Finite Element Modeling of Flash Thermographic Detection of Defect - - PowerPoint PPT Presentation

NSF REU EMCoR@NCAT Grant # ACI-1560385

Finite Element Modeling of Flash Thermographic Detection of Defect in a Steel Plate

REU Fellow: Eric Feigin Mentor: Dr. Mannur Sundaresan

July 27, 2018

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Motivation

▪ Defects are inevitable in structures that are made of metal, composites

• r hybrid

▪ During flight – runway debris or foreign particles ▪ During fabrication or maintenance – dropping tools, walking over parts ▪ To identify these defects Non Destructive Evaluation (NDE) technologies are used ▪ Nondestructive testing is a wide group of analysis techniques used in science and technology industry to evaluate the properties of a material, component or system without causing damage. ▪ NDE techniques are Acoustic Emission, ultrasonic testing, thermography etc.

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Motivation

▪ Flawless manufacturing techniques are rarely achieved in the field, so inspection techniques are vital. ▪ Pulse or flash thermography is used to display differences in the decay of heat throughout a surface from a normal thermal flow in order to detect subsurface defects. ▪ Difficulties arise in simulating multiple differences in multiple materials to generate accurate results that can be experimentally calibrated

▪ Used to compare to field results to identify unknown defects and their dimensions. ▪ Takes a lot of time to create multiple thermographic profiles examining temperature with relation to time/length of profile

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Motivation

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Literature Review

▪ Sundaresan, M., & Sripragash, L. (2016). A normalization procedure for pulse thermographic nondestructive. NDT&E International, 14-23.

▪ Thermography

▪ Inspector uses infrared camera (IR) in a passive mode to measure the steady state surface temperature of a component ▪ Assesses large areas of structures in a relatively short duration of time, with heat applied to the surface of the test object by an external energy source (IR camera). ▪ Due to heat pulse, temperature rises instantly resulting in change of surface temperature.

▪ Rate of change of surface temperature is function of time.

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Literature Review

▪ Krishnapillai, M., Jones, R., Marshall, I., Bannister, M., & Rajic, N. (2006). NDTE using pulse thermography: Numerical modeling. Composite Structures, 241-249. ▪ Pulse Thermography

▪ A flash or pulse of heat is introduced for a fraction of a second ▪ Used to display differences in the decay of heat throughout a surface from a normal thermal flow.

▪ Thermography Technique Benefits

▪ Scanning broad areas ▪ Efficient prediction capabilities for anomalies underneath surfaces

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Literature Review

▪ Sundaresan, M., & Sripragash, L. (2016). A normalization procedure for pulse thermographic nondestructive. NDT&E International, 14-23. ▪ Temperature Contour Plot

▪ Measures temperature of pixels with respect to time and length along the profile ▪ Shows heat flow through surface with curved lines showing effect of defect

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Literature Review

▪ Sundaresan, M., & Sripragash, L. (2016). A normalization procedure for pulse thermographic nondestructive. NDT&E International, 14-23. ▪ 𝛽 is the diffusivity of the plate given by

𝛽 =

𝜆 𝑞𝑑

▪ With 𝜆 as thermal conductivity, 𝑞 as density, and 𝑑 as heat capacity of the plate.

𝑢∗ =

𝑀2 𝜌𝛽

▪ With L as the thickness of the plate

t*

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Literature Review

▪ Shepard, S. M., Lhota, J. R., Rubadeux,

• B. A., Wang, D., & Ahmed, T. (2003).

Reconstruction and enhancement of active thermographic image sequences. Optical Engineering.

▪ Thermographic Signal Reconstruction

▪ Time-derivative images help identify subsurface defects ▪ A logarithmic scale is used to compare the time history of every pixel in the field of view, as changes from ideal behavior are easily identifiable

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Literature Review

▪ Thermographic Signal Reconstruction (cont.) ▪ Use time of first peak in second derivative of natural log of time versus natural log

• f pixel temperature to

determine defect depth 𝑢𝑒 =

𝑀𝑒2 𝜌𝛽

▪ In which td is the time and Ld is the defect depth

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Literature Review

▪ Mucha, M. (n.d.). Heat transfer analysis in ABAQUS. Kraków, Poland. ▪ Bathe, K.-J. (2009). Why to Study Finite Element Analysis. Cambridge, Massachusetts, United States

▪ The thermography process can be simulated using Finite Element Analysis tools such as ABAQUS.

▪ Finite element analysis is a numerical method for solving problems of engineering and mathematical physics. ▪ It subdivides a large problem into smaller, simpler parts that are called finite elements. ▪ The simple equations that model these finite elements are then assembled into a larger system of equations that models the entire problem.

▪ Abaqus is a Finite Element Analysis software that can simulate incredibly complicated components, structures and systems under a wide variety of situations and loading conditions.

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Research Goals

▪ Examine effect of defect on steel plate through plots for thermographic profile

▪ Use AISI 1018 Mild/Low Carbon Steel ▪ Use Abaqus to build a model and simulate transient heat transfer with pulse thermography ▪ Have temperature contour plot displaying effect of defect on heat flow ▪ Create graph of time versus temperature for nodes

▪ First and Second Derivative plots as well to measure expected numerically calculated depth to actual defect depth

▪ Be able to use my results in application to different metals, composites, hybrids whose properties are known

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Methodology

▪ Abaqus takes inputs for boundary conditions, such as the geometry of the surface or part, the type of load applied, the amplitude of the load, which set the load will be applied across, etc. ▪ Mesh is an arrangement of finite elements defined on an FEA model ▪ Once Abaqus has defined elements, it can solve multiple differential equations

▪ The user does not have to perform any physical calculations, Abaqus evaluates the temperature between steps of time (either user-inputted or automatically generated).

