Spaces:
Sleeping
Sleeping
Daniel Varga
commited on
Commit
•
e8def5c
1
Parent(s):
e733d30
copied supplier
Browse files- v2/architecture.py +4 -12
- v2/supplier.py +0 -1
- v2/supplier.py +136 -0
v2/architecture.py
CHANGED
@@ -13,10 +13,12 @@ STEPS_PER_HOUR = 12
|
|
13 |
|
14 |
# SOC is normalized so that minimal_depth_of_discharge = 0 and maximal_depth_of_discharge = 1.
|
15 |
# please set capacity_Ah = nominal_capacity_Ah * (max_dod - min_dod)
|
|
|
|
|
16 |
class BatteryModel:
|
17 |
def __init__(self, capacity_Ah, time_interval_h):
|
18 |
self.capacity_Ah = capacity_Ah
|
19 |
-
self.efficiency = 0.
|
20 |
self.voltage_V = 600
|
21 |
self.charge_kW = 50
|
22 |
self.discharge_kW = 60
|
@@ -94,7 +96,6 @@ class Decision(IntEnum):
|
|
94 |
class Decider:
|
95 |
def __init__(self):
|
96 |
self.input_window_size = STEPS_PER_HOUR * 24 # day long window.
|
97 |
-
self.output_window_size = STEPS_PER_HOUR # only output decisions for the next hour
|
98 |
self.random_seed = 0
|
99 |
|
100 |
# prod_cons_pred is a dataframe starting at now, containing
|
@@ -123,16 +124,6 @@ class DummyPredictor:
|
|
123 |
return prediction
|
124 |
|
125 |
|
126 |
-
'''
|
127 |
-
bess_nominal_capacity: float = 330 # [Ah]
|
128 |
-
bess_charge: float = 50 # [kW]
|
129 |
-
bess_discharge: float = 60 # [kW]
|
130 |
-
voltage: float = 600 # [V]
|
131 |
-
maximal_depth_of_discharge: float = 0.75 # [dimensionless]
|
132 |
-
energy_loss: float = 0.1 # [dimensionless]
|
133 |
-
bess_present: bool = True # [boolean]
|
134 |
-
'''
|
135 |
-
|
136 |
# this function does not mutate its inputs.
|
137 |
# it makes a clone of battery_model and modifies that.
|
138 |
def simulator(battery_model, supplier, prod_cons, prod_predictor, cons_predictor, decider):
|
@@ -215,6 +206,7 @@ def simulator(battery_model, supplier, prod_cons, prod_predictor, cons_predictor
|
|
215 |
else:
|
216 |
consumption_from_network_to_bess = 0
|
217 |
|
|
|
218 |
soc_series.append(battery_model.soc)
|
219 |
consumption_from_solar_series.append(consumption_from_solar)
|
220 |
consumption_from_network_series.append(consumption_from_network)
|
|
|
13 |
|
14 |
# SOC is normalized so that minimal_depth_of_discharge = 0 and maximal_depth_of_discharge = 1.
|
15 |
# please set capacity_Ah = nominal_capacity_Ah * (max_dod - min_dod)
|
16 |
+
#
|
17 |
+
# TODO efficiency multiplier is not currently used, where best to put it?
|
18 |
class BatteryModel:
|
19 |
def __init__(self, capacity_Ah, time_interval_h):
|
20 |
self.capacity_Ah = capacity_Ah
|
21 |
+
self.efficiency = 0.9 # [dimensionless]
|
22 |
self.voltage_V = 600
|
23 |
self.charge_kW = 50
|
24 |
self.discharge_kW = 60
|
|
|
96 |
class Decider:
|
97 |
def __init__(self):
|
98 |
self.input_window_size = STEPS_PER_HOUR * 24 # day long window.
