Astm stp 1092 1990

SUNDER Collection and Evaluation of Fatigue and Fracture Mechanics Data According to the European High Temperature Materials Databank Petten Standard-- HANS-HELMUT OVER AND BRUNO BUCH

STP 1092

Applications of Automation

Technology to

Fatigue and Fracture Testing

Arthur A Braun, Noel E Ashbaugh, and Fraser M Smith, editors

AsT

1916 Race Street

Philadelphia, PA 19103

Library of Congress Cataloging-in-Publication Data

Applications of automation technology to fatigue and fracture testing/

Arthur A Braun, Noel E Ashbaugh, and Fraser M Smith,

editors

(STP: 1092)

Papers presented at a symposium held in Kansas City, Missouri,

22-23 May 1989, sponsored by ASTM Committees E9 on Fatigue

and E24 on Fracture Testing

"ASTM publication code number (PCN) 04-010920-30" T.p

verso Includes bibliographical references and index

ISBN 0-8031-1401-X

1 Materials Fatigue Testing Congresses 2 Materials

Fracture Testing Congresses I Braun, Arthur A.,

1953- II Ashbaugh, Noel E., 1940- III Smith,

Fraser M., 1959-

IV ASTM Committee E9 on Fatigue V ASTM Committee E24

on Fracture Testing VI Series: ASTM special technical

publication; 1092

Copyright 9 1990 by the American Society for Testing and Materials All rights reserved9

No part of this publication may be reproduced, stored in a retrieval system, or transmitted,

in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise,

without the prior written permission of the publisher

NOTE The Society is not responsible, as a body, for the statements and opinions advanced in this publication

Peer Review Policy

Each paper published in this volume was evaluated by three peer reviewers The authors

addressed all of the reviewers" comments to the satisfaction of both the technical editor(s)

and the ASTM Committee on Publications

The quality of the papers in this publication reflects not only the obvious efforts of the

authors and the technical editor(s), but also the work of these peer reviewers The ASTM

Committee on Publications acknowledges with appreciation their dedication and contribution

of time and effort on behalf of ASTM

Printed in Ann Arbor, MI November 1990

Foreword

The papers in this publication, Applications of Automation Technology to Fatigue and

1989 The symposium was sponsored by ASTM Committees E9 on Fatigue and E24 on

Fracture Testing Arthur A Braun, MTS Systems Corporation, Noel E Ashbaugh, Uni-

versity of Dayton, and Fraser M Smith, Sarcos Research Corporation, presided as co-

chairmen of the symposium and are coeditors of this publication

A PC-Based Data Acquisition System for Composite Material Fracture

Testing GLENN E COLVIN, JR., AND STEPHEN g SWANSON

A Test System for Computer Controlled Theromechanical Fatigue Testing

Simulation of Mechanical and Environmental Conditions in Fatigue Crack Growth

Fatigue Crack-Growth Measurement Using Digital Image Analysis Technique

ALEX S REDNER, ARKADY S VOLOSHIN, AND ARVIND NAGAR

M e t h o d - - v A L E R I O BICEOO, D1NO LIVIERO, CARLO FOSSATI, AND

Use of Automated Ball Indentation Testing to Measure Flow Properties and

Estimate Fracture Toughness in Metallic MaterialS FAHMY M HAGGAG,

RANDY K NANSTAD, JOHN T HUTYON, DAVID L THOMAS, AND

LABORATORY SYSTEMS AND INFORMATION MANAGEMENT

Automation Software for a Materials Testing Laboratory MICHAEL A McGAW AND

PETER J BONACUSE

Automated Fatigue and Fracture Laboratory with Multiple Load Frames and Single

Host Computer System g SUNDER

Collection and Evaluation of Fatigue and Fracture Mechanics Data According to

the European High Temperature Materials Databank (Petten) Standard

HANS-HELMUT OVER AND BRUNO BUCHMAYR

A Data Acquisition and Control Program for Axial-Torsional Fatigue Testing

SREERAMESH KALLURI AND PETER J BONACUSE

Overview

STP1092-EB/Nov 1990

The continuing advancement of computer and software technology has allowed for the

automation of materials testing systems and processes to become commonplace Automa-

tion, which was at first a very expensive and complicated accessory to a materials testing

system, is now a inexpensive and often necessary subsystem Many test techniques now

require the speed, consistency, and computational capability inherent in these systems

Hardware cost~ have continued to spiral downward in conjunction with incredible increases

in computational bandwidth, display technology performance, and mass storage capacity

and speed Software technology, the real key to forward progress, has improved significantly,

allowing for shorter application development time with higher application performance This

is especially true in the area of real-time systems software which is critical for testing system

control and data acquisition

This symposium is the third in a series of symposia concerned with the advancement of

the state of the art in automated fatigue and fracture testing The first was the Use of

Computers in the Fatigue Laboratory held in New Orleans, Louisiana in November of 1975

The proceedings were published in STP 613 The second symposium on this topic was entitled

Automated Test Methods for Fracture and Fatigue Crack Growth held in Pittsburgh, Penn-

sylvania during the Fall E9/E24 meeting in November of 1983 The proceedings of this

symposium were published in STP 877 This current symposium was organized in order to

conduct a state of the art review of the technology The symposium was driven by the work

of the task group E9.04.01 on Automated Testing which is a task group of the E9 committee

on Fatigue and its' subcommittee on Apparatus and Test Methods The intent of this task

group is to conduct such a technology review on a three to four year time interval thus

keeping pace with the rapid advances in computing and software engineering technology as

they apply to fatigue and fracture testing

There are a number of areas where automation technology enhances fatigue and fracture

testing The emphasis of this symposium was placed upon the issues of test system imple-

mentation, test techniques, applications of networking and information management within

a testing laboratory, control and data acquisition techniques, and applications or imple-

mentations where the computer provided enhanced analysis or simulation capability These

areas of interest were selected to focus on tasks in the fatigue and fracture testing process

that reside at different levels within this process

Automated systems implementation and test techniques are closest to the actual tasks

of acquiring materials property data In this arena, concerns are primarily on compute

bandwidth and real-time software efficiency Fatigue and fracture tests, being dynamic tests,

require higher data acquisition and compute bandwidth than many common real-time systems

possess The task of determining the crack length in a fatigue-crack growth test via the

compliance technique for example requires data acquisition speed, simultaneity, and com-

pute speed for online crack length calculations from resultant compliance data Often the

testing task requires parallelism in the system implementation to allow for control, data

acquisition, and online conditional processing to be performed in the course of the test

This requires multitasking executive software or highly efficient single tasking environments

that allow for prioritized interrupt driven system services or polling implementations with

sufficient speed to handle all of the tasks at hand A number of systems implementation

oriented papers were presented in the first session of the symposium The range of solutions

2 AUTOMATION IN FATIGUE AND FRACTURE TESTING

was broad It should be noted that the hardware options ranged from simple personal computers to multitasking engineering workstations The Colvin and Swanson paper on

"The Development of a Low Cost PC-Based Data Acquisition System" epitomized the trend toward using cost effective yet high performance personal computers to automate mechanical tests At the other end of the spectrum; McKeighan and Hillberry's paper on "Fatigue and Fracture Testing Using a Multitasking Minicomputer Workstation" is an example of the use

of a high-performance engineering workstation where the benefits of using a multitasking executive greatly enhance the utility of such a system in the laboratory by allowing, for example, analysis and network transactions to be performed concurrently with a executing test

The next level of application, which is actually a step back from the hardware and software details, revolves around the utilization of the technology to allow a new or unique technique

to be developed Fatigue-crack growth near threshold testing may be performed manually without computer automation but with the aid of automation, the system efficiency, test repeatability, and data quality are enhanced significantly The determination of Jf~ can be

a rather arduous chore when using multiple specimens and nonautomated analysis tech- niques The computer controlled single specimen adaptation of this test is a much less labor intensive task and is the norm for this fracture toughness test The paper by Bicego et al illustrates the state of the art for this test and the computer automation that is becoming is integral to this test The natural extrapolation of all of this as performance in hardware and software increases is the ability to perform true calculated variable control tests where a calculated parameter either directly or via a cascade control approach is used to maintain some specimen condition

