Load Flow First?

[hidalgoe]

The power engineers whom I work with have been debating whether to do the usual load flow study first, then the short circuit study or the other way around. One group favors load flow first because it gives you the pre-fault and angle values to use in your short circuit study. Others say a voltage of 1pu and angle of zero is adequate for most studies.

Who is correct?

Also, if your client wants a fairly inclusive system analysis (SC, load flow, transient analysis, motor starting, etc) for a fairly comlex system (2000 buses of typical HV feeders, LV feeders, motors, non linear loads,etc.), how long does it take to produce the study assuming you have all the data you need and a good computer program?

 

[mpparent]

My opinion is that typically, it is ok to have 1pu/0deg for an initial short circuit study. If you feel (mostly by experience), if there needs to be special considerations, then you could do load flow first.

If the comprehensive study is all that you're doing, maybe 6 - 9 months. Honestly, it's taken me almost 5 yrs to do my plants studies (still not done), but I also have other responsibilities.

Mike

 

[rbulsara]

Actually it takes only minutes to run a program, if you have all the 'data'. It's not running theprogram that takes time, but 'gathering the data and inputing them correctly and then verifying that you are setting up the correct scenario, checking for errors (if you know you may have made an error)etc. is takes the time.

Hoenstly, you need to calculate anything to accuracy required to make a decision or conclusion. So accuracy of input data is also accordingly vary. You need to decide whethere you are looking for precsion or accuracy. I usally go for the accuracy and for that assuming prefault voltage as 1.0pu is sufficient and more than adequate and is conservative.

For example, if you come up with SCC of 45000A, you do not pick a device of 46000 AIC but will go for 65000kAIC or somehting which is more standard. So you can determine what you are after.

 

[dpc]

There is little advantage in using the results of a power flow analysis to set the pre-fault conditions. Many software packages do not even link the two.

Any power flow computation is a snapshot of your system at one particular loading condition. Faults can occur at any time, including extremely light loads.

I generally try to determine what the highest typical value of incoming utility grid voltage is, then use that as the pre-fault voltage at the source.

 

[Bung]

It depends on what you want to use the results for. You decide what you want the studies to tell you, then that will dictate what studioes are done, in what order. But generally, if the order of doing the studies matters, then it is an iterative process anyway. You can't do the load flows if the system is no good from a fault level perspective, and you can't do the fault studies if the suystem is no good for power flow.

My thinking is that you start from a power flow perspective, and work from there. But then my background was in utility planning before I got into protection (also in a utility). Industrial tend to be much more cavalier with fault studies and load flows (I did a few years in heavy industry before I got into the utilities). But only in the sense they only have to cover their own fundamental orifices, whereas utilities are expected to cover everybody's.




Bung
Life is non-linear...

 

[Alex68]

I believe that a rule does not exist!
It depends on what is the aim of your work: designing a new plant or verifying an existing one.

In the first case, the L.F. studies and the S.C. calculations are mutually affected and proceed parallely.

In the second case, the two calculations may be indipendent.

About S.C., I think that you should apply a standard which imposes the plant conditions (voltages, angles and so on). Otherwise if you apply the superposition method, the initial condition must be specified. Anyway you should calculate the maximum fault currents for the components sizing and the minimum fault currents for the protection setting.

 

[DanDel]

It is only important that you consider the most conservative (worst-case) values for each study.
In the case of a short circuit study, you should consider the highest possible contributions to get the worst-case fault current, and for the load flow study, the lowest possible contributions (and/or switching arrangements) for the worst-case voltage drops.

 

[Alex68]

Sorry DanDel, but it is not sufficient.

In case of LF, also the minimum load condition is very important: the voltage must be within the high and low limits, so you should verify also the minimum load condition with the maximum voltage of your source to adjust the tap changer positions, for instance.

In case of SC, the minimum current is fundamental for the protection settings.

