Considerations Regarding the Behavior of Damaged Structures

G.R. Gillich1, D. Amariei2, E.D. Birdeanu1 and V. Iancu1

1Faculty

of Engineering, Eftimie Murgu University, Resita 320085, Romania

gr.gillich@uem.ro (G.R. Gillich), e_d_birdeanu@yahoo.com (E.D. Birdeanu), v.iancu@uem.ro (V. Iancu)

2Center

of Advanced Research, Design and Technology, Eftimie Murgu University, Resita 320085, Romania d.amariei@uem.ro (D. Amariei)

Abstract. Researches concerning structures safety improvement have increased significantly in the past decades, structural health monitoring becoming an important support in taking decision regarding the need of these structures remediation. Various techniques, from visual inspection to nondestructive evaluation or vibration-based damage detection are used. The last ones use information regarding changes of natural frequencies, damping ratios, flexibility or mode shapes. Even if numerous results have been achieved in this field, several problems still have to be solved for specific applications before their implementation in practice. Our study approaches the methods viability of damage detection in simple structures using only one accelerometer by analytical method, finite element method and experimental test. The obtained results by all three methods showed that this approach can be used only in high damage rates cases. Finally, are pointed some issues which have to be further worked out.

1 Introduction
Non-destructive testing, including monitoring and diagnostics, is nowadays widely practiced in mechanical and civil engineering. It influences our approach of designing of new structures and monitoring and maintenance of existing ones. From the series of now available non-destructive testing methods, the dynamic ones grow in importance comparing with conventional diagnostic methods, like the once based on visual inspection or ultrasonic analysis. This is motivated by the global perspective offered by dynamic techniques (tests developed over time are able to indicate the appearance of possible damages in the structure), comparing with the local nature of conventional methods (which require good preliminary knowledge about the position of the area where the damage is located). Another advantage of the dynamic methods is linked to the fact that these techniques do not requested access to the damaged area, while conventional methods require direct accessibility to the area to be inspected. However, the two methods can be used complementary; they do not exclude each other [1]. Numerous researches have been performed in order to develop and prove the viability of dynamic methods. In this context, important progresses made on the technological field have generated accurate and reliable experimental methods, while the interpretation of measurement results remains a bit behind. Morassi and Vestroni [2] and Zhou [3] presented a series of vibration based damage detection techniques, for structures and simple elements, based on: natural frequency shifts, damping, change of mode shapes, flexibility of structures and model updating. Some authors focused their researches on simple elements, like Choi et al. [4] on simple supported beams or Rucka [5], Barbosa et al. [6] and Xia et al. [7] on cantilever beams. It has to be mentioned that all methods base on processing data of change of natural frequency and shape modes. Similar researches have been done by the authors [8] in order to show that it is possible to detect damages in beams, without concentrating on the localization or characterization of the damage. The

researches developed last time, presented in this paper, had the intention to determine the sensitivity of various types of structures (beams) by different types and locations of damages.

2 Theoretical Investigations
The aim of our research is to determine the natural frequencies for a steel beam in undamaged state, supported in different ways, and to compare it with the corresponding harmed beam. This helps us to find out if, for the analyzed cases, the shift in frequency provides enough information to identify and locate damages. In our theoretical investigations the analytical method and the finite element method were used. 2.1 Analytical Method At first a steel cantilever beam, clamped on one end (point - as shown in figure 1) and free on the other (point in figure 1), is considered as testing structure. The beam has the length l = 1000 mm, wide b = 50 mm and height h = 5 mm and consequently a cross-section A = 250 10 6 m2 and moment of inertia I = 520 .833 10 12 m4 respectively. The producer indicates following material parameters: mass density = 7850 ratio = 0.3. kg/m3, Youngs modulus E = 2.01011 N/m2 and the Poissons

Fig. 1. Configuration of the cantilever beam

Afterwards similar beams, simple supported and clamped on both ends (points and ) respectively, are considered, like it is presented in figure 2. The accelerometer was mounted in both this cases close at the middle of the beam (point ).

a.

b. Fig. 2. Scheme of the simple supported beam (a.) and double clamped beam (b.)

To find out the analytical calculated natural frequencies for the undamaged cantilever beam, we started from the beam equations [9]:

4 v A 2 v + =0 x 4 EI t 2

(1)

where v is the vertical displacement of the beam at distance x measured from the free end. Considering that v depends on distance x and time t, and the evolution in time is harmonically, the expression of displacement is:
v = X ( x ) T (t ) = X sin t

(2)

After derivation and substitution in relation (1), one obtains:

X IV
with the solution:

A2 X =0 EI

(3)

X = A sin x + B cos x + C sinh x + D cosh x

where we noted

(4)

A2 = 4 . EI
After three derivations of the solution (4) results the system:

(5)

X = A s in x + B c o s x + C s in h x + D c o s h x X = ( A c o s x B s in x + C c o s h x + D s in h x ) 2 X = ( A s in x B c o s x + C s in h x + D c o s h x ) X = 3 ( A c o s x + B s in x + C c o s h x + D s in h x
1 + cos cosh = 0 ,

(6)

Putting the boundary condition for the clamped cantilever beam (to find the constants A, B, C and D) X (0) = X (0) = 0 , and X (l ) = X (l ) = 0 , one obtains the equation (7)

with = l , which permits to calculate the i values for i vibrations modes. Multiplying relation (5) with l 4 and substituting the values of i , results the analytic calculated angular frequencies i , and consequently the natural frequencies of the undamaged cantilever beam:

fi =
the first six being presented in table 1.