Taking yet another step back from the testing system and the local testing techniques leaves one in the laboratory environment A key to competitive success be it in the research laboratory or the industrial design allowables laboratory is in the ability to take the results from automation enhanced testing instruments and rapidly and efficiently analyze the raw data and make these results available to the design function, test requestor, manufacturing organization, materials supplier or materials user Integration of the testing laboratory with the rest of a given organization is becoming a very important consideration Organizing test results and materials data into easily accessed formats and making this information accessible are the key issues The technologies through which this is accomplished are networking, database technology, Laboratory Information Management Systems (LIMS), and common access and analysis applications software These concepts were addressed in the last session

of the symposium The paper by McGaw and Bonacuse and the paper by Sunder illustrate the trend to tie all.of the testing automation subsystems within a test laboratory together via a network to facilitate the rapid movement of test information and results to points accessible by other segments of an organization This scenario will become more common with time The paper by Over and Buchmayr is concerned with the difficult task of organizing materials data in a database and then providing the tools to allow for easy access to the information As database software technology improves and the tools for data extraction improve (via the standardization of query tools such as SQL-structured query language), the sharing of materials data wilI be enhanced thus reducing replication of test work or allowing for more critical experimental work to be carried out

To summarize, this symposium and the resultant proceedings are intended to provide an update on the applications of automation across the field of computer assisted fatigue and fracture testing It is the intention of the Automation task group in E9 to revisit this area every three to four years to keep track of the utilization of advancing hardware and software technology Further advances in artificial intelligence, advanced software development tools, direct digital control, networking technology, and information management systems will

OVERVIEW 3

promote new test capability and test techniques and allow for further efficiencies to be obtained in the process of materials characterization Integration of testing laboratories with manufacturing, materials development, inspection capability, and the design functions via automation technology will enhance structural reliability and shorten time to market for new products,

The symposium cochairmen would like to acknowledge the efforts of all the authors and the ASTM staff members which made the symposium and the resultant publication possible

Arthur A Braun

MTS Systems Corporation, Minneapolis,

MN 55424; symposium cochairman and coeditor

Systems Implementation

Randy K Nanstad, t David J Alexander, 1 Ronald L Swain, 1

John T Hutton, 2 and David L Thomas 2

A Computer-Controlled Automated Test

System for Fatigue and Fracture Testing

REFERENCE: Nanstad, R K., Alexander, D J., Swain, R L., Hutton, J T., and Thomas,

D L., "A Computer-Controlled Automated Test System for Fatigue and Fracture Testing,"

Applications of Automation Technology to Fatigue and Fracture Testing, A S T M STP 1092,

A A Braun, N E Ashbaugh, and E M Smith, Eds,, American Society for Testing and Materials, Philadelphia, 1990, pp 7-20

ABSTRACT: A computer-controlled system consisting of a servohydraulic test machine, an

in-house designed test controller, and a desktop computer has been developed for performing automated fracture toughness and fatigue crack growth testing both in the laboratory and in hot cells for remote testing of irradiated specimens Both unloading compliance and dc-po- tential drop can be used to monitor crack growth The test controller includes a dc-current supply programmer, a function generator for driving the servohydraulic test machine to re- quired test outputs, five measurement channels (each consisting of low-pass filter, track/hold amplifier, and 16-bit analog-to-digital converter), and digital logic for various control and data multiplexing functions The test controller connects to the computer via a 16-bit wide photo- isolated bidirectional bus The computer, a Hewlett-Packard Series 200/300, inputs specimen and test parameters from the operator, configures the test controller, stores test data from the test controller in memory, does preliminary analysis during the test, and records sensor calibrations, specimen and test parameters, and test data on flexible diskette for later recall and analysis with measured initial and final crack length information During the test, the operator can change test parameters as necessary

KEY WORDS: computers, fatigue (materials), crack growth, fracture toughness, test con-

troller, compliance, dc-potential, J-integral, multichannel test controller, J-R curve, analog-

to-digital converter, photo-isolation, complementary metal oxide semiconductor, clip gage, automation, interface, fracture testing, fatigue testing

T h e a u t o m a t i o n of materials testing e q u i p m e n t is certainly not a recent concept R e - searchers have applied various degrees of a u t o m a t i o n o v e r the years with the general ob- jectives of increasing productivity, efficiency, and consistency T h e a d v e n t of desktop laboratory c o m p u t e r systems capable of m a c h i n e control and data acquisition was the real catalyst in this area, and the result has b e e n an explosion of c o m p u t e r a u t o m a t i o n This is evident f r o m the relatively narrow technical field for which this symposium was d e v e l o p e d

T h e applications for c o m p u t e r a u t o m a t i o n span the entire range of technology T h e char- acteristics of a u t o m a t e d systems are as diverse as the applications, reflecting the particular needs of the user S o m e are hard, dedicated, inflexible systems that p e r f o r m precisely the same set of tasks for every operation, while others have a high d e g r e e of flexibility and adaptability T h e needs of the fracture and fatigue testing c o m m u n i t y span that range T h e

1 Group leader, metallurgical engineer, and principal technologist, respectively, Metals and Ceramics Division, Oak Ridge National Laboratory, Oak Ridge TN 37831-6151

z Development engineer and instrument technician, respectively, Instrumentation and Controls Di- vision, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6151

7

8 AUTOMATION IN FATIGUE AND FRACTURE TESTING

rapid evolution of sophisticated desktop computers, peripheral devices, and other electronic

hardware, in terms of speed and memory capacity, has likewise allowed for increasingly

greater flexibility and capability in the software for test control, data acquisition, storage,

and analysis

The Fracture Mechanics Group of the Metals and Ceramics Division at Oak Ridge Na-

tional Laboratory (ORNL) began a computer automation activity in 1978 for the purpose

of conducting elastic-plastic fracture mechanics tests The system has evolved markedly since

then, particularly in terms of speed The need for test control and rapid data acquisition

during fatigue crack growth testing spurred the development of a high-speed, multichannel

test controller This paper describes the computer-automated system, test and analysis pro-

cedures, and some test results

Background

Automated testing for evaluation of fracture resistance was largely spurred by develop-

ments in elastic-plastic fracture mechanics Starting with the concept of the J-integral by

Rice [1] and the description of a practical means for estimating J versus crack extension in

test specimens by Rice et a| [2], the advantages of computer involvement were apparent

It was the development of the unloading compliance test method [3], however, that forced

the incorporation of computers in test systems The unloading compliance test procedure

requires excellent test control, high-precision data acquisition capability, and rapid calcu-

lation The use of computers for automated unloading compliance testing has been described

Although the unloading compliance technique is an accepted procedure for determining

results is not an easy task Because the unloading compliance technique involves a fairly

large number of periodic unloading cycles, usually with a hold period at the start of each

cycle to allow for load relaxation in the system, the testing time can be on the order of 1

h In many instances, especially those involving remote testing for irradiated specimens in

a hot cell, the high expense of facilities and equipment mandate that all feasible reductions

in testing time be effected Because the unloading compliance test requires high-precision

measurements of displacement during each unloading cycle, the extensometer (usually a

clip-on displacement gage) is very important The extensometer must be carefully calibrated

and must be seated in such a way that effects of error sources such as friction and vibration

are minimized Testing at low and high temperatures adds temperature shifts in extensometer

calibration as another source of error The ability to accurately infer crack length without

resorting to unloading the specimen (with the associated extensometer and sources of error)

has made the dc-potential drop (dc-pd) method for determining the crack length a widely

tests Dc-pd is an important aspect of the testing system and analysis procedures described

herein

Description of Test System

Figure 1 shows a block diagram of the major components of the interactive fracture

mechanics test system The computer is a Hewlett-Packard Series 200/300 with 4 MB of

random-access memory and Hewlett-Packard technical BASIC operating system and runs

a 1.024-MB test control/data acquisition and analysis program developed in-house specifi-

cally for this application The computer connects to the test controller via a 16-bit-wide

photo-isolated bidirectional control/data bus The test controller (prototype shown in Fig

