Implement start penalty functionality #291
Conversation
…f usage that each start should be equivalent to.
- Add (guessed) explanations for the start constraints. - Move the start variable constraints that define it out of the set_def_constant if statement. - FIx the penalty, original was invalid LLM slop.
…"free" if the load is already running - leads to oscillation.
|
Solves #261 |
…nd, including that the constant functionality has not been working properly.
|
Hi @werdnum, is this ready to merge? |
|
I think it works in the configuration file. I’ll have to check to make sure.
Will take a look at documentation.
— Andrew Garrett
…On Tue, 28 May 2024 at 23:11, David ***@***.***> wrote:
Hi @werdnum <https://github.com/werdnum>, is this ready to merge?
Probably some documentation needed somewhere.
From what I see this is only usable from the runtime parameters
dictionary, but not from the configuration file?
—
Reply to this email directly, view it on GitHub
<#291 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AACCFXV7AF5J4772D6EEAJDZER7BFAVCNFSM6AAAAABIEWH3ROVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCMZVGE4DGNZRG4>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
|
Ok, I'm merging. But from what I see this will only works when passing |
|
Hi, this was merged but not correctly reviewed I guess. Let's start with this code for This Same comment for the |
|
Hi David, thanks for taking a look and sorry I haven't got to the
documentation yet!
Happy to help you understand the code and of course if there's something
I've done that's hard to understand, please let me know how I can make it
better.
I am relatively confident in the basic functionality because of the
expanded unit tests, but there could be small bugs of course.
— Andrew Garrett
On Mon, 10 Jun 2024 at 05:34, David ***@***.***> wrote:
Hi, this was merged but not correctly reviewed I guess.
I was trying to update the documentation an stumbled on this.
There are some important things that I don't fully understand:
Let's start with this code for def_start_penalty:
# Add term penalizing each startup where configured
if ("def_start_penalty" in self.optim_conf and self.optim_conf["def_start_penalty"]):
for k in range(self.optim_conf["num_def_loads"]):
if (len(self.optim_conf["def_start_penalty"]) > k and self.optim_conf["def_start_penalty"][k]):
objective = objective + plp.lpSum(
-0.001 * self.timeStep * self.optim_conf["def_start_penalty"][k] * P_def_start[k][i] *\
unit_load_cost[i] * self.optim_conf['P_deferrable_nom'][k]
for i in set_I)
This self.optim_conf["def_start_penalty"] is supposed to be a list of
floats.
But it seems to me that you are treating this a list of bool.
For example here if ("def_start_penalty" in self.optim_conf and
self.optim_conf["def_start_penalty"]).
This is a common idiom in python - a nonzero value evaluates to true, a
zero value (or empty string, None, etc) evaluates to false. I'm checking
whether there is any value (so I don't just assume it's a list and crash if
there's nothing there). This is what's recommended by the Google style
guide (
https://google.github.io/styleguide/pyguide.html) but happy to follow your
lead if you have another style preference.
Another term I don't understand is len(self.optim_conf["def_start_penalty"])
> k. If that parameter is a list of floats with a single float for each
deferrable then when is this condition supposed to True?
Here I'm trying to check whether the list of start penalties is long enough
to have a value for the current one. I don't want to try to get a staff
penalty value for load 2 (3rd in the list) if there are only 2 items in the
list. Len() returns the number of values, and k is the zero based index of
the deferrable load. So if there are 2 deferrable loads, K could be 0 or 1.
And again this is treating self.optim_conf["def_start_penalty"][k] as if
it were a bool.
Right, I don't want to do any processing for start penalty if it's set to
zero.
Same comment for the def_current_state parameter.
This condition can never hold if that parameter is a list of bool with a
single bool for each deferrable load: if ("def_current_state" in
self.optim_conf and len(self.optim_conf["def_current_state"]) > k):
Remember K will take all values from 0 to num_deferrable_loads exclusive,
so this will be false only when the def_current_state list if under sized.
…
I'm a bit lost here.
—
Reply to this email directly, view it on GitHub
<#291 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AACCFXTTMMWZHAP6RPLSWSTZGSU2XAVCNFSM6AAAAABIEWH3ROVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNJWG42TEMZVGQ>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
How to use:
def_start_penaltyto a list of floats. For each deferrable load with a penaltyP, each time the deferrable load turns on will incur an additional cost ofP•P_deferrable_nom•cost of electricity at that time. Intuitively, how many extra hours would the load have to run to avoid starting up again.def_current_stateinruntime_paramsto avoid incorrectly penalizing a start if a forecast is run when the load is already running.I also made a couple of other supporting changes:
test_optimization.py, plus one for theset_def_constantfunction. In doing so, I found a missing constraint forset_def_constantand added it to the optimizer (P_def_bin2was not constrained to be zero whenP_deferrablewas zero).