def test_dispatch1(dcline_net):
net = dcline_net
pp.create_piecewise_linear_cost(
net, 0, "ext_grid", array([[-1e12, -0.1 * 1e12], [1e12, .1 * 1e12]]))
pp.create_piecewise_linear_cost(
net, 1, "ext_grid", array([[-1e12, -0.08 * 1e12], [1e12, .08 * 1e12]]))
net.bus["max_vm_pu"] = 2
net.bus["min_vm_pu"] = 0 # needs to be constrained more than default
net.line[
"max_loading_percent"] = 1000 # does not converge if unconstrained
pp.runopp(net)
consistency_checks(net, rtol=1e-3)
rel_loss_expect = (net.res_dcline.pl_kw - net.dcline.loss_kw) / \
(net.res_dcline.p_from_kw - net.res_dcline.pl_kw) * 100
assert allclose(rel_loss_expect.values, net.dcline.loss_percent.values)
p_eg_expect = array([-5.00078353e+02, -8.05091476e+05])
q_eg_expect = array([7787.55773243, -628.30727889])
assert allclose(net.res_ext_grid.p_kw.values, p_eg_expect)
assert allclose(net.res_ext_grid.q_kvar.values, q_eg_expect)
p_from_expect = array([500.0754071])
q_from_expect = array([7787.45600524])
assert allclose(net.res_dcline.p_from_kw.values, p_from_expect)
assert allclose(net.res_dcline.q_from_kvar.values, q_from_expect)
p_to_expect = array([-0.0746605])
q_to_expect = array([-627.12636707])
assert allclose(net.res_dcline.p_to_kw.values, p_to_expect)
assert allclose(net.res_dcline.q_to_kvar.values, q_to_expect)
def test_get_costs():
""" Testing a very simple network for the resulting cost value
constraints with OPF """
# boundaries:
vm_max = 1.05
vm_min = 0.95
# create net
net = pp.create_empty_network()
pp.create_bus(net, max_vm_pu=vm_max, min_vm_pu=vm_min, vn_kv=10.)
pp.create_bus(net, max_vm_pu=vm_max, min_vm_pu=vm_min, vn_kv=.4)
pp.create_gen(net, 1, p_kw=-100, controllable=True, max_p_kw=-5, min_p_kw=-150, max_q_kvar=50,
min_q_kvar=-50)
pp.create_ext_grid(net, 0)
pp.create_load(net, 1, p_kw=20, controllable=False)
pp.create_line_from_parameters(net, 0, 1, 50, name="line2", r_ohm_per_km=0.876,
c_nf_per_km=260.0, max_i_ka=0.123, x_ohm_per_km=0.1159876,
max_loading_percent=100 * 690)
pp.create_piecewise_linear_cost(net, 0, "gen", np.array([[-150, 300], [0, 0]]))
# run OPF
pp.runopp(net, verbose=False)
assert net["OPF_converged"]
assert net.res_cost == 2 * net.res_gen.p_kw.values
def test_contingency_sgen(base_net):
net = base_net
pp.create_sgen(net,
1,
p_kw=-100,
q_kvar=0,
controllable=True,
max_p_kw=-5,
min_p_kw=-150,
max_q_kvar=50,
min_q_kvar=-50)
# pwl costs
# maximize the sgen feed in by using a positive cost slope
# using a slope of 1
# | /
# | /
# | /
# |/
#-------------------------------------------
# p_min_kw /|
# / |
# / |
pp.create_piecewise_linear_cost(
net, 0, "sgen",
array([[net.sgen.min_p_kw.at[0], net.sgen.min_p_kw.at[0]], [0, 0]]))
pp.runopp(net)
assert abs(net.res_cost - net.res_sgen.p_kw.at[0]) < 1e-5
# minimize the sgen feed in by using a positive cost slope
# using a slope of 1
# \ |
# \ |
# \ |
# \|
#-------------------------------------------
# p_min_kw |\
# | \
# | \
net.piecewise_linear_cost.f.at[0] *= -1
pp.runopp(net)
assert abs(net.res_cost - net.res_sgen.p_kw.at[0] * -1) < 1e-5
try:
net.piecewise_linear_cost = net.piecewise_linear_cost.drop(index=0)
except:
net.piecewise_linear_cost = net.piecewise_linear_cost.drop(0)
# first using a positive slope as in the case above
pp.create_polynomial_cost(net, 0, "sgen", array([1, 0]))
pp.runopp(net)
assert abs(net.res_cost - net.res_sgen.p_kw.at[0]) < 1e-5
# negative slope as in the case above
net.polynomial_cost.c.at[0] *= -1
pp.runopp(net)
assert abs(net.res_cost - net.res_sgen.p_kw.at[0] * -1) < 1e-5
def test_cost_piecewise_linear_load_uneven_slopes():
""" Testing a very simple network for the resulting cost value
constraints with OPF """
# boundaries:
vm_max = 1.05
vm_min = 0.95
# create net
net = pp.create_empty_network()
pp.create_bus(net, max_vm_pu=vm_max, min_vm_pu=vm_min, vn_kv=10.)
