• Tidak ada hasil yang ditemukan

Limitations of The SHM Method

Dalam dokumen CALIFORNIA INSTITUTE OF TECHNOLOGY (Halaman 141-144)

Conclusions and Future Work

4.2 Limitations of The SHM Method

expansion of the probability integrals was determined to be sufficient for the examples considered in this work, it may not be in all cases. Thus, a thorough study of how well the coarse approximation works in different scenarios should be conducted. Other computational issues which should be investigated are the improvements in approximation achievable with the fine approximation method, and the error introduced by the iterative minimization scheme.

4.1.3 Automation

In light of the fact that the ultimate goal of this work is to create a system for use on real structures, many operational questions need to be addressed. For example, on-line determination of the modal parameters from measured data is a fundamental requirement if the SHM system is to be fully automated. Also, criteria for setting the maximum size of the monitoring cycle window used in determining the sequence of probabilities of variation need to be established.

Such criteria should consider the trade-off between the increased computation and the potentially increased ability to detect low levels of damage by having longer windows. As a final matter, the interface between the SHM analysis and the end user needs to be considered.

are noted as such.

Types of Damage

A very basic difficulty in SHM comes from establishing what types of changes in a structure can actually be detected. One of the fundamental assumptions for performing global SHM, as mentioned in Section 1.3, is that changes in the structure will affect some measured data to a sufficient extent to be able to characterize the changes. Certain types of changes may not have such effects.

For instance, plastic deformation due to high loading conditions might be considered a type of "damage" that is of importance to detect. However, once the high loading event has passed, a plastically deformed structural member will behave elastically once again, and quite possibly not alter measurements of the dynamic response of the structure. As another example, consider a highly redundant structure. If a single structural member is damaged in some fashion, the local effect might be significant, but the global effect may not be. Unless measurements are made in the specific area of the member, the damage could go undetected. Sensing these types of damaging events is a serious challenge to a global SHM method since the measured data is not significantly effected by the damage.

Model Problems

The highly redundant structure example brings up another limitation of global model-based SHM methods: The degree to which damage can be located depends on the nature of the measured data. The number of measured DOFs and significantly expressed modes of vibration in the recorded data dictate how many model parameters can be identified without uniqueness problems.

If very few model parameters can be uniquely identified, the model for the structure may not accurately reflect the true behavior. Thus, the second

assumption made in Section 1.3 could be violated. That assumption involved treating changes in the model identified from different data sets as proxies for changes in the real structure.

The method presented in this work uses substructuring to reduce the num- ber of parameters so that they can he uniquely identified. As shown in Chap- ter 3, substructuring can have the effect of smearing elemental level damage over an entire substructure, so the sensitivity to damage may be reduced.

This is a limitation to the presented method which, as mentioned in Sec- tion 4.1, could be addressed through adaptive suhstructuring (Hjelmstad and Shin 1997) or selective parameter updating (Farhat and Hemez 1993).

Another result of using models with few parameters is that even if the existence and location of damage can be found, the degree of damage will be extremely difficult to determine. This is a fundamental limitation of global SHM methods. The best way to overcome this limitation is to use the global SHM to establish existence and possible location of damage, and apply local SHM techniques to actually find and determine the extent of the damage.

Problems From Using Modal Data

The advantages of using modal parameters in the SHM process have already been commented upon in Section 1.3. There are, however, disadvantages. The modal parameters themselves are identified using some system identification technique to fit the parameters of a modal model to measured data. Thus, uncertainties are introduced which were ascribed to "noise" in this work. The uncertainties are passed on to the model parameters identified using the modal parameters. Suppose a significant portion of the variation in the modal pa- rameters when there is no damage is due to the identification technique and not the underlying measured data. Performing SHM using the measured time history data directly may then offer some advantages.

Fooling a SHM Method

The final problem with SHM occurs when what appears to be damage may not be damage. Following an extreme loading event (e.g. earthquake), many concrete structures exhibit a decrease in their frequencies of vibration. This effect is believed to be caused by a combination of loosening of the concrete, the soil, and the non-structural elements in the structure. Over time, some of the frequencies may increase slightly, but they generally stay lower than the original values. Any reasonable SHM technique would interpret the change as damage and identify the possible damage locations. In reality, however, the structure may not be truly damaged. Thus, interpreting damage in concrete structures may present difficulties.

Dalam dokumen CALIFORNIA INSTITUTE OF TECHNOLOGY (Halaman 141-144)