UPS reactive power supply ? 3

[tomad]

Per manufacturers' data, the output power for a UPS is limited by its apparent power output [kVA] and its real power output [kW].
EX. for a 100 kVA rated UPS, its real power output is listed as 80 kW. That would be equivalent to assume that the UPS can deliver a maximum of 60 kVar. Where does that reactive power come from ? Adjustable capacitors on the output ?

 

[sibeen]

The reactive power comes from the load.

If you have a look at the schematic of a standard inverter, in many cases what isn't shown is a 'hidden' device included in the switches (IGBT) of the inverter. This is an anti-parallel diode across the IGBT.

Any reactive current that is flowing uses the diodes in the inverter as there current path.

Some UPS systems also place extra capacitors, above and beyond what is needed for filtering, on the output of the inverter. These can be used to offset any inductive current. This keeps the current in the inverter and output transformer (if used) to a minimum.

 

[tomad]

Sibeen,

Thank you for your response. But you mention that the reactive power comes from the load. When the manufacturers indicate a power factor for the load, they usually mean a lagging (inductive) power. In order to cover that reactive (inductive) power requirement of the load, this power (reactive, inductive) has to come from somewhere, and it cannot come from the same load, because the load has an (inductive) power factor. The character of the load can be either inductive or capacitive (lagging or leading).

 

[sibeen]

tomad, just because something is rated to be able to support a certain load, it doesn't mean that it has to give that sort of load. When a manufacturer gives a rating of 100 kVA @ 0.8 P.f, it means that up to this rating there is no need to de-rate the UPS.

If the load is purely resistive then the UPS can support a load up to 80 kW; in this case there will be no reactive current as the load doesn't draw any.

 

[tomad]

sibeen, the initial (internal) discussion started when we were requested to decide if a certain UPS meets the user's requirements. The manufacturer's documents listed as limits the kVA rating and the kW rating. Since the tests were performed using a resistive load bank (pf = 1) the question arose how to get the UPS loaded up to its rated output current. Usually the output current is the value used to estimate overload, so they were not able to achieve that. So we indicated to load up the UPS using the available load bank until the output current reaches the rated value. This triggered a controversy that in this case "something" will be overloaded, but nobody of the consulted people were able to point out what exactly would be overloaded in this situation. We tried to find a model describing that requirement and assumed that, if there was a component limiting the active current ( current originating in the DC part), the rest of the current (to account for the difference between the value corresponding to the 0.8 pf and the total current determined from the kVA rating) has to come from something exterior to the inverter part(i.e. a capacitor). [I mentioned outside the inverter part, not outside the UPS assembly].
We could not clarify if the resulting power factor was a maximum or a minimum one, so, considering the maximum given values for the apparent power [kVA] and real [kW] we concluded that in this case there should be a maximum reactive power source [kVar].
So it has not been implied that it has to be a reactive current for any load, but, if a power factor of 0.8 was indicated (for a certain kVA rating) it has to exist a reactive power source (as sqrt((kVA)2 - (kW)2)), otherwise the statement of a power factor makes no sense.

 

[jghrist]

I'm no expert, but I'd look at it the same as a generator. There is a real power limit based on the prime mover capacity. In the case of a UPS, the prime mover would be the battery that has a maximum power flow capability. There is a kVA limit based on the ability to handle the current. The power factor or kvar limits are calculated from the real power and current limits.

You usually think of battery capacity in terms of energy (amp-hours) instead of power, however, so my analysis may be all wet. Maybe the real power limit is based on the battery amp-hour capacity and the design time of battery operation.

