Astm stp 1231 1994

The broad range of topics describe how advancements in digital computer hardware and software have opened up new opportunities in mechanical testing, modeling of physical processes, data

Trang 2

STP 1231

Automation in Fatigue

and Fracture: Testing

and Analysis

Claude Amzallag, Editor

ASTM Publication Code Number (PCN):

04-012310-30

ASTM

1916 Race Street Philadelphia, PA 19103 Printed in the U.S.A

Trang 3

Library of Congress Cataloging-in-Publication Data

Automation in fatigue and fracture: testing and analysis / Claude

Amzallag, editor

(STP: 1231)

"ASTM publication code number (PCN) 04-012310-30."

Includes bibliographical references and index

Photocopy Rights

Authorization to photocopy items for internal or personal use, or the interna~ or personal use

of specific clients, is granted by the AMERICAN SOCIETY FOR TESTING AND MATERIALS for users registered with the Copyright Clearance Center (CCC) Transactional Reporting Service, provided that the base fee of $2.50 per copy, plus $0.50 per page is paid directly to CCC, 222 Rosewood Dr., Danvers, MA 01923; Phone: (508) 750-8400; Fax: (508) 750-4744 For those organizations that have been granted a photocopy license by CCC, a separate system of payment has been arranged The fee code for users of the Transactional Reporting Service is 0-8031-1985-2/94 $2.50 + 50

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 to time and effort on behalf of ASTM

Printed in Fredericksburg, VA December 1994

Downloaded/printed by

University of Washington (University of Washington) pursuant to License Agreement No further reproductions authorized

Trang 4

Foreword

The International Symposium on Automation in Fatigue and Fracture: Testing and Analysis,

was held 15-17 June 1992 in Paris, France It was cosponsored by the: Societe Francaise de

Metallurgie et de Materiaux (SF2M), Committee on Fatigue, France; and American Society

for Testing and Materials (ASTM), Committee E9 on Fatigue, USA

Also offering valuable cooperation were the: Society of Automotive Engineers (SAE);

Fatigue Design and Evaluation Committee, USA; Engineering Integrity Society (EIS), UK;

and National Research Institute for Metals (NRIM), Japan

The Symposium was an extension of the series of International Spring Meetings of SF2M

This publication is a result of this symposium Claude Amzallag, IRSID-Unieux, France, is

the editor

Acknowledgment

The Organizing Committee, who helped develop the program and provide session chairmen

and reviewers, are acknowledged for their assistance Ms Gail Leese, (PACCAR Technical

Center, USA) and Dr Dale Wilson (Tennessee Technical University, USA) helped shape the

symposium, provide reviewers, and graciously offered their time in reviewing papers

In addition to the help of the technologists cited above, the editor wishes to express gratitude

to the staff members of SF2M and ASTM, particularly Yves Franchot, SF2M, who handled

the administration of the symposium

Trang 5

A Sampling of Mechanical Test A u t o m a t i o n Methodologies Used in a Basic

Research Laboratory -G A HARTMAN, N E A S H B A U G H , A N D D J B U C H A N A N 36

C o m p u t e r Applications in Full-Scale A i r c r a f t Fatigue Tests -R L HEWITT AND

Microprocessor-Based Controller for A c t u a t o r s in S t r u c t u r a l Testing R SUNDER

A n A u t o m a t e d I m a g e Processing System for the M e a s u r e m e n t of S h o r t Fatigue

C r a c k s a t Room and Elevated T e m p e r a t u r e s - - L Yl, R A, SMITH,

C o m p u t e r - A i d e d L a s e r I n t e r f e r o m e t r y for F r a c t u r e Testing A K MAJI AND

A u t o m a t e d D a t a Acquisition and Data B a n k Storage of Mechanical Test Data:

A n I n t e g r a t e d Approach -G BRACKE, J BRESSERS, M STEEN, AND H H OVER 108 Sampling Rate Effects in A u t o m a t e d Fatigue C r a c k G r o w t h R a t e T e s t i n g - -

P r o c e d u r e for A u t o m a t e d Tests of Fatigue C r a c k P r o p a g a t i o n - - v BACHMANN,

A u t o m a t i o n of Fatigue C r a c k G r o w t h Data Acquisition for C o n t a i n e d and

Through-Thickness C r a c k s Using E d d y - C u r r e n t and Potential

Trang 6

A Computer-Aided Technique for the Determination of R-Curves from

Center-Cracked Panels of Nonstandard Proportions G R SUTTON,

FATIGUE UNDER VARIABLE AMPLITUDE LOADING The Significance of Variable Amplitude Fatigue Testing D SCH~3TZ

Spectrum Fatigue Life Assessment of Notched Specimens Using a Fracture

Mechanics Based Approach M V O R M W A L D , P HEULER, AND C KRAE 221 Spectrum Fatigue Testing Using Dedicated Software c MARQUIS AND J SOLIN 241

A Computerized Variable Amplitude Fatigue Crack Growth Rate Test Control

Automated Fatigue Test System for Spectrum Loading Simulation of

High-Cycle Fatigue of Austenitic (316L) and Ferritic (A508) Steels Under

Gaussian Random LoadingwJ.-P GAUTHIER, C A M Z A L L A G , J.-A LE DUFF,

Crack Closure Measurements and Analysis of Fatigue Crack Propagation

Under Variable Amplitude Loading c AMZALLAG, J.-A LE DUFF,

A Fatigue Crack Propagation Model Under Variable Loading J GERALD AND

Sensitivity of Equivalent Load Crack Propagation Life Assessment

to Cycle-Counting Technique E LE PAUTREMAT, M OLAGNON,

FATIGUE AND F R A C T U R E A N A L Y S I S AND S I M U L A T I O N

Fatigue Life Prediction Under Periodical or Random Muitiaxial Stress States

Nenber-Based Life Prediction Procedure for Mnltiaxially Loaded Components

Fatigue Test Methods and Damage Models Used by the SNCF for Railway

Load Simulation Test System for Agricultural Tractors K NISHIZAKI 419 Applying Contemporary Life Assessment Techniques to the Evaluation of

Urban Bus Structures M M DE FREITAS, N M MAIA, J MONTALVAO E SILVA,

Trang 7

Fatigue and Fracture Analysis of Type 316L Thin-Walled Piping for

Heavy Water Reactors: Crack Growth Prediction Over 60 Years

(With and Without Stratification) and Flawed Pipe Testing A B POOLE

A Rule-Based System for Estimating High-Temperature Fatigue L i f e - -

7010 Alloys Subjected to Aeronautical Spectra -c B L E U Z E N ,

Using Maximum Likelihood Techniques in Evaluating Fatigue Crack Growth

Curves -s E C U N N I N G H A M AND C G ANNIS, JR 531 Advances in Hysteresis Loop Analysis and Interpretation by Low-Cycle

Fatigue Test Computerization -G DEGALLA1X, P HOTTEBART, A SEDDOUKI,

An Automatic Ultrasonic Fatigue Testing System for Studying Low Crack

Growth at Room and High Temperatures -T wu, J NI, AND C BATHIAS 598 Database for Aluminum Fatigue DesigneD KOSTEAS, R ONDRA, AND

