Load Flow

[Cerkit]

Hi,

Just trying to get my head around load flow analysis. Can someone explain what it means when the software makes a comparison between the calculated value and the actual value and keeps doing iterations until this value falls with a given range? I can understand that we can get calculated values by doing nodal analysis at each node and get the current flowing throughout the network.

But what do they mean by the 'actual values'? What are these 'actual values' that the software is comparing with, I thought that other than impedance values and characteristics of the sources everything else was an unknown, so what is it comparing exactly?

Sorry if this is a silly question.

Thanks

 

[aussiejohn2]

G'day Cerkit,
Loadflow is solving a nonlinear set of equations by starting with an initial guess, and then using an algorithm that generates successively better guesses (Google "Newton's method"). The process can in principle be continued for an infinite number of steps, improving every time (if you use infinite precision variables), but the improvements get smaller and smaller each time. Infinite accuracy (precision) would take infinitely many steps, taking infinite time and requiring infinite precision variables. This is neither needed, affordable, nor justified by the input data so the process is stopped once the improvements - i.e. the difference between previous guess and current guess - are considered "small enough".
In short, the comparison is not between the calculated values and the actual values (the actual values are unknown). It is between the current guess, and the previous guess.
John.

 

[Cerkit]

Hi,

Makes sense thanks for the response.

 

[davidbeach]

Are you just looking at study cases or are you dealing with some form of state estimator? A state estimator would have actual field values at some locations and would be calculating the rest. The calculations need to produce values very close to the actual values for the rest of the solution to be viable.

 

[Cerkit]

Hi,

Just looking at classic load flow calculation used by most packages, Newton Raphson iterations.

 

[waross]

The process can in principle be continued for an infinite number of steps, improving every time (if you use infinite precision variables), but the improvements get smaller and smaller each time. Infinite accuracy (precision) would take infinitely many steps, taking infinite time and requiring infinite precision variables.

This statement may be true with the constraints stated stated.
However if the intent is to discover a numerical answer, I found it much more efficient and very much quicker to discover digits rather than use very small steps.
I played with this method back in the days of 8080 based personal computers and basic programming.
I tried a test problem with very small steps and it took about 40,000 iterations to generate a solution to the 8 digit precision of my computer.
Then I abandoned small steps and wrote a routine that discovered DIGITs.
I started with one digit.
10 initial digit x = 1
20 solve for x
30 Too large an answer; subtract 1 from the least significant digit and add 1/10 the least significant digit. go to 20,else
40 Too small an answer, add one to the least significant digit go to 20
I got an answer to 8 figures in about 40 iterations.
That would be possible with hand calculations.
Think outside the box.


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

 

[aussiejohn2]

Hi Bill,
The method does not use small constant steps. It is an innate property of the method that the steps get smaller as the solution progresses. Your method will work, but you will find NR (or friends such as secant, gauss-seidel, fast-decoupled) is faster to reach a given precision.
Regards,
John.