Detection of multiple defects based on structural health monitoring of pipeline using guided waves technique
Monitoring and inspecting the health condition and state of the pipelines are significant processes for an early detection of any leaking or damages for avoiding disasters. Although most Non Destructive Test (NDT) techniques are able to detect and locate damage during the maintenance intervals, inte...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English English |
| Published: |
2018
|
| Subjects: | |
| Online Access: | http://eprints.uthm.edu.my/188/1/24p%20HATEM%20MOSTAFA%20ELWALWAL.pdf http://eprints.uthm.edu.my/188/2/HATEM%20MOSTAFA%20ELWALWAL%20COPYRIGHT%20DECLARATION.pdf http://eprints.uthm.edu.my/188/3/HATEM%20MOSTAFA%20ELWALWAL%20WATERMARK.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Monitoring and inspecting the health condition and state of the pipelines are significant processes for an early detection of any leaking or damages for avoiding disasters. Although most Non Destructive Test (NDT) techniques are able to detect and locate damage during the maintenance intervals, interrupted services could result in high cost and lots of time consumed. In addition, most NDTs are utilized to detect and locate single damage such as axial crack, circular crack, or vertical crack only. Unfortunately, these NDTs are unable to detect or localize multi-type of damages, simultaneously. In this research, the proposed method utilizes the Structural Health Monitoring (SHM) based on guided wave techniques for monitoring steel pipeline continuously in detecting and locating multi-damages. These multi damages include the circumference, hole and slopping cracks. A physical experimental works as well as numerical simulation using ANSYS were conducted to achieve the research objectives. The experimental work was performed to validate the numerical simulation. An artificial neural network was used to classify the damages into ten classes for each type of damage including circumference, hole and sloping cracks. The obtained results showed that the numerical simulation was in agreement with the experimental work with relative error of less than 1.5%. In addition, the neural network demonstrated a feasible method for classifying the damages into classes with the accuracy ranged from 75% to 82%. These results are important to provide substantial information for active condition monitoring activities. |
|---|