2) includes a dc-current programmer for controlling a Kepco Model ATE6-100M power

supply operated in current-regulation mode, a ramp and sine output function generator for

NANSTAD ET AL ON A COMPUTER-CONTROLLED TEST SYSTEM 9

HEWLETT-PACKARD 200/300 SERIES COMPUTER 4MB RAM HP TECH BASIC ORNL SOFTWARE 1.024 MB

1

ORNL TEST CONTROLLER PHOTOISOLATION CMOS LOGIC CURRENT "FPD ACTIVE I PD REFERENCE STRAIN ADC

SUPPLY }VOLTAGE ADO I VOLTAGE ADO 16-BIT

PROGRAMMER / 16-BIT I 16-BIT

programming an MTS or Instron servohydraulic test machine, five measurement channels

(each consisting of eight-pole low-pass filter, track/hold amplifier, and 16-bit successive-

approximation analog-to-digital converter) for reading the test machine signal conditioner

and low-noise dc-pd amplifier outputs, and complementary metal oxide semiconductor

(CMOS) digital logic

The dc-current programmer consists of a 12-bit digital-to-analog converter and a trans-

former-isolated output amplifier and can set the output of the dc-current supply over a range

of 0 to 100 A

Function generator ramp outputs (used in J-R testing) are obtained by using a 12-bit

digital-to-analog converter to set input current to an operational amplifier integrator: this

results in ramp outputs that are smooth and stepless even at low ramp rates The range of

ramp rates available is split into three subranges (0.00005 to 0.1 dc V/s, 0.1 to 10 dc V/s,

and 10 to 1000 dc V/s) by switching to different subrange capacitors in the integrator; this

results in reasonable resolution in selection of nominal ramp rate When the computer orders

the ramp output to hold, the digital-to-analog converter current is disconnected from the

integrator, and the integrator output voltage remains constant An auxiliary 12-bit digital-

to-analog converter is used to null integrator input offset currents that would cause integrator

output voltage drift during holds The test controller has memory locations for up to eight

ramp segments (consisting of ramp rate and endpoint voltage) that can be loaded from the

computer to generate arbitrary waveforms (such as triangle, trapezoidal, etc.) Function

generator sine outputs (used in FCG testing) are obtained by feeding a triangle waveform

(the sine argument) from the integrator to an integrated circuit sine wave generator Am-

plitude and mean value of the sine wave are programmed by 12-bit digital-to-analog con-

verters

The test controller reads the servohydraulic test machine load and extension conditioner

10 AUTOMATION IN FATIGUE AND FRACTURE TESTING

FIG 2 - - O R N L test controller that comprises a muhi-output fimction generator, dc-power supply

programmer and CMOS logic on the top plane, and five channels" of analog-to-digital conversion and

low-pass filters on the lower plane

output voltages by means of 16-bit analog-to-digital converters fed through eight-pole low-

and 200 Hz for a 20-Hz FCG test) and precision track/hold amplifiers The track/hold

amplifiers snapshot the respective filter outputs at the start of each analog-to-digital con-

version cycle so that the signals are not changing during analog-to-digital conversion Active

and reference dc-pd voltages are amplified by in-house developed low-noise instrumentation

NANSTAD ET AL ON A COMPUTER-CONTROLLED TEST SYSTEM 11

amplifiers and then read with the same type of low-pass filter, track/hold amplifier and

analog-to-digital converter circuitry used for load and extension channels The analog-to-

digital converters are all triggered simultaneously (which eliminates time skew between

channels), and the readings are saved in latches (which allows the next analog-to-digital

conversion cycle to start immediately in high-frequency FCG testing) and then multiplexed

back to the computer The sequence of channel readings multiplexed back to the computer

for each analog-to-digital converter cycle (the data frame) is loaded to the test controller

by the computer before the test starts During the test, load, extension, active dc-pd voltage,

and reference dc-pd voltage readings are automatically sent back to the computer in the

data frame at rates up to 2500 frames/s

The test controller contains a cycle counter (advanced by the sine output of the function

generator) and cycle counts can be included in the data frame Trigger modes built into the

test controller include demand (low speed, used in J-R tests), pacer (high speed, used in

FCG tests), and waveform maximum/minimum The test controller is designed to be cali-

brated and adjusted entirely by the computer and an external digital voltmeter; no zero or

range potentiometers for the various digital-to-analog and analog-to-digital converters are

included or needed CMOS digital logic was chosen for low power consumption and adequate

speed (1-MHz clock rate)

Acceptable operation of the test system in test control and data acquisition depends upon

the minimization of noise pickup, primarily 60 Hz A number of features provide electrical

isolation of system components so that a single-point grounding system can be implemented

The test controller incorporates photo-isolation on all input/output lines to the computer

and programming lines to the dc-current supply, the extensometer is electrically isolated

from the specimen by using Teflon or ceramic spacers between the razor blades and the

specimen, and the specimen grips are electrically isolated from the load frame by ceramic

inserts in the load train ball joints The single-point ground chosen is that of the servohy-

draulic test machine electronics Heavy-gage grounding cables connect the test controller

data acquisition common, the dc-current supply output, and the low-noise dc-pd amplifier

common to the single-point ground Since all system components are initially isolated, no

ground loops (with the inherent 60-Hz noise pickup) are formed by the single-point grounding

procedure Electrical isolation of the specimen grips also prevents shunting of dc-current

around the specimen The grounding scheme has proved to be effective in minimizing noise

pickup

Description of Extended Range Clip-On Displacement Gage

For many tests a commercially produced clip-on displacement gage meets all requirements

for an extensometer, but testing at higher temperatures or over extended displacement

ranges necessitates the use of an in-house developed clip-on displacement gage The in-

house gage has an operating temperature range of - 196 to 300~ and calibration (at 25~

over a 1.27 to 20.3 mm (0.050 to 0.800 in.) displacement range shows a very linear behavior

Because the clip-on displacement gage is used over a wide temperature range and because

the unloading compliance test technique requires high precision, an experiment was un-

dertaken to examine calibration shifts as a function of temperature Calibrations were per-

formed at - 1 9 6 , - 7 3 , 25,100,200, and 288~ over a 3.78 to 16.5 mm (0.150 to 0.650 in.)

displacement range Third degree polynomial fits were calculated for the voltage versus

displacement data at the different temperatures All residuals between the data and fits were

2.5 • 10-4 mm (10 5 in.) or less Comparison of the fits revealed that first order coefficients

for low and high temperatures varied less than 4% from that at room temperature A second

calibration at 25~ following the high and low temperature calibrations agreed with the

12 AUTOMATION IN FATIGUE AND FRACTURE TESTING

original calibration within about 1% Because the temperature effects were found to be relatively insignificant, most testing and analysis is done with calibrations performed at room temperature, not at test temperature It should be noted that the term "calibration" as used here refers to a process of relating binary data values to corresponding engineering unit results using a higher-order polynomial function, not to a determination of strain errors based on a linear function, as in ASTM Method of Verification and Classification of Ex- tensometers (E 83-85)

DC-Potential Drop Calibration

For accurate determination of crack length with the dc-pd technique, an accurate cali- bration relating dc-pd voltages and crack length is essential Johnson [8] developed a rel- atively general relationship used by many researchers, and other calibrations are presented

in the literature for various fracture mechanics specimens [9,10], including an experimental

and numerical analysis [4] of the compact specimen For this program, however, a study was undertaken to examine the effects of potential drop and current input lead locations

on the calibration using both experimental and numerical methods This paper summarizes that study; a comprehensive discussion will be published separately

Aluminum sheet specimens 3.17 mm thick (0.125 in.) were fabricated with overall planar dimensions eight-times those of the standard 1/2T compact (C(T)) specimen in ASTM Test Method for Jic, a Measure of Fracture Toughness (E 813-87) The dc-pd voltages across the specimen slot, using various combinations of current input lead and dc-pd voltage (both active and reference) lead locations, were measured for a series of specimens having very

narrow slots of lengths (ratio of crack length to width, a / W ) from 0.25 to 0.95 The probes

for each specimen were spot-welded in place Figure 3 shows the finite-element mesh and one combination of dc-pd lead locations investigated for the 1/2T C(T) specimen configu- ration The A D I N A T finite-element code was used to numerically determine calibrations for the same combinations of variables A comparison of the experimental and numerical

