Modeling a PV inverter for fault contribution into distribution system 3

[rockman7892]

Does anyone have experience with modeling a PV inverter in order to model short circuit fault contribution into an existing distribution system?

From what I have looked at I see that the rule of thumb for an inverter's fault contribution is 1.5 - 2 times its continuous rating. Has anyone used these numbers or similar numbers?

What is the best model representation for the inverter interfacing with a distribution system? I'm using SKM and the first thing that comes to mind is using a utility source to model the inverters contribution although I'm not sure if this is the best option. Is there some better option for modeling the inverter fault contribution.

Also I don not have the system I am interfacing with modeled so I am going to provide a fault magnitude and X/R ratio at the POCC to the rest of the system. Can I use the cable and transformer impedances between the inverter and the POCC as the impedance component of the fault calculation at the POCC bus or is the fact that the inverter fault contribution may be dynamic change things somewhat?

Any thoughts on how this effects coordination in the distribution system as well? I am dealing with a 250kW inverter connecting to the distribution system at 480V through a 400A thermal-mag breaker. Is there any coordination considerations between this breaker and the POCC pane's main breaker? I know that an inverter is cable of disconnecting itself very quickly during a fault in some cases 1-2ms which then may not require coordination?

 

[davidbeach]

Welcome to the club. If you ever find a good answer to your question, please share.

As you've found, there are no good tools for modeling inverters, or other power electronic driven/controlled sources, in system analysis programs. The only modeling I've seen that seems to work has involved the inverter manufacturer supplying the modeler with proprietary control algorithms under tight non-disclosure requirements. I've seen the results, but not the actual model itself.

 

[rockman7892]

Thanks davidbeach

I know what your saying, I've been searching for days now and cant seem to find a decent technique or explanation for modeling.

In the case where an inverter is only sourcing into the utility do you believe the amount of current will be dependent on the output of the solar arrays in the case of a PV system?

 

[davidbeach]

I doubt it, but don't know for sure. Given some things I've seen, I suspect that at least the initial fault response all comes from energy already in the dc link. Sustained fault feeding may depend on the arrays, but then again there is very little power in a fault, lots of current, but next to no power, so it may not take much input to allow sustained fault current sourcing either.

 

[redfurry]

The problem with inverters is that they don't act like a rotational machine. I've also seen numbers of about 1.5 - 2 FLA for maximum fault infeed, but it appears that this is very transitory and only lasts a cycle or so. Like David, I suspect this is due to capacitors in the inverter circuit and not really due to the power being fed into the inverter at all.

In studies I've seen in the past the consultant used the 2.0 x FLA to determine an equivalent impedance and then modeled the inverter as a simple generator. This gives a very liberal representation of the fault infeed (almost certainly overstates any realistic fault contribution).

Even under these circumstances, the relatively low infeeds from inverters generally mean very low fault contribution to a distribution system (especially once the step-up transformer is considered).

If your purpose is to determine fault infeed back into the system, you may be able to get away with using this as a worst-case approximation. If the fault contribution is still of concern, you will need to go back to the inverter manufacturers and get additional ammunition.

 

[cranky108]

Some modeling programs include models for current limited (machines). I have used the model but I don't have much experence to know if it works right.

We just don't have that many faults on our underground.

 

[rockman7892]

redfurry

I think the approach you mentioned for modeling as 2x FLA is sufficient for the purpose of worst case magnitude.

Would you say it is better to convert the 2x FLA to an impedance and respresnt the inverter as a simple generator or utility source? In either case do you hold the voltage of the source constant at the rated inverter output and then enter the converted impedance as the source impedance? What would you use as an X/R for the source?

I assume this is all before the step up transformer that you mentioned and that after the transformer the fault levels may be near insignifigant.

Is there a reference that lists the 2x FLA for inverter fault current? I have seen this rule of thumb in several articles but I not seen this defined in ay standards. Anyone know of this listed in any standards.

 

[Marmite]

Attached is a paper prepared for the UK government on the impact of DG on fault levels, which may be of interest.
Regards
Marmite

 

[redfurry]

Regarding the maximum fault infeed of 2x FLA. I do not believe there are any standards that address this. IEEE 1547 doesn't mention anything like this because it covers all DG technologies, not just inverters.

I have seen the maximum short circuit current listed on different manufacturers technical guides. From memory I remember infeeds in the range of 1.2 to 1.4 FLA (I probably have some PDFs I could dig up when I get home).

If you know the type of inverter unit you're working with, the manufacturer should be able to give you this info. Some have this info on their web sites.

How you perform the study is up to you. I'm more familiar with using software to model the entire utility distribution system, and using realistic voltages.
In the spirit of 'worst-case', I would use whatever X/R ratio gives you the highest infeeds. You are unlikely to find accurate info about this.

 

[jraef]

I've wondered this too. But in a PV system, wouldn't the inverter maximum fault current contribution be inherently limited to the PV supply current capacity? Every PV string will have an "I_sc" rating, being the maximum Short Circuit current that the string can deliver. Wouldn't the maximum output of the inverter then be limited to an aggregate of the input string I_sc currents? Article 690 of the NEC has you do this for selecting the proper OCPD, I would think the same holds true for fault current contribution as well.