Material Data Banks: Design and Use, an Example in the Automotive

Hypertext and Expert Systems Application in Fatigue Assessment and Advice

A Software System for the Enhancement of Laboratory Calculations A GALTIER 648

Trang 8

Overview

STP1231-EB/Dec 1994

In the diverse and complex technology of fatigue and fracture, it is increasingly important for societies and engirieers to exchange information of mutual interest It is thus critical to provide forums, such as the subject symposium, to allow for open exchange With knowledge

of the needs of industry, researchers gain insight valuable in assuring their focus is on meaningful topics Armed with the latest developments from the research community, engineers, in turn, are able to apply and validate these concepts and findings from the research community The goal of the Symposium on Automation and Fatigue and Fracture: Testing and Analysis,

was to be just such a forum on an international scale Developers of testing methodology, researchers and scientists who evaluate and predict materials response, and engineers who apply the results to current day challenges in industry joined together to reflect on recent achievements in the areas of:

1 Automated testing systems and methods,

2 Models and methods for predicting fatigue life under complex loading,

3 Fatigue and fracture analysis and simulation, and

4 Applications and prediction methods

This collaboration resulted in the presentation of 45 papers to an audience of around 150 technologists, representing more than 18 countries and 5 continents The broad range of topics describe how advancements in digital computer hardware and software have opened up new opportunities in mechanical testing, modeling of physical processes, data analysis and interpretation, and, finally, applications in engineering environments

This volume is offered as a valuable source of information for all those interested in deepening their understanding of fatigue and fracture phenomena It is the hope of all involved that this may spawn yet further ideas and innovations in applying multidisciplinary technologies

to testing and analysis automation, which in turn may open new doors of understanding

C Amzallag IRSID-Unieux, France;

symposium chairman and editor

Trang 9

Automated Testing Systems and

Trang 10

A r t h u r A B r a u n t

A Historical Overview and Discussion of

Computer-Aided Materials Testing

REFERENCE: Braun, A A., "A Historical Overview and Discussion of Computer-Aided

Materials Testing," Automation in Fatigue and Fracture: Testing and Analysis, ASTM STP

pp 5-17

ABSTRACT: Consistency of test data has always been a key concern in any materials testing

application Test technique or method, operator skill and experience, and capabilities of the

apparatus are all parameters that affect the consistency of the desired information The arrival

of testing automation has contributed significantly to improving the consistency of materials

testing apparatus, modifying existing test methods, creating new test methods due to enhanced

capability, and improving the productivity of testing systems

This paper surveys the development of computer-aided testing over the last 20 to 25 years

and includes a discussion of current systems implementations and the emerging area of labora-

tory-wide automation The rapid development of materials testing automation capability has

generally tracked the trends in the computer industry Advances in microprocessor hardware

technology have driven testing automation by allowing for embedded intelligence in key test

system components and by allowing for high-performance supervisory computer subsystems

to control or supervise the overall test rig Software technology advances in concert with

expanding hardware capability have provided truly useful real-time operating environments,

more efficient applications development tools, and higher productivity through more intuitive

user interface technology All together, these technology improvements have allowed for more

sophisticated, consistent, and higher performance testing automation Further improvements

will be realized through the true utilization of the emerging digitally based systems architectures

and emerging networking technology This discussion concludes with a brief look at where

emerging capabilities such as these will allow for new types of experiments to be performed

and where information management will be enhanced, thus allowing for greater productivity

in the test laboratory

KEY WORDS: materials testing, test automation, controls, data acquisition, historical survey,

fatigue (materials), fracture (materials), data analysis, testing methods

This paper describes the historical development of automation applied to fatigue and fracture

testing Automation capability for servohydraulic mechanical testing systems appeared in the

late 1960s with the advent of lower-cost minicomputer capability and software options that

allowed for the demanding real-time requirements of fatigue and fracture tests to be addressed

As computer hardware and software improved, gains in increased test control and data acquisi-

tion performance as well as options to use the automation facility for new types of tests

emerged This evolution occurred in several phases, which will be discussed here

The first phase of early implementations was concerned primarily with interfacing lower-

cost minicomputers with the system analog controls for data acquisition and program generation

Group manager, Applications Engineering, Aerospace Structures and Materials Testing, MTS Systems

Corporation, Eden Prairie, MN 55344

Trang 11

6 AUTOMATION IN FATIGUE AND FRACTURE

(drive signal or function generation) This allowed for greater efficiency and some of the first

generation of calculated variable control tests (tests in which the computer was used to control

a secondary or indirectly calculated parameter such as true strain or stress intensity range by

adjusting the primary control parameter, such as force or strain, during the course of the test)

The second phase was really a transition Prior to the transition, the minicomputer solution

was optimized for higher performance.with more capable hardware and software The transition

began with the availability of very low cost personal computers (PCs) and the beginnings

of microprocessor technology use in distributed system functions such as in data displays,

servocontrollers, and control of peripheral devices such as temperature controllers Applications

software quickly used these enhanced capabilities and many types of tests were created that

used computer control

The third phase is the period we are currently experiencing where there has been a re-

integration of system control functions with data acquisition, function generation, and peripheral

control in the current digital control systems coupled with the use of higher-performance PC

or workstation hardware and modern software technology The emphasis is shifting from

hardware orientation to software The applications possibilities of some of these totally software-

based systems remain to be realized in third-generation applications software It is believed

that the extension of this phase will be not necessarily in radical changes to the automation

of the test system but rather in the connection of the test system to design, manufacturing,

and modeling functions within a given enterprise through networking and enhanced software

data sharing capability Also, the software-based nature of the control systems will be utilized

to implement truly adaptive control (autotuning systems or systems that optimize the control

parameters in response to changes in the test specimen) and to implement new tests based

upon the ability to use calculated parameters to control tests Each of these periods wilt be

discussed in more detail in terms of hardware, software, applications, and performance

Early Implementations (1965 to 1975)

Servohydranlic test system technology emerged in the late 1950s and early 1960s with

applications in structural testing and simulation being the first requirements These systems

used analog control based upon vacuum tube technology [ 1 ] By the mid-1960s, servohydraulic

test systems were becoming widely used for fatigue and fracture tests Several evolutions of

electronics technology were required before the vacuum tube-based controls were replaced

first by discrete transistor logic and then by integrated circuit technology Initial attempts using

analog computers for test automation provided significant enhancements to the basic closed-

loop capability [2] The desire to utilize an easier-to-program digital computer could not

be satisfied, however, until cost-effective digital computer hardware and software became

commercially available By the end of the decade, the commercial availability of minicomputer

systems provided the first opportunity to marry computer control to these electrohydraulic sys-

tems

These first implementations interfaced the minicomputer to the analog controller through

an analog interface in which digital-to-analog (D/A) converters were typically used as a

command reference (program source or function generator source) for the system and analog-

to-digital (A/D) converters were used to acquire data (measure and store forces, strains,

displacements, etc.) from the system Figure 1 illustrates the typical system architecture func-

tionally Figure 2 shows a typical system configuration from this period The computers used

were, by today's standards, limited The typical PDP 8 system manufactured by Digital

Equipment Corporation utilized limited ferrite core memory typically in the 4 to 8-k word

range, had limited processing power, and required a paper tape for program input and storage

Disk and tape technology usage became more viable as costs for these devices were reduced

Trang 12

BRAUN ON HISTORY OF COMPUTER-AIDED MATERIALS TESTING 7

I

L~

Trang 13

FIG 2 Example of a first-generation PDP 8 automated test system

The A/D and D/A converters used were typically 12-bit devices providing one part in 4096

resolution over - 1 0 V The software in the earliest systems was either machine language

based making programming the system a major ordeal, or a specialized assembly developed

for materials testing The "MTL" language developed by MTS Systems Corporation is an

example of one of these proprietary languages Much progress was made in this mode as

exemplified by the work of Conle and Topper [3], Richards and Wetzel [4], and Martin and