FIG 3 Finite-element mesh used to perform numerical calibration for dc-potential drop calibration

for C(T) specimen Current input and potential voltage lead locations are also shown for one particular combination

NANSTAD ET AL ON A COMPUTER-CONTROLLED TEST SYSTEM 13

calibrations is shown in Fig 4 The results are in excellent agreement and validate the use

of the numerical analysis techniques to provide reliable calibrations for different geometries and probe locations All of the results are based on ratios of pd-active and pd-reference voltages This approach cancels out the effects of slow changes in de-current, changes in specimen electric resistance caused by slow changes in test temperature, different specimen materials, and different specimen sizes; one fourth-order polynomial fit to the aluminum

sheet specimen data shown is used for all J-R and FCG testing

Elas•-Plastic Fracture Mechanics Testing

Fracture mechanics tests are performed to evaluate plane-strain fracture toughness, K~c,

ductile fracture initiation toughness, Jt,., J-R curve behavior, and an elastic-plastic deter-

mination of toughness at the onset of cleavage, Kjc The Kjc is determined by calculating

the J-integral at the onset of cleavage, Jc, then using the equation, Kjc = ~v/(EJc), where E

is Young's modulus The test specimen is usually the C(T) geometry, although modifications

to the software can be implemented to allow for virtually any specimen geometry providing the operative analytical equations are known The system has been used with 89 kN MTS,

445 kN MTS, and 890 kN (20, 100, and 200 kip, respectively) Instron servohydraulic test machines

Test Procedures

Before the test starts, the computer prompts the operator for information regarding servohydraulic test machine in use, specimen dimensional and material information, test temperature, crack length determination method (unloading compliance or dc-pd or both),

and displacement rates for loading and unloading The J-R tests are usually conducted in

FIG 4. Comparison of numerical and experimental de-potential drop calibrations for the C(T)

specimen design and lead locations shown in Fig 3

14 AUTOMATION IN FATIGUE AND FRACTURE TESTING

strain control with feedback to the servohydraulic test machine derived from a clip-on displacement gage seated on razor blades usually positioned at the specimen load line, although other locations can be accommodated A series of computer prompts specifies initial settings for all servofiydraufic test machines switches and controls and the correct sequence for mounting the specimen in the grips, turning on hydraulic pressure, etc

After the specimen is equilibrated at test temperature, but prior to the actual testing, the specimen is cycled between high and low loads (both loads are chosen to be less than the maximum load used in precracking the specimen) twice to seat the clip-on displacement gage on the razor blades The program displays a plot of specimen load versus displacement and prompts the operator regarding satisfactory gage seating The next step involves the acquisition and averaging of 60 dc-pd readings (taken over about a 1.5-s interval) to deter- mine the initial values of pd-voltages that will be used to calculate the dc-pd value for initial crack length The specimen is similarly cycled a third time to obtain the initial value of compliance that, together with the previously entered estimate for initial crack length, is used to calculate an effective modulus (as specified in ASTM E 813-87) Additionally, a correction factor that forces the extrapolated load-displacement trace through the origin is calculated and applied to all subsequent displacement data

If the operator accepts the value for compliance, the test starts The initial slope of the load-displacement trace is used to determine the 95% secant offset line as described in ASTM Test Method for Plane-Strain Fracture Toughness of Metallic Materials (E 399-83)

As the load increases, the computer continually calculates the corresponding secant offset line displacement When three successive specimen displacements exceed that of the secant fine, assuming the specimen had not failed during loading, the load at which the trace crossed the secant line is designated Po Loading is continued until a load corresponding to 1.1 x

Po is reached, and a hold period, generally 10 to 20 s long, is initiated to allow for load relaxation in the system The target load of 1.1 x Po was selected based on the provision

in ASTM E 399-83 that a maximum load exceeding that value negates the determination

of a valid Kjc The intent is to preserve the purity of a test that experiences brittle fracture under conditions that might result in a valid K~ During the hold period, the computer acquires and stores 60 readings of both pd-active and pd-reference voltages with full current and zero current The zero-current readings are used to subtract out electronic drift and thermal electromotive force (emf) effects At the completion of the hold period, the spec- imen is unloaded, then reloaded and another hold period started The amount of unloading

is dependent on the accuracy of the compliance determined during the previous unloading (a statistical analysis of the unloading data is performed) If the current-cycle compliance accuracy is unsatisfactory, the unload-reload sequence will be repeated with a greater amount

of unloading, The maximum unloading allowed is 50% of the load at the end of the first hold period and is generally about 15%

During the second hold, the pd-active and pd-reference voltages with full and zero current are used to calculate the dc-pd value for crack length and the compliance and effective modulus are used to calculate the unloading compliance value for crack length These results, the load, extension, area under the load-displacement curve, and J-integral at the end of the cycle are all output to the printer, and a plot of J versus crack extensions is displayed

on the CRT before continuing the test

The displacement increment for the next test cycle is calculated as a percentage (specified

by the operator as one of the test parameters) of the displacement at the first unloading point (1.1 x Po) Use of that increment continues until the J versus crack extension curve crosses the second exclusion line as defined in ASTM E 813-87 After that, each succeeding displacement increment is calculated as a percentage of the previous displacement so that the displacements become progressively greater as the test continues The load-line dis-

NANSTAD ET AL ON A COMPUTER-CONTROLLED TEST SYSTEM 15

placements are also corrected for changes in specimen geometry with large openings [I1]

The test is stopped automatically when the specimen fractures or the target end displacement

is reached, or when the operator intervenes Then the computer does final calculations for

estimated crack length, crack extension, toad, displacement, area under the load versus

displacement curve, and J-integral If the specimen did not fracture, the crack front is

marked For steels, the specimen is heat-tinted at about 300~ until the surfaces are obviously

discolored, and then broken open after being cooled in liquid nitrogen For other materials,

such as aluminum, the crack front is marked by fatigue cycling for several thousand cycles

and then broken open

Analysis Procedures

The length of the fatigue precrack and the final crack length are measured, as prescribed

in ASTM E 813-87, using a digital measuring microscope The computer analysis routine

uses the measured initial crack length value and the compliance determined at the initial

test unloading to calculate an effective modulus (the equation in ASTM E 813-87 is used)

that is used for the balance of the analysis If dc-pd data are to be analyzed, a plot of pd-

voltage ratio versus displacement is displayed on the CRT and the operator prompted to

determine the point at which crack opening effects were complete (usually the start of an

initial linear relationship between the pd-voltage ratio and the displacement), and the dc-

pd polynomial function is normalized to give the measured initial crack length for the pd-

voltage ratio at that point This procedure differs from that described by Lowes and Fear-

nehough [12], and used by others [13-15], and investigations to determine which of the

procedures gives the more reliable results are continuing

Any combination of three different J-integral formulations may be calculated: (1) the

deformation J given in ASTM Test Method for Determining J-R Curves (E 1152-87), (2)

the so-called modified J of Ernst [16] (formulated to account for crack extension), and (3)

the Merkle-Corten J [17] (formulated to account for the tensile component in compact

specimens) The selected J values, crack length information, etc., are printed in SI or English

units or both for each test cycle Portions of the load versus displacement curve showing PQ

and maximum load can be displayed on the CRT and dumped to the printer if the operator

chooses A value of K o is calculated according to ASTM E 399-83, and J values are calculated

at maximum load and the point of unstable fracture (J~) The value of K o is evaluated for

ASTM E 399-83 validity and, if appropriate, designated a valid Kk Final measured crack

length is compared to those predicted by unloading compliance and dc-pd analyses, and the

differences in crack extension printed The [3~c adjustment described by Merkle [18], to

account for loss of constraint in the transition region of ferritic steels, is used to adjust the