|
|
|
99 |
self.random_seed = 0
|
100 |
|
101 |
# prod_cons_pred is a dataframe starting at now, containing
|
|
|
124 |
return prediction
|
125 |
|
126 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
127 |
# this function does not mutate its inputs.
|
128 |
# it makes a clone of battery_model and modifies that.
|
129 |
def simulator(battery_model, supplier, prod_cons, prod_predictor, cons_predictor, decider):
|
|
|
206 |
else:
|
207 |
consumption_from_network_to_bess = 0
|
208 |
|
209 |
+
supplier.(consumption_from_network)
|
210 |
soc_series.append(battery_model.soc)
|
211 |
consumption_from_solar_series.append(consumption_from_solar)
|
212 |
consumption_from_network_series.append(consumption_from_network)
|
v2/supplier.py
DELETED
@@ -1 +0,0 @@
|
|
1 |
-
../supplier.py
|
|
|
|
v2/supplier.py
ADDED
@@ -0,0 +1,136 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# modeling an energy supplier for the purposes of peak shaving
|
2 |
+
|
3 |
+
import numpy as np
|
4 |
+
import pandas as pd
|
5 |
+
import datetime
|
6 |
+
import unittest
|
7 |
+
|
8 |
+
|
9 |
+
class Supplier:
|
10 |
+
# price [HUF/kWh]
|
11 |
+
# peak_demand kW
|
12 |
+
# surcharge_per_kw [HUF/kW for each 15 minute timeframe]
|
13 |
+
def __init__(self, price):
|
14 |
+
self.hourly_prices = np.ones(168) * price
|
15 |
+
self.peak_demand = np.inf # no demand_charge by default
|
16 |
+
self.surcharge_per_kw = 0
|
17 |
+
|
18 |
+
# start and end are indices of hours starting from Monday 00:00.
|
19 |
+
def set_price_for_interval(self, start, end, price):
|
20 |
+
self.hourly_prices[start:end] = price
|
21 |
+
|
22 |
+
# start and end are indices of hours of the day. for each day, this interval is set to price
|
23 |
+
def set_price_for_daily_interval(self, start, end, price):
|
24 |
+
for day in range(7):
|
25 |
+
h = day * 24
|
26 |
+
self.set_price_for_interval(h + start, h + end, price)
|
27 |
+
|
28 |
+
def set_price_for_daily_interval_on_workdays(self, start, end, price):
|
29 |
+
for day in range(5):
|
30 |
+
h = day * 24
|
31 |
+
self.set_price_for_interval(h + start, h + end, price)
|
32 |
+
|
33 |
+
def set_demand_charge(self, peak_demand, surcharge_per_kw):
|
34 |
+
self.peak_demand = peak_demand # [kW]
|
35 |
+
# the HUF charged per kW of demand exceeding peak_demand during a 15 minutes timeframe.
|
36 |
+
self.surcharge_per_kw = surcharge_per_kw # [HUF/kW]
|
37 |
+
|
38 |
+
@staticmethod
|
39 |
+
def hour_of_date(date):
|
40 |
+
hours_since_midnight = (date - datetime.datetime(date.year, date.month, date.day, 0, 0, 0)).total_seconds() / 3600
|
41 |
+
# weekday() calculates from sunday morning:
|
42 |
+
hungarian_weekday = (date.weekday() + 0) % 7
|
43 |
+
hours_elapsed_in_previous_days = hungarian_weekday * 24
|
44 |
+
return int(hours_since_midnight) + hours_elapsed_in_previous_days
|
45 |
+
|
46 |
+
def price(self, date):
|
47 |
+
return self.hourly_prices[self.hour_of_date(date)]
|
48 |
+
|
49 |
+
# demand is the maximum demand in kW during a 15 minute interval
|
50 |
+
def demand_charge(self, demand):
|
51 |
+
if demand <= self.peak_demand:
|
52 |
+
return 0.0
|
53 |
+
else:
|
54 |
+
return (demand - self.peak_demand) * self.surcharge_per_kw
|
55 |
+
|
56 |
+
# demand_series is pandas series indexed by time.
|
57 |
+
# during each time step demand [kW] is assumed to be constant.
|
58 |
+
def fee(self, demand_series):
|
59 |
+
prices = [self.price(date) for date in demand_series.index]
|
60 |
+
prices_series = pd.Series(data=prices, index=demand_series.index)
|
61 |
+
# prices are HUF/kWh, demand is kW. note the missing h.
|
62 |
+
|
63 |
+
step_in_hour = demand_series.index.freq.n / 60 # [hour], the length of a time step.