pp.create_bus(net, max_vm_pu=vm_max, min_vm_pu=vm_min, vn_kv=.4)
pp.create_load(net, 1, p_kw=100, controllable=True, max_p_kw=150, min_p_kw=50, max_q_kvar=0,
min_q_kvar=0)
pp.create_ext_grid(net, 0)
pp.create_line_from_parameters(net, 0, 1, 50, name="line2", r_ohm_per_km=0.876,
c_nf_per_km=260.0, max_i_ka=0.123, x_ohm_per_km=0.1159876,
max_loading_percent=100 * 690)
pp.create_piecewise_linear_cost(net, 0, "load", np.array([[0, 0], [75, 51], [150, 101]]))
# run OPF
with pytest.raises(OPFNotConverged):
pp.runopp(net, verbose=False)
assert net["OPF_converged"]
assert abs(net.res_cost - net.res_load.p_kw.values / 1.5) < 1e-3
def test_dcopf_pwl(simple_opf_test_net):
# create net
net = simple_opf_test_net
# pp.create_polynomial_cost(net, 0, "gen", np.array([-100, 0]))
pp.create_piecewise_linear_cost(
net, 0, "gen", np.array([[-200, -20000], [-100, -10000], [0, 0]]))
# run OPF
pp.rundcopp(net, verbose=False)
assert net["OPF_converged"]
# check and assert result
logger.debug("test_simplest_voltage")
logger.debug("res_gen:\n%s" % net.res_gen)
logger.debug("res_ext_grid:\n%s" % net.res_ext_grid)
logger.debug("res_bus.vm_pu: \n%s" % net.res_bus.vm_pu)
assert abs(100 * net.res_gen.p_kw.values - net.res_cost) < 1e-3
def test_mixed_p_q_pwl():
vm_max = 1.05
vm_min = 0.95
# create net
net = pp.create_empty_network()
pp.create_bus(net, max_vm_pu=vm_max, min_vm_pu=vm_min, vn_kv=10.)
pp.create_bus(net, max_vm_pu=vm_max, min_vm_pu=vm_min, vn_kv=.4)
pp.create_gen(net,
1,
p_kw=-100,
controllable=True,
max_p_kw=-5,
min_p_kw=-150,
max_q_kvar=50,
min_q_kvar=-50)
pp.create_ext_grid(net, 0)
pp.create_load(net,
1,
p_kw=20,
controllable=False,
max_q_kvar=50,
max_p_kw=100,
min_p_kw=50,
min_q_kvar=-50)
pp.create_line_from_parameters(net,
0,
1,
50,
name="line2",
r_ohm_per_km=0.876,
c_nf_per_km=260.0,
max_i_ka=0.123,
x_ohm_per_km=0.1159876,
max_loading_percent=100 * 690)
# testing some combinations
pp.create_piecewise_linear_cost(net, 0, "gen",
np.array([[-150, 150], [150, -150]]))
pp.create_piecewise_linear_cost(net,
0,
"gen",
np.array([[-150, 150], [150, -150]]),
type="q")
pp.runopp(net, verbose=False)
assert net["OPF_converged"]
assert net.res_cost == -net.res_gen.p_kw.values + net.res_gen.q_kvar.values
def test_cost_mixed():
""" Testing a very simple network for the resulting cost value
constraints with OPF """
# boundaries:
vm_max = 1.05
vm_min = 0.95
# create net
net = pp.create_empty_network()
pp.create_bus(net, max_vm_pu=vm_max, min_vm_pu=vm_min, vn_kv=10.)