Churchill [5] Significant advances were made in performing strain-controlled fatigue tests

with calculated variable limit programming for load, strain, or inelastic strain A significant advance common to all of these works was the introduction of the computed variable control

capability discussed previously A good example of this approach is the tests that were developed

for axial strain control where the axial strain was calculated from the diametral strain [6]

The most significant limitations of these early systems were the severe memory limitations and the primitive programming environment for creating testing applications programs By

the middle of the 1970s, metal oxide semiconductor memory, MS I (medium scale integration),

and the use of higher level languages such as BASIC and FORTRAN brought about the next

phase of development in testing automation

Trang 14

BRAUN ON HISTORY OF COMPUTER-AIDED MATERIALS TESTING 9

The Minicomputer Refinement Period (1972 to 1980) and the Transition to Personal

Computers (1980 to 1985)

The advent of cheaper memory and higher performance processors as exemplified by the early members of Digital's PDP 11 family of minicomputers allowed for higher level language use on these systems to become feasible Languages such as FORTRAN and especially interpretive BASIC required another level of performance in the computer This additional performance was not required in the machine language/assembly language implementations This added complexity also required more memory in addition to a more powerful processor The early 1970s brought hardware meeting these requirements from companies such as Digital Equipment, Data General, and Hewlett Packard Mass storage had developed to the point where magnetic tape and disk subsystems were usable and the paper tape based systems were disappearing Computer manufacturers were also providing "operating systems" that managed system peripherals and memory and provided a structure upon which to build and use higher level programming tools

The basic system hardware architecture of the systems implementation did not change radically during this time A "processor interface" continued to bridge the space between the analog control system and the computer There were, however, some attempts to eliminate the analog controls also in some of the earliest direct digital control (DDC) systems at this time [7] Processor performance, however, severely limited the sample rate of these systems and forced the majority of implementations to use analog controllers A/D and D/A resolution initially was limited to 12 bits but increased to 14 and 16 bits in the late 1970s and early 1980s as higher-resolution higher-performance components became available Improvements

in function generation were developed that provided more localized hardware control of the D/A converter such as "segment generation" (where a local clock steps the D/A through a wave table and provides scaling), thus off-loading the computer from generating every D/A step and freeing up time for other tasks Similar developments were provided through local clocking of A/D input channels Also, other hardware features were developed for the "processor interface." Computer-controlled control mode switching, system monitoring, voltage sensing, digital input/output (I/O) logic, and computer hydraulic system shutdown capabilities were refined and then put under software control through callable library routines accessible in the high level programming language used with these systems

The most notable advances were accomplished in the software environment where higher level programming languages with built-in function calls to assembly language hardware control routines were used to make the task of developing test software somewhat easier The work of Donaldson et al described in Ref 8 is typical of the state of the art in the mid 1970s These systems at first were typically single station, that is, there was one computer and processor interface per test system Graphics capability emerged in the early 1970s allowing for data acquired to be plotted on a terminal screen and for plots to be outputted to plotter and hard copy units for reporting Figure 3 shows a typical system from this period of refinement The programming languages typically had a set of callable routines for graphics that allowed for "on-line" graphics to be shown during the course of a test To obtain the best real-time response possible, the operating systems for these computers were typically memory resident, nonswapping, and did not dynamically reallocate memory Digital's RTI l operating system was a typical example of this type of operating system This changed as hardware, peripheral, and memory performance increase toward the end of this period

During this time, test technology advanced with the enhanced computer power being utilized

to perform multiaxial test control with data acquisition [9] and stress intensity range controlled fatigue-crack growth tests [10] among many others The hallmark of this period, however, was that software technology was expanding to use the higher performance processors, addi-

Trang 15

FIG 3 Automated test system from the early 1970s

tional memory and more readily available mass storage capability while, in general, maintaining

the need for a single computer to be dedicated to a single test system The predominant

computer suppliers were Digital Equipment Corporation, Data General, and Hewlett Packard

Transition Period

Increased performance in minicomputers, additional memory, and less expensive higher

performance mass storage facilitated the transition from single-station systems to multistation

and multiuser systems This is the culminating period of the development and use of minicomputer systems in materials testing applications The subsequent availability of microprocessor

technology caused the next real evolution to occur It is interesting to note that during the

period from the late 1960s to the early 1980s the emphasis consisted of using a single processor for all tasks on a single system and then on multiple systems Processor interface technology,

programming languages, and operating systems concentrated on this philosophy

The multistation/multiuser systems that evolved in the late 1970s and early 1980s used the highest performance minicomputer technology available The Digital PDP 11/34 became, for example, a common platform upon which to implement some of these systems Figure 4 shows

a typical system configured to control five test stations performing fatigue-crack growth tests Extended addressing allowing for increased memory (the 11/34, for example, used 18-bit memory addressing), faster disk drives (allowing for swap oriented operating systems operating systems to be usable in real time), and operating systems designed for real time multi-user activity allowed the extension to multistation systems

Applications software did not necessarily change greatly during this time but rather was refined to utilize the higher performance The availability of microprocessor technology prior

Trang 16

FIG 4 -Multistation/multiuser system o'pical of the late 1970s and early 1980s

to desk-top personal computers allowed these systems to reach their ultimate form Figure 5 shows a service simulation system in which the multiuser minicomputer system controls two test systems as well as two microprocessor-based temperature/humidity chamber controllers while performing network transactions with a laboratory host computer As performance requirements began to exceed the capabilities of these multiuser implementations, microprocessor technology was used in a new generation of processor interfaces to unload the beleaguered minicomputer Distribution of some functions such as environmental chamber control or auxiliary data acquisition was also used to off load the central minicomputer It was becoming obvious that the performance of these multiuser systems was very difficult to predict and that

it was very easy to over commit the resources of the minicomputer and its operating system despite the sophistication in these systems To perform multiple real-time tests with function generation, data acquisition, on-line calculated variable control, graphics, remote control of peripheral controllers such as chamber controllers, and carry out network transactions is about

as much as these single-processor architectures could handle and is a testimonial to the ingenuity

of the minicomputer vendors and those who utilized the technology

The final form of the minicomputer-based system was essentially a distributed system; with central lhnction generation, data acquisition, and test control still the domain of the supervisory minicomputer but with ancillary functions distributed elsewhere, Examples were microprocessors embedded in the processor interface further refining function generator and data acquisition local control, use of external intelligent peripheral controls for environmental systems, and

Trang 17

FIG 5 Multistation test system with environmental simulation control

"smarter" analog servocontrollers with functions such as mode switch or data display processing

controlled by local embedded microprocessors

This was the state of the art during the period 1978 to 1983 ASTM STP 710 on Computer

Automation o f Materials Testing from 1978 provides a wide overview of the state of the art

in this period [11] The essence of the symposium proceedings revolves around getting the

most out of the minicomputer architectures, their associated operating systems, and software

programming environments This period overlaps the availability of the first large-scale integra-

tion devices such as Digital's LSI-11 processors, but the design approach is the same These

systems were complex It was very difficult to predict the performance of the multiuser

environment Applications programming remained the responsibility of the user with little off-

the-shelf test software really available until the mid-1980s Thus, the system is only as easy

to use as the skill of the programmer would allow Some implementations with lower cost

single-user desk-top computers were beginning to emerge in response to the complexity of

the multiuser solution The further development of microprocessors essentially ended the era

of minicomputers on materials testing systems

Personal Computers and Materials Testing

The further development of microprocessor technology brought about the inevitable develop-

ment of cost-effective single-board computers, desk-top computers, and eventually personal

computers This evolution brought about a significant change in testing automation and acceler-

