It would be reasonable to add minimum SOC to the battery model, at least with home batteries one is not recommended to fully discharge.

I am interested in using this project, but not clear what the license is. Could you add a license? Thank you.

Hello,

I am trying to install this package and seems there is an issue with contourpy package and its build process. The error message indicates that there is a problem parsing the output of vswhere.exe.
I am on Python3.10 and have just installed Visual Basic (didn't have it before).

Thanks.

Hi, I just wanted to check if it was ok for me to reference your code on
https://github.com/openclimatefix/solar-and-storage.

This is a repo for modelling batteries, using cxvpy and has the option of adding a solar panel.

I feed in some simple prices and simple battery setup and I'm confused about the output.

I'm using prices of [100, 200] so I expect the optimal battery profile would be to import at full capacity (1MW) in the first period and then export at full capacity in the second period.

Instead, see below, it's importing at 1MW in period[0] but exporting at 0.90909091MW in period[1]. Is there an inefficiency being modelled in the battery?

Finally, the charge state at the end is 0.0 - which is either 10% loss of efficiency??

>>> import energypylinear as epl
>>> model = epl.Battery(power=1, capacity=2)
>>> info = model.optimize([100, 200, 0])
>>> pprint(info)
[{'Actual [$/5min]': 8.333333333333334,
  'Charge [MWh]': 0.0,
  'Export [MW]': 0.0,
  'Forecast [$/5min]': 8.333333333333334,
  'Forecast [$/MWh]': 100,
  'Gross [MW]': 1.0,
  'Import [MW]': 1.0,
  'Losses [MW]': 0.0,
  'Net [MW]': 1.0,
  'Prices [$/MWh]': 100},
 {'Actual [$/5min]': -13.63636365,
  'Charge [MWh]': 0.083333333,
  'Export [MW]': 0.90909091,
  'Forecast [$/5min]': -13.63636365,
  'Forecast [$/MWh]': 200,
  'Gross [MW]': -0.90909091,
  'Import [MW]': 0.0,
  'Losses [MW]': 0.090909091,
  'Net [MW]': -0.818181819,
  'Prices [$/MWh]': 200},
 {'Actual [$/5min]': None,
  'Charge [MWh]': 0.0,
  'Export [MW]': None,
  'Forecast [$/5min]': None,
  'Forecast [$/MWh]': 0,
  'Gross [MW]': None,
  'Import [MW]': None,
  'Losses [MW]': None,
  'Net [MW]': None,
  'Prices [$/MWh]': 0}]

Following along with the README and I get:

info = model.optimize(prices, forecast=forecasts, initial_charge=4)

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: optimize() got an unexpected keyword argument 'forecast'

I may miss something, but I feel like the power constraint are not considered in the current version.
I would add these constraints:
self.prob += imports[i] <= self.power
self.prob += export[i] - losses[i] <= self.power

So, the current output is an ordered list of Dicts. In this list, objects [0:-2] describe the periods; the starting charge, the import/export operation throughout the period and so on. Then the final object in the list is largely null as it doesn't have import/export throughout the period, and just describes the end state of the battery.

It might be more useful if:

1. The operations and the end state were called out into separate dictionaries
2. The operations were indexed/keyed by the period ID

This would result in something like (some keys removed for brevity, but the structure is there):

{
    "operations": {
        "0": {
            "Import [MW]": 2.0,
            "Export [MW]": 0.0,
            "Charge [MWh]": 0.0
        },
        "1": {
            "Import [MW]": 0.0,
            "Export [MW]": 2.0,
            "Charge [MWh]": 1.0
        }

    },
    "final-state": {
        "Charge [MWh]": 0.0
    }
}

This allows for a more explicit expression of what each part of the output is, and allows for extension of the output without breaking backwards compatibility. For example, the summary could be extended to contain price $/MWh across the whole run, or each operational period could have additional values like DateTimeZone if actual times are being used. Or, the output could also be extended to output something other than a final-state and a list of operations, maybe confirmation of the Battery config, or ...

The point is, enclosing output into a named dictionary key, allows for this expansion without breaking existing setups.

Hello

My name is Andreas Slyngstad, and I am a software developer, currently working on a business case for batteries in Germany.

The business case is based around price arbitrage, and of the limits for this business case is number of battery cycles performed in a given simulation period.
Here, a battery cycle includes discharge -> charge for some predefined limit of depth of discharge.

My question is:
Is there any possible way to enforce the MILP solver constrain the problem to only be able to do
a maximum of n number of cycles, lets say max 15 for some simulation period.