2 i 2

EI , Al 4

(8)

Similar, by putting the corresponding boundary conditions for the undamaged simple supported and double clamped beam respectively, one obtains the natural frequencies presented also in table 1.
Table 1. Natural frequencies of undamaged beams, differently supported, calculated analytically

Mode 1 Cantilever beam 4.077

Natural frequency fi [Hz] Simple supported beam 11.444

Double clamped beam 26.021

2 25.549 3 71.539 4 140.189 5 231.742 6 346.182 2.2 Finite Element Method

45.776 102.996 183.105 286.101 411.985

71.727 140.614 232.441 347.227 484.970

The FEM computation was made on 3D model beams, using the first ten modal structural analysis types, because some bending vibrations modes on orthogonal direction and torsion vibrations modes are interfered. For meshing, it was used a 5 mm element size; the weight of the accelerometer was neglected. The first and fifth mode shapes for the undamaged beams (cantilever, simple supported and double clamped) are presented in figure 3, while their first six natural frequencies in vertical direction for the intact and damaged state are presented in table 2. Following notations are used: - U for the undamaged beam - D1 for the beam with one damage at point - figure 1 - D2 for the beam with two damages, at point and point - figure 1

a.

b.

c.

d.

e.

f.

Fig. 3. Mode shapes for undamaged cantilever beam (a. first mode and b. fifth mode), simple supported beam first mode and d. fifth mode) and double clamped beam (e. first mode and f. fifth mode)

(c.

Comparing tables 1 and 2, it is obvious that the natural frequencies of beams determined by the FEM are close to them determined analytically, so we concluded that the used methodology was correct.
Table 2. Natural frequencies of undamaged and damaged beams, differently supported, determined with FEM

Mode

1 2 3 4 5 6

Natural frequency fi [Hz] Cantilever beam Simple supported U D1 D2 U D1 4.091 4.079 4.0548 11.444 11.289 25.634 25.302 25.100 45.779 45.775 71.780 71.767 71.360 103.010 101.650 140.690 138.940 138.620 183.140 183.070 232.650 232.530 229.720 286.150 282.450 347.690 343.440 343.070 411.960 411.640

Double clamped U D2 26.241 25.990 72.317 72.036 141.770 139.850 234.380 234.260 350.220 345.660 489.330 488.840

3 Experimental Researches
The experimental tests performed an analysis of the steel beams dynamic behavior for two situations: undamaged and with damage on its middle (point in figure 4). The cantilever beam described in chapter 2 was mounted in a rigid support. The measurement system (figure 4), composed by a Toshiba laptop, a NI cDAQ-9172 compact chassis with NI 9234 four-channel dynamic signal acquisition modules and a Kistler 8772 accelerometer has been used for signal acquisition.

1 3

Fig. 3. Experimental stand for frequency analysis of cantilever beams The LabVIEW programming environment was used to develop the virtual instrument which acquired the time history of acceleration and realized the spectral analysis. A window presenting the acquired and processed signal is shown in figure 5.

To measure the accelerations on vertical direction for the free end of the beam (point in figure 4) a vertical force was applied to bring the mechanical system out of its equilibrium position. Suppressing the force, the beam started to vibrate. We recorded and stored the acceleration values using the acquisition system, for the undamaged beam; repeating the process for nine times. Afterwards damages on the middle (point in figure 4) of the beam were produced by saw cuts, like those presented in detail A in figure 1, and a new series of measurement realized.

Fig. 5. Acquired and processed signal (time history of acceleration and spectral analysis)

All signals were processed in order to obtain their frequency spectrum. Using the pick detector we extracted the natural frequencies of the undamaged and damaged steel cantilever beam respectively. The mean values for the first six natural frequencies are presented in table 3.
Table 3. Natural frequencies of undamaged and damaged cantilever beam

Mode

1 2 3 4 5 6

Natural frequency fi [Hz] Cantilever beam U D1 4.000 4.000 24.000 24.000 70.000 69.000 134.000 131.000 226.000 222.000 335.000 329.000

4 Centralized Results
Results, obtained by simulations and measurements, are centralized in tables 4 to 7, in order to provide an overview about the behaviour of damaged beams and the utility of the investigation methods applied. Because the methodologies are focused on the shift of the natural frequency, also presented in these tables, we tried to find out relevancy for detection, characterisation and localization of damages.

