Good afternoon Phil Corso and ControlsGuy25 .

I´m not comparing apples and oranges.

First things first.

As I´ve reported, I´m interested on the SIR - synchronous inertial response - of our generators.

I´m fully aware of the governors different technologies and speed responses.

But, for now, I just want to address the SIR and the resulting RoCoF.

Our grid is small, isolated from the mainland. There are no ties that can help us on outages and prevent frequency excursions.

Since we have only around 30 synchronous generators ( fuel, natural gas, steam and hydro ) I can easily calculate the expected RoCoF after an outage of the instantaneous highest power output generator (N-1).

The issue is that, performing "post-morten" outage events analysis, the measured RoCoF is always lower than the expected calculated one.

We are a small tourist island with no synchronous motors on the grid.

I´m aware that induction motors on the grid will provide some inertia, lower and slower than synchronous machines.

We also have several wind generators farms that, specially the oldest ones, will also provide some inertia.

It´s a quest for the lost inertia.

But I know we could consider a frequency-dependent load damping effect. I´m focusing on that issue. Normally it gets neglected but I want to demonstrate that it exists and maybe justifies the calculated and measured RoCoF differences.

My goal is to demonstrate to my Dispatch Center colleagues that the inertia they consider, is not provided by the old, high inertia constant, fuel generators.

One of our old diesel generators has almost the same kinetic energy than one of the brand new natural gas driven generators.

The lost inertia is elsewhere. I´m looking for it.

The governors response will be addressed in due time.

All your comments will be greatly appreciated.

Best regards.

Carlos Melim

 

Controlsguy25!
I apologize for offending you!
Phil

Phil Corso,
It was just a misunderstood, we are humans and I know that communication throw forum that can lead to misinterpretation.
James.

 

Good afternoon Phil Corso and ControlsGuy25 .

I´m not comparing apples and oranges.

First things first.

As I´ve reported, I´m interested on the SIR - synchronous inertial response - of our generators.

I´m fully aware of the governors different technologies and speed responses.

But, for now, I just want to address the SIR and the resulting RoCoF.

Our grid is small, isolated from the mainland. There are no ties that can help us on outages and prevent frequency excursions.

Since we have only around 30 synchronous generators ( fuel, natural gas, steam and hydro ) I can easily calculate the expected RoCoF after an outage of the instantaneous highest power output generator (N-1).

The issue is that, performing "post-morten" outage events analysis, the measured RoCoF is always lower than the expected calculated one.

We are a small tourist island with no synchronous motors on the grid.

I´m aware that induction motors on the grid will provide some inertia, lower and slower than synchronous machines.

We also have several wind generators farms that, specially the oldest ones, will also provide some inertia.

It´s a quest for the lost inertia.

But I know we could consider a frequency-dependent load damping effect. I´m focusing on that issue. Normally it gets neglected but I want to demonstrate that it exists and maybe justifies the calculated and measured RoCoF differences.

My goal is to demonstrate to my Dispatch Center colleagues that the inertia they consider, is not provided by the old, high inertia constant, fuel generators.

One of our old diesel generators has almost the same kinetic energy than one of the brand new natural gas driven generators.

The lost inertia is elsewhere. I´m looking for it.

The governors response will be addressed in due time.

All your comments will be greatly appreciated.

Best regards.

Carlos Melim

Good afternoon Carlos Duarte,

I am glad that you can go further, on you research and case studies .

What is meaning of Lost Inertia ? Can you tell us more about?
Thank you for your comments.

Also I started a conversation here, can you reply to my message?

Best regards,
James.

 

Good night James.

I´ve downloaded the book from www.engineeringbookspdf.com

I can calculate the inertia of our grid and predict the initial rate of change of frequency (ROCOF).

But, after a real outage event, performing a "post-mortem" analysis of our data, I can measure the real RoCoF.

I´ve studied several outage events and, without exceptions, the real RoCoF figures are smaller than the predicted, calculated ones.

The grid has more inertia than I consider in my calculations. I don´t know where is comes from.

That´s why I say that I´m on a quest for the lost inertia.

Best regards.

Carlos Melim

 

Good night James.

I´ve downloaded the book from www.engineeringbookspdf.com

I can calculate the inertia of our grid and predict the initial rate of change of frequency (ROCOF).