Example Battery specs

• Power/Energy 1MW/2MWh
• Efficiency: 98%

If possible (but not critical), is there any way to set a min/max range of SOC (state of charge) ?
Lets say state of charge can be in the range of 10% - 90%.

Best regards

Andreas Slyngstad

Hello,

I am trying to calculate battery dispatch simulations for my battery and I notice some problems with the SOC calculations. After carrying out the calculations for yearly data I see that SOC column is always empty. And when I calculate it myself I see that SOC reaches over the batteries maximum capacity. Wouldn't it be best for the battery to work in a daily format? Where the batteries capacity at the end of the day should be at 10% SOC.

And I have encountered rows where charging and discharging occurs, but if battery is charging at full capacity of the inverter then there is no room for the battery to discharge. (uploading example screenshot) The battery is 4 MWh capacity and the power is 3.2MW initial and final charge variables are set to 0) But in the battery object I input 3.6 MWh capacity, because it is stated by the producers of the battery that it should never reach below 10% SOC levels. So after the calculations I add to the SOC column 0.4 MWh thus achieving full battery charge.

When feeding in a forecast as well as a price we get an assertion error:

>>> import energypylinear as epl
>>> model = epl.Battery(power=2, capacity=4)
>>> prices = [10, 50, 10, 50, 10]
>>> forecasts = [10, 50, 10, 50, 10]
>>> model.optimize(prices, timestep='30min', forecasts=forecasts)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/thomasharvey/.pyenv/versions/linear_value/lib/python3.6/site-packages/energypylinear/battery/battery.py", line 87, in optimize
    assert len(forecasts) == len(prices)
AssertionError

At download and install with pip git+, I get two errors, I think related to contourpy (I work on Python 3.10):

1. ERROR: could not register module: No module named 'wrapt', even though I have wrapt==0.15.0 installed
2. meson-python: error: Could not find meson version 0.63.3 or newer, found pip_system_certs: ERROR: could not register module: No module named 'wrapt'
   meson-python: error: 1.2.1
   meson-python: error:

, even though I have meson-python==0.12.0 and meson==1.2.1 installed

Do you have any suggestions?

I'm now seeing failures in test_power_capacity_initial_charge():

When input params are [2, 4, 0]:

(Pdb) gross
[2.0, 2.0, 1.4545455, 2.0, 2.0, 2.0]
(Pdb) net
[2.0, 2.0, 1.60000005, 2.2, 2.0, 2.2]

And in test_batt_efficiency.

E       First has 1, Second has 0:  -1.0
E       First has 0, Second has 1:  -0.33333334000000003

Are you seeing failures as well - these are fairly new tests. I'd dig into it, but I think it's in the workings of the solution.

If I input prices of [100, 200] I expect the results to inform me of how the battery should operate in each of these 2 price periods.

Instead, it shows one period of valid import/export instruction and then the end state with summary of end charge and null import/export.

If I add a third price, I get the expected results for the 2 periods and the results are the same regardless of the 3rd price fed into the solver.

Here https://github.com/ADGEfficiency/energy-py-linear/blob/master/energypylinear/battery/battery.py#L29 the battery is defaulting to a 90% round trip efficiency.

This isn't clear in the docs.

Considering what would be best for quick off-the-shelf use of the library; would a default 100% efficiency be better and/or to make it clear that efficiency is built into the battery model?

Just creating a ticket for this task :)

• Create a PyPi account
• Add a LICENSE.txt
• Add a build system (circleCI or Jenkins)
• Profit! 🎉

See https://medium.com/@joel.barmettler/how-to-upload-your-python-package-to-pypi-65edc5fe9c56 for some guidance.

I am trying to apply the library for home battery optimization scenario. To take hourly prices of electricity and get charge/discharge schedule.

It would be reasonable to include cost of charge/discharge cycle to optimization model. So the model would not charge/discharge if the cost of cycle would be higher than the profit from charge.discharge.

I'm running energy-py-linear in a container with a read-only file system. I've noticed that the library fails in a read-only file system because of this section of code:

Current Behaviour

When using a read-only file system, the library fails as it attemps to create a directory called logs even when enable_file_logging is set to false.

Suggested Fix

Move the pathlib.Path("logs").mkdir(exist_ok=True) line into the conditional for enable_file_logging.

@johnfeldhausen - I've started work on bringing the custom objective function terms into epl.Accounts here #65.

The initial idea is something like:

simulation = site.optimize(verbose=4, objective={"terms": terms})
accounts = epl.get_accounts(simulation.results, terms=terms)
print(accounts.custom.cost)

Will see how it goes!

pip install throws with Python 3.11 throws an error:

0.1.0 Requires-Python >=3.10, <3.11

Is the code really dependent on such narrow Python versioning?