Re: DCOPF not converging for large system

[Ray Zimmerman]

Is the initial power flow feasible? My guess is that the initial power flow is 
violating one of the generator limits, so when you enforce them with the OPF, 
the solution is no longer feasible.

   Ray

> On Aug 8, 2017, at 3:45 AM, prem panigrahi <prempanigr...@outlook.com> wrote:
> 
> 
> Dear all
>I am facing problem in DCOPF. I need help. In my power grid model 
> i have fixed the capacity of transmission line in RATE A column by 110% of 
> initial power flow. Then increase the load by a factor (1.01) and run the 
> dcopf. But matpower showing infeasible model due to which it is not 
> converging. I am not able to understand the problem. Here load and generation 
> is also well balanced. Now which parameter constraint to converging of my 
> test case. I stuck here for a long time, so please help me
> 
> 
> 
> The result shown below is after increasing the load by 1.01. 
> 
>  
> MATPOWER Version 6.0, 16-Dec-2016 -- DC Optimal Power Flow
> Gurobi Version 7.0.2 -- automatic LP solver
> Optimize a model with 19515 rows, 4968 columns and 47500 nonzeros
> Coefficient statistics:
>   Matrix range [2e-03, 9e+05]
>   Objective range  [1e+02, 2e+02]
>   Bounds range [5e-03, 1e+02]
>   RHS range[3e-17, 4e+02]
> 
> Concurrent LP optimizer: dual simplex and barrier
> Showing barrier log only...
> 
> Presolve removed 2 rows and 17 columns
> Presolve time: 0.06s
> 
> Solved with dual simplex
> Solved in 0 iterations and 0.07 seconds
> Infeasible model
> 
> >>>>>  Did NOT converge (0.17 seconds)  <<<<<
> 
> 
>  
> 
>  
> Checking connectivity ... single fully connected network
> Elapsed time is 0.051284 seconds.
> 
> Full
>System   
> Number of:   -- 
>   buses  3989   
>   loads  2227   
> on   2227   
> off -   
> fixed2227   
> dispatchable-   
>   on-   
>   off   -   
>   generators  979   
> on979   
> off -   
>   shunt elements  247   
>   branches   8096   
> on   8096   
> off -   
> ties (off)  -   
> 
> Load
>   active (MW)   
> dispatched  85879.4 
>   fixed 85879.4 
>   dispatchable  -   
> nominal 85879.4 
>   on85879.4 
>   off   -   
>   fixed 85879.4 
>   dispatchable  -   
> on  -   
> off -   
>   reactive (MVAr)   
> dispatched  23668.6 
>   fixed 23668.6 
>   dispatchable  -   
> nominal 23668.6 
>   on23668.6 
>   off   -   
>   fixed 23668.6 
>   dispatchable  -   
> on  -   
> off -   
> 
> Generation  
>   active (MW)   
> dispatched  86451.8 
> max capacity   213706.7 
>   on   213706.7 
>   off   -   
> min capacity   -72665.3 
>   on   -72665.3 
>   off   -   
>   reactive (MVAr)   
> dispatched  24069.5 
> max capacity69809.7 
>   on69809.7 
>   off   -   
> min capacity   -50352.2 
>   on   -50352.2 
>   off   -   
> 
> Shunt Injections
> active (MW) -   
> reactive (MVAr) 10848.8 
> 
> Branch Losses   
> active (MW) -   
> reactive (MVAr) -   
> 
> DC line 
>   export (MW)   
> dispatch-   
> max capacity-   
>   on-   
>   off   -   
> min capacity-   
>   on        -   
>   off   -   
> 
> Reference Buses 
>   num of ref buses  1   
>   ref bus numbers3292   
> 
> 
> From: bounce-121637002-78651...@list.cornell.edu 
> <bounce-121637002-78651...@list.cornell.edu> on behalf of Ray Zimmerman 
> <r...@cornell.edu>
> Sent: Monday, July 3, 2017 9:33:26 PM
> To: MATPOWER discussion

Re: DCOPF not converging for large system

[prem panigrahi]

Dear all
   I am facing problem in DCOPF. I need help. In my power grid model i 
have fixed the capacity of transmission line in RATE A column by 110% of 
initial power flow. Then increase the load by a factor (1.01) and run the 
dcopf. But matpower showing infeasible model due to which it is not converging. 
I am not able to understand the problem. Here load and generation is also well 
balanced. Now which parameter constraint to converging of my test case. I stuck 
here for a long time, so please help me



The result shown below is after increasing the load by 1.01.


MATPOWER Version 6.0, 16-Dec-2016 -- DC Optimal Power Flow
Gurobi Version 7.0.2 -- automatic LP solver
Optimize a model with 19515 rows, 4968 columns and 47500 nonzeros
Coefficient statistics:
  Matrix range [2e-03, 9e+05]
  Objective range  [1e+02, 2e+02]
  Bounds range [5e-03, 1e+02]
  RHS range[3e-17, 4e+02]

Concurrent LP optimizer: dual simplex and barrier
Showing barrier log only...

Presolve removed 2 rows and 17 columns
Presolve time: 0.06s

Solved with dual simplex
Solved in 0 iterations and 0.07 seconds
Infeasible model

>>>>>  Did NOT converge (0.17 seconds)  <<<<<





Checking connectivity ... single fully connected network
Elapsed time is 0.051284 seconds.

