You are on page 1of 1

Fluid Structure Interactions Research Group

Thermoelastic Stress Analysis for Damage Assessment


Richard Hebb rih103@soton.ac.uk - School of Engineering Sciences Supervisors Prof. J. M. Barton, P. Tatum (AWE) Aims and Objectives
To develop understanding of stresses around a crack tip using Thermoelastic Stress Analysis (TSA). Locate and identify both internal and external damage on small scale pipes. Develop a non-contact method of excitation for TSA.

Analysis of Data
The genetic algorithm is used to fit a cardioids to extracted isopachics from the data. The cardioids can be seen to rotate anti-clockwise with increasing distance from the crack-tip which the first order Westergaard equation [3] is unable to account for.
Cardioid Number
1 2

Thermoelastic Stress Analysis


The relationship between the change in stresses around a crack tip and the temperature change in the material (under adiabatic conditions) is given by the Williams expansion:
AS 2 K I2 2 r
2 K II

x (pixels (mm))

y (pixels (mm)

r0 (pixels (mm)

2 (radians)
0.289 0.393

Fitness 0.35978 0.49101

45.75 (13.7) 89.30 (26.8) 89.02 (26.7) 41.99 (12.6) 80.79 (24.2) 53.86 (16.2)

cos 2

K II tan KI
-1

Ts

2 BI

2 II

rcos 2

B II tan BI
-1

O r

3 2

[1]

3
4 5

40.23 (12.1) 76.82 (23.0) 37.98 (11.4)


39.41 (11.8) 77.39 (23.2) 29.48 (8.8) 39.30 (11.8) 77.11 (23.1) 23.82 (7.1)

0.338
0.442 0.497

0.61657
0.58084 0.55064

A = Constant for given loading and boundary conditions BI,II = Constants S = Signal produced by the detector for given r KI and KII = SIFs O = Higher order terms

Ts = T-stress constant term = Coefficient of linear thermal expansion Cp = Specific heat at constant pressure = Mass density

It can then be shown the Williams expansion can be related to the temperature change induced by:
T T AS [2] C p

Results for fitting a cardioid to the extracted isopachics

It has been shown that the previously omitted higher order terms are the cause of the extra rotation.

Therefore measuring the temperature change allows the Stress Intensity Factors (SIFs) to be calculated. Ignoring the higher terms and re-arranging shows that a curve of constant signal takes the form of a cardioid: 2 KI2 KII r 1 cos 2 [3] 2 2 A S

Experimental Technique
A crack is grown in a Dural plate (with a centrally located spark eroded starter slot) by fatiguing the plate below its fracture toughness. For Dural this is 19.0 MPa m0.5 which can be used to calculate the applied stress required by: k = 19.0 MPa m k 1C a = semi-crack length = 20mm [4] C = Geometry dependant constant = 1.004 C a
1C 0.5

This relates to an applied stress of 75.5 MPa for the plates, or an applied load of 49.15 kN. The crack is grown to approximately 30mm, with TSA being performed at regular intervals. Once the crack has grown, a smaller plate is cut from the specimen with the grown crack at the centre at an angle of 15, 30, 45 and 60 degrees, thus giving specimens containing mixed-mode, centrally located cracks. TSA is then performed on the mixed-mode specimens.

Plot showing contours of first order (black) and higher order (red) Williams expansion with the crack line (blue).

Simulated Data and Higher Order GA


As opposed to fitting a cardioid to isopachics, more accurate SIF calculation is possible by fitting the full Williams expansion to the entire field. Simulated fields are being created both as a control mechanism for the modified GA and to gain knowledge of the influence of the higher order terms by varying individual values.

Schematic of the specimens and mounts. Only the top mount has been shown for clarity.

Analysis Technique
Isopachics extracted from the data are fed into a genetic algorithm which fits a cardioid curve to the raw data. The fitness of the fit is determined by the inverse of the mean square error for the curve fit: 100 N 2 M ri rie [5] 2 N r i1 Fitness = M-1as SIFs then calculated by simultaneously solving: 2 K II T Cp r0 2 2 tan [7] K I K II [6] 2 T KI

Actual data set

A simulated set of data

Work is currently ongoing to determine the physical meaning and values of the higher order terms.

Further Work
Completion of the genetic algorithm. Alternate excitation methods of the plate (ultrasound, shaker at natural frequency, transient load). FEA analysis of pipe work to find natural frequencies, expected response from TSA. Experimental work on pipe work, including alternate non-contact excitation methods.

You might also like