Trang 18

ated the end of the minicomputer as the materials testing automation solution of choice

Essentially, the change was economic These new microcomputers provided minicomputer

performance at very low cost Most of the minicomputer manufacturers realized this when it

was too late to respond and lost the market to a new group of suppliers Miller et al provide

an eloquent discussion of the merits of the central computer time-sharing implementation

versus distributed control [12] Their solution is the natural progression of the distribution of

intelligence discussed in the previous section

The availability of personal computers from IBM, Apple, and others caused a radical change

to the nature of materials testing automation in the mid 1980s The PC revolution alone was

not sufficient, however In addition, as in the previous evolution with the minicomputer, suitable

I/O hardware, operating systems, and programming tools had to become available to allow

the process to begin again with the new generation of computing hardware In some ways, a

step backwards was made in terms of hardware and software sophistication Much of the

I/O hardware available for PCs at this time was 12 bit rather than 16 bit, which was the

standard on minicomputer implementations In addition, many PC I/O modules did not support

simultaneous sample/hold data acquisition or allow simultaneous input and output in direct

memory access (DMA) mode The software callable routines for data acquisition, signal output,

and general purpose I/O lacked the customization for materials tests that had been engineered

into their minicomputer-based counterparts Finally, the operating systems were limited in

terms of how they supported concurrency and managed memory The proceedings of the

ASTM Symposium on Automated Test Methods for Fracture and Fatigue Crack Growth held

in Pittsburgh in November 1983 show the beginning of the transition [13] There is a mix of

papers with the majority discussing work performed with minicomputer architecture systems,

but there are also a number of papers discussing the use of Apple II, Hewlett Packard Desktop,

and other PC systems illustrating the beginning of the change It would take several more

years for the performance of these PC systems to increase significantly and for the quality

of the I / 0 hardware and software tools to be good enough to displace the minicomputer

implementations The prime driver, however, was the low cost of these systems making them

very attractive for laboratory automation, and this alone forced the transition

Personal Computers, Workstations, Digital Control, and Laboratory Automation

The period from 1985 to the present has seen the maturing of the PC platform In addition,

the required changes have happened with respect to available high performance plug in I/O

hardware for PCs Software operating systems, development systems, and user interfaces have

become available allowing for the proper utilization of the hardware and the implementation

of easy to use applications software for common materials tests and the creation of generic

test programs that may be used as tool boxes for addressing nonstandard tests without forcing the

user to write traditional software The industry acceptance of the IBM-compatible architecture as

a de facto standard has driven hardware costs down to incredibly low levels while providing

hardware performance that far exceeds the performance of the most costly minicomputer

systems available as recently as three years ago Microprocessor technology has advanced

along with software tools to allow for multiprocessor implementations We no longer have to

rely on a single computer to accomplish the task but rather multiple processors may be almost

casually assembled into systems for advanced control and data acquisition implementations

Two systems architectures are currently in use The most common architecture remains the

analog controller supervised by a computer with suitable I/O hardware for function generation,

data acquisition, and ancillary control The new analog controllers use the latest in low-noise

integrated circuitry and are much less expensive than their predecessors They are designed

with the intent of PC automation The state of the art with respect to systems of this type is

Trang 19

somewhat captured in the proceedings of the 1989 ASTM Symposium entitled Applications of Automation Technology to Fatigue and Fracture Testing [14] By this time the IBM architecture

dominates the examples of systems presented The Apple II implementations have disappeared

to be replaced by the Apple Macintosh One paper uses a UNIX workstation [15] for its

multitasking features and graphical user interface This type of implementation is a forerunner

of the more advanced systems discussed later

The other system implementation that is more recent in design is the digital control system These systems continue to utilize analog signal conditioning, and servovalve driver electronics but the control loop, function generation, and data acquisition are implemented in software Due to the high sample rate required to maintain reasonable system accuracy and the parallelism inherent in materials testing systems, most if not all of these digital control systems tend to

be multiprocessors It should be noted that at this point the journey has been made from the design where a single processor controlled multiple systems to the point where multiple processors (each singly more powerful than the single processor used for the multistation implementation) control a single system The process of distribution of processing power in these systems is becoming fully developed In these systems, the PC becomes that supervisor, the development environment, the host for the system user interface, and the site for data analysis, networking, and report generation The multiprocessor has multiple high-performance processors distributed and networked via a network or bus in a manner such that processor power may be dedicated to each specific function within the system One such system on the market currently uses nine processors connected through 20 million bit/s links to perform digital control, data acquisition, and system housekeeping The same 20 mbit/s connection connects an industry standard PC to the multiprocessor completing the system A block diagram for this digital control system is shown in Fig 6 A typical system as actually implemented

in hardware is shown in Fig 7 Communication within the multiprocessor and with the PC becomes a key design consideration along with operating system choice and user interface

9 User Interface * DDC Servo Control

9 Test Execution 9 Data Acquisition

9 Development 9 Function generation

9 Readout

9 Hydraulic System Control

i ~ - " ' - ~ '.,'~.i

ir ' i!

"'r i : i.~

Local Control Panel

9 Specimen Loading

9 Local Status Display

9 Hydraulic Control

9 Run/Stop Control

FIG 6 Digital system with PC user interface functional

Trang 20

FIG 7 Typical test system with current-generation digital controls

design The state of the art seems to be a very high speed communications facility used within the multiprocessor and as a link to the host computer, a multitasking, virtual memory, operating system that allows multithreaded execution and preemptive priority scheduling, and a graphical user interface or GUI to allow for intuitive operation of the system A history of the development

of such a system is discussed in Ref 16 Many of the tradeoffs made to obtain a solution such

as the one identified earlier are discussed in Ref 16

The software-based nature of a digital system allows for greater flexibility and a high degree

of reconfigurability This is especially true of a random access memory (RAM)-based system where the operating software for the control system is downloaded from the host In this case,

a system can be updated with new software and the new features may be utilized the next time the multiprocessor is booted from the host A software-based user interface (such as one implemented via a graphical user interface, or GUI) is desirable over hard controls such as buttons or knobs In the same manner, the user interface may be redesigned or modified and the changes may be used without modifying or altering hardware This type of system implementation ultimately provides lower maintenance and update cost and, hence, lower cost of ownership A modular hardware architecture that allows for control system software modifications and user interface modifications through software updating rather than hardware

Trang 21

modification will provide the necessary platform for advanced controls and user interface

technology as they become available

Applications software at this time appears to be taking two paths One path provides the

traditional standard tests in turnkey applications software These packages typically address

tests that are well defined by existing standards The other route is a general-purpose application

program that may be configured ("programmed" but at a very high level) to perform a variety

of control, data acquisition, and logical tasks allowing for nonstandard tests to be accomplished

Most vendors of test equipment now provide applications programs of these two types Program-

ming tools exist for these systems as in the past but the occurrence of user programming is

becoming less with time as the expectations of the user community become higher concerning

software for materials testing Expectations with respect to ease of use are much higher due

to the widespread use of GUIs such as OS/2's Presentation Manager, Microsoft's Windows,

Apple's Macintosh OS, and Motif on top of UNIX It is often joked that if a user has to read

a manual, the software is substandard This is becoming less of a joke and more of a test of

the quality of the software as we move forward

Futures

It seems logical that the emergence of the software-based systems that have been discussed

will allow for enhancement of both machine control capabilities and of applications program-

ming Since the servocontroller, function generator, and other system components are now

programs or software objects, it seems that these objects will evolve and that other objects

with additional capability will be created One merely has to look at National Instruments'

LabView product [17] and their concept of a virtual instrument, built in a graphical editor on

a Macintosh, to understand the potential of software-based digital materials test system control-

lers Some areas where one could expect development would be."

