Unable to Enact External AGC Signals on REGF1

[talisha-g]

Dear Dr. Cui, Jinning and ANDES developers,

I'm Talisha, a graduate student at the University of British Columbia. I'm using ANDES and ANDES Gym as part of my research in RL for secondary frequency control in grids with inverters, similar to Dr. Cui's Andes_gym paper but with REGF1 modules added, modifying REGF1 control values in order to implement AGC.

AGC works successfully for TGOV1N via paux0 as per the ANDES examples. My question is: how do I change Pref or Paux signals to REGF1 such as in scheduling and in AGC control signals?

My understanding is that REGF1 is supposed to accept external AGC signals as per the EPRI specification stating it should support Paux/Qaux inputs. I have tried changing Paux, Pref, a few variables in the related PV, as well as tried changing Id0 as per your solution in this related question. and looked at the source code for inspiration. With every attempt, I re-inspected the values and they indeed successfully changed - but it did not seem to have an effect in the actual simulation.

I have placed some code snippets of things I've tried in the comments.

Thank you in advance for your help and time. ANDES has been fun to learn, and I hope I can get through this challenge and continue developing my model for a future publication.

Kind regards,

Talisha

[talisha-g]

More details:

I have attempted multiple approaches to inject AGC signals into REGF1, listed below for quick reference:

Method 1: Direct Pref modification using direct vector, alter() or set()

case.REGF1.Pref.v[regf1_idx] = new_value
case.REGF1.alter('Pref', regf1_idx, new_value)
case.REGF1.set(src='Pref', idx=regf1_device_idx, value=new_value, attr='v')

Result: Pref changes, but Pe (power output) unchanged

Method 2: Direct Paux modification
case.REGF1.Paux.v[regf1_idx] = new_value
Result: Blocked by algebraic equation e_str='0 - Paux' forcing Paux = 0

Method 3: Modify gammap to indirectly change Pref

cur_gammap = case.REGF1.gammap.v[regf1_idx]
p0s = case.REGF1.P0s.v[regf1_idx]
new_gammap = cur_gammap + additional_gammap
case.REGF1.gammap.v[regf1_idx] = new_gammap

Result: gammap changes but Pref and Pe unchanged

Method 4: Modify StaticGen linked to REGF1

gen_idx = case.REGF1.gen.v[regf1_idx]
case.StaticGen.set(src='p0', idx=gen_idx, value=cur + val, attr='v')

Result: No effect on REGF1 power output

Method 5: Hack via Id0 (ConstService) - inspired by REGCA1
Similar to the REGCA1 workaround where Ipcmd0_LVG.v[:] = desired_Pe / voltages, I attempted:

new_Id0 = new_pref / v_mag
case.REGF1.Id0.v[regf1_idx] = new_Id0

Result: Id0 changes successfully, but Pe still unchanged

All modifications persist but have zero effect on power output.

[cuihantao]

Hi, thank you for posting your questions and the steps you've tried. I will have some time after a few deadlines on 10/31. Meanwhile, you can trace how the power set point signal is set in the REGF1 model and modify the constant associated with that equation, which might look like P - Pref0 = 0 (pref0 is the constant in this example).

[talisha-g]

Hi Dr. Cui,

Thank you for your guidance. I believe I've found a solution for enabling external AGC signals in REGF1, but I'd like to confirm if this approach is correct or if there's a better way.

The changes I made:

In the REGF1 model, I modified the Paux and Qaux algebraic constraint from:

self.Paux = Algeb(v_str='0', e_str='Paux')
self.Qaux = Algeb(v_str='0', e_str='Qaux')

To:

self.Paux = Algeb(info='Auxiliary active power signal',
                  tex_name='P_{aux}',
                  v_str='Paux0',
                  e_str='Paux0 - Paux')
self.Qaux = Algeb(info='Auxiliary reactive power signal',
                  tex_name='Q_{aux}',
                  v_str='Qaux0',
                  e_str='Qaux0 - Qaux')

And updated the PI controller initialization:

self.PIplim = PIController(u='Pref + Paux - Psig_y',
                           kp=self.KPplim, ki=self.KIplim,
                           x0='Pref')

self.PIqlim = PIController(u='Qref + Qaux - Qsig_y',
                           kp=self.KPqlim, ki=self.KIqlim,
                           x0='Qref')

This appears to work - I can now successfully control REGF1 inverters via case.REGF1.Paux0.v[idx] = value for AGC, similar to TGOV1N's paux0.

Could you please confirm if this is the correct approach for implementing external AGC signals in REGF1? Are there any potential issues or a better method I should consider?

Thank you again for your time and help!

Kind regards,

Talisha

[cuihantao]

This is the correct approach! I'm glad you found it out. Just make sure that you are doing in-placement assignment to the "v" field of Paux0 when you update the values.

…

On Tue, Oct 28, 2025, 2:15 PM TG ***@***.***> wrote: Hi Dr. Cui, Thank you for your guidance. I believe I've found a solution for enabling external AGC signals in REGF1, but I'd like to confirm if this approach is correct or if there's a better way. The changes I made: In the REGF1 model, I modified the Paux and Qaux algebraic constraint from: self.Paux = Algeb(v_str='0', e_str='Paux') self.Qaux = Algeb(v_str='0', e_str='Qaux') To: self.Paux = Algeb(info='Auxiliary active power signal', tex_name='P_{aux}', v_str='Paux0', e_str='Paux0 - Paux') self.Qaux = Algeb(info='Auxiliary reactive power signal', tex_name='Q_{aux}', v_str='Qaux0', e_str='Qaux0 - Qaux') And updated the PI controller initialization: self.PIplim = PIController(u='Pref + Paux - Psig_y', kp=self.KPplim, ki=self.KIplim, x0='Pref') self.PIqlim = PIController(u='Qref + Qaux - Qsig_y', kp=self.KPqlim, ki=self.KIqlim, x0='Qref') This appears to work - I can now successfully control REGF1 inverters via case.REGF1.Paux0.v[idx] = value for AGC, similar to TGOV1N's paux0. Could you please confirm if this is the correct approach for implementing external AGC signals in REGF1? Are there any potential issues or a better method I should consider? Thank you again for your time and help! Kind regards, Talisha — Reply to this email directly, view it on GitHub <#621 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABSNZA3N3QAIAN6NXO7FLSL3Z6XE7AVCNFSM6AAAAACKCLFLQGVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTIOBQG4YTANA> . You are receiving this because you commented.Message ID: ***@***.***>