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Methodology

▪ Abaqus Use and Model Creation ▪ Following (Mucha) in steps for building model and running heat transfer analysis ▪ Geometry – 1.5 inch length by .5 inch height initial plate (cross section)

▪ .5 inch length by .4 inch height defect from bottom left corner

▪ Properties of AISI 1018 Mild/Low Carbon Steel used – density, conductivity, specific heat capacity

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Methodology

▪ Increment minimum of .0001 seconds and maximum of .05 ▪ Field/History Output request – only thermal data extracted ▪ Amplitude – temperature rises to 100000 degrees Celsius, remains for 1 millisecond, drops to zero for duration of job ▪ Meshed with approximate size of .0025 meters for each element.

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Methodology

▪ Numerical Modeling ▪ Abaqus, a finite element analysis software, finds the temperature change over time for each individual element/node ▪ Data is exported to Excel through Abaqus’ Excel plug-in ▪ First set of graphs

▪ Get rid of repeated data in Excel (time is repeated for each set of nodal temperatures) ▪ Use MATLAB

▪Natural log of time versus natural log of temperature using log() command ▪Use diff () function to create first and second derivatives of ln(temperature) with respect to ln(time) and plot versus ln(time)

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Methodology

▪ Contour Plot

▪ Create vector in MATLAB of nodal distance – distance divided by number of nodes with successive distance along path corresponding to node number ▪ Use contour and contourcbar commands in MATLAB to create contour plot of temperature with respect to distance along surface as well as natural log of time ▪ Download Mapping Toolbox for MATLAB if not working with full version ▪ Plot shows effect of defect not only for time but for how temperature changes along the surface in the x direction

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Results

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Results

▪ Abaqus results – see change of heat wave throughout surface ▪ Different colors at end as compared to animation – nodal temperature values very similar at end of simulation

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Results

▪ Time versus Nodal Temperature plot – different color for each node ▪ Temperature rises instantaneously due to flash, cools uniformly at once, then effect of defect is visible ▪ First and final nodes begin to split in temperature around t = .26 seconds

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Results

▪ Same plot with middle nodes removed to more clearly see thermal behavior of nodes ▪ Top line is initial node – top left corner of plate, bottom line is final node ▪ Exhibits behavior resembling that

• f plot in literature review with

node directly above defect remaining higher in temperature at first.

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Results

▪ Although choppier than the first derivative plot shown prior, the graph fits the general shape of the expected first derivative plot. ▪ With normalization, a flatter shape prior to ln(t) = 0 could be seen ▪ Sharp line in the beginning is just due to bounds, examining behavior directly after the pulse

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Results

▪ This graph was generated in the last week of the REU program. ▪ Clearly, it does not resemble the second derivative plot included earlier from prior research. ▪ However, the first derivative plot resembles the shape it should, so with more time spent working in MATLAB, a better second derivative plot could be created. ▪ Had a more accurate graph been generated, the first peak of the first node’s temperature line could be used to check the accuracy of our model by using 𝑢𝑒 =

𝑀𝑒2 𝜌𝛽 to estimate the defect depth.

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Results

▪ This figure shows the effect of the defect on temperature along the profile. ▪ Temperature decreases nearly linearly for most nodes, until the nodal temperatures split around e^2 to e^3 seconds into the experiment. ▪ Then, temperature is seen to drop following a logarithmic pattern resembling the previous literature plot. ▪ The curved lines aren’t in the exact spots but fit the general pattern as

• ther defects also would.

North Carolina Agricultural and Technical State University EMCOR@NCAT NSF Grant

Research Goals and Outcomes

▪ Examine effect of defect on steel plate through plots for thermographic profile

▪ Natural Log Graph of Time vs. Temperature

▪ Successful – matched behavior of ideal graph ▪ Defective area resisted cooling, eventually approached same temperature as sound area

▪ First Derivative of Natural Log Graph of Time vs. Temperature

▪ Mainly Successful – Choppy but fits ideal patterns

▪ Second Derivative of Natural Log Graph of Time versus Temperature

▪ Not Successful – Issues with MATLAB, not enough time to create working plot

▪ Temperature Contour Plot

▪ Successful – Fits ideal patterns, slightly different location of curves possibly due to change in defect size

Conclusion/Discussion

NSF REU EMCoR@NCAT Grant # ACI-1560385

▪ Research Goals ▪ Be able to use my results in application to different metals, composites, hybrids whose properties are known

▪ My work has incorporated computer aided design modeling into thermographic nondestructive evaluation, with results that should increase the use of CAD for generating thermographic profiles. ▪ My results show that the plots necessary for building a thermographic profile to identify defects can be created. ▪ Although the second derivative plot was not successful, the general shape

• f the natural log plot of nodal temperatures was correct, as was that of the

first derivative.

Future Work

NSF REU EMCoR@NCAT Grant # ACI-1560385

▪ The plots could be fine-tuned for the second derivative with more work in MATLAB, and/or possibly taking more precise data in Abaqus with smaller step values. ▪ Once updated, my model’s graphs could be used as a standard thermographic profile for comparing to field results for this specific defect (probably have to be normalized)

▪ Would lead to generating hundreds or thousands of defect combinations for thermographic profiles.

▪ An extension of my research would most likely involve Python scripting which I did not use due to time constraints and the low number of models created.

▪ Spent majority of time learning Abaqus, but actual model building and data extraction was quick, already indicative of efficiency of process