1 Truly adaptive control algorithms that learn from the system and test specimen

2 Calculation of and control from virtual channels where mathematical operations may

be performed on incoming data streams and the resultant virtual data used for control

or other decision making

3 Feedback linearization and scrvovalve linearization Where the acquired feedback and

the valve drive output are compensated by system for nonlinearities inline on the

executing system

4 Advanced function generation capability for forcing system compliance through over

programming in random loading applications

In the user interface area, similar capability may be utilized to alter the look and feel of

the system to better meet user requirements Multitasking capability and interprocess communi-

cation could be used to teach models on-line through the observation of data returning from

an executing materials test Similarly, executing test software could, directly through laboratory

networking and database technology, update a designer at another location to facilitate simulta-

neous engineering of components and structure The author feels that most of these concepts

are achievable on the current platforms and that the most significant gains over the next few

years will be in the area of software enhancements to these new control architectures and the

integration of these systems into enterprise-wide modeling, manufacturing, and design activities

S u m m a r y

The historical development of materials testing automation as it applies to fatigue and fracture

testing has been discussed covering the period from 1965 through 1992 Some speculation has

Trang 22

also been presented regarding the direction of development in this area over the next three to five years This survey has been created to essentially review the progress in this area and provide background information to supplement the proceedings of the symposium of which this paper was a part It is presented with the thought that often you cannot understand or see where you are going unless you know where you have been, and how you got there The reader is strongly encouraged to investigate the references in this paper as they are the true documentation for the developments in this area

References

[1] Johnson, H C "Mechanical Test Equipment in the Sixties: A Decade of Radical Change," Closed Loop Magazine of Mechanical Testing, MTS Systems Corporation, Minneapolis, MN, Fall/Winter

1974, pp 15-21

[2] Richards, E D and Wetzel, R M., "Mechanical Testing of Materials Using an Analog Computer,"

Materials Research and Standards, Feb 1971, pp 19-22

[3] Conle, E A and Topper, T H., "Automated Fatigue Testing," Closed Loop, MTS Systems Corpora-

tion, Minneapolis, MN, Spring 1971, pp 8-10

[4] Richards, E D and Wetzel, R M., "Application of Analog and Digital Computers to Fatigue Testing," Ford Motor Company Scientific Research Staff Report SR 71 - 138, Ford Motor Company, Dearborn, MI, Sept 1971

[5] Martin, L M and Churchill, R W., "Interfacing the Computer to a Materials Test System,"

Proceedings, Spring Meeting, Society for Experimental Stress Analysis, May 1969

[6] Slot, T., Stentz, R H., and Berling, J T in Manual on Low Cycle Fatigue Testing, ASTM STP

465, American Society for Testing and Materials, Philadelphia, 1969

[7] Boggs, B C., Mondol, N K., McQuown, R E., and Anderson, J G., "Digital and Analog Computer

Equipment and Its Application to In-House Testing," Use of Computers in the Fatigue Laboratory, ASTM STP 613, American Society for Testing and Materials, Philadelphia, 1976, pp 2-26

[8] Donaldson, K H., Dittmer, D E, and Morrow, J., "Fatigue Testing Using a Digital Computer-

Based System," Use of Computers in the Fatigue Laboratory, ASTM STP 613, American Society

for Testing and Materials, Philadelphia, 1976, pp 27 49

[9] Penn, R W., Fong, J T., and Kearsley, E A., "Experience in Data Acquisition and Reduction for

a Biaxial Mechanical Testing Program," Use of Computers in the Fatigue Laboratory, ASTM STP

613, American Society for Testing and Materials, Philadelphia, 1976, pp 78-93

[10] Kaisand, L R and LeFort, P., "Digital Computer Controlled Threshold Stress Intensity Factor Fatigue Testing," Use of Computers in the Fatigue Laboratory, ASTM STP 613, American Society

for Testing and Materials, Philadelphia, 1976, pp 142-159

[11] Computer Automation of Materials Testing, ASTM STP 710, B C Wonsiewicz, Ed., American

Society for Testing and Materials, Philadelphia, 1978

[12] Miller, N R., Dittmer, D E, and Socie, D E, "New Developments in Automated Materials Testing Systems," 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 9-26

[13] 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

[14] 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, 1989

[15] McKeighan, P C., Evans, R D., and Hillberry, B M., "Fatigue and Fracture Testing Using a Multitasking Minicomputer Workstation," 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, 1989

[16] Braun, A A., "The Development of a Digital Control System Architecture for Materials Testing Applications," Proceedings, 17th International Symposium for Testing and Failure Analysis, Los

Angeles, Nov 1991, American Society for Metals International, Metals Park, OH, 1991, pp 437-

444

[17] "LabView 2 Reference Documentation," National Instruments, Austin, TX, 1990

Trang 23

S o m a s u n d a r a m D h a r m a v a s a n t a n d S a r a h M C P e e r s t

General Purpose Software for Fatigue

Testing

REFERENCE: Dharmavasan, S and Peers, S M C., "General Purpose Software for Fatigue

Testing," Automation in Fatigue and Fracture: Testing and Analysis, ASTM STP 1231, C

Amzallag, Ed., American Society for Testing and Materials, Philadelphia, 1994, pp 18-35

ABSTRACT: Recently, most manufacturers of servohydraulic materials test equipment have

developed digital control systems Digital systems have several advantages over the more mature analog controllers, in particular, the availability of features that allow more realistic tests to be carded out Also, the ability to interface and communicate with computers easily makes it possible to perform not only realistic but also very complex control and data acquisition tasks However, the software required to define these complex tasks and communicate with digital controllers is demanding

The development of digital controllers has been made possible by the tremendous advances

in microprocessor technology These developments have also given rise to computers with ever- increasing capabilities at lower costs The software architectures controlling these computers have also given rise to sophisticated developments in order to make the computers easier to use Most computers now have some form of graphical user interface (GUI) The downside of most GUIs is that the development of application software is much more complex In addition, most of these operating systems are not ideally suited for real-time control that is absolutely vital for automated fatigue testing

The difficulties associated with developing software for specific fatigue tests may be somewhat alleviated by the development of suitable tools within an integrated architecture that allow the scientist/engineer to define the required test in the form of functional blocks This type of general-purpose software can then take the test definition provided by the scientist and convert

it into a form suitable for the digital controller

A suitable test description paradigm is proposed in terms of general requirements; a test- description language with elements such as objects, functional blocks, time, events, and data presentation; and some specific requirements for a software development environment From this, general-purpose software for automated fatigue testing can be developed and used at a high level of abstraction In addition, the paper reviews the development of an integrated general- purpose program for automated fatigue testing that implements the functional blocks required but without the full test description language The paper also uses the example of testing offshore structural components using realistic sea-state sequences, such as WASH, to illustrate the problems that must be solved

Some implications of this approach in future fatigue testing applications are also discussed

in the paper

KEY WORDS: automated fatigue testing, digital control systems, servo-hydraulic test equipment, data acquisition, alternating current potential drop, crack growth monitoring, fracture (materials), fatigue (materials), testing methods, data analysis, test automation

The increasing complexity o f structures and components are made possible by the availability

of advanced modeling and analysis tools such as finite element methods Although the use o f these tools has resulted in the design of efficient components, they are still prone to fatigue