Table 4. Shift of natural frequencies for the cantilever beam, determined by using the FEM

Mode 1 2 3 4 5 6 U 4.091 25.634 71.780 140.690 232.650 347.690 D1 4.079 25.302 71.767 138.940 232.530 343.440

Natural frequency fi [Hz] Change U-D1 D2 Change D1-D2 -0,293 % 4.0548 -0.588 % -1,295 % 25.100 -0,798 % -0,018 % 71.360 -0,576 % -1,244 % 138.620 -0,230 % -0,052 % 229.720 -1,208 % -1,222 % 343.070 -0,108 %

Change U-D2 -0,880 % -2,083 % -0,585 % -1,471 % -1,259 % -1,329 %

Table 5. Shift of natural frequencies for the cantilever beam, determined by measurements

Mode 1 2 3 4 5 6 U 4.000 24.000 70.000 134.000 226.000 335.000

Natural frequency fi [Hz] D1 Change U-D1 3.975 -0,625 % 24.000 0,000 % 69.000 -1,429 % 131.000 -2,239 % 222.000 -1,770 % 329.000 -1,791 %

Table 6. Shift of natural frequencies for the simple supported beam, determined by using the FEM

Mode 1 2 3 4 5 6 U 11.444 45.779 103.010 183.140 286.150 411.960

Natural frequency fi [Hz] D1 Change U-D1 11.289 -1,354 % 45.775 -0,009 % 101.650 -1,320 % 183.070 -0,038 % 282.450 -1,293 % 411.640 -0,078 %

Table 7. Shift of natural frequencies for the simple supported beam, determined by using the FEM

Mode 1 2 3 4 5 6 U 26.241 72.317 141.770 234.380 350.220 489.330

Natural frequency fi [Hz] D1 Change U-D1 25.990 -0,957 % 72.036 -0,389 % 139.850 -1,354 % 234.260 -0,051 % 345.660 -1,302 % 488.840 -0,100 %

5 Conclusions
Previous studies present various methods to the structures integrity dynamic evaluation, without specifying the conditions for which the methods are recommended to be applied. This implies, by structures dynamic analysis, a previous good experience for engineers in the field of structural health monitoring. Our research, partially presented in the actual paper, intends to bring novelty in this field,

providing recommendations about the methods to be used for different cases in dynamical evaluation of structures. Examining the results for the cantilever beam, presented in tables 1 to 3, we remark a good concordance between the natural frequency values obtained by analytical calculus, deduced with the FEM and the experimental ones. This proves that correct theoretical algorithms and high-accuracy measurement equipment and methodologies have been used. On the other side the results presented in tables 4 to 7 show that the shifts in frequency for the analyzed structures, even for damages which reduce dramatically the moment of inertia, dont exceed 3%, being for almost the large majority of the cases lower than 1.5%. It is also to be mentioned that increased number of damages does not increase dramatically the shift in frequency. This makes methods based alone on the shift of frequency not suitable, the results have to be completed with other information about the structures behaviour. As next researches the authors plan to extend the experiments on trusses as well as on plates too.

Acknowledgement
The authors acknowledge the support of the Romanian Ministry of Labor, Family and Social Protection by the Managing Authority of Sectorial Operational Programme for Human Resources Development (AM POSDRU) for creating the possibility to perform these researches by Grant POS DRU 5159 - PhDStudents for Innovation and Competitiveness support, co-financed by structural funds.

References
1. Balageas D., Fritzen C.P., Gemes, A.: Structural health monitoring. London: ISTE Ltd, 2006 2. Morassi, A., Vestroni, F.: Dynamic Methods for Damage Detection in Structures, CISM Courses and Lectures, Vol. 499, Springer Wien New York (2008) 3. Zhou, Z.: Vibration-based damage detection of simple bridge superstructures, Doctoral Thesis, University of Saskatchewan, Saskatoon (2006) 4. Choi, F.C., Li, J., Samali, S., Crews, K.: An Experimental Study on Damage Detection of Structures Using a Timber Beam, Journal of Mechanical Science and Technology 21 (2007) 903-907 5. Rucka, M.: Experimental and numerical studies of guided wave damage detection in bars with structural discontinuities, Arch Appl Mech. (2009) Published online 6. Barbosa, F.S., Cury, A.A., Vilela, A.R., Borges, C.C.H., Cremona, C.: Damage evaluation of structures using dynamic measurements, Proceedings of the Experimental Vibration Analysis for Civil Engineering Structures EVACES 05 Bordeaux (2005) 7. Xia, Y., Hao, H., Brownjohn, J., Xia, PQ.: Damage identification of structures with uncertain frequency and mode shape data, Earthquake Engineering and Structural Dynamics 31 (2002) 10531066 8. Gillich, G.-R., Birdeanu, E.D., Gillich, N., Amariei, D., Iancu, V., Jurcau, C.S.: Detection of damages in simple elements, Proceedings of the 20th International DAAAM Symposium, Volume 20 No. 1 (2009) 0623-0624 9. Buzdugan, Gh.: Material strenght (Rezistena materialelor in Romanian in original). 10th edition, Editura Tehnica Bucuresti (1974) 448-453