Kj~ value to a K~c

The operator can invoke a power-law regression routine that fits the selected J versus

crack extension results The routine displays the results and the power-law fit on the C R T

and prompts the operator to accept the fit; if the fit is rejected, any suspect test results may

be excluded and the regression repeated Once the fit is accepted, the power-law fit plot

(including ghosted symbols for any excluded results) is dumped to the printer, J~c is deter-

mined in accordance with ASTM E 813-87, and the tearing modulus [19] is calculated

The agreement between final measured and predicted crack extensions is generally within

5% and rarely exceeds the 15% criterion in ASTM E 813-87 This is true for both unloading

compliance and dc-pd analyses Figure 5 shows a representative example of a J-R test result

obtained with this system and test procedure At the 538~ (1000~ test temperature,

suspected pin rotation problems caused very poor crack length prediction by the unloading

compliance technique, but the dc-pd predictions were excellent Future research in this area

16 AUTOMATION IN FATIGUE AND FRACTURE TESTING

FIG 5. Comparison of J versus crack extension results obtained simultaneously during a test of a

C( T) specimen using unloading compliance and dc-potential drop crack measurement techniques

will involve the use of dc-pd techniques to infer crack extension in a high-rate J-R test

Some of the known problems associated with high-rate loading are discussed in the following

section on fatigue crack growth rate testing

Fatigue Crack Growth Testing

A primary motivation for the development of the test controller was the desire to obtain

precision measurements of load, displacement, and dc-pd voltages at high rates during FCG

testing Many examples of computer-controlled fatigue crack growth rate testing are available

in the literature [20,21], and the methods for acquiring and analyzing the data are varied

Likewise, the methods for determining crack length are varied, not only in terms of the

measurement technique (unloading compliance and dc-pd), but also in terms of data and

waveform analysis In the case of dc-pd, for example, some experimenters read the peaks

of the voltage waveform during cycling, while others choose to collect data over a specific

region on the waveform All the techniques are chosen with the goal of obtaining the most

representative dc-pd voltages for the crack length at the time of the reading

One of the primary problems associated with this type of cyclic testing is the occurrence

of large fluctuations in the dc-pd signals These fluctuations have been identified by various

investigators [22,23] to be due to inverse magnetostriction effects, the so-called Villari effect

This effect exists when strain applied to ferromagnetic material induces anisotropy of mag-

netic properties Davis and Plumbridge [24] conducted an extensive investigation of these

effects They reported that the fluctuations are proportional to strain rate and that crack

closure effects arc minor by comparison

During experiments for this program using a 1T C(T) specimen instrumented as described

for J-R testing and a multi-channel digital oscilloscope, similar large fluctuations were re-

corded at load cycle frequencies ranging from 0.1 to 30 Hz The oscilloscope traces were

taken with full current and zero current, with and without cyclic loading Figure 6 shows

NANSTAD ET AL ON A COMPUTER-CONTROLLED TEST SYSTEM 17

CURRENT-OFF VOLTAGE ( M Y ) CURRENT-ON VOLTAGE (MV)

FIO 6 Typicat waveforms ob~medfromdc~otentialdrop experimen~ with fuH currentand zero

current, wi& and wi~out &ad, wi~ and wi~out ~cling

PD ERROR (PERCENT)

one example of the waveforms obtained from such an experiment The use of high-capac-

itance electrical filters to eliminate the fluctuations was rejected because of the waveform

distortion and interchannel phase shift problems caused by that approach; instead, various

data manipulation strategies were investigated The result of these studies was the devel-

opment of a technique using pd-voltage sampling over complete load cycles with full and

zero dc-current The full and zero current technique (which avoids the need for bipolar dc-

current supplies or current-reversing switch contacts in the current path) cancels errors due

to thermal emf and slow dc-electrical drift in the pd-voltage amplifiers Integration of the

pd-voltage readings from a load cycle gave mean values for pd-active and pd-reference

voltages that were within 0.5% of the voltages read in the static load case

After incorporating the sampling and analysis techniques in computer software that could

exercise the test controller, the test system described for ]-R testing was used to perform

similar experiments and to verify that the test controller is capable of conducting tests at

load cycle frequencies up to 30 Hz For each pd-voltage sampling, the load, pd-active and

pd-reference voltage waveforms are read twice, once with full current and once with zero

current A b o u t 100 readings are taken for each waveform to be sure that at least one load

cycle will be covered from start to end The zero current readings are subtracted from the

corresponding full current readings (correspondence is determined by reference to the load

waveform), and the results averaged to obtain one reading for each voltage For a test

conducted at 20 Hz, the test variables are sampled (read with full current, dc-current supply

18 AUTOMATION IN FATIGUE AND FRACTURE TESTING

programmed for zero current, read again, and dc-current supply programmed for full current) and manipulated within about 1.5 s The software contains logic to determine if a data point

is significant based on the change from the last crack extension value If a point is determined

to be significant, the load, pd-active and pd-reference voltage waveform data for that cycle are saved along with cycle number and the crack extension data point The testing performed

so far has yielded very promising results Software for the test system and analytical tech- niques described is being developed for various types of FCG tests (constant load, constant

K, decreasing K, constant Kmax-increasing R ratio, etc.), and specific test results will be reported separately

Summary

A computer-controlled system has been developed for J-R and F C G testing The test system is comprised of a laboratory desktop computer running calibration/test/analysis software, test controller, de-current supply, and servohydraulic test machine Development

of the computer software and the test controller in-house ensures an open system that can

be adapted to meet changing J-R and FCG test requirements A modular approach in software development ensures that operator interaction, error checking and recovery, and test controller utility subprograms can be readily applied to other testing such as tensile, crack-arrest, fatigue, etc The test controller provides high-speed, high-precision test control and data acquisition Both unloading compliance and dc-pd techniques can be used to monitor crack extension for J-R tests Dc-pd is used for uninterrupted FCG testing at load cycle rates up to 30 Hz Test errors, such as those due to inverse magnetostriction effects (Villari effect) and thermal emf effects, are eliminated by a full current/zero current, full waveform sampling, and averaging technique

References

[1] Rice, J R., "A Path Independent Integral and the Approximate Analysis of Strain Concentrations

by Notches and Cracks," Journal of Applied Mechanics, Transactions, American Society of Me- chanical Engineers, Vol 35, 1968, pp 379-386

[2] Rice, J R., Paris, P C., and Merkle, J G., "Some Further Results of J-Integral Analysis and Estimates," Progress in Flaw Growth and Fracture Toughness Testing, ASTM STP 536, J G

Kaufman, Ed., American Society for Testing and Materials, Philadelphia, 1973, pp 231-245 [3] Clarke, G A., Andrews, W A., Paris, P C., and Schmitt, D W., "Single Specimen Tests for Jlc Determination," Mechanics of Crack Growth, ASTM STP 590, J R Rice and P C Paris, Eds., American Society for Testing and Materials, Philadelphia, 1976, pp 27-42

[4] McGowan, J J and Nanstad, R K., "A Direct Comparison of Unloading Compliance and

NANSTAD ET AL ON A COMPUTER-CONTROLLED TEST SYSTEM 19

Potential Drop Techniques in J-Integral Testing." Proceedings, 1984 SEM Fall Conference, Com-

puter-Aided Testing and Modal Analysis, Milwaukee, 4-7 Nov 1984, Society for Experimental

Mechanics, Brookfield Center, CT, 1984

[5] Clarke, G A and Brown, G M., "Computerized Methods for J~c Determination Using Unloading

Compliance Techniques," Computer Automation of Materials Testing, ASTM STP 710, B C

Wonsiewicz, Ed., American Society for Testing and Materials, Philadelphia, 1980, pp 110-126

[6] Joyce, J A and Gudas, J P., "Computer-Interactive J~o Testing of Navy Alloys," Elastic-Plastic

Fracture, ASTM STP 668, J D Landes, J A Begley, and G A Clarke, Eds., American Society

for Testing and Materials, Philadelphia, 1979, pp 451-468

[7] Halliday, M D and Beevers, C J., "The dc Electrical Potential Method for Crack Length

