Analysis of simulated C-Scans in Steel pins

ResumeSam Parikh (ESM) and Fernando Das Neves (Computer Science)

Introduction

Steel pin details are present in bridge structures throughout the United States. These pins are fracture critical members that are impossible to inspect visually1. Perhaps the most infamous instance of the need for inspection occurred on June 8, 1983 when portions of US Interstate-95 in Greenwich, CT collapsed killing three persons. The inspection of steel pins is complicated because of limited access to the areas of the pin where critical cracks typically occur. Added to this complication is the fact that wear due to surface abrasion will often cause grooves to develop in these same locations. The grooves although an unwanted form of deterioration do not pose the same level of concern as fatigue cracks. Consequently, detecting pin deterioration and differentiating between wear grooves and cracks are essential to assuring safe use of the associated bridge structure.

The Radford Bridge, one of the many bridges constructed using steel pins.

A closer look at the Radford Bridge, showing one of the steel pins.

In addition, once the presence of a crack has been established it is desirable to monitor the rate of crack growth for situations where the crack length is below the critical value that would result in immediate replacement. Several attempts have been made to use ultrasonic inspection to detect and distinguish cracks in steel pins, yet no standard approach has been developed for solving this important problem2-11 . In order to more thoroughly evaluate a pin in the field a procedure for performing an ultrasonic C-scan has been developed and demonstrated in the field12. This procedure provides high resolution images of imperfections detected using manual ultrasonic A-scan inspection. The interpretation of these images is not straight forward, since the ultrasound interaction with the pin geometry as well as the uncertain nature of the imperfection introduce complications. In order to assist with the interpretation of these images a computer algorithm to model the ultrasound interaction with the pin has been developed.

The Project:

Ultrasonic C-scans scans ares one of the methods available to analyze the state of steel pins in bridges.A C-Scan is the plotting of a collection of gated amplitudes of ultrasonic A-Scan signals. An A-Scan is the amplitude of the signal versus time/depth, given that in steel pines the time is directly proportional to the speed of sound. In the field, when C-Scans are done for special cases, the C-Scans do not give us geometric description or geometric placement or crack/ware grove within the solid steel pins. Since the speed of sound if constant within a homogeneous steel pin the time can be directly related to the depth.

Image of a C-Scan of a Steel pin, showing a crack. The scale above indicates amplitude.

This picture shows the probe mounted over a pin, ready to scan.

Plotting of C-Scans is done in color-directly proportional (linear) to the flaw amplitude compared to a given amplitude within a certain range. The problem with analyzing this plotting to extract flaw information (depth dimension) within the pin is that, due to the sound field propagating in a conical zone (the beam spread), a reflection can occur anywhere within that zone. Even if the beam is picked by the probe, that does no put the flaw that cause the reflection immediately beneath the probe. The current analysis of C-Scans relates the crack size and dimension to what appears on the C-Scan, and put the crack in cross-section plane parallel to the crack image that was pictured in the C-Scan.

Scope of this project:

The scope of this project is to use a ray-trace model to get A-Scans after inserting a mathematical model of cracks within a model of a solid steel pin. The A-Scans will then be reconstructed into C-Scans and will show what that particular crack would look like on the C-Scan image.. We take data generated by the mathematical model inputting various cracks at different depths and geometric dimensions, and to compare the C-Scan images generated. We also would like to try out different beam spreads and to see whether that makes any worthwhile difference.

Proposal:

We will create a program that implements the mathematical model and will use AVS to manipulate the data created by the model. We will write new AVS modules to manipulate the scatter data generated by the program. A further extension of the project is to analyze the 3D dataset by displaying it on the CAVE via NCSA's Crumbs. We will write a module that allows to ouput the volumetric data from AVS directly in Crumbs format.