Running Abaqus in Matlab environment

In some cases, we need to perform a parametric study: how certain parameters affect the stress/strain/displacement at certain point or certain region in a model, or doing certain structural optimization. For such a circumstance, combining Abaqus with an additional programming language , which has an optimization capability would be unavoidable. Following shows you a snip how to perform the analysis repeatedly in Matlab.

for i = 1 : 10
istatus = dos('abq661 job=filename interactive'); % Run Abaqus,
if ~(istatus) exit; end
% Call a function to parse the necessary outcome.
% Call an optimization algorithm. See. help optim
% Call a function to modify the Abaqus input file
end

FeG-

Matlab: conv.m and FFT Convolution

There are two ways in solving a discrete integral convolution in Matlab: ‘conv.m’ and by means of ‘fft.m’. The difference of both functions is as follow:

Consider two input vectors: x and y, and both vectors have the same length of N. The FFT provides the convolution by z = real( ifft( fft(x) .* fft(y) ) ), where z is the result of the convolution. The FFT convolution, in this example, produces the vector z having a length of N. For the same case, the operation z’ = conv(x, y), a time domain convolution, will produce z’ having a length of (2N-1). That is the difference of both functions. The similarity is the first N data of z and z’ are the same.

FeG-