But, after a real outage event, performing a "post-mortem" analysis of our data, I can measure the real RoCoF.

I´ve studied several outage events and, without exceptions, the real RoCoF figures are smaller than the predicted, calculated ones.

The grid has more inertia than I consider in my calculations. I don´t know where is comes from.

That´s why I say that I´m on a quest for the lost inertia.

Best regards.

Carlos Melim

Good Night Carlos,

Thank you for the link, I downloaded it too.
The handbook looks, very good and contains lot of interesting informations.

I found this formula related to lost inertia , you may check and tell us if is matching with your studies research:

This is calculated as follows (based on the background
previously provided) and linearised over the small
frequency disturbance range:
Ekin.(min) =Pcont * Fn /2*RoCof+ Ekin (cont)(5)
where:
Ekin.(min) = Minimum system inertia required (MW.s);
Pcont = Worst case size of largest credible multiple
contingency (MW));
fn = System frequency (50 Hz);
RoCoF = Pre-defined acceptable RoCoF (Hz/s); and
Ekin.(cont.) = Amount of system inertia lost in credible
multiple contingency.

This was for the South African grid studies , can you confirm that formula in your studies?

Thank you for your answer,
Hope this can help,
Best regards,
ControlGuy25.

 

Hello everybody

I have data from a programmed outage of one of our generators. I´ve recorded the frequency excursion.

I want to calculate the RoCoF from the acquired data. I have 1.25 ms samples of the grid frequency.

I´ve worked them in Excel and came out with the following graph:



The generator 13 outage occurred at 0.5 s.

I´ve used the Excel LINEST function: Least Squares Method with a sliding window of 12.5 cycles, 250 ms.

The highest value of RoCoF detected was -0.50 Hz/s.

I´ve analysed values recorded during a 500 ms time period. From outage instant, time 0.5 s, to time 1 s.

The highest RoCoF calculated result seems quite low.

Any comments on the applied method?

Is 500 ms a full time analysis period enough?

Is 12.5 cycle least mean square sliding window a recommend calculation method?

Please consider the following 5 s period graph:



As you can see, frequency gets completely arrested only around time 1.16 s.

But has a local recovery around time 1.06 s.

I´m interested in RoCoF calculation before the governors responses.

If I extend my analysis period to 1.5 seconds I will get a RoCoF of -0.63 Hz/s.

But will it be trustable?


I am looking forward to hearing from you.

Best regards.

Carlos Melim

 

Hello everybody

I have data from a programmed outage of one of our generators. I´ve recorded the frequency excursion.

I want to calculate the RoCoF from the acquired data. I have 1.25 ms samples of the grid frequency.

I´ve worked them in Excel and came out with the following graph:

View attachment 285

The generator 13 outage occurred at 0.5 s.

I´ve used the Excel LINEST function: Least Squares Method with a sliding window of 12.5 cycles, 250 ms.

The highest value of RoCoF detected was -0.50 Hz/s.

I´ve analysed values recorded during a 500 ms time period. From outage instant, time 0.5 s, to time 1 s.

The highest RoCoF calculated result seems quite low.

Any comments on the applied method?

Is 500 ms a full time analysis period enough?

Is 12.5 cycle least mean square sliding window a recommend calculation method?

Please consider the following 5 s period graph:

View attachment 286

As you can see, frequency gets completely arrested only around time 1.16 s.

But has a local recovery around time 1.06 s.

I´m interested in RoCoF calculation before the governors responses.

If I extend my analysis period to 1.5 seconds I will get a RoCoF of -0.63 Hz/s.

But will it be trustable?


I am looking forward to hearing from you.

Best regards.

Carlos Melim

Carlos Duarte,


Thank you so much for this studies results.

My first question before to forget is did you find the " lost inertia " finaly?



The method you are using, for RoCof Calculation seems quite good.

Can you please , tell us how you can disctinguish that frequency is recovering at 1.06 on the graph for "grid response"

I would not say same statement ( for me frequency is recovering lately between 3.50 s and 3.60 s but I may not read good the graph.) Please clarify.

You stated that:
"If I extend my analysis period to 1.5 seconds I will get a RoCoF of -0.63 Hz/s."
I think that is quiet trustable .

Again "felicitations/congratulations" for your study/work thats a good case study.

Thank you for your answer.
Controls Guy25