pp.create_bus(net, max_vm_pu=vm_max, min_vm_pu=vm_min, vn_kv=.4)
pp.create_gen(net, 1, p_kw=-100, controllable=True, max_p_kw=-5, min_p_kw=-150, max_q_kvar=50,
min_q_kvar=-50)
pp.create_ext_grid(net, 0)
pp.create_load(net, 1, p_kw=20, controllable=False, max_q_kvar=50, max_p_kw=100, min_p_kw=50,
min_q_kvar=-50)
pp.create_line_from_parameters(net, 0, 1, 50, name="line2", r_ohm_per_km=0.876,
c_nf_per_km=260.0, max_i_ka=0.123, x_ohm_per_km=0.1159876,
max_loading_percent=100 * 690)
# testing some combinations
pp.create_polynomial_cost(net, 0, "gen", np.array([0, 1, 0]))
pp.runopp(net, verbose=False)
assert net["OPF_converged"]
assert net.res_cost == - net.res_gen.p_kw.values
net.polynomial_cost.c.at[0] = np.array([[1, 0, 0]])
pp.runopp(net, verbose=False)
assert net["OPF_converged"]
assert net.res_cost - net.res_gen.p_kw.values**2 < 1e-5
net.polynomial_cost.c.at[0] = np.array([[1, 0, 1]])
pp.runopp(net, verbose=False)
assert net["OPF_converged"]
assert net.res_cost - net.res_gen.p_kw.values**2 - 1 < 1e-5
net.load.controllable.at[0] = True
pp.runopp(net, verbose=False)
assert net.res_cost - net.res_gen.p_kw.values ** 2 - 1 < 1e-5
pp.create_piecewise_linear_cost(net, 0, "load", np.array([[0, 0], [100, 100]]), type="p")
pp.runopp(net, verbose=False)
assert net.res_cost - net.res_gen.p_kw.values ** 2 - 1 - net.res_load.p_kw.values < 1e-5
def _create_costs(net, ppc, gen_lookup, type, idx):
if ppc['gencost'][idx, 0] == 1:
if not len(ppc['gencost'][idx, 4:]) == 2*ppc['gencost'][idx, 3]:
logger.error("In gencost line %s, the number n does not fit to the number of values" %
idx)
pp.create_piecewise_linear_cost(net, gen_lookup.element.at[idx],
gen_lookup.element_type.at[idx],
ppc['gencost'][idx, 4:], type)
elif ppc['gencost'][idx, 0] == 2:
if len(ppc['gencost'][idx, 4:]) == ppc['gencost'][idx, 3]:
n = len(ppc['gencost'][idx, 4:])
values = ppc['gencost'][idx, 4:] / power(1e3, array(range(n))[::-1])
else:
logger.error("In gencost line %s, the number n does not fit to the number of values" %
idx)
pp.create_polynomial_cost(net, gen_lookup.element.at[idx], gen_lookup.element_type.at[idx],
values, type)
else:
logger.info("Cost mode of gencost line %s is unknown." % idx)
def test_cost_pwl_q_3point():
# We have a problem with the cost value after optimization of 3 point q cost functions! It returns the amount of q at the EG, but not the costs!
# Also, the q result is not the optimum!
vm_max = 1.05
vm_min = 0.95
# create net
net = pp.create_empty_network()
pp.create_bus(net, max_vm_pu=vm_max, min_vm_pu=vm_min, vn_kv=10.)
pp.create_bus(net, max_vm_pu=vm_max, min_vm_pu=vm_min, vn_kv=.4)
pp.create_sgen(net,
1,
p_kw=-100,
controllable=True,
max_p_kw=-5,
min_p_kw=-150,
max_q_kvar=50,
min_q_kvar=-50)
pp.create_ext_grid(net, 0)
pp.create_load(net, 1, p_kw=20, controllable=False)
pp.create_line_from_parameters(net,
0,
1,
50,
name="line2",
r_ohm_per_km=0.876,
c_nf_per_km=260.0,
max_i_ka=0.123,
x_ohm_per_km=0.1159876,
max_loading_percent=100 * 690)
pp.create_piecewise_linear_cost(net,
0,
"sgen",
np.array([[-50, 50], [0, 0], [50, 50]]),
type="q")
# run OPF
pp.runopp(net, verbose=False)
assert net["OPF_converged"]
def test_dcopf_pwl():
# create net
net = pp.create_empty_network()
pp.create_bus(net, vn_kv=10.)
pp.create_bus(net, vn_kv=.4)
pp.create_gen(net,
1,
p_kw=-100,
controllable=True,
max_p_kw=-5,
min_p_kw=-150,
max_q_kvar=50,
min_q_kvar=-50)
pp.create_ext_grid(net, 0)
pp.create_load(net, 1, p_kw=20, controllable=False)
pp.create_line_from_parameters(net,
0,
1,
50,
name="line2",
r_ohm_per_km=0.876,
c_nf_per_km=260.0,
max_i_ka=0.123,
x_ohm_per_km=0.1159876,
max_loading_percent=100)
# pp.create_polynomial_cost(net, 0, "gen", array([-100, 0]))
pp.create_piecewise_linear_cost(
net, 0, "gen", array([[-200, 20000], [-100, 10000], [0, 0]]))
# run OPF
pp.rundcopp(net, verbose=False)
assert net["OPF_converged"]
# check and assert result
logger.debug("test_simplest_voltage")
logger.debug("res_gen:\n%s" % net.res_gen)
logger.debug("res_ext_grid:\n%s" % net.res_ext_grid)
logger.debug("res_bus.vm_pu: \n%s" % net.res_bus.vm_pu)
assert abs(100 * net.res_gen.p_kw.values - net.res_cost) < 1e-3
def test_3point_pwl():
vm_max = 1.05
vm_min = 0.95
# create net
net = pp.create_empty_network()
pp.create_bus(net, max_vm_pu=vm_max, min_vm_pu=vm_min, vn_kv=10.)
