0% found this document useful (0 votes)
4 views1 page

Radovic 2014

The document discusses the organization and analysis of stress data from a bridge structure using Finite Element Analysis (FEA). It details the categorization of elements into groups, the determination of optimal histogram bin sizes for stress ranges, and the application of the Tucker3 tensor decomposition method to analyze stress distributions under varying loads. The study focuses on the bottom flange and cross-frame elements of the bridge to assess their structural response to loading conditions.

Uploaded by

Abhishek B s
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
4 views1 page

Radovic 2014

The document discusses the organization and analysis of stress data from a bridge structure using Finite Element Analysis (FEA). It details the categorization of elements into groups, the determination of optimal histogram bin sizes for stress ranges, and the application of the Tucker3 tensor decomposition method to analyze stress distributions under varying loads. The study focuses on the bottom flange and cross-frame elements of the bridge to assess their structural response to loading conditions.

Uploaded by

Abhishek B s
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

dimensions in the array.

Stress ranges were used because it


would be impractical to capture all possible stress values of
each element in the structure. In this way, the data emulates
the form of stress histograms for each loading increment.
The data extracted from the FEA was organized into
element groups representing different structural members of
the bridge in different locations in the bridge structure. For
example, cross-frame elements that are connecting Girder 1
and Girder 2 (see Fig. 4) are labeled XG1; cross-frame
elements that are connecting Girder 2 and Girder 3 are labeled
XG2; and cross-frame elements that are connecting Girder 3
and Girder 4 are labeled XG3. Girder 1 bottom flange
elements are labeled G1BF, Girder 2 bottom flange elements
are labeled G2BF, and so on. At the end of this step, a total of
7 element groups are formed (X1G, X2G, X3G, G1BF, G2BF,
G3BF and G4BF), each one containing histogram data for 51
stress ranges and 17 load magnitudes, representing data from
280,976 locations in the structure.
The optimum histogram bin size was determined using the
mean integrated squared error procedure [9]. Mean integrated
square error algorithm is used to measure the goodness-of-the-
fit of a histogram by computing the estimated errors for
C. FEA Loading Conditions several candidate bin sizes, and the bin size that yields the
smallest estimated error is selected. The stress ranges
(histogram bins) were kept constant for all element groups as
The basic function of the modified Riks method is that it
Fig. 4. Bridge 7R: photograph of actual structure on the top and cross- well as being constant between load increments. As a result,
sectional view of FEA on the bottom. for the data set presented in this paper, the optimal bin size
was determined to be 733 psi and a total of 51 bins were used.
Note that von Mises stresses [7] are used in this work, so all
associated structural response metrics such as stress) that
stress values have a positive sign.
satisfy equilibrium at each magnitude of loading. The Riks
method is ideally suited to analyzing the behavior after such a Because the girder bottom flanges and cross-frames have
peak loading is reached. Such algorithms exist in commercial significantly different numbers of elements, comparing counts
FEA software, such as Abaqus 6.12 [7], which is used in this of the frequencies of elements in each stress bin would not
work. In this case, the load is expressed in terms of load reveal the desired information. Therefore, the number of
proportionality factors (a multiple of the input load). For this elements in each stress range in each element group is
study, the input load is a HS-20 vehicle load commonly used normalized (divided) by the number of elements in the
in bridge design and evaluation [8]. Since each LPF is a corresponding group. For example, if 665 is the number of
multiple of the input load and the input load is a HS-20 loading bottom flange elements in Girder 4 that are in a given stress
vehicle, the results of the Riks method are expressed in terms range and because the total number of elements in the bottom
of number of these design vehicles. For the scenario being flange of Girder 4 is 3838, then the resulting value in this
described here, for which non-linear geometry was prescribed, stress range (bin) would be 0.173 (665/3838). The procedure
the maximum LPF at which convergence is obtained is 17. is repeated for each element group and for each loading.
D. Data Preprocessing for Tensor Decomposition This preprocessing results in creating element group stress
In order to carry out Tucker3 decomposition, the data must histograms as shown in Fig. 5. All preprocessing was executed
be organized in a specific three-way format. In this work, the by a custom-made Matlab code [10].
goal was to investigate the stresses in different element E. Tensor Decomposition Procedure using Tucker3 Method
groups representing different structural members under
Tensor decomposition of the resulting three-way array,
increasing loads. Since the girders of Bridge 7R were simply
mathematically expressed as xϵR (7×51×17), was performed
supported composite beams, their bottom flanges will
using the N-way Toolbox in Matlab R2014. The first mode of
experience the highest stress. Thus, only the bottom flange
the array is made up of 7 element groups (4 girder element
element stresses were considered for the girders in this
groups and 3 cross-frame element groups organized in rows),
example. It follows that girder bottom flange and cross-frame
the second mode consists of 51 stress ranges (51 stress
element groups, stress magnitude (expressed in terms of the
histogram bins organized in columns), and the third mode
number of elements experiencing the same range of stress at a
consists of the 17 loading increments also known as LPFs.
given load) and increasing loads were selected to represent the

63

You might also like