/PERTURB/PART/SHELL
Block Format Keyword This option can be used to study the robustness of a design by generating different thickness values for every shell element in the specified part group. The random noise scale factors can have either a normal (Gaussian) distribution or random distribution.
Format
(1)  (2)  (3)  (4)  (5)  (6)  (7)  (8)  (9)  (10) 

/PERTURB/PART/SHELL/ID  
perturb_title  
F_Mean  Deviation  Min_cut  Max_cut  Seed  Idistri  
grpart_ID  parameter 
Definitions
Field  Contents  SI Unit Example 

ID  Perturb
identifier. (Integer, maximum 10 digit) 

perturb_title  Perturb
title. (Character, maximum 100 characters) 

F_Mean  Mean value of the random
noise scale factor. (Real) 

Deviation  Standard deviation for the
normal distribution. Only used when Idistri=2. (Real) 

Min_cut  Minimum value for random
noise scale factor. Must be entered when using Idistri=1, random distribution. If random value is < Min_cut, random value is replaced with Min_cut. (Real) 

Max_cut  Maximum value for random
noise scale factor. Must be entered when using Idistri=1, random distribution. If random value is > Max_cut, random value is replaced with Max_cut. (Real) 

Seed  Seed used for random noise
generation.
(Integer) 3 

Idistri  Distribution type.
(Integer) 4 

grpart_ID  Part group
identifier. (Integer, maximum 10 digits) 

parameter  Parameter identifier.
(Character, maximum 20 characters. Starting at any location in the input field) 
Example
#12345678910
/PERTURB/PART/SHELL/1
set Random Noise with normal distribution on Shell Thickness (model unit : gmmms)
# Mean_value Deviation Min_cut Max_cut Seed Idistri
1.8 0.02 0.7 1.2 7182 2
#grpart_ID parameter
123456 THICK
#12345678910
/PERTURB/PART/SHELL/2
set Random Noise with random distribution on Shell Thickness
# Mean_value Deviation Min_cut Max_cut Seed Idistri
2.1 0.9 1.1 7182 1
#grpart_ID parameter
123457 THICK
#12345678910
/GRPART/PART/123456
test_grpart
1 2
#12345678910
/GRPART/PART/123457
test_grpart
3
#12345678910
#ENDDATA
Comments
 Two computations with the same values, Seed will return exactly to the same results. Currently the only valid parameter is THICK.
 Due to discretization, the random noise in the model will not exactly match the input random noise values. However, the model’s actual random noise values are calculated and output in the Starter output file.
 Seed is an integer used to initialize random number generation for the random noise. The Seed can be changed to study sensitivity or model robustness. If Seed=0, a random (different) seed value is calculated for each run and written to the Starter output. To rerun an analysis with Seed=0, use the Seed value written to the Starter output in the original analysis as input in the analysis rerun.
 If Idistri=1, the distribution of the scale factors will be
random.If Idistri=2, the normal distribution of the scale factors will have the following probability density function.