Measurement," The Measurement of Crack Length and Shape During Fracture and Fatigue, C J

Beevers, Ed., Engineering Materials Advisory Services, Ltd., London, 1980, pp 85-112

[8] Johnson, H H., "Calibrating the Electric Potential Method for Studying Slow Crack Growth,"

Materials Research Standards, Vol 4, 1965, pp 442-445

[9] Ritchie, R O and Bathe, K J., "On the Calibration of the Electrical Potential Technique for

Monitoring Crack Growth Using Finite Element Methods," International Journal of Fracture, Vol

15, No 1, 1979, pp 47-55

[10] Freeman, B L and Neate, G J., "The Measurement of Crack Growth During Fracture at Elevated

Temperatures Using the dc-Potential Drop Technique," The Measurement of Crack Length and

Shape During Fracture and Fatigue, Engineering Materials Advisory Services, Ltd., Warley, Eng-

land, 1980, pp 435-459

[11] Gray, R A., Jr., Loss, E J., and Menke, B H "'Development of J-R Curve Procedures," NRL-

EPRI Research Program (RP 886.2), Evaluation and Prediction of Neutron Embrittlement in

Reactor Pressure Vessel Materials, Annual Progress Report for CY 1978, NRL Report 8327, Naval

Research Laboratory, Washington, DC, Aug 1979

[12] Lowes, J M and Fearnehough, G D., "The Detection of Slow Crack Growth in Crack Opening

Displacements Using the Electrical Potential Method," Engineering Fracture Mechanics, Vol 3,

1971, pp 103-108

[13] Vassilaros, M G and Hackett, E M., "J-Integral R-Curve Testing of High Strength Steels Utilizing

the Direct Current Potential Drop Method," Fracture Mechanics: Fifteenth Symposium, ASTM

STP 833, R J Sanford, Ed., American Society for Testing and Materials, Philadelphia, 1984, pp

535-552

[14] Hackett, E M., Kirk, M T., and Hays, R A., An Evaluation of J-R Curve Testing of Nuclear

Piping Materials Using the Direct Current Potential Drop Technique, NUREG/CR-4540, U S

Nuclear Regulatory Commission, Washington, DC, Aug~ 1986

[15] Wilkowski, G M and Maxey, W A., "Review and Applications of the Electric Potential Method

for Measuring Crack Growth in Specimens, Flawed Pipes, and Pressure Vessels," Fracture Me-

chanics: Fourteenth Symposium, Vol II: Testing and Applications, ASTM STP 791, J C Lewis

and G Sines, Eds., American Society for Testing and Materials, Philadelphia, 1981, pp II-266- II-294

[16] Ernst, H A., "Material Resistance and Instability Beyond J-Controlled Crack Growth," Elastic-

Plastic Fracture: Second Symposium, Vol I: Inelastic Crack Analysis, ASTM STP 803, C F Shih

and J P Gudas, Eds., American Society for Testing and Materials, Philadelphia, 1983, pp 1-

191-1-223

[17] Merkle, J G and Corten, H., "A J-Integral Analysis for the Compact Specimen, Considering

Axial Force as Well as Bending Effects," Journal of Pressure Vessel Technology, Vol 96, Nov

1974, pp 286-292

[18] Merkte, J G., An Examination of Size Effects and Data Scatter Observed in Small Specimen

Cleavage Fracture Toughness Testing, NUREG/CR-3672 (ORNL/TM-9088), Oak Ridge National

Laboratory, Oak Ridge, TN, April 1984

[19] Paris, P., Tada, H., Zahoor, A., and Ernst, H., "The Theory of Instability of the Tearing Mode

for Elastic-Plastic Crack Growth," Elastic-Plastic Fracture, ASTM STP 668, J D Landes, J A

Begley, and G A Clarke, Eds., American Society for Testing and Materials, Philadelphia, 1979,

pp 5-36

[20] Burgers, A and Kemper, P D., "Automatic Crack Length Measurement by the Electrical PD

Method with Computer Control," Advances in Crack Length Measurement, C J Beevers, Ed.,

Engineering Materials Advisory Services, Ltd., London, 1982, pp 325-342

[21] Vecchio, R S et al., "Development of an Automatic Fatigue Crack Propagation Test System,'" Automated Test Methods for Fracture and Fatigue Crack Growth, ASTM STP 877, W H Cullen,

R W Landgraf, L R Kaisand, and J H Underwood, Eds., American Society for Testing and Materials, Philadelphia,' 1985, pp 44-66

20 AUTOMATION IN FATIGUE AND FRACTURE TESTING

Services, Ltd., London, 1982, pp 139-158

Aluminum Alloy," Ph.D thesis, University of Bristol (as referenced in Davis and Plumbridge, see Ref 24)

Glenn E Colvin, Jr.,1 and Stephen R Swanson I

A PC-Based Data Acquisition System for Composite Material Fracture Testing

REFERENCE: Colvin, G E., Jr., and Swanson, S R., "A PC-Based Data Acquisition System

for Composite Material Fracture Testing," Applications of Automation Technology to Fatigue and Fracture Testing, ASTM STP 1092, A A Braun, N E Ashbaugh, and E M Smith~ Eds., American Society for Testing and Materials, Philadelphia, 1990, pp 21-37

ABSTRACT: Automated data acquisition is a necessity for characterizing composite materials due to the nonlinearities associated with matrix-dominated properties and measurement of multiple strains A low-cost data acquisition system based on a Zenith 286 PC-compatible computer has been developed for characterizing the mechanical response of composite ma- terials subjected to a variety of loading arrangements The system utilizes a Data Translation DT2821 card that provides 16 channels of analog-to-digital conversion An inexpensive, low- noise strain gage conditioner and amplifier was built to provide ten conditioned DC output channels for strain gage measurements A menu-driven software package, written in the C programming language, runs the system and creates an easy-to-use operator interface The software allows the user to access standard test configurations, standard or customized data reduction routines, and port data to an Apple Macintosh The results of an investigation measuring the resolution and accuracy of the system are reported Data acquisition examples are presented for testing composite I-beams in three-point bending, composite cylinders under uniaxial compressive loads, and composite tubes under biaxial loading

KEY WORDS: automation, fatigue testing, fracture testing, data acquisition, personal com- puter, composite materials, system verification

Personal computers have pervaded many aspects of engineering since their introduction

in the late 1970s This is especially true in the testing community as the use of an automated data acquisition system for certain classes of tests has currently become a necessity For instance, the anisotropic nature and nonlinearities of fiber-dominated, resin-based composite materials require a multitude of data channels that can only be monitored with a data acquisition system

The pervasiveness of personal computers has led many people to view PC-based data acquisition systems as readily available technology that are easy to construct, and indeed numerous books have been written on the subject [1-4] However, these books often em- phasize the general details involved in constructing data acquisition hardware while ignoring the proliferation of low-cost commercial hardware, the difficulties associated with developing data acquisition software, or the specific details required to verify system integrity The intent of this paper is to describe an inexpensive data acquisition system oriented towards the special requirements of composite materials testing Unlike metals, test methods for mechanical testing of composite materials have not been widely standardized resulting

in a lot of on-going research into fundamental composite test method development Further,

Research assistant and professor, respectively, Mechanical Engineering Department, University of Utah, Salt Lake City, UT 84112

21

22 AUTOMATION IN FATIGUE AND FRACTURE TESTING

compared to metals testing, composite materials testing often requires a large number of channels since monitoring load and displacement in most cases does not provide a sufficiently detailed assessment of what has occurred within the specimen

System Development and Description

The critical first step in developing an effective data acquisition system is defining the system requirements The basic data acquisition system requirements for the University of Utah Mechanics of Composites Laboratory were:

at least three other analog-to-digital ( A / D ) channels at 12-bit resolution, including sign

to use different programming languages for data reduction

Once the system requirements were defined, the individual hardware and software com- ponents were selected

