Third-party - Abaqus

Import the necessary libraries.

import glob
import pickle
import time
import os
from UQpy.distributions import Normal, Uniform
from UQpy.run_model.RunModel import RunModel
from UQpy.run_model.model_execution.ThirdPartyModel import ThirdPartyModel
from UQpy.sampling import MonteCarloSampling

calling_directory = os.getcwd()
t = time.time()

Building the model There are two probabilistic input variables, the fire load density and the yield strength.

var_names = ['qtd', 'fy']

Create the model object

m = ThirdPartyModel(model_script='abaqus_fire_analysis.py', input_template='abaqus_input.py',
                    output_script='extract_abaqus_output.py', var_names=var_names,
                    model_dir='SFE_MCS', )
abaqus_sfe_model = RunModel(cores_per_task=1, ntasks=24, model=m)
print('Example: Created the model object.')

Towards defining the sampling scheme The fire load density is assumed to be uniformly distributed between 50 \(MJ/m^2\) and 450 \(MJ/m^2\). The yield strength is assumed to be normally distributed, with the parameters being: mean = 250 \(MPa\) and coefficient of variation of \(7%\).

Creating samples using MCS.

d_n = Normal(loc=50, scale=400)
d_u = Uniform(location=2.50e8, scale=1.75e7)
x_mcs = MonteCarloSampling(distributions=[d_n, d_u], samples_number=100, random_state=987979)

Running simulations using the previously defined model object and samples

sample_points = x_mcs.samples
abaqus_sfe_model.run(samples=sample_points)

The outputs from the analysis are the values of the performance function.

qois = abaqus_sfe_model.qoi_list

Save the samples and the qois in a dictionary called results with keys ‘inputs’ and ‘outputs’.

results = {'inputs': sample_points, 'outputs': qois}

Pickle the results dictionary in the current directory. The basename and extension of the desired pickle file:

res_basename = 'MCS_results'
res_extension = '.pkl'

Create a new results file with a larger index than any existing results files with the same name in the current directory.

res_file_list = glob.glob(res_basename + '_???' + res_extension)
if len(res_file_list) == 0:
    res_file_name = res_basename + '_000' + res_extension
else:
    max_number = max(res_file_list).split('.')[0].split('_')[-1]
    res_file_name = res_basename + '_%03d' % (int(max_number) + 1) + res_extension

res_file_name = os.path.join(calling_directory, res_file_name)

Create a new results file with a larger index than any existing results files with the same name in the current directory.

with open(res_file_name, 'wb') as f:
    pickle.dump(results, f)
print('Saved the results to ' + res_file_name)

print('Example: Done!')
print('Time elapsed: %.2f minutes' % float((time.time() - t) / 60.0))