 

[waross]

You can build a reactive load with a variac and a transformer.
Short circuit the secondary of the transformer and power it with a variac from the UPS. The current will be almost all reactive. Balance this against the conventional load bank to get maximum kW at maximum KVA. Use the %impedance rating on the transformer to calculate the voltage needed to develop the amount of reactive current that you need.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[tomad]

jghrist, you are perfectly right with the generating set limitations due to the prime mover.
But in this case, since the UPS is to be run normally (when the power grid is on) connected to the grid (on-line), the power flow is from grid via rectifier-inverter-load, the battery kicks in only if the grid is down.
Indeed, the battery capacity is usually given in Ah, its voltage being determined by the type of batteries used, but in the end is the energy V*A*h considered.
But exactly as you would discharge a battery by connecting the load between its terminals; the connected resistance determinates the current, which, having the voltage gives you the discharge power. The maximal discharge time is based on up to where do you want to dischage the battery. But the discharge power is limited only by the cable, since you can short circuit the terminals if you have a thick cable. In this case the resistance is low, the current high and the available time low.

 

[tomad]

Waross, I am assuming that by the proposed solution in your response, if I understood correctly, to feed a Variac (autotransformer ?) from the UPS terminals and with its output connected to the primary of a normal transformer with short-circuited secondary, a reactive load could be simulated. Where do you mean to connect the load bank when you state "balance it against a conventional (I assume resistive) loadbank ? It cannot be connected in parallel to the short-circuited transformer secondary. Do you have a diagram with the proposed solution that you could attach to your response for clarification ?

 

[sibeen]

tomad, unfortunately in this case there is no simple or all-encompassing answer. It depends upon the UPS topology and the design considerations taken into account by the manufacturer.

A quick hstory lesson. Up until approximatly 10 years ago computer power supplies always (OK, nearly always) would draw a lagging current and so UPS manufacturers therefore designed their UPS to supply this load profile. This is why we have the standard 0.8 PF for a UPS. What wasn't specifically mentioned was that the load was assumed to have a lagging power factor and the design of te UPS took this in mind.

Then about 10 years ago computer power supplies began to improve. The power factor of these supplies began to move closer to unity and in some cases actually began to go slight (0.95) leading. This then began to cause problems for some UPS systems and there was many cases of systems goign into overload or shuting down due to overheating even though the indicated load may have only been 80%. As I stated in my first post many manufacturers used additional capacitors in the output to offset any inductive current in the output. If the load then became capacitive, the inverter and output transformer then had to provide the capacitive current to the load as well as supply the capacitive current to its own intergral output capacitor.

Some manufacturers have addressed this issue and now have UPSs that are still rated at 0.8 PF but rate this at both capacitive and inductive loads. These UPSs can then supply full rated current from 0.8 leading to 0.8 lagging, noting that there is still a kW limit on the UPS.

As to your situation, you need to know how the UPS is rated and the manufacturer or consultant should be able to provide this information. You also need to know what your load profile is going to be. If your load is going to be capacitive there is no advantage in testing the UPS with an inductive load.

 

[tomad]

Sibeen, when you say "from 0.8 lagging to 0.8 leading" do you mean (- for lagging and + for leading) (as an example)-0.8 , -0.9, 1, +0.9, +0.8 ? all at maximum apparent output power (kVA)? What is the reason for a 80 kW limit in this case ?

 

[tomad]

Sibeen, the load characteristic was 0.96 inductive.

 

[waross]

Hi;
The current drawn by a short circuited transformer is mostly reactive.
Hook the variac in parallel with the resistive load bank.
Use the load bank to supply the resistive load, and use the variac driving a shorted transformer to supply the reactive load. The shorted transformer will draw some real current to supply the resistive losses. If you have instruments to measure the power factor you can vary your resistive load ank with the inductive load bank so that the Watts consumed by the resistive load bank plus the Watts consumed by the shorted transformer equal the Watt rating of the UPS under test, and the reactive current is adding enough current to reach the KVA limit.
With a UPS rated for 100 KVA and 80 kW you will want 60 KVAR. I would run the reactive load bank up to about 50 or 60 kW. Then add the shorted transformer. Increase the variac until your instrumentation indicates 50 KVARs.Then add enough resistive load so that your total kW = 80.
At this point your transformer should have a combined load of 80 kW and 100 KVA.
If you want to test with a leading power factor, Connect capacitors to the variac instead of a transformer.


Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[tomad]

Waross, I agree with your explanation and the fact that a short circuited transformer represents an inductive load. We were looking to find an answer where the UPS provides the reactive power. The load (with an inductive power factor of 0.8) had to get that reactive power somewhere (from a capacitor inside the UPS ?).

 

[sibeen]

tomad, what happens if you plug your load into the mains or onto a generator?

If plugged into a generator the inductive current, caused by the load, will flow in the windings of the generator.

In the UPS these currents will flow in the inverter of the UPS. As mentioned in one of my posts, this is why the switching devices in the inverter have an anti-parallel diode.

 

[waross]

Don't get hung up on the mysterious appearance of reactive currents and loads.
What is happening is that the load characteristics are such that the current is either lagging or leading the voltage. There is no "Reactive Power Generator" in the load. Reactive currents and reactive power are a way of describing and quantifying the phase shift between the voltage and the current.
If you want to test the functioning of a supply with a power factor rating other than unity, use a load or combination of loads that draws current at the appropriate phase angle.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[tomad]

Waross, what I meant is that between 2 points A (source) and B (load), the direction of the reactive power flow can be either from A to B or from B to A. The direction (from A to B) is generally considered to be positive (+) while from B to A it is considered as negative. Positive is referenced as generated power while negative direction is referenced as consumed power. In our case the source is considered as the UPS while the load part is represented by the load bank + variac + transformer in case of testing and the servers, etc under normal circumstances. If the character of the load is inductive, that reactive power has to be provided by the source (UPS) if not a battery of capacitors connected in parallel with the UPS terminals (either inside the UPS or as a separate battery)

 

[tomad]

Sibeen, if the (synchronous) generator is connected to a power grid with imposed frequency and voltage, the generator will generate power at those forced values. If you change the excitation to a value over of under the one required for a "real/ohmic" load (pf=1) at these values, it will generate either inductive power or capacitive power (the power factor at which it generates will be for instance 0.9 ind or 0.9 cap). Since in most cases in the network there is a requirement to provide inductive power (which would require an increased excitation power (i.e. excitation current), to avoid overloading the excitation windings there are special generators that are used solely for providing reactive power. These generators can basically have smaller diameter shafts, etc since there is no real power (torque) requirement (at the level at which it would be if they would provide real power). In case of a stand-alone generator with a fixed load, the voltage of the generated power will increase/decrease if the excitation deviates for that particular load characteristic.
If you connect an inductive load(i.e.normal induction motor) to a power network it will need reactive power for its magnetization for operating and the network (either grid or standalone) will provide it, as described above.
If, assuming that all current (corresponding to the max kVA rating of the UPS) flows through the inverter, and this total current is independent of the load power factor, what will be overloaded in this case? If the inverter components are sized only for the active current ( in our case corresponding to 80 kW), what provides the 60 kVar that it is guaranteed to provide at the rated values 80 kW/100 kVA ? As you mentioned earlier it has to come from capacitors.
What is that antiparallel diode system supposed to do ? If it is to prevent reactive currents from flowing into the UPS inverter, the circuit is closed via those diodes, no reactive power is provided to the load (i.e. induction motor).

 

[waross]

Real power is Real. Reactive power is imaginary. If you have a load that causes the current to shift in relation to the voltage it is convenient to imagine something called "reactive power" and it may be used to accurately calculate current and voltage relationships.
Don't try to determine where reactive power comes from, it is imaginary.
Phase shift. That's all. When we get this worried about where "Reactive power" is produced it's time to stop and remember that "reactive power" is a convenient way to describe a phase shift.
The UPS must accept a load that may cause a phase shift.
If you want to test a UPS' ability to accept loads that may cause a phase shift, don't go looking for a "reactive power generator" to rent. Use a load that causes a phase shift. That's what you want to test, the ability of the UPS to accept any load within its rating.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[davidbeach]

Amen brother Bill.