Hardware

To maintain compatibility with a sister data acquisition system dedicated to the unique requirements of short duration impact events, a Zenith 2 286 IBM PC-compatible computer was used to build the data acquisition system The principal benefits of PC-compatible machines are their low cost, expandability, and relatively fast processor speeds Computer performance was enhanced with an Inte180287 math coprocessor and 512 kbytes of expansion RAM Peripheral devices included a 20 Mbyte internal hard disk, an Epson LQ 500 dot matrix printer, and an HP 7475A six pen plotter A sketch of the system is given in Fig 1

A Data Translation DT 2821 acquisition and control card was installed that had 16 single ended A / D channels with 12-bit resolution, including sign, and a 50-kHz maximum sampling rate Sixteen channels were necessary because composite testing requires that strain mea- surements be made along a variety of axes in a number of different locations For instance, ASTM Test Method for Tensile Properties of Fiber-Resin Composites (D 3039-76 (1982)) recommends the use of three longitudinal strain gages for longitudinal tension testing of unidirectional composite laminae Additional gages are required if the Poisson's ratio is to

be determined The University of Utah biaxial composite laminate test specimen [5] requires

a minimum of six strain gages along with two load channels Strain gage conditioning was provided by a specialized 10-channel 120-ohm strain gage conditioner unit The details of the strain gage conditioning unit are presented in the appendix

Software

The Microsoft C, version 5.1, compiler in conjunction with the Data Translation A T L A B library was used to write the software that drives the data acquisition card The A T L A B library is a collection of functions that drive the DT2821 interface card and frees the pro-

2 Many manufacturers make data acquisition products The products discussed in this paper were selected based on the constraints of the project and laboratory and do not constitute an endorsement

of the particular manufacturer over another

COLVlN AND SWANSON ON A PC-BASED DATA ACQUISITION SYSTEM 23

grammer from assembler or machine language details For example, the simple program

that follows uses four A T L A B functions to initialize the A T L A B subroutines, select ac-

quisition board 1, reset the board, and return the digital value of channel 1 at a gain of 2:

aLinitialize ();

aLselect~_board (1);

al_reset 0 ;

al_adc_value (1, 2, &value);

Using the A T L A B library was a compromise between a commercial data acquisition package

and programming the data acquisition card at the machine language level

Two software selectable data acquisition modes are available: a burst mode and an in-

termittent mode The discernible difference between the two modes is the relationship

between channel sample time and data group sample time as illustrated in Fig 2 A data

group represents a single scan of the requisite data channels Data group values are often

COLVIN AND SWANSON ON A PC-BASED DATA ACQUISITION SYSTEM 25

treated as occurring at the same instance in time when plotting data points during the data

reduction process

A burst mode scans the requisite data channels at a given frequency such that the number

of channels times the channel sample time equals the data group sample time This invalidates

the data group simultaneity assumption and creates a significant problem when dealing with

composite testing where ten unique data channels are not uncommon in a data group, Figure

2a shows that the time between the last sample of one data group and the first sample of

the next data group is less than the time between the first and last samples within a data

group when many channels are being monitored Simultaneous sample and hold A / D hard-

ware can alleviate this problem but is expensive; thus, a means of lowering costs is to use

a single A / D converter with a multiplexer and placing a sample time delay between data

groups This delay is what the intermittent mode uses to allow the assumption of simultaneity

within a data group

The intermittent mode, shown in Fig 2b, scans the requisite channels at the maximum

data acquisition card frequency of 50 kHz, writes the data to disk, and checks for a user

interrupt before repeating A t 50 kHz, the channel sample time is 20 Ixs while the data

group sample time varies from approximately 0.2 s for a three sample data group to 0.9 s

for a ten sample data group

Writing the data to disk substantially increases the data acquisition system's cycle time;

however, it adds a certain degree of flexibility First, it simplifies data acquisition consid-

erations by allowing the use of a simultaneity assumption Second, the system was designed

for use with monotonic tests where high data acquisition rates are not an absolute require-

ment Lastly, future software modifications will principally deal with data reduction func-

tions To make these modifications easier for future users, intermittently writing the raw

test data in A S C I I integer form to disk allows users to read the data with another program-

ming language such as Basic or Pascal to reduce the data themselves resulting in flexible

data reduction software The data could have been saved in binary form; however, many

users are uncomfortable with binary form because it cannot be directly viewed on the monitor

nor can it be read directly Thus, while intermittent writing of data to disk increases the

cycle time of the data acquisition system, it offers other advantages by allowing the use of

the simultaneity assumption during data reduction and adding flexibility to the data reduction

process

The data acquisition software consists of a master menu allowing the o p e r a t o r to choose

among several options:

1 Channel zeroing

2 Channel calibration

3 General, user-specified test configurations

4 Specific test configurations

The zeroing and calibration options allow the test operator to check and inspect the requisite

data channels prior to running a test User specified test configurations allow the test operator

to use the data acquisition system for customized, unique acquisition needs during test

method development while the specific test configurations improve operator performance

when using established test methods

Figure 3 provides the flow of the software functions for all the data acquisition options

A T L A B functions are used at various points in the flow for DT2821 card initiation, channel

scans, channel gain modifications, sampling frequency definition, and DT2821 card termi-

nation The user is required to input a data file name early in the flow for data storage

Data is written to disk in one of two ways For a burst mode, data is written after a predefined

26 AUTOMATION IN FATIGUE AND FRACTURE TESTING

ENTER FROM MAIN MENU

I

RELEASE DATA BUFFER)

I

TERMINATE DATA ) ACQUISITION CARD

I

AUTOMATED DATA REDUCTION

I

~ RETURN TO MAIN MENU~

FIG 3. Core function for data acquisition software

data buffer in R A M has been filled through the use of a D M A capability provided by the

DT2821 card For an intermittent mode, data is written between data group samples A t

the end, program control is passed to the standard data reduction algorithms that are to be

completed prior to returning to the main menu

Porting

Once the data acquisition system acquires the data and converts it to a usable form, the

data often needs to be transferred from the Zenith-286 based data acquisition system to an

Apple Macintosh that is used for report generation The incompatibility of the disk drives

COLVIN AND SWANSON ON A PC-BASED DATA ACQUISITION SYSTEM 27

between the two computers necessitated the use of some sort of hard link between them

Networking is expensive; hence, the Traveling Software package Lap Link was used that

does an ASCII file transfer from the 286 serial port to the Macintosh modem port This

solution cost 30% of the networking approach and maintained a free expansion slot in the

Zenith 286 for future use

Verifying System Performance

The most critical step in developing a data acquisition system is determining its final

performance characteristics and providing for effective system troubleshooting VanDoren

[1] defines the three most critical regimes to overall system performance:

1 the transducer and circuitry before the signal conditioner input,

2 the single conditioner, and

3 the single conditioner ouptut through the analog to digital conversion

Once the data is in digital form, it becomes a matter of carefully storing the data and properly

reducing it

Of the basic signals that must be dealt with by the Mechanics of Composites Laboratory

in the first two regimes, most involve commercial test machines in which the specifications

governing their performance are known The combined noise for regimes one and two of

these machines can also be checked with an oscilloscope The newer machines have signal

noise bands as low as -+5 mV peak-to-peak while some of the older machine controllers

have signal noise bands of approximately -20 mV peak-to-peak

The major component in regimes one and two that is not a commercial piece of equipment

is the strain gage conditioning unit described in the appendix The strain gages and their

leads make up the first regime, while the conditioning unit itself makes up the second regime

Measurements Group, Inc., supplies the specification for each lot of strain gages regarding

gage behavior, Additionally, the strain gage leads are short enough that loss of temperature

compensation and signal attenuation is not a problem [6], and the strain gages are allowed

to warm up prior to Wheatstone bridge balancing to eliminate concerns about self-heating

The final conditioning unit noise, after amplifying the completed bridge signal, was found

to be high frequency low amplitude noise (0.2 mV peak-to-peak at a gain of 100) that was

less than the least significant bit (LSB) of the A / D converter The conditioner gains were

verified accurate to within 0.2%

The third regime in the data acquisition system defined by VanDoren [1] deals with the