 

[cflatters]

I have to say that I use SKM software for loadflow and short circuit studies, the first emphasis is on ensuring that the load information is input and verified

Once I have the model as I want and am happy with the output I will run the fault study (IEC 60909). The model has the option to use loadflow voltages as the pre-fault voltages which I will do if I know the values are correct (ie verified from site), otherwise I tend to use the 'c' factors for voltage.

 

[rbulsara]

cflatters:

What if a fault occurs when there is no load or low load condition???

read what dandel and dpc said again which is a good summary..

load flow has little to do with SCC study.

 

[rbulsara]

In fact instead of worrying about slightly low voltage conditions, if you are so picky, ..what about slightly overvoltage conditions which regularly occur on systems close to substations and/or under low load condtions on the grid, specially during night...so using prefault voltage as 1.0 is as good as any and least controversial and adequately conservative.

 

[Bung]

I'm not sure that I agree with Alex68 that minimum conditions are the most important for protection settings. For grading margins, you want maximum fault level conditions.

Bung
Life is non-linear...

 

[AntonNel]

For Loadflow Analyses, I recommend using cable impedance etc, calculated at about 40 deg celc.(Higher impedance = larger voltage drope).

From my point of view, I recommend you keep the load flow and sc study seperate.

I use a system temperature of about 25 deg celc. There are a couple of recommended protection analyses methods. I prefer the IEC60909. The fault levels are calculated using 1.1 pu voltage above 1 kV. This compensates for motor contributions, etc.
If you know the maximum system voltages, use that + a safety factor. i.e max voltage = 1.05pu + SF.

Be practical in what you want to achieve.

I recommend you spend some time to ensure the data and model is accurate. Normally, we get excited to start with a system study. Only to find out later that there was some major error. Take time to get the model right. This is more important than figures you don't trust.

 

[Alex68]

Sorry AntonNel, but during the design phase of a distribution grid or of a BOP of a power plant you have to define the tap position of the NoLoadTapChangers. This tap position becomes the actual transformer ratio to be considered in all of your studies, including the SC.
So the two studies are not independent each other!
Moreover I suggest to base the SC studies on the IEC 60909-2001, which recommands the study of the minimum and maximum fault currents.

Dear Bung, I respect your opinion but I disagree. Your O.C. protections must also be able to detec and remove a 2ph (or 1ph to ground = very high probability) fault at the end of long lines, not only the 3ph fault (= very low probability) at the busbars of the switchgears! So, if the max fault current are important for the component sizing, the minimum fault current are very important for the protection settings.

An example: some years ago I had to design a little power plant connected to the distribution grid. The voltage grid varied in a big range while I had to limit strictly the contribution of the plant to a grid fault. I had to determine the range of the OLTC and the Xt of the Step Up transformer to satisfy the two conditions.
Do you think that the LF and SC studies were indipendent?
Absolutely not!

An universal rule does not exist!

 

[Guzzy89]

Hidalgoe,
Looks like this one has been beaten to death, however, I 'll add a few comments. I would think that you would want to perform a load flow study first. By load flow study, I mean getting actual load data from the field. If you have long Radial lines emanating from the trunk of that system the load could be very close to your fault current. Typically you want a four to one (Fault current to load current) ratio. This does not always happen. Once you have collected your load data, model the system, complete with load data, in a software package and perform a fault analysis. You will need the load data and the fault analysis on order to implement proper protection / coordination.
Moving on - even if your lines are close to your source, I would still collect load flow data first. When a fault analysis is performed, with load data present, the software will take in to account the load current, as well as, the fault current, giving you accurate results.
If you have not chosen a software package, I would suggest PSS/Adept. This is what I use. This package has great functionality.

Regards

 

[hidalgoe]

Guzzy89,

Yes, it looks like it might have been beaten to death, but I suspect that we have not settled this one, yet. That is why our field of endeavor is both an art and a science.

PSS/Adept, ETAP, SKM, Cyme, et al are all great programs.

My agency has settled on EDSA because of the diversity of modules offered and because it has been accepted (to the best of my knowledge) as official programs by many US and foreign agencies.