"Dear future generations: Please accept our apologies. We were rolling drunk on petroleum."
— Kilgore Trout (via Kurt Vonnegut)
For the best use of Eng-Tips, please click here -> faq731-376

 

[cranky108]

I would think the fault current would be limited by the PV supply. However, how would you know this, or if it had changed?
The assumption should be made of an infinite source into the inverter, and the inverter as the limiting factor. After all the inverter is less likely to change than the number of PV panels.

In the case of a recent PV system installed near here, there were two inverters feeding a common step-up transformer. Then I believe there were three or four of these to give about 5 MW. (I don't remember the exact size, and I don't feel like looking it up).

There was also a grounding transformer to provide a ground reference (like for a breaker operation on this feeder).

 

[LionelHutz]

I wouldn't expect any PV inverter to be capable of much more than about 125% of it rated output before current limiting or trips activate. Inverter stages are designed closely matched to the application these days and manufacturers just don't give away much if any excess capability. So, using a value somewhere between 150% to 200% should be safe.

I'm not sure that modeling it as a fixed source with a high impedance would be correct. A varying source with a current limit would likely be a better representation.

 

[rockman7892]

As LionelHutz mentioned a varying source such as a generator as opposed to a utility source would be a better representation as others have also mentioned above.

I've also read some papers that mentioned modeling the generator set at PQ. I'm assuming this means setting the generator equal the the kW of the inverter and then setting the generator at some power factor? I believe I've also seen somewhere that inverters have set power factors so any thoughts on what to set the generator pf to? Are inverters typically capable of supplying reactive power with a high X/R ratio such as in a fault?

When modeling the inverter as a generator how should we input the Ansi fault contribution data for the generator X", X', X etc... Should we leave this at the default for the size generator or somehow manipulate this to provide the required 2X FLA current. I would think if we left it at default it would provide far in excess of 2X current.

Any thoughts on what to consider for coordination between the inverter output breaker located at the POCC panel and the main breaker in the POCC panel on a 480V system?

 

[rockman7892]

I finally heard back from SKM with their advice for modeling this inverer.

They also stated that a generator model was the best representation and that the generator voltage and kW rating should match the inverter, and then the positive sequence impedance should be manually manipulated until the desired ouput current current is reached. So to achieve a 2X fault current rating the per unit positive sequence impedance can be changed to .5. He didn't have much to say about what to set the p.f. of the generator at.

He also mentioned that a utility source could be used with the current value of 2x FLA entered for the fault contribution.

 

[jghrist]

You might try modeling it as a UPS. From SKM Help:

At present the UPS component can be used in DAPPER Demand Load, Load Flow, Comprehensive Fault, A_FAULT and IEC_FAULT studies. A UPS can also be used in the Unbalanced/Single Phase studies.

 

[rockman7892]

One other thing that comes to mind is the possible presence of filtering capacitors at the output of the inverter.

Do the capacitors need to be taken into consideration for their possible discharge current contribution into the fault, thus possibly raising the instantaneous level of the fault?

 

[cheezemanrich]

PV utility interactive inverters which are certified to UL1741 require a Short-circuit test (47.3) as part of the abnormal tests.
The maximum output short circuit current and duration are to be measured during the output short circuit condition.
The DC input used must be 2x the rated input current so this test is worst case because the PV modules are current limited and always sized less than the rated input current.
The manufacturer should have this information available and will usually print it in the manual under specifications or on the specifications sheet for the product.
The inverters I work with have a fault current equal to the maximum continuous current and a duration in the microseconds.
The inverters also must comply with IEEE1547 voltage and frequency limits as well as anti-islanding tests so usually in a fault condition they will trip off as soon as the voltage or frequency goes out of range.

You are correct also that the PV input is current limited and is also limited by the irradiance at the time of the fault.
A fault at 9:00am may only be 5-10 percent of the rating but at high noon on a clear cool day it could be 100 percent.

The power factor is by default typically >0.99 but some inverters can be adjusted for leading or lagging, usually as high as +0.80 to -0.80, but only in coordination with the utility.
The power factor can also be gathered from the datasheet or the manual if in question.

 

[ReRoy]

I have models for two different modeling systems (PSEE and PSLF) which my company had done by a third party engineering group.

These are positve sequence models - they are 3ph-delta connected and have no zero or negative sequence components. The models are based on the fact the inverter monitors the voltage applied at its mains terminals and creates an equivalent voltage waveform via PWM techniques which "follows" the grid voltage.

The current fed to the grid is modeled as a current source, which has a power capability limited by the irradiance level of the PV array. The current source is active only when the terminal voltages are within the limits prescribed by UL1741/IEEE1547.

Thus the inverter is not at all similar in operation to a rotating shaft generator. Rotating machinery is voltage based - it generates a voltage when energized and has a Voc, but an inverter based device does not - its terminal voltage will always reflect the connected grid and it will inject current in phase with that voltage unless commanded to operate with a pf different than 1 (typically 0.995, but capable of being commanded to +/- 0.90 pf.)

 

[jghrist]

The inverters I work with have a fault current equal to the maximum continuous current and a duration in the microseconds.

The current probably is from discharging the dc link capacitors. Considering the low level and very short time, I think it would be appropriate to model the inverter the same way that you would model a shunt capacitor for fault studies. That is, ignore it.

 

[Chebahtah]

GE has a quick study on modeling inverter DG on their PSLF site.

Of course, the study is mostly geared towards performing steady-state and dynamic studies in PSLF but it has some info on low voltage ride through and a quick bit on fault analysis.

Regards,
Justin Chebahtah