DT707 connector card and DT2821 card Figure 4 reports the results of a detailed constant

voltage check of all ranges at various voltage levels Each data point represents an arithmetic

mean of 2000 data points sampled at a 4 kHz frequency In the worst case, a five percent

difference between the input voltage and the measured voltage was found for a gain of 1

at 10 mV (approximately two LSBs) At 48.8 mV, which corresponds to 19 LSBs, the percent

difference has dropped to a percent or less for all DT2821 gains The percent difference is

less than 0.2% for the rest of the range

VanDoren's regime breakdown for determining system performance is also effective from

a maintenance and troubleshooting standpoint The first and second regimes can normally

be checked to see if they fall within predefined specifications independently of regime three

Calibration switches on the test machines and the strain gage conditioners allow those

respective components to be checked while constant voltage checks of the A / D converter

can be performed to determine the accuracy of its calibration The first application in the

next section illustrates this procedure

28 AUTOMATION IN FATIGUE AND FRACTURE TESTING

|

I n p u t V o l t a g e

FIG 4. Constant voltage A / D conversion accuracy

The last point regarding system performance is the use of automated calibration [1] By grounding or shorting the inputs to an unused channel of the A / D converter, a reference ground can be determined to check the A / D converter offset Likewise, connecting the input leads to a constant, known voltage source, which can be as simple as a battery, provides

a check of the A / D converter gain While these checks can be done frequently during a test, the software uses these devices to check the A / D converter calibrations prior to a test

to determine the offset (normally one half LSB) and to adjust all the subsequent test data

by

Sample Applications

Three applications of the data acquisition system will be described to illustrate the variety

of uses and the details regarding data acquisition in composite materials testing The first application simply monitors the actuator load and displacement in a three-point bend test

of a composite I-beam and illustrates a means of dealing with and isolating noise in the data The second application uses the data acquisition system for measuring actuator load and displacement, extensometer strain, and six strain gage channels in a uniaxial compression test Lastly, the use of the data acquisition system in a biaxial tension-compression test of

a 10.16-cm (4-in.) tube is described in which six strain gage channels and two load channels are monitored

I - B e a m Test

One simple use of the data acquisition system was to monitor the load and stroke of a composite I-beam test The load-displacement plot in Fig 5a shows the response of the I-

COLVIN AND SWANSON ON A PC-BASED DATA ACQUISITION SYSTEM 29

beam in three-point bending However, this plot was not directly obtainable from the raw

data as the MTS 406 controller stroke conditioner had a lot of unexpected noise Figure 5b

shows a plot of the raw data

Since the A / D calibration checks (Regime 3) before and then after the test showed no

problems with the A / D converter, an oscilloscope was used to check the noise level on both

the load and stroke channels of the 406 controller A n approximate +40 mV peak-to-peak,

10 kHz noise signal was isolated on the stroke channel The frequency corresponds to that

of the LVDT master oscillator excitation leading to the general conclusion that something

was wrong with the signal filter Because the actuator could not respond to a 10 kHz signal,

the test was considered to be valid; hence, in addition to repairing the conditioner, it was

also necessary to "clean up" the test data

For data reduction, the raw data was ported to an A p p l e Macintosh where it was analyzed

with the Cricket Graph software package A n equation for the stroke data versus sample

number, which can be correlated with time, was determined with a y-based least squares

regression, see Fig 5c A corrected displacement was then calculated for each sample number

to provide the data to produce the load displacement curve in Fig 5a

Thus, while the resolution of an A / D converter can permit the detection of noise in an

analog signal, many numerical means exist to filter the data either during the acquisition

process or, as in this case, during the data reduction process

Uniaxial Compression Test

The behavior of fiber-dominated, resin-based composite laminates under compressive

loadings is not well understood Contributing to this problem is the lack of commonly

accepted test specimens for laminate compressive behavior as evidenced by the lack of an

A S T M standard for compressive testing of composite laminates A tubular specimen was

developed for testing composite laminates because it provides inherent buckling stability

and eliminates the free edge effect [7] The test method uses the data acquisition system

for measuring actuator load and displacement, extensometer strain, and six strain gage

channels Three 0/90 strain gages are used for measuring the average axial and hoop strain

of the specimen

The multitude of strain/displacement measurements allows separation of localized ma-

terial nonlinearities from test configuration induced effects, such as bending due to misa-

ligned grips This example application of the data acquisition system for monitoring nine

individual channels also illustrates the importance of the intermittent mode discussed earlier

in providing for the use of a simultaneity assumption in reducing data

A plot of the acquired actuator load-displacement data points, such as given in Fig 6a,

shows the post-failure load carrying capability of the specimen and indicates no significant

change in the loading rate prior to failure

The superimposed plots of the individual axial strains in Fig 6b show that variability in

the gages is minimal for most of the test and increases as the specimen approaches failure

If bending were a problem in the test, such consistent gage readings for the first half of the

test would not be possible The divergence of the gage readings indicates that localized

nonlinear material behavior might be present and might be a potential source of failure

Extensometers are not as widely used in composite material tests as in metal tests One

reason is the lack of large plastic deformations that make it easier to use foil type strain

gages Because each has its advantages, both are used in the uniaxial compression test

method The average strain of the three gages shown in Fig 6b is plotted along with the

extensometer strain in Fig 6c The two strain measuring devices show relatively good

30 AUTOMATION IN FATIGUE AND FRACTURE TESTING

'eq%~ e% % ~ eeeo ~

(N)I) peo"l

32 AUTOMATION IN FATIGUE AND FRACTURE TESTING

34 AUTOMATION IN FATIGUE AND FRACTURE TESTING

agreement, that is, the difference is less than the variability between the individual strain

gage readings

Biaxial Tension-Compression Test

Many failure theories exist to explain the catastrophic failure of fiber-dominated, resin-

based composite laminates under multiaxial loadings; however, limited empirical data is

available for comparing the relative merits of the individual theories A 10.16-cm (4-in.)

tubular specimen was developed to generate data for composite laminates subjected to biaxial

tension-tension and tension-compression loadings [5] The data acquisition system is used

to monitor six strain gage channels, the axial load applied to the tube, and the internal

pressure of the tube

Figure 7 shows the measured laminate stresses versus the product of the experimental

laminate stiffnesses and the experimental strains for a biaxial tension-compression test These

relationships from linear classical laminated plate theory should ideally result in 45 ~ line

Figure 7a provides the tensile hoop response that is basically linear due to the stiffening of

fibers in tension compensating for the nonlinear behavior of transverse plies The axial

compressive response is shown in Fig 7b for the same test The compressive response

becomes nonlinear near failure due to the nonlinear behavior of the 45 and 90 ~ plies not

being offset by fiber stiffening

Summary

This paper described a data acquisition system recently developed for the specialized

requirements of composite materials testing The system combined low cost, commercial

hardware with customized software The software was constructed in a modular fashion to

provide for future growth Many compromises had to be made in the software dealing with

when and how to save the data and the timing of the acquisition process The principal

requirement of the compromises was that data points within a data group could be treated

as occurring simultaneously

The system was constructed in a manner that facilitated system validation and system troubleshooting This was accomplished by breaking the verification process down into three

regimes dealing with the transducer and circuitry prior to signal conditioner input, the signal

conditioner, and the signal conditioner output through the A / D conversion

Several composite testing examples were presented A simple load displacement test of

a composite I-beam was presented to illustrate how the data acquisition system aids in the

isolation of noise in the test equipment Two examples were then presented that illustrate

some of the special requirements of composite materials testing The lack of standardized

specimens means test instrumentation must satisfy the dual requirements of measuring material behavior and verifying general test method behavior The former dealing with the

anisotropic, nonlinear behavior of composite materials, and the latter dealing with such

issues as grip-induced bending

Acknowledgments

The hardware for this proiect was purchased with a State of Utah Center for Excellence Grant and the development was performed under a contract from the Lawrence Livermore National Laboratory and the Army Ballistic Research Laboratory The support of these

organizations is greatly appreciated Additionally, travel funds provided by the University

of Utah Mechanical Engineering Department Chairman, Dr David W Hoeppner are also

COLVIN AND SWANSON ON A PC-BASED DATA ACQUISITION SYSTEM 35