Full
   System
Number of:   --
  buses  3989
  loads  2227
on   2227
off -
fixed2227
dispatchable-
  on-
  off   -
  generators  979
on979
off -
  shunt elements  247
  branches   8096
on   8096
off -
ties (off)  -

Load
  active (MW)
dispatched  85879.4
  fixed 85879.4
  dispatchable  -
nominal 85879.4
  on85879.4
  off   -
  fixed 85879.4
  dispatchable  -
on  -
off -
  reactive (MVAr)
dispatched  23668.6
  fixed 23668.6
  dispatchable  -
nominal 23668.6
  on23668.6
  off   -
  fixed 23668.6
  dispatchable  -
on  -
off -

Generation
  active (MW)
dispatched  86451.8
max capacity   213706.7
  on   213706.7
  off   -
min capacity   -72665.3
  on   -72665.3
  off   -
  reactive (MVAr)
dispatched  24069.5
max capacity69809.7
  on69809.7
  off   -
min capacity   -50352.2
  on   -50352.2
  off   -

Shunt Injections
active (MW) -
reactive (MVAr) 10848.8

Branch Losses
active (MW) -
reactive (MVAr) -

DC line
  export (MW)
dispatch-
max capacity-
  on-
  off   -
min capacity-
  on-
  off   -

Reference Buses
  num of ref buses  1
  ref bus numbers3292



From: bounce-121637002-78651...@list.cornell.edu 
<bounce-121637002-78651...@list.cornell.edu> on behalf of Ray Zimmerman 
<r...@cornell.edu>
Sent: Monday, July 3, 2017 9:33:26 PM
To: MATPOWER discussion forum
Subject: Re: DCOPF not converging for large system

If you have a solved AC power flow, that’s a good start. Normally, the first 
step I try when faced with an OPF that won’t converge is to eliminate the 
branch flow limits by setting the RATE_A column of the branch matrix to zero 
(or by increasing the values by some factor). In your case, you can check which 
flows in the power flow solution exceed their limits and only relax those to 
begin with.

If I were you, I would relax any branch, voltage and generator limits as 
required to make the power flow solution you have feasible for the AC OPF 
problem. It should solve. If it does, then you can begin to eliminate those 
relaxations to find the source of the infeasibility.

   Ray


On Jun 23, 2017, at 10:55 AM, Mort D. Webster 
<md...@psu.edu<mailto:md...@psu.edu>> wrote:


Dear Matpower Folks:

I have larger network that is a solved load flow case from an ISO.

The AC power flow converges fine, but neither AC nor DC OPF solve.

I supplied my own estimated generator costs, and perturbed them so that none 
are identical.

I get the following error:


>> mpc = iso2015sum

mpc =

 

Re: DCOPF not converging for large system

[Ray Zimmerman]

If you have a solved AC power flow, that’s a good start. Normally, the first 
step I try when faced with an OPF that won’t converge is to eliminate the 
branch flow limits by setting the RATE_A column of the branch matrix to zero 
(or by increasing the values by some factor). In your case, you can check which 
flows in the power flow solution exceed their limits and only relax those to 
begin with.

If I were you, I would relax any branch, voltage and generator limits as 
required to make the power flow solution you have feasible for the AC OPF 
problem. It should solve. If it does, then you can begin to eliminate those 
relaxations to find the source of the infeasibility.

   Ray


> On Jun 23, 2017, at 10:55 AM, Mort D. Webster  wrote:
> 
> Dear Matpower Folks: 
> 
> I have larger network that is a solved load flow case from an ISO. 
> 
> The AC power flow converges fine, but neither AC nor DC OPF solve. 
> 
> I supplied my own estimated generator costs, and perturbed them so that none 
> are identical. 
> 
> I get the following error: 
> 
>> >> mpc = iso2015sum 
>> 
>> mpc = 
>> 
>>  version: '2' 
>>  baseMVA: 100 
>>  bus: [7108x13 double] 
>>  gen: [720x25 double] 
>>   branch: [8930x13 double] 
>> bus_name: {7108x1 cell} 
>>   dcline: [4x17 double] 
>>  gencost: [720x6 double] 
>> 
>> >> results = rundcopf(mpc) 
>> 
>> opf_setup: Warning: Multiple reference buses. 
>>For a system with islands, a reference bus in each island 
>>may help convergence, but in a fully connected system such 
>>a situation is probably not reasonable. 
>> 
>> 
>> MATPOWER Version 6.0b2, 01-Nov-2016 -- DC Optimal Power Flow 
>> The interior-point-legacy algorithm uses a built-in starting point; 
>> ignoring supplied X0. 
>> Exiting: One or more of the residuals, duality gap, or total relative error 
>>  has stalled: 
>>  the primal appears to be infeasible and the dual unbounded since 
>>  the dual objective > 1e+10 
>>  and the primal objective > -1e+6. 
>> 
>> >  Did NOT converge (7.52 seconds) < 
>> 
>> 
>> results = 
>> 
>>  version: '2' 
>>  baseMVA: 100 
>>  bus: [7108x17 double] 
>>  gen: [720x25 double] 
>>   branch: [8930x21 double] 
>> bus_name: {7108x1 cell} 
>>   dcline: [4x17 double] 
>>  gencost: [720x6 double] 
>>order: [1x1 struct] 
>>   om: [1x1 opf_model] 
>>x: [7700x1 double] 
>>   mu: [1x1 struct] 
>>f: 2.1192e+06 
>>  var: [1x1 struct] 
>>  lin: [1x1 struct] 
>>   et: 7.5220 
>>  success: 0 
>>  raw: [1x1 struct] 
> 
> The system has several complex features: 
> 
> 4 HVDC lines 
> 
> several storage resources as generators 
> 
> 4 buses as reference/slack bus type 
> 
> many "isolated" bus types ("4"). 
> 
> Can you suggest where in the results I can look to diagnose the problem? 
> 
> Thanks so much! 
> 
> Mort Webster
> -- 
> Mort Webster
> Associate Professor of Energy Engineering
> Dept of Energy and Mineral Engineering
> Pennsylvania State University
> University Park, PA 16802
> (814) 863-1640
> m...@psu.edu