pp.create_bus(net, max_vm_pu=vm_max, min_vm_pu=vm_min, vn_kv=.4)
pp.create_sgen(net,
1,
p_kw=-100,
q_kvar=0,
controllable=True,
max_p_kw=-100,
min_p_kw=-100.5,
max_q_kvar=50,
min_q_kvar=-50)
pp.create_ext_grid(net, 0)
pp.create_load(net, 1, p_kw=20, controllable=False)
pp.create_line_from_parameters(net,
0,
1,
50,
name="line2",
r_ohm_per_km=0.876,
c_nf_per_km=260.0,
max_i_ka=0.123,
x_ohm_per_km=0.1159876,
max_loading_percent=100 * 690)
pp.create_piecewise_linear_cost(net,
0,
"sgen",
np.array([
[-100, 1],
[0, 0],
[100, 1],
]),
type="q")
# creating a pwl cost function that actually is realistic: The absolute value of the reactive power has costs.
pp.runopp(net, verbose=False)
# assert abs( net.res_sgen.q_kvar.values ) < 1e-5
# this is not the expected result. the reactive power should be at zero to minimze the costs.
# checkout the dirty workaround:
# the first sgen is only representing the positive segment of the function:
net.piecewise_linear_cost.p.at[0] = np.array([[0, 1]])
net.piecewise_linear_cost.f.at[0] = np.array([[0, 1]])
net.sgen.min_q_kvar.at[0] = 0
# what we can do instead is modelling a second sgen on the same bus representing the negative segment of the function:
pp.create_sgen(net,
1,
p_kw=0,
q_kvar=0,
controllable=True,
max_p_kw=0.01,
min_p_kw=-0.01,
max_q_kvar=0,
min_q_kvar=-10)
pp.create_piecewise_linear_cost(net,
1,
"sgen",
np.array([
[-100, 100],
[0, 0],
]),
type="q")
# runOPF
pp.runopp(net, verbose=False)
assert abs(sum(net.res_sgen.q_kvar.values)) < 1e-5
# et voila, we have the q at zero. sure, we do have two seperate sgens now and this is very dirty. but it's working.
# let's check if we can handle overvoltage
net.bus.max_vm_pu = 1.041
pp.runopp(net, verbose=False)
assert abs(max(net.res_bus.vm_pu.values) - 1.041) < 1e-5
def test_contingency_load(base_net):
net = base_net
pp.create_load(net,
1,
p_kw=-100,
q_kvar=0,
controllable=True,
max_p_kw=150,
min_p_kw=5,
max_q_kvar=50,
min_q_kvar=-50)
# pwl costs
# minimze the load by using a positive cost slope
# using a slope of 1
# | /
# | /
# | /
# |/
# -------------------------------------------
# p_min_kw /|
# / |
# / |
pp.create_piecewise_linear_cost(
net, 1, "load",
array([[0, 0], [net.load.max_p_kw.at[1], net.load.max_p_kw.at[1]]]))
pp.runopp(net)
assert abs(net.res_cost - net.res_load.p_kw.at[1]) < 1e-5
# maximize the load in by using a negative cost slope
# using a slope of 1
# \ |
# \ |
# \ |
# \|
# -------------------------------------------
# p_min_kw |\
# | \
# | \
net.piecewise_linear_cost.f.at[0] *= -1
pp.runopp(net)
assert abs(net.res_cost - net.res_load.p_kw.at[1] * -1) < 1e-5
# poly costs
try:
net.piecewise_linear_cost = net.piecewise_linear_cost.drop(index=0)
except:
# legacy fix
net.piecewise_linear_cost = net.piecewise_linear_cost.drop(0)
# first using a positive slope as in the case above
pp.create_polynomial_cost(net, 1, "load", array([1, 0]))
pp.runopp(net)
assert abs(net.res_cost - net.res_load.p_kw.at[1]) < 1e-5
# negative slope as in the case above
net.polynomial_cost.c.at[0] *= -1
pp.runopp(net)
assert abs(net.res_cost - net.res_load.p_kw.at[1] * -1) < 1e-5