1Lecturers, NDE Centre, University College London, London WC1E 7JE, UK

18

Copyright* 1994 by ASTM International www.astm.org

Trang 24

DHARMAVASAN AND PEERS ON GENERAL PURPOSE SOFTWARE 19 damage The modeling of fatigue behavior in test samples and components is the subject of

intense research For example, in the offshore industry, the tubular welded joint, a commonly

used structural component, has been the focus of attention for several years [1] The type of

testing carried out on these joints has progressed from simple constant-amplitude tests to

complex wave forms that simulate the random wave loading experienced by offshore structures

In addition, the tests have been carried out in a seawater environment with cathodic protection

The data acquisition requirements are also quite demanding and have included crack-depth

measurement using an alternating current potential drop (ACPD) system [2]

The limiting factor in increasing the load carrying capacity or efficiency is the availability

of suitable materials For example, structural ceramics are capable of withstanding extremely

high temperatures and are being actively researched for use in high-efficiency engines The

development of nonmetallic materials requires new mechanical test methods that are capable

of quantifying the strength characteristics for use in analysis methods

The majority of the tests need some form of computer-controlled testing due to the complexity

of loading or data acquisition requirements and, in view of this, most manufacturers of

servohydraulic test equipment have developed digital control systems, or in other words,

machine controllers that are programmable by a host computer Digital systems have several

advantages over the more mature analog controllers as they have immediate (within one clock

cycle) response times and, hence, step changes can be made to parameter values In contrast,

some of these features are not possible with analog controllers Also, the ability to interface

and communicate with computers easily makes it possible to perform not only realistic but

also very complex control and data acquisition tasks However, the software required to define

these complex tasks and communicate with the digital controllers is demanding in terms of

performance The development of the software is also made difficult by the real-time nature

of the problem and requires a good understanding of the behavior of the controller as well as

the requirements of the test

The development of digital controllers has been made possible by the tremendous advances

in microprocessor technology These developments have also given rise to computers with

ever-increasing capabilities at lower costs The software architectures controlling these comput-

ers have also given rise to sophisticated developments to make the computers easier to use

Most computers now have some form of graphical user interface (GUI) The downside of

most GUIs is that the development of application software is more complex In addition, most

of these operating systems are not ideally suited for real-time control that is absolutely vital

for automated fatigue testing

The difficulties associated with developing software for specific fatigue tests may be some-

what alleviated by the development of suitable tools within an integrated architecture that

allow the scientist or engineer to define the required test in the form of a suitable high-level

description This type of general-purpose software can then take the test definition provided

by the scientist and convert it into a form suitable for the digital controller This paper will

describe a suitable test description paradigm that has been implemented to work with a digital

controller and the problems that had to be overcome Some implications of this approach in

future fatigue testing applications are also discussed

A Fatigue Test Description Paradigm

Functional Blocks

A laboratory fatigue test can be abstracted to a sequence of well-defined processes that may

or may not be carried out depending on conditions or circumstances A test can then be

described as a sequence of functional blocks, each of which has to have certain properties

Trang 25

defined Examples of functional blocks are sine waveform generation, collection of data from

strain channel, limit setting, etc

Functional blocks may be carried out in parallel In particular, the processes of data acquisition

and of control are independent of one another in that they may be carried out, and usually

are, in parallel

Requirements for the Test-Description Language

1 In common to any programming language, the primary requirement is that of complete-

ness The language should allow all elements of the test to be described fully In addition,

it should be flexible so that it does not constrain the design of a test

2 Real-time language As any fatigue test is a demanding real-time application, it should

have constructs that define time constraints In addition, there should be facilities to

recover from situations where there is insufficient time to carry out a task

3 Modularity Tests are often repeated in modified forms, hence, the ability to reuse

functional blocks of any test is important for practical (reduced time in developing a

test) and reliability reasons (fully tested functional blocks increase reliability) This is

best done by allowing the definition of hierarchies of functional blocks, that is, blocks

within blocks

4 Predictability and reliability The failure of a fatigue test may lead to the possibility

of a dangerous situation arising Thus, the language should be, as a minimum, predictable,

that it should not be possible for unexpected sequences of events to occur For further

safety, the language should provide a way of describing safe or expected behavior and

of defining error recovery actions This could then allow the recognition of potentially

dangerous situations, that is, of behavior outside the defined safe behavior that would

then trigger an appropriate recovery action

5 Built-in or stored knowledge reduces design time Examples of such knowledge include

materials and their properties, test-piece geometries, graphs and report forms for the

display of data, etc Also information on the restrictions imposed by hardware limitations

must be made available to the test designer For instance, the rate of data acquisition

is often limited by the hardware Tests could be stored as a templates, which define

the test completely except for only a few parameters to be input at the time of rerunning

of the test

6 As more knowledge is built in to the system, it may be possible to predict the performance

of the system to ensure that dangerous situations do not occur Storage of past test

results in the appropriate form requires intelligence The appropriate form is one that

allows future inductive or analogical reasoning about the behavior of similar tests

Elements of the Language

The description language needs to have the following elements in order to fully describe a test

Objects or Entities "Objects" or entities are derived from the object-oriented programming

paradigm that includes specialization, inheritance, and the representation of relationships

between objects [3] Each type or "class" of object will have expected properties and will be

used in particular ways

Examples of objects are: the test-piece, materials, test hardware, computer records and files

for data acquisition, data structures (including arrays), graphs, databases, and functional blocks

Trang 26

DHARMAVASAN AND PEERS ON GENERAL PURPOSE SOFTWARE 21

Functional Blocks, Processes, and Scripts A process can be defined as a specialized object with a particular role A functional block, another specialized object, provides a "script" of processes or other functional blocks to be performed and defining the objects that are involved

in the processes The concept of a script is borrowed from the field of artificial intelligence [4], but here it is used merely as a practical way of describing a procedure to be carried out The designed test itself is again a functional block

The base process classes are:

1 data acquisition,

2 control/loading, and

3 input and output to the user interface

To fully define the script requires control constructs for sequences, parallelism, iteration and repetition, and conditional logic

Time As fatigue tests can last from a few seconds to several months, a clock capable of working at different resolutions must be maintained In addition, a virtual clock may be necessary for the running of simulated tests for the purposes of checking the overall test design

Events The description of an occurrence can be made clearer and more concise if it is possible to give an abstract description of the event Hence, the implementation of the language must provide recognition and identification of given events For example, an event may be given as "the failure of the test-piece," and there must be a way therefore of describing how and when the test piece is said to be failed A complete definition of "failure" is not always possible and hence recognition is required

For recognition, pattern matching or fuzzy logic or both [5] for the use of qualitative information may be employed However, the use of such concepts does not always coincide with the required predictability of the language Whatever recognition process is used must

be well-defined and therefore, such advanced techniques may not be appropriate for this field

at present

Data Analysis and Presentation During and after the completion of a test, there will be

a need to analyze the data obtained from the test Therefore, an ability to define calculations and also to link to external calculation libraries is required The link to external libraries is important as there is a wealth of validated and tested programs already available and it will

be impossible to incorporate all of this into a limited language

Some of the features required are:

1 data acquisition and signal processing functions,

2 general mathematical functions,

3 statistical functions,

4 random number generators,

5 numerical techniques such as interpolation, and

6 graphical presentation of data, including conventional forms for the display of fatigue test data

Software Development Environment

For ease of development of the test, various forms of the test description must be allowed The ideal is that of a mixture of graphical elements with textual input The description of a

Trang 27

test is essentially procedural to which traditional flow-charts are ideally suited Hence, a description of the test could be carried out using graphical flow-charting symbols and arrows with appropriate text and references to entities Such a graphical flowchart would then need

to be translated into low-level test control instructions

In defining the requirements for data acquisition in a test, it is easier to be able to view a predicted graph of results as given by a preliminary definition of the process to be able to judge that sufficient data will be acquired Furthermore, a facility that would allow the modification of the amount and type of data to be acquired by pointing and selecting appropriate sections of the graph would encourage increased effectiveness of the designed test

Testing of a design is vital One method is that of simulation, although there are difficulties

in providing true simulation due to the uncertainties associated with damage mechanisms of different materials and components Therefore, a pseudo-simulation where the user triggers certain actions is probably the only way in which testing prior to carrying out the test can be achieved However, the interpretation of the simulation results can be difficult particularly when only presented as numerical data Simulation of the test shown graphically (stick diagrams

of expected effects of loading on test-pieces and graphs of data) is useful to allow the user

I

!

