Civil Engineering Reference
In-Depth Information
The BCSIMS system is set up to use various techniques of damage detection at different locations. In addition to the
techniques already implemented in the system, part of the mandate of the BCSIMS scope is to conduct research on new
techniques and implementation of other existing techniques. The first of the implemented algorithms is in the SIMS2
module. It is a statistical algorithm [
6
]. As with many algorithms, the method observes damage as changes to modal
parameters. Consequently, the first requirement of the method is to obtain a series of data sets from the same structure, to
obtain a baseline model from which to observe potential changes.
One desirable feature for this algorithm is its ability to detect damage even in the presence of noise, and common
environmental effects such as temperature changes. An additional advantage is with regards to the concept of the sub-
structuring or clustering of the model for the analysis. This is important for the speed of the algorithm, and for the accuracy
of the damage identification. In addition, parts of the structure, which are not expected to be damaged, can be removed from
the analysis to improve on speed and accuracy.
A second set of algorithms has been implemented to run in an offline mode through the SIMS3 machine. Five of those can
be considered as mode shape based methods. These methods were compiled and used in a well-known study by [
7
]. These
algorithms are:
1. Damage index method - [
8
]
2. Mode shape curvature method - [
9
]
3. Change in flexibility method - [
10
]
4. Change in uniform load surface curvature - [
11
]
5. Change in stiffness method - [
12
]
The primary advantage of these methods is the relative ease of formulation, speed of use and simplicity of output results.
However, the disadvantage is in the requirement for measurement of the mode shapes in more detail on the structure. This
can present a problem in many situations when limited instrumentation is available.
The last set of methods is the flexibility-based methods, of which two are implemented also in an offline mode on SIMS3.
The flexibility-based methods use the concept of changes in assembled flexibility matrix to identify and locate damage in a
structure. The flexibility matrix can be obtained for stochastic (output-only) data by manipulations of the results from time
domain system identification. Previous work [
13
] utilized two variations of stochastic flexibility methods. These methods are:
1. Stochastic Damage Locating Vector (SDLV) Technique - [
14
], based on the Damage Locating Vector (DLV) Technique -
[
15
,
16
].
2. Proportional Flexibility Matrix Technique - [
17
]
Finite element models of the monitored structure are another important element in the overall data analysis capabilities.
Software is setup on the SIMS3 for the analysis; both during initial calibration and during routine and triggered events. The
models will be updated using data from the measured structure, and are used for analysis such as:
1. Stress and load
2. Fatigue
3. Damage location and quantification
4. Prognosis (life expectancy)
For most monitoring system cases, a preliminary FEM will be created and an on-site ambient vibration test will be
performed. The FEM will then be manually or automatically updated based on the obtained results. The model can then be
used to:
1. Design the permanent monitoring system
2. Evaluate potential damage detection methods through simulation
3. Perform real-time analysis
4. Be used for scheduled structural analysis based on updated models
Automated FEM updating is performed through the SIMS3 PC using the FEMTools commercially available software.
Once a new set of modes is placed in the database, an automated process on the SIMS3 machine triggers an updating run
using the software; the updated modal comparison matrix and parameter changes are placed in the database. The new FE
model is placed in a folder labeled with the event ID and time. The new updated model can be used for further analysis. The
process requires a preliminary manual updating of the model, usually done with data from a more detailed ambient vibration
test. The settings from this 'manual' update are used to create command files for the automated process.
