Trouble with indexing in Gurobi

Answered

Maximilian Sletbakk

• October 17, 2022 11:10

Hello!
I am trying to optimize a "microgrid" with regards to cost, the model will take a fixed demand, a flexible demand, spot price for electricity, along with fixed production data for a solar power system. Additionally power can be stored in a battery to be discharged during optimal times, as well as charged. The time period is 48 hours.

I have most of this defined and working, but I am struggling with getting the battery to work, I have tried different ways of defining it, but i am getting keyerrors due to my indexing:
Here I have tried to define the battery as a constraint, the charge and discharge models seem to work, they both have values for the 48 time-steps in model t. I have also tried to define it using the same formula, except using batterylevel[t] = everything else of [t-1]. Sadly none of this works. How can I implement this? Any help would be greatly appreciated!

3

• 

  Jaromił Najman

  Gurobi Staff

  • October 17, 2022 12:15

  Hi Maximilian,

  Could you please provide a minimal reproducible example showing the issue? This means that in your case, you would need to at least define 3 variables with an index t. Could you also show the error message you get?

  Best regards, 
  Jaromił

  2
• 

  Maximilian Sletbakk

  • October 17, 2022 13:29

  

  

  

  Thank you for your reply!
  So my variables are:The objective function is to minimize the sum of net_transformer_load (pictured below) multiplied by the spot price.
  There are also many more constraints

  

   

  The error message is the one below, probably refering to the first picture I sent, also attached below

  

  

   

   

  2
• 

  Jaromił Najman

  Gurobi Staff

  • October 17, 2022 13:45

  Please note that posting screenshots is not a good way to provide a reproducible example. In particular, the value of \(\texttt{model.t}\) is missing.

  Anyway, I suppose that the maximum value for \(\texttt{model.t}\) is 47 and you are trying to execute

  battery_test_rule(model,47)

  which leads to to the code trying to access

  model.batteryLevel[48]

  which is not defined.

  1
• 

  Maximilian Sletbakk

  • October 17, 2022 13:58

  

  

  

  # Set representing the time steps, 48 hours
  model.t = Set(initialize=np.arange(0, 48), ordered=True)
  # Set representing the time steps until all vehicles must be fully charged, 32 hours
  model.c = Set(initialize = np.arange(0,32), ordered=True)

  The model c refers to vehicles, this is the flexible demand that I mentioned previously. 
  So yes it is trying to access a value that it is not indexed for I suppose. I am wondering how I can fix this, as I need the value of the battery level for the previous time-step to carry over to the beginning of the next time step. The usual way I would solve this with an array does not seem to work inside the model.

  1
• 

  Jaromił Najman

  Gurobi Staff

  • October 17, 2022 14:18

  You could just call the \(\texttt{battery_test_rule}\) method only for \(t < 47\).

  0
• 

  Maximilian Sletbakk

  • October 17, 2022 14:25

  

  

  

  What is the best way to express this in the constraint format? 

  0
• 

  Jaromił Najman

  Gurobi Staff

  • October 17, 2022 14:47

  What is the best way to express this in the constraint format? 

  Could you elaborate what exactly you mean?

  You could go with

  if t < 47:
    model.battery_level2 = Constraint(model.t, rule = battery_test_rule)
  else
    model.battery_level2 = None

  0
• 

  Maximilian Sletbakk

  • October 17, 2022 15:15

  

  

  

  Thanks! That fixed the keyerror! 
  However my 

  model.battery_level

  values do not change, they stay at 1000  which is the upper bound
  
  

   def battery_level_rule(model,t):                                             
          return aml.inequality(0,model.batteryLevel[t],1000)
      model.battery_level = Constraint(model.t, rule = battery_level_rule)

  How can I assign the variable 

   model.batteryLevel = Var(model.t, within = NonNegativeReals)

  to be a result of other variables? I suppose my constraint

   def battery_test_rule(model,t):
         return model.batteryLevel[t+1] == model.batteryLevel[t]+model.batteryCharge[t]-model.batteryDischarge[t]
         if t < 47:
             model.battery_level = Constraint(model.t, rule = battery_test_rule)
         else:
             model.battery_level = None

  Is just checking if the equality is true, so how can I get 

  model.batteryLevel

  to update according to 

  model.batteryLevel[t+1] = model.batteryLevel[t]+model.batteryCharge[t]-model.batteryDischarge[t]

  Thanks in advance!

  0
• 

  Jaromił Najman

  Gurobi Staff

  • October 17, 2022 15:28

  Is just checking if the equality is true, so how can I get 

  If you add the constraint to your model, it is not "just checking" whether the equality holds. When the constraint is added to your model, every feasible solution point has to satisfy this constraint.

  I suggest, you write your model to an LP file via

  model.write("myLP.lp")

  This will generate a file called \(\texttt{myLP.lp}\) which you can inspect and check whether all constraints look as you expect them to.

   

  0
• 

  Maximilian Sletbakk

  • October 17, 2022 15:58

  

  

  

  Okay, I am suspecting that the errror is due to the fact that the "battery" can both charge and discharge at the same time, so I will try to fix that
  Thank you very much for your time!

  0

Please sign in to leave a comment.