General Purpose Interface Bus link

FIG l Schematic diagram of test setup

Trang 28

behind FLAPS is to design the required test off-line The designed test is then used by FLAPS

to carry out the test The general structure of the program is shown in Fig 2

System Setup -The System Setup phase is used to select system-dependent information, such as type of controller being used and the type of crack measurement system, in addition

to setting up the various parameters necessary before running a test such as calibration, selection

of engineering units, and other parameters This information is used during the Run Phase to control the tests in the most appropriate manner

The ability to store and recall setup data rapidly is useful in cases where there is a need to change from one type of testing to another The main advantage of digital systems is the ability to restore the overall system to a predetermined state This aids with repeatability of tests

Design The Design phase allows the setting up of templates known as Control Files to run specific applications The various waveform generation, data collection, and program control options are chosen and the sequence of operations is defined This process is carried out prior to testing and provides a library of test routines

It was recognized that for sophisticated tests, the sequence of events had to be considered

in detail and this information then programmed It is in the programming of complex sequences that the Design phase provides a simplified method by providing several high-level block types that can be linked in any sequence These high-level functional blocks will be described

in the next section Using traditional programming languages, the user would have to program each step

Test Preview One of the main problems in programming complex sequences for fatigue testing is that in the event of a mistake a valuable sample could be destroyed or substantial time lost In order to minimize this risk, a Test Preview program was developed This program

Trang 29

reads in the Control File as input and displays events that make up the test graphically, such

as data acquisition actions, loading changes with profiles, conditional tests on input or output data, etc The waveforms and the points at which data acquisition will occur are displayed on

a time or cycle axis A sample screen is shown in Fig 3 Visual representation of real-time sequences or simulation has been attempted by several workers [8] but, invariably for complex problems, the visual representation can be difficult to interpret Work is being carried out at University College London (UCL) on a hierarchical graphical representation of real-time processes

calibration and units information from the settings file, to control a test This module has a detailed knowledge of the controller in use and sends the appropriate instructions to the controller The current range of digital controllers handle a substantial amount of real-time processing thereby freeing up the host computer to concentrate on analyzing and presenting data in addition to controlling the test

including presentation-quality graphics The analysis part may be performed with commercially available spreadsheet packages as facilities were developed for transferring data from the FLAPS database structure to other programs However, for numeric intensive processing, spreadsheets can be very inefficient For example, a simple least squares fit can take several

FIG 3 -An example of a test sequence

Trang 30

DHARMAVASAN AND PEERS ON GENERAL PURPOSE SOFTWARE 25 seconds using a commercial spreadsheet whereas when written using a traditional compiled

language such as FORTRAN or C the same result can be achieved in a fraction of a second

It is possible with modern operating systems to dynamically link user-written programs with

this module

Waveform Generation and Control

The different types of waveform required to provide a comprehensive range of fatigue test

simulations are as follows:

Simple Waveforms For most applications, a combination of constant amplitude waveforms

such as sine, square, and triangular are adequate These waveform types are generated by the

digital controller by sending relatively simple commands It is also possible to link several

blocks of different amplitudes, mean level, and frequency together to perform simple block

loading tests

Ramp-type loading is also possible using the ramp generator built into the controller

Service Data Playback Several industries have developed standardized load sequences

that are representative of the load or stress experienced by a typical component The WASH

(Wave Action Standard History) is such an example and is used in the offshore industry The

information on the sequence is stored in the form of turning-point magnitudes (that is, the

peak and trough value) In this way, the amount of information to be stored is substantially

reduced while maintaining the characteristics of the sequence

The peak and trough values are sent to the digital controller, and the controller then fits a

haversine or straight lines between the peak and trough values This type of architecture

substantially reduces the load on the computer allowing time for more data acquisition and

supervisory tasks

Alternatively, the actual digitized data points could be stored However, this information

must be pre-processed after recording to remove spikes and other spurious data

Random Load History from Power Spectral Density Another more generalized way of

generating realistic sequences that contain not only the characteristic amplitude but also the

frequency component is to define a characteristic "power spectral density" (PSD), and then

to obtain the time series from this PSD definition

For each characteristic power spectrum, ~,(to), a digital filter, h,(t), can be found This filter

is the discrete inverse Fourier transform of H,(to), the transfer function of qbx(co) Therefore

where dp~(to) is the power spectrum of a white noise (uniform spectrum) Function hx(t)

effectively amplifies all the desired frequencies so that unwanted frequencies remain but only

as an insignificant part of the time history The desired time history is therefore given by

f x

"qx(t) = hx(t)e(t - "r)d'r

0

(2)

The white noise sources, e(t), are generated by the pseudorandom binary shift (PRBS) register

technique This technique makes use of a register containing a series of digits, 0 and 1 At

every clock pulse (signal output time), all of the digits are shifted one place to the right The

Trang 31

last digit is abandoned and the first is formed from either a two-way or a four-way programmable

feedback loop (Fig 4)

The advantage of the PRBS technique is the excellent frequency control as is demonstrated

in Fig 5 that shows the target and generated spectrum

The frequency content is considered one of the more important factors in corrosion fatigue;

therefore, the PRBS method is considered necessary

The random history within a sea-state can be non-Gaussian for some cases This behavior

can be simply modeled by raising the generated time history to a certain power More sophisti-

cated models using windowing techniques are also available [9] However, these techniques

require more information concerning the load history than just the power spectra This extra

information is still very limited and, therefore, this extension has not been implemented

Data Acquisition

One of the main requirements in data acquisition for fatigue testing applications is the ability

to collect data at specified times or events In the past, it was difficult to achieve this and, as

a result, a vast amount of data was collected in order not to miss any important event However,

the digital controller provides several ways in which the actual occurrence of an event can be

detected and the data collected on the occurrence of the specified event In this manner, data

can be collected intelligently and in manageable amounts

The following types of data can be collected using FLAPS:

1 position, load, or strain feedback,

2 peak and trough data, and

3 ultimate peak or trough

In addition, it is possible to commence data acquisition on an increment of feedback or

increment of time or cycles (linear or logarithmic) with this information It is also possible

to collect data at the instant o f failure o f the sample

Conditional Logic

As explained previously, FLAPS attempts to provide the control structures necessary to

design a complex sequence of testing In order to achieve this, several block types have been

developed to provide actions that will allow the required flexibility

The most basic of these constructs is the Unconditional Loop feature that allows the test

to loop back to a predetermined stage In addition, the looping can be combined with certain

conditions being achieved This is possible by the provision of logical operators such as "less

than" or "greater than." These operators can be applied to collected data or other internal

counters or timers to provide extremely powerful control features

Some of the actions that may be performed on a condition being met are:

One of the major problems faced in designing a general-purpose program for engineering

applications is the diversity and complexity of calculations that may need to be carried out