|
64 |
+
# for each step the product tells the fee IF the step was 1 hour long. it's actually step_in_hour long:
|
65 |
+
consumption_charge = demand_series.dot(prices_series) * step_in_hour
|
66 |
+
|
67 |
+
# 15 minutes (the demand charge calculation interval) should be a multiple of the series time step.
|
68 |
+
assert 15 % demand_series.index.freq.n == 0
|
69 |
+
time_steps_per_demand_charge_evaluation = 15 // demand_series.index.freq.n
|
70 |
+
# fifteen_minute_peaks [kW] tells the maximum demand in a 15 minutes timeframe:
|
71 |
+
fifteen_minute_peaks = demand_series.resample('15T').max()
|
72 |
+
demand_charges = [self.demand_charge(demand) for demand in fifteen_minute_peaks]
|
73 |
+
total_demand_charge = sum(demand_charges)
|
74 |
+
return consumption_charge + total_demand_charge
|
75 |
+
|
76 |
+
|
77 |
+
class TestSupplier(unittest.TestCase):
|
78 |
+
|
79 |
+
def setUp(self):
|
80 |
+
self.constant_price = 10
|
81 |
+
self.supplier = Supplier(self.constant_price)
|
82 |
+
|
83 |
+
def test_hourly_prices(self):
|
84 |
+
expected_hourly_prices = np.ones(168) * self.constant_price
|
85 |
+
self.assertTrue(np.array_equal(self.supplier.hourly_prices, expected_hourly_prices))
|
86 |
+
|
87 |
+
def test_set_price_for_interval(self):
|
88 |
+
self.supplier.set_price_for_interval(0, 24, 20)
|
89 |
+
expected_hourly_prices = np.ones(168) * self.constant_price
|
90 |
+
expected_hourly_prices[0:24] = 20
|
91 |
+
self.assertTrue(np.array_equal(self.supplier.hourly_prices, expected_hourly_prices))
|
92 |
+
|
93 |
+
def test_price(self):
|
94 |
+
increased_price = 20
|
95 |
+
self.supplier.set_price_for_interval(0, 24, increased_price)
|
96 |
+
|
97 |
+
date = datetime.datetime(2023, 4, 30, 12, 0, 0) # Sunday noon
|
98 |
+
expected_price = self.constant_price
|
99 |
+
self.assertEqual(self.supplier.price(date), expected_price)
|
100 |
+
|
101 |
+
date = datetime.datetime(2023, 5, 1, 12, 0, 0) # Monday noon
|
102 |
+
expected_price = increased_price
|
103 |
+
self.assertEqual(self.supplier.price(date), expected_price)
|
104 |
+
|
105 |
+
date = datetime.datetime(2023, 5, 2, 12, 0, 0) # Tuesday noon
|
106 |
+
expected_price = self.constant_price
|
107 |
+
self.assertEqual(self.supplier.price(date), expected_price)
|
108 |
+
|
109 |
+
def test_fee(self):
|
110 |
+
start = pd.Timestamp('2021-04-28')
|
111 |
+
end = start + pd.Timedelta(days=1)
|
112 |
+
freq = '5T' # 5 minutes
|
113 |
+
time_index = pd.date_range(start=start, end=end, freq=freq, inclusive='left')
|
114 |
+
constant_demand = 100
|
115 |
+
demand_in_kw = [constant_demand] * len(time_index)
|
116 |
+
|
117 |
+
demand_series = pd.Series(data=demand_in_kw, index=time_index)
|
118 |
+
# 24 because it's a 24 hour period with constant demand:
|
119 |
+
self.assertEqual(self.supplier.fee(demand_series), constant_demand * 24 * self.constant_price)
|
120 |
+
|
121 |
+
extreme_demand = 1000
|
122 |
+
demand_series[12:24] = extreme_demand # in second hour we set extreme demand.
|
123 |
+
|
124 |
+
expected_fee = (constant_demand * 23 + extreme_demand) * self.constant_price
|
125 |
+
self.assertEqual(self.supplier.fee(demand_series), expected_fee)
|
126 |
+
|
127 |
+
# now the (1000-500) kW above 500 kW is surcharged for (1000-500 kW) * 10 HUF/kW/15mins, for 1 hour,
|
128 |
+
# that is 500*10*4=20000 demand_charge.
|
129 |
+
self.supplier.set_demand_charge(peak_demand=500, surcharge_per_kw=10)
|
130 |
+
expected_fee += 20000
|
131 |
+
self.assertEqual(self.supplier.fee(demand_series), expected_fee)
|
132 |
+
|
133 |
+
|
134 |
+
|
135 |
+
if __name__ == '__main__':
|
136 |
+
unittest.main()
|