def test_shift_scalar(self):
#shift
c = Talbot(f=f3, n=24, shift=1.0, vectorized=False)
assert_allclose(c(1), np.exp(1))
def test_shift2_scalar(self):
#shift
d = Talbot(f=f4, n=24, shift=1.2, vectorized=False)
assert_allclose(d(1.2,args=(2,)), 1.2*np.sin(2*1.2), atol=1e-6)
def test_two_t_scalar(self):
a = Talbot(f=f1, n=24, shift=0.0, vectorized=False)
#single value of t:
assert_allclose(a([1,2]), np.exp(np.array([-1,-2])))
def test_args_scalar(self):
#args
b = Talbot(f=f2, n=24, shift=0.0, vectorized=False)
assert_allclose(b(1, args=(1,)), np.exp(-2))
def test_shift2(self):
#shift
d = Talbot(f=f4, n=24, shift=1.2)
assert_allclose(d(1.2,args=(2,)), 1.2*np.sin(2*1.2), atol=1e-6)
def test_single_t_scalar(self):
a = Talbot(f=f1, n=24, shift=0.0, vectorized=False)
#single value of t:
assert_allclose(a(1), np.exp(-1))
def test_args(self):
#args
b = Talbot(f=f2, n=24, shift=0.0)
assert_allclose(b(1, args=(1,)), np.exp(-2))
def test_shift(self):
#shift
c = Talbot(f=f3, n=24, shift=1.0)
assert_allclose(c(1), np.exp(1))
def test_two_t(self):
a = Talbot(f=f1, n=24, shift=0.0)
#single value of t:
assert_allclose(a([1,2]), np.exp(np.array([-1,-2])))
def test_single_t(self):
a = Talbot(f=f1, n=24, shift=0.0)
#single value of t:
assert_allclose(a(1), np.exp(-1))
def test_t_is_zero_error(self):
a = Talbot(f=f1, n=24, shift=0.0)
#t=0 raise error:
assert_raises(ValueError, a, 0)
def laplace_inverse(self, f, tvar, args, **opt):
ilt = Talbot(f, **opt)
val = ilt(tvar, args)
return val, 0.0
def cosenzaandkorosak2014(z, t, theta, v, tpor=None, L = 1, kv = 1, mv = 0.1, gamw = 10,
ui = 1, nterms = 100):
"""Secondary consolidation of Clay as an anomalous diffusion process
An implementation of [1]_.
Features:
- Single layer, soil properties constant over time.
- Instant load uniform with depth.
- Vertical flow.
- Similar to Terzaghi 1d consolidation equation but with additional
fractional time derivative that retards pore pressure dissipation
as a possible creep mechanism.
- Uses Laplace transform in solution.
Parameters
----------
z : float or 1d array/list of float
Depth.
t : float or 1d array/list of float
Time.
theta : float or array/list of float
Parameter in [1]_.
v : float or array/list of float
Parameter in [1]_.
tpor : float or 1d array/list of float
Time values for pore pressure vs depth calcs.
L : float, optional
Drainage path length. Default H = 1.
kv : float, optional
Vertical coefficient of permeability. Default kv = 1.
mv : float, optional
Volume compressibility. Default mv = 0.1.
gamw : float, optional
Unit weight of water. defaule gamw = 10.
ui : float, optional
Initial uniform pore water pressure. Default ui = 1.
nterms : int, optional
Maximum number of series terms. Default nterms= 100.
Returns
-------
por : 2d array of float
Pore pressure at depth and time. ppress is an array of size
(len(z), len(t)).
avp : 1d array of float
Average pore pressure between depth H and depth Z.
settlement : 1d array of float
Surface settlement at depth z.
Notes
-----
The article [1]_ only has single values of `theta` and `v`.
Here I've simply added more.
References
----------
Code developed based on [1]_
.. [1] Cosenza, Philippe, and Dean Korosak. 2014. 'Secondary Consolidation
of Clay as an Anomalous Diffusion Process'. International Journal
for Numerical and Analytical Methods in Geomechanics:
doi:10.1002/nag.2256.
"""
# def F(s, v, theta, lam):
# return (1+theta*s**(v-1))/(s + theta*s**v+lam)
def F(s, v, theta, lam):
numer = np.ones_like(s)
denom = s+lam
for v_, the_ in zip(v,theta):
numer += the_*s**(v_-1)
denom += the_*s**v_
return (numer/denom)
# def Bn(r, lam, the, v, t):
# numer = lam * the*r**(v-1)*cmath.sin(np.pi*v)
# denom = (lam - r + the*r**v*cmath.exp(1j*np.pi*v))**2
#
# return math.exp(-r * t)*numer/denom
z = np.atleast_1d(z)
t = np.atleast_1d(t)
theta = np.atleast_1d(theta)
v = np.atleast_1d(v)
if tpor is None:
tpor=t
else:
tpor = np.atleast_1d(t)
M = ((2 * np.arange(nterms) + 1) * np.pi)
# an = 2 / M
dTv = kv / L**2 / mv / gamw
lamda_n = M**2*dTv
Z = (z / L)
# por = np.zeros((len(z), len(tpor)))
# doc = np.zeros(len(t))
a = Talbot(F, n=24, shift=0.0)
Bn = np.zeros((len(tpor), nterms), dtype=float)
for j, lam in enumerate(M):
Bn[:, j] = np.array(a(tpor, args=(v, theta, M[j])), dtype=float)
if np.allclose(tpor, t): #reuse Bn for Doc
Bn_ = Bn.copy()
Bn = Bn[np.newaxis, :, :]
An = 2/M[np.newaxis, :] * np.sin(M[np.newaxis, :] * Z[:, np.newaxis])
An = An[:, np.newaxis, :]
por = np.sum(An*Bn, axis=2)*ui
#degree of consolidation
An = 2*2**2/M[np.newaxis, :]**2
if np.allclose(tpor, t):
Bn = Bn_
else:
Bn = np.zeros((len(t), nterms), dtype=float)
for j, lam in enumerate(M):
Bn[:, j] = np.array(a(t, args=(v, theta, M[j])), dtype=float)
doc = 1 - np.sum(An*Bn, axis = 1)
return por, doc
