Here I would like to compare and talk about the option of these two schemes from perspective of source of technology and performance reliability, and hope to give readers some inspiration and help.

I. Definition and source on framework of two bypass schemes Modularized UPS, as the name suggests, refers to the method of dividing high-power UPS system into multiple sub-modules in parallel, achieving online expansion upgrade, maintenance of the system, and greatly improving system reliability, usability and energy conservation effect and reducing customers’ maintenance costs through the optimization of system control. In recent years, it has gradually become the first choice of mainstream customers. The analysis on the typical 300kVA system in the market based on ten 30kVApower modules is listed below.

(1)Decentralized bypass framework

For decentralized bypass framework, in addition to rectifier, inverter and battery conversion and other parts, each power module also contains the static bypass which has the capacity equivalent to that of power module. It can be considered a UPS without LCD monitoring. Multiple modules compose the system in parallel and the relationship between the modules is similar to that of traditional multi-parallel UPS system. When system switches to bypass power supply, the load realizes parallel-operation power supply with reliance on all power module decentralized bypass within all power modules. System framework diagram is shown in Figure 1.

Figure 1. Framework diagram of decentralized bypass

(2)Centralized bypass framework

For centralized bypass framework, the system merely has a bypass module which has the equivalent capacity with system capacity. Power module contains only rectifying, inverter and battery conversion circuit, each part has independent controller, the parallel operation between modules is no longer the UPS parallel-operation system, but the logic which contains complex inverter current-sharing, bypass control and monitoring etc. The system framework diagram is shown in Figure 2.

Figure 2. Framework diagram of centralized bypass

(3)Development sources of two technical schemes

The concept of modularized UPS originated in customers’ demands on simplification of system maintenance. They hope that the key business will not be affected in the case of fault and a simple replacement operation can immediately restore the system. It naturally occurs to manufactures that UPS parallel system can be designed to be of a modularized structure, which is the source of the decentralized bypass scheme.

Advantages of decentralized bypass system include simple controllability and small development difficulty, merely required transplantation of the original UPS parallel operation system and optimization of the monitoring part; low cabinet cost; relatively lower costs for smaller capacity of bypass devices; multi-path redundancy possessed by static bypass.

Centralized bypass scheme is the technical route which is developed after decentralized bypass.It has undertaken huge changes in aspect of parallel-connection current sharing control, system logic coordination and fault-tolerant capability when compared to traditional parallel-operation UPS system. It can be said to be a brand-new technical field with great development difficulties.

The difference brought by two technical routes in aspect of performance and reliability will be introduced in the following text.

II. Performance difference between two schemes

The importance of static bypass which is taken as the last barrier of UPS power supply is self-evident. Common bypass power supply includes the following circumstances: Inverter failure, Inverter overload or over-temperature, output short circuit. It can be seen that most of operating conditions of bypass power supply are extreme operating conditions and evaluation of devices should be more rigorous.

(1)Steady-state operating conditions

In the case of bypass power supply, centralized bypass scheme is easy to be understood. There is only a bypass which provide all electric current. Bypass capacity is designed according to maximum system capacity. It is not related to quantity of module configuration and there exists no problem.

Decentralized bypass scheme utilizes multi-circuit low-power static bypass to bear the load. Since bypass circuit is a low-impedance loop, multi-loop current sharing couldn’t be controlled with utilization of software method, the current sharing between modules depends entirely depend on the following factors:

1) Difference between individual devices. It mainly refers to differentiated break-over voltage drop and the inevitable dispersion of device manufacturers.

2) Difference of loop impedance. It mainly lies in that the length of each circuit cable couldn’t be ensured to be the same. Moreover, the impedance of each connection point of cables couldn’t be controlled for the reason of technique control etc.

In general, even according to the most optimistic estimation, the current sharing difference is unlikely to be less than 20%, that is to say, there exists the risk of excessive current in part of modules. It is very dangerous in rigorous application.

Due to this uncontrolled current-sharing capability, partial manufacturers come up with the “solution”----bypass current-sharing inductor. The principle is extremely simple. That is, each bypass circuit connects an inductor in series (as shown in Figure 3), uses impedance of inductor to balance the current of sub-circuit (also the conventional parallel system method). Regardless of the larger system loss brought by 10% individual difference of inductance value, this scheme also has an insurmountable barrier in transient performance.

Figure 3. Bypass current-sharing inductor of a certain manufacturer

(2) Transient operating conditions

The operation condition of inverter switching to bypass generally is urgent operation condition. It has extremely high requirement on time sequence of switching, otherwise it easily incurs break-off of key load. As to switching in the case of large load or fault current, transient operating current may be several times of rated current of the system. It is the reason why the design of static bypass requires larger margin.

Main parameter of transient-current impact resistance of static bypass device is I2t. It is the current integration within short period (generally less than 10ms). If I2t is too large, device is likely to be burnt up. In performance parameters of UPS, according to common regulations, 1000% maintained 10ms of bypass overload capability means that bypass need to provide not less than ten times of rated current within protection time of power-distribution switch (10ms). Next, we will take 300kVA system as example to analyze the difference of impact resistance capability of different devices.

For decentralized static bypass device, the maximum current level of its single device is 70A for the reason of current technological capability. According to device specifications of a certain well-known manufacturer, the provided maximum I2t is 7200A2S (<10ms) and 300k system can be deemed as 10-cricuit device for parallel operation.

Centralized static bypass uses SCR module, the most mainstream manufacturer is SEMIKRON in Germany. Let’s have a look at I2t parameters of SKKT323/16E. As to the same 10ms conditions for the 450000 A2S, the difference between the two attains more than 60 times under the same 10ms condition.

As to 300kVA system, we calculate the I2t requirement for a common 1000% overload 10ms.

That is to say, single SCR module of centralized bypass completely can provide 10ms protection capability of over ten times the volume of rated current. However, static bypass based on discrete devices is far from enough even though that unavailable current sharing of device is not taken into consideration.

Current-sharing control of transient switching is related to devices, impedance of each circuit. Moreover, it is also related to control. Since each module has its own controller, due to the influence of several factors such as processing speed of each processor, communication delay and differentiated modules, the actual switching of each module must have a different time-delay, which result in that the first module switched to bypass is likely to withstand 100 times the volume of module capacity rated current. Since it is transient high current, even the series-connection bypass current-sharing inductor couldn’t help to limit current. It is a task which is impossibly to be accomplished by any device. This kind of switching is tantamount to the circumstance of in-situ explosion. Schematic diagram of short circuit fault current is shown in Figure 4.

Figure 4. Short circuit fault current diagram

Of course, manufacturers also know this truth and also they provide a corresponding "solution" that only inversion maintains 200ms in the case of short circuit, and then do not switch to bypass and directly shut down.

Corresponding explanation is listed below. The operating conditions which has ten times the volume of rated current is more frequent than that of output short circuit. In the case that the inverter cannot provide enough current for fault analysis (usually 3 times the rated maintenance 200ms), the system will switch to bypass power supply, use low impedance high current of bypass to withstand protection device of short circuit point ( switch or fuse),which is required to be taken in to consideration in power distribution design. In the case of a properly designed distribution system, protection design of sub-circuit should not result in override protection, i.e., the lower-stream fault should not cause the upper-stream switch operation, the worst case of the system is that it is switched to the bypass, and then uses the bypass overload capacity to start the lower-stream protection device, which is the source of anti-impact requirements.

If the system which uses decentralized bypass is forced to be switched to bypass, devices will indisputably be damaged and system outage will occur for the reason of insufficient impact resistance and asynchronous switch-over. Thus, manufacturer can only adopt the design which prohibits to be switched to bypass. It can be imagined that if there is a branch occurs short circuit fault in a complex machine room or factory, it will result in the unavailability of the whole system. It will not be accepted in any case in practical application, but it is concurrently the inherent problem which couldn’t be resolved by decentralized bypass.

III. System reliability analysis

The alleged advantage of decentralized bypass is bypass redundancy. Centralized bypass is considered to exist single point of failure. Relevant analysis is indicated below.

(1)Analysis from perspective of device selection

From perspective of device selection, the reliability of single high-power SCR is much higher than that of the system composed by large quantities of small devices. Centralized bypass module has simple functions. It is merely required to consider the effect of devices and few peripheral drive circuits, while decentralized bypass concurrently is affected by numerous devices within module for it is distributed within power module. The engineering personnel who have maintenance experience all know that, malfunction of rectifying circuit and inverter circuit probably incur malfunction of other partial electric circuits for the reason of splashing spark etc. That is to say, static bypass faces relatively many uncertain risks. If centralized bypass can be said to be single malfunction, decentralized bypass can be possibly referred to as multi-node malfunction.

(2)Analysis from perspective of system capacity

From perspective of system capacity, capacity design of centralized bypass is designed in accordance with cabinet and it is not related to quantity of configured modules, while static bypass capacity of decentralized bypass is decided by module capacity. In other words, the system will lose corresponding static bypass capacity when module fault occurs. A more extreme example is that when cabinet is configured with two power modules, if the load rate is about 55%, when a module occurs malfunction, the remaining one module will be in operating condition of 110% overload and the final result is system power failure. The same conditions for centralized bypass is not a problem at all. Centralized bypass module equipped with the advantage in device capacity and even some manufacturers can provide 125% long-term overload capability, which all provide absolute guarantee to system reliability.