Trang 33

In the case of FLAPS, it is very difficult to define the type of analysis that needs to be carried

out on the collected data One way of overcoming this would be to export the data to other

analysis programs This has been done to some extent in that there is a facility to export data

to commercially available spreadsheets This will provide simple postprocessing capabilities

For complex and intensive calculations, the spreadsheet approach is not practical due to the

penalties imposed in terms of speed of calculation Also the implementation of some of the

algorithms that may already be coded in a high-level language such as FORTRAN or C in

the dialect of the spreadsheet may be difficult Therefore, a feature for linking calculation

libraries that are user written at run-time has been developed

Application to Fatigue Testing of Tubular Joints

Tubular-welded joints are a commonly used structural component in fixed offshore platforms

These joints are subject to fatigue damage and, as a result, several major research programs

have been carried out [1] Due to the geometric complexity of these components, most of the

early work concentrated on producing stress-life (S-N) information Crack growth data, on the

other hand, were obtained from testing small-scale and tubular joint specimens in air and a

corrosive environment (seawater) under constant amplitude loading Therefore, it has become

necessary to test tubular joints under a realistic load history in order to correlate realistic

behavior with that observed from simpler load histories (such as constant-amplitude sine-

wave loading)

The development of realistic service stress histories and how these are implemented in

FLAPS is described subsequently

Trang 34

Realistic Service Stress History

The realistic history is formulated from information extracted from extensive in-service

load history monitoring projects [10] The monitored results, however, will be unique to the

location, platform dimensions, configuration, payload, foundation behavior, and other factors

Therefore, it is necessary to extract from these lengthy records, the most salient features

relevant to fatigue At the same time, as many characteristics of the load history as possible

should be incorporated in order to avoid omitting any hidden factors related to fatigue, which

may not be evident from current knowledge

The in-service load history was found to behave like a random sequence of short sea states;

therefore, the long-term root mean square (zero mean) of stress/strain varies continuously (Fig

6) The short sea states were found to be generally stationary, and their frequency content can

be described by broad-band, double-peak power spectra (Fig 7) In the majority of cases, the

random load history can be approximated as a Gaussian process The most noticeable exception

is the loading on small-diameter secondary members near the mean sea level The non-Gaussian

effect is principally caused by the nonlinear drag response of the structural members

Wirsching proposed a series of eleven sea states for fatigue reliability analysis of offshore

structures [11] This series makes use of an equation combining the Bretschneider wave spectra

with a nominal response peak to describe the sea-state structural response spectra Based on

the same idea, Hartt and Lin [12] developed a six sea-state (Fig 8) sequence suitable for

fatigue testing The random sea-state sequence is dependent on the long-term occurrence

statistics (fractions of time) and is generated by a Markov chain technique

An international committee was set up to carry out a detailed study of all the recent North

Sea monitoring results with the objective of producing a realistic wave-loading sequence The

proposed standard known as WASH was based on the concept developed by Hartt but also

takes into account statistical sea-state durations The resulting sequence contains a series of

twelve sea states [13]

In the recommended WASH standard history, the two highest sea states have been combined

with the third highest state because of the very small occurrence probability of the former In

addition, in order to compress the lengthy history into a reasonable size suitable for laboratory

testing, the lowest two states are omitted, and the probability of occurrence of the third lowest

state has been reduced to 4.7% Therefore, the WASH standard load history comprises the

"most relevant" 20% of year-round history Further development to incorporate the non-

Gaussian/nonlinear effect is still continuing

The WASH sequence has been generated using the random load generator described earlier

and the resulting time series converted to peak-trough playback data compatible with FLAPS

A sample of the WASH sequence as seen in the FLAPS Test Preview screen is shown in Fig 9

Crack Growth Data in Air and Seawater

The crack growth in the tubular joints was monitored using the ACPD technique [14] Fixed

probes were attached around the welded intersection and connected to the ACPD instrument

through a multiplexer that is controlled by FLAPS A schematic of the test setup is shown in

Fig 10

This arrangement allowed the crack shape evolution to be monitored A sample of the crack

shape data obtained is shown in Fig 11

Future Developments

The software described here, FLAPS, has proved to be very versatile although only part of

the proposed fatigue test description paradigm is used in the implementation However, as

Trang 36

Qa

._

Frequency (8z)

FIG 7 Typical strain power spectrum

computers become more powerful and the amount of features that are at present carried out

by the host computer are moved to the digital controller, it will be possible to implement the

complete description language

Other developments that are taking place in artificial intelligence and object-oriented pro-

gramming technology will also allow the encapsulation of further knowledge of materials

testing within the computer At present, this can be achieved by using dedicated knowledge-

based system development tools to create a test sequence for a particular type of test For

example, the user could specify that a low-cycle fatigue test on a new alloy was required

The system could create a task list with appropriate parameters using embedded knowledge

about FLAPS and the type of test

The Role of Standards

The increasing use of computers in controlling materials test machines, data acquisition

equipment, and other measurement devices makes it important to provide standards so that

equipment from different manufacturers may be used and data interchanged freely At present,

due to the device or computer specific nature of computer-based systems, it is not always

possible to interchange data easily In addition, some of the algorithms implemented to analyze

data depends on the software implementation This has obvious implications in comparing

test data obtained from different sources In view of this, an ISO Working Group has been

set up to draft a suitable standard

Conclusions

A fatigue test description paradigm was proposed in this paper so that general purpose

software for automated fatigue testing can be developed and used at a high level of abstraction

This paper has reviewed the development of an integrated general-purpose program for

automated fatigue testing that implements the functional blocks required but without the full

test description language Even with this limitation, the program was found to be versatile

One of the main findings in using such a program is that the time required to set up and

carry out a complex fatigue test is reduced substantially In addition, due to the high level of

abstraction, the engineer/scientist can concentrate on the application rather than the details of

the controller or device

Trang 38

DHARMAVASAN AND PEERS ON GENERAL PURPOSE SOFTWARE 33 The program has been used for corrosion fatigue testing on tubular-welded joints with

complex service load simulation and extensive crack growth data acquisition in order to

characterize the crack shape evolution in both air and seawater

Trang 39

3 4 AUTOMATION IN FATIGUE AND FRACTURE

FIG lO -Schematic of corrosion fatigue test setup

Trang 40

[2] Topp, D A and Dover, W D., "Review of ACPD/ACFM Crack Measurement Systems," Review

of Progress in Quantitative NDE, Vol 10A, D O Thompson and D E Chimenti, Eds., Plenum

Press, New York, 1991, pp 301-308

[3] Stefik, M and Bobrow, D G., "Object-oriented programming: Themes and Variations," The AI Magazine, Vol 6, 1984, pp 40-62

[4] Minsky, M., "A Framework for Representing Knowledge." The Psychology of Computer Vision,

McGraw-Hill, New York, 1975

[5] Zadeh, L., Fuzzy Sets and their Applications to Cognitive and Decision Processes, U.S.-Japan

Seminar on Fuzzy Sets and their Applications, Berkeley, CA, 1974

[6] Dharmavasan, S., Broome, D R., Lugg, M C., and Dover, W D., "FLAPS A Fatigue Laboratory

Applications Package," Proceedings, 4th International Conference on Engineering Software, Adey,

R A., Ed., Springer-Verlag, London, 1985

[7] Instron 8500 Digital Controller Command Set Manual, Instron Corporation, Canton, MA, 1990

[8] Jones, S., Graphical Interfaces for Knowledge Engineering: An Overview of Relevant Literature,

The Knowledge Engineering Review, Vol 3, No 3, Cambridge University Press, Cambridge, UK,

[11] Wirsching, P H., "Preliminary Dynamