def test_production4(self):
'''
Test (stochasitc) average throughput with degradation.
'''
repair_lower = 5
repair_upper = 15
system = maintsim.System(process_times=[5],
failure_mode='degradation',
failure_params={
'failed state': 10,
'degradation rate': 0.2
},
repair_params={
'CM':
stats.randint(repair_lower,
repair_upper + 1)
})
time_horizon = 500
# calculate true expected ttf (expected time to absorption)
def expected_ttf(Q):
M = Q[:-1, :-1]
I = np.identity(len(M))
N = np.linalg.inv(I - M)
ones = np.ones(shape=(len(M), 1))
t = np.matmul(N, ones)
return t
E_ttf = expected_ttf(system.machines[0].degradation)[0][0]
# expected ttr
E_ttr = (repair_lower + repair_upper) / 2
A = E_ttf / (E_ttf + E_ttr)
E_production = A * (time_horizon / system.machines[0].process_time)
reps = 30
production_samples = []
for _ in range(reps):
system.simulate(sim_time=time_horizon, verbose=False)
production_samples.append(system.machines[0].parts_made)
# conduct one sample t-test
# H_0: average production sample = expected production
# H_1: average production sample != expected production
_, p_value = stats.ttest_1samp(production_samples, E_production)
message = (f'Mean production is {np.mean(production_samples)}' +
f' from {len(production_samples)} samples.' +
f' Expected mean: {E_production}')
self.assertGreaterEqual(p_value, 0.10, msg=message)
def test_production2(self):
'''
Deterministic production volume of one machine.
'''
time_horizon = 500
cycle_time = 10
expected_production = time_horizon / cycle_time
system = maintsim.System(process_times=[cycle_time])
system.simulate(sim_time=time_horizon, verbose=False)
self.assertEqual(system.machines[-1].parts_made, expected_production,
'Should be {}'.format(expected_production))
def test_production3(self):
'''
Deterministic production volume of a two-machine line.
'''
time_horizon = 500
cycle_times = [5, 3]
expected_production = time_horizon / max(cycle_times)
system = maintsim.System(process_times=cycle_times)
system.simulate(warmup_time=100, sim_time=time_horizon, verbose=False)
self.assertAlmostEqual(system.machines[-1].parts_made,
expected_production,
msg=f'Should be {expected_production}+/-1',
delta=1)
def test_degradation1(self):
'''
Test average time to failure of a single machine.
'''
system = maintsim.System(process_times=[5],
failure_mode='degradation',
failure_params={
'failed state': 10,
'degradation rate': 0.2
},
repair_params={'CM': stats.randint(2, 3)})
reps = 1
warmup_time = 100
time_horizon = 5000
ttf_samples = np.array([])
for _ in range(reps):
system.simulate(warmup_time=warmup_time,
sim_time=time_horizon,
verbose=False)
# gather ttf samples from simulation
ttfs = system.maintenance_data[system.maintenance_data['activity']
== 'failure']['duration'].values
ttfs = ttfs[ttfs != 'NA']
ttf_samples = np.append(ttf_samples, ttfs)
# calculate true expected ttf (expected time to absorption)
def expected_ttf(Q):
M = Q[:-1, :-1]
I = np.identity(len(M))
N = np.linalg.inv(I - M)
ones = np.ones(shape=(len(M), 1))
t = np.matmul(N, ones)
return t
E_ttf = expected_ttf(system.machines[0].degradation)[0][0]
# conduct one sample t-test
# H_0: average ttf sample = expected ttf
# H_1: average ttf sample != expected ttf
_, p_value = stats.ttest_1samp(ttf_samples, E_ttf)
message = f'Mean TTF is {np.mean(ttf_samples)} from {len(ttf_samples)} samples'
self.assertGreaterEqual(p_value, 0.10, msg=message)
def test_ppl1(self):
'''
Test permanent production loss (PPL) measurement of one machine.
'''
system = maintsim.System(process_times=[5],
failure_mode='degradation',
failure_params={
'failed state': 10,
'degradation rate': 0.2
},
repair_params={'CM': stats.randint(2, 3)})
system.simulate(sim_time=1000, verbose=False)
repairs = system.maintenance_data[system.maintenance_data['activity']
== 'repair']
total_repair_time = sum(repairs['duration'])
self.assertEqual(total_repair_time, system.machines[0].total_downtime,
f'Should be {total_repair_time}')