Evolution of the 480v static switch
Large static transfer switches are an accepted part of the power system design of modern data processing facilities. The facilities use redundant power sources to supply their data processing equipment with clean continuous power. Many different power architectures exist for these power systems, but in all of them, one or more STS are situated between the basic power sources (UPS, on-site generators, raw utilities) and the data processing equipment. The STS’s function in each of these architectures is to provide a redundant power source to the DP equipment in case there is a failure in the primary path.
As an integral part of that power delivery system, the STS must have many qualities and features that are required by the facility manager and the IT manager. The STS must be:
> robust and have fault tolerant design for very high availability
> Tolerant of the behavior of the upstream power sources and the downstream loads
> small as possible with minimum footprint.
> cost effective
> efficient in its power use, with minimal impact on the A/C requirements
In the past, the older versions of the 480v STS would have met all these requirements except one. That one exception was - it was not tolerant of the situation where the upstream power systems feeding the two sides of the switch were not “in phase” with each other when the STS was required to make a transfer. The underlying reason was that the 480v STS almost always has at least one output power transformer used to reduce the 480v power to usable 120/208v for the DP loads. But when the 480v switch is called upon to transfer when its two sources are out of phase, the earlier STS would still make a rapid ¼ cycle transfer. The out of phase power condition would cause large inrush transient currents to be drawn by the transformer for a few electrical cycles. This large current transient is caused the same underlying transformer physics that causes the familiar “cold inrush” current of transformers. Many times the resulting inrush current would overload the UPS that suddenly received the load. In other cases a circuit breaker between the UPS and the STS would trip open. These effects obviously defeat the entire purpose of having a redundant power system.
This lack of tolerance to out-phase 480v transfers has been addressed in a number of different ways in the past .
> UPS manufacturers have devised complex UPS controls to keep the two UPS in phase at almost all times. These go by various names such as Load-Bus-Synch, or Hot-Synch. However these systems are common mode failure points in the UPS systems, and they cannot compensate for all possible off-design conditions that a data center can experience.
> Consultants have specified STS using 208v SCR switching. This approach puts a full power transformer on each source upstream of the SCR switching. But that adds a second entire power transformer and increases size and cost of the STS, and deceases its efficiency, just to insure that the STS will not create an out-phase inrush.
> Some UPS manufacturers contend that there is no need for two independent power sources at all in the DP facility - and thus there is no need for the entire STS function. But this approach fails to account for A) the failure rate of the circuit breakers in the apparatus between the basic power system and the DP servers and B) the human error factor of personal who are working on the distribution system panel during maintenance.
> Lastly there are more “dual corded” DP loads, which at first glance would seem to eliminate the need for an STS. These “dual cords” also take various forms and some even have means of accepting an external DC input as a third power source. It is acknowledged by leading consultants that these “dual cords” can complement an overall power system that also employs STS, rather than supplant the STS. Moreover there are many individual pieces of DP equipment, including legacy equipment, that are not dual corded and will still need direct access to redundant power paths that STS can provide.
Also as a caution, the user must check that a given dual corded load has full internal provisions to insure that its two AC inputs can never be internally connected together with an inadvertent internal fault since that might endanger the entire DP room. Also check that the facility manager can designate the prime power source of the dual cord in normal operation, so that normal load balancing of the power distribution system can be accomplished. Both these functions are done routinely with an STS.
The new 480v STS
Now there is a way to make the 480v STS more phase tolerant, without sacrifice of its other qualities. This is new software feature called Inrush restraint.
Inrush restraint is available as an optional feature on SS3. As a software feature it does not impact the footprint or the efficiency of the SS3.
The Inrush restraint feature is an intelligent control that accounts for the phase difference between the sources, and the loading on the switch to make the optimum choice of a deliberately inserted small delay in the SCRs transfer control. This feature is only used for a case where an out-phase transfer must be made.
This feature differs from other SCR switching methods that insert only a fixed switching delay. The fixed delay that does not account for the exterior parameters of the power system- phase and loading- that are so necessary for proper suppression of inrush currents under all conditions. Nor does it optimize the suppression of inrush.
With Inrush restraint, the normal DPS patented ¼ cycle SCR transfer will be still made as long as the source-to-source phasing is within a phase window that the customer can adjust. Inrush restraint is only used for SCR transfer control when
A) it is selected to be used and
B) when the sources move outside the customer selected phase window.
The exact duration of the switching delay is carefully adjusted in real time for each transfer to insert the shortest possible delay into the SCR switching. Inrush restraint will still suppress the inrush currents by an order of magnitude, and at the same time hold the total 480v output voltages of the switch well within the ITIC power quality envelope.
An example of Inrush restraint can be seen in the two figures at the end.
In the first figure is of a 480v STS rated at 400 amps making a typical ¼ cycle transfer between two sources that are 120 electrical degrees apart at the time of transfer. The output current traces at the bottom of the figure show that peak transient inrush current to a 225 kva downstream transformer is 4500 amps. This is an entire order of magnitude greater than the normal rating of the switch and it is sixteen times the rated full load current of the transformer. Because of the great short-term overload capability of the STS itself, it is not damaged in any way by this inrush. Similarly in this case the receiving upstream source was a stiff utility feed, so there were no adverse consequences.
But consider that the same magnitude of inrush current would require approximately a 2.0 megawatt UPS system to be able to stay “on line” while supplying that magnitude of transient current. Also consider the complexity of breaker coordination of the UPS output breaker and power distribution panel breakers that is required to permit such short-term overloads. Considering that this inrush came for one single 225 kva transformer when switched out of phase, it is easy to see why the power system designers had been driven to look for other solutions. But in doing so, they had to accept the drawbacks of those solutions.
Now consider the second figure. It shows the same 400 amp switch use the Dynamic Delay option to transfer between the same 120 deg out of phase sources. To increase the scope of the problem, DPS used a 300 kva downstream transformer in this particular figure. Note that there is now approximately an 8 millisec (½ cycle) delay inserted at calculated times into the SCR transfer process. Note that the peak inrush currents in this case are now suppressed to approximately 500 amps, not 4500 amps.
This lower of transient inrush current would not disturb 200 kva UPS. Nor would it present any breaker coordination issues. Yet the voltage power quality of the STS is well within the CBEMA limits that allow a full 20 millisec of total outage.
Performance
In all cases the inrush restraint option will complete the SCR transfer in less than the 20 millisec allowed by the ITIC envelope. In most cases the inserted delay will be much shorter since it is controlled in real time for each transfer.
The suppression of out-phase inrush currents will be at least of 8/1 over what those same transient inrushes would have been if a normal fast ¼ cycle transfer had been made under identical circumstances with the same transformer (or transformers) on the switch output with the same loading on the those transformers. In many cases the suppression will be much larger.
Summary
The DPS SS3 switch with Inrush restraint now offers all the features required by a 480 volt STS. It is
> robust and has fault tolerant design for high availability, as all DPS switches
> now tolerant of the behavior of the upstream power sources and the downstream
loads. Out of phase transfers will suppress the inrush currents.
> small in footprint since it needs only one power transformer
> cost effective since the switching is done at the 480v, lower current stage.
> efficient because there it avoids the no load losses of a second transformer, and
does it’s switching at lower currents.
Figure 1 fast ¼ cycle transfer into a 225kva transformer
Figure 2 Inrush restraint transfer into a 300kva transformer
end 2/24/05
JEM C:\………\white papers\Advantages of 480v STS with Dynamic Delay rev 1