Uninterruptible power supplies (UPSs) can prevent loads from seeing voltage sags, surges, or other electrical disturbances. UPSs can also power the load with a battery source during complete outages until the generators start. In fact, they can even transfer the load to a secondary stable source using the internal static transfer switch (STS) during abnormal circumstances. Perhaps most importantly, transformerless UPSs are designed to prevent downstream loads from ever seeing an interruption to their power source.
In the late 1990s to early 2000s, large facility data centers were constructed around the United States due to the dotcom era rush. A large number of these facilities purchased transformer-based UPS modules. These modules are now nearing the end of their useful lives. Fortunately, transformerless UPSs are becoming the standard of data center design all across the world. These UPS systems can provide harmonic mitigation and power factor correction with active front-end controls. They can operate with very little output voltage harmonics and a superior step load response. Transformerless UPSs can provide these great characteristics, while providing higher efficiencies over transformer-based topologies.
Data center operators sometimes question the UPS module’s ability to handle fault scenarios and ask how these scenarios affect the facility operation. This issue occurs due to varying industry standards as to how the UPS will function during these critical situations. The causes of data center outages can range from human error to catastrophic component failure. The UPS cannot prevent human error. However, how it reacts to device failures and faults should be based on expected results of the system.
A typical UPS module will see a multitude of faults in its 15- to 20-year life span in a data center. The causes of faults can range from the utility, the load, the battery, or even the UPS module itself. It is best to categorize these in two groups based on what the UPS module can do to alleviate these problems.
The utility can cause fluctuations in voltage and frequency that would normally inflict damage to the critical load. As a result, the transformerless UPS module must react to these fluctuations by making minor corrections to the converter to offset the increase or decrease in the voltage. These corrections maintain steady internal DC buss voltages during such fluctuations. The steadier the DC buss voltage remains, the easier it is for the UPS to maintain its output voltage. The UPS can also accept a wide frequency window. This is not a common problem on utility, but the frequency can shift during generator operation due to a faulty sync module or during upstream utility to generator transfers and returns. The effects of varying frequency will also cause the UPS to modify the control of the converter section. The main concern with frequency fluctuations is the UPSs’ ability to synchronize to the bypass source during these frequency changes. If the source goes out of the desired range, the UPS module will be forced to operate in an asynchronous mode. This means the UPS will follow its own internal clock and produce a very precise 60.0Hz sine wave instead of the irregular frequency the utility/generator is providing.
Another source of issues that can be caused upstream of the UPS module is harmonics from other equipment on the same utility source. One of the benefits of transformerless UPS modules is their insulated-gate bipolar transitor- (IGBT) based control of the converter sections, which allows for the UPS to correct these incoming harmonics. They can effectively neutralize the current harmonic content of the incoming wave and help prevent their harmful effects on other upstream equipment. The converter will also prevent these harmonics from reaching the critical load by canceling them out before they reach the internal DC bus.
The UPS module can experience a severe voltage spike due to a lightning strike or any other transient causing event. The events can exceed device voltage limits and cause destruction to devices if the UPS module cannot isolate itself from the utility source before its limits are exceeded. There are physical limitations to the speed at which this can occur. The input isolation device, which is typically a contactor, can only be operated so quickly. The UPS can detect rising input voltages in the micro seconds and begin the process of isolation by commanding the contactor to open. The contactor will operate as fast as mechanically allowed, which can take anywhere from 20 to 50 msecs. In this time period, the transients can often exceed 600 VAC or higher. It is common place to provide a transient voltage surge suppressor internal to the UPS cabinet to help suppress these spikes in voltage.
DOUBLE CONVERSION TRANSFORMERLESS UPS
The online double conversion transformerless UPS module can prevent upstream events from affecting the operation of the downstream critical components. If the UPS module is transformerless, line interactive UPSs may not be able to prevent fast-acting transients from affecting the load. If these were the only issues that could provide problems for the UPS module, life would be simple. However, there are many things that can cause downstream catastrophes.
If equipment fails, it is not always as simple as the device just ceasing to work. There are numerous issues that can cause the feeds to the downstream critical load to see large currents due to short circuits or ground faults. These currents can be high enough to force the UPS module to transfer to the static bypass path on an instantaneous overcurrent. If the fault event is long enough, it will typically cause the breaker feeding the bypass path to trip on an overcurrent. If the overcurrent is short lived and the upstream breaker does not trip before the fault current level returns to normal, then the UPS will transfer back from the bypass path to the inverter.
If the fault is phase-to-ground, the current may not exceed the limits of the output of the UPS. This type of fault may be sustained for long periods of time with relatively little current flowing through the breakers. The breakers will not trip without ground fault protection. If the breakers do not have this type of fault detection, a potential safety hazard may be sustained. This is where the UPS ground fault detection comes in to play. The UPS module will detect a voltage offset between its zero reference and its ground. This zero reference is commonly referred to as a virtual neutral. It allows the UPS module to ensure that the output is not floating vs. ground and also is used to detect ground faults. Once the UPS module detects the ground fault occurring, the UPS module will go into alarm and begin to determine whether the fault is a DC ground fault, an internal UPS fault, or external to the UPS module. If the fault is determined to be an internal or external fault, the UPS module will transfer to bypass via the static transfer switch to protect itself from feeding the fault. This function can be disabled if a high resistance ground (HRG) is utilized in the system. In this scenario, the HRG would be causing the ground fault to be limited to 5 to 7 A, and it would be expected for the UPS module to maintain inverter operation until the fault is found and corrected locally.
While the UPS is running on battery, the phase-to-ground faults will act slightly different. Since the UPS is running on battery, the main input will be isolated from the utility neutral-to-ground bond. This will prevent the downstream current from returning to that source. Instead, the ground current will return through the UPS virtual neutral-to-ground connection. This connection will act similar to an internal HRG. The power flow is limited through this path to approximately 7 A. The UPS module will still detect the ground current flowing, and it will go into an alarm state. Once the source returns, either from the utility or generator, the fault return path will now be through the utility or generator neutral-to-ground bond. Once this path returns, the fault current will no longer be limited to 7 A by the UPS virtual neutral. It will typically increase to a high current fault, which may cause the UPS module to transfer to its bypass path on an instantaneous overload. This will allow the fault to be cleared by the upstream circuit breaker. If the fault current still remains low enough once the neutral-to-ground bond has been introduced, the UPS will stay in alarm and begin to determine the location of the ground fault. Phase-to-phase faults while on battery operation will produce large amounts of current and cause an instantaneous overload of the UPS module, forcing the unit to transfer to bypass.
The battery system can also cause fault conditions on the UPS module. As batteries degrade, they can begin to leak from their terminals. The acid is conductive, so if the post of the battery establishes an electrical connection to the battery cabinet chassis or rack ground by way of the acid, a fault can occur. These can start as low current positive-to-ground or negative-to-ground fault conditions. The current created by this fault will create a DC offset between the virtual neutral and the UPS ground. The UPS will detect this offset and go into a “DC Ground Fault Alarm.” The UPS module will stop the DC chopper circuit from charging the batteries while they are in this ground fault state. If the fault continues after the chopper has been stopped, the UPS module will open its DC breaker, isolating itself from the ground fault. The isolation of the battery string will stop the current flow through the ground fault, since the battery string is floating. It will also help to prevent thermal runaway situations. If the UPS continues to see the ground fault occurring with the DC breaker open, it will assume the fault may be occurring inside the UPS module. In this instance, the UPS will initiate an instantaneous transfer to bypass to prevent any further damage to itself or to the critical load.
A short circuit between the battery terminals will cause a high current fault. If the fault is located after the battery breaker, the current will cause the breaker to trip. If it is before the breaker the battery will begin to deplete itself at a rapid pace, discharging from the positive terminals to the negative terminals. In both events the UPS will continue to operate on inverter, and there will be no damage to the charging circuit of the UPS module.
These are the basic faults that can occur to the UPS module from external sources. The UPS module will function as its design is intended under most events. However, understanding how the UPS functions during adverse events will enable data center users to achieve a higher comfort level with their systems and may allow for faster recovery times and shorter down time. Hopefully, this new knowledge will also allow decision-makers to move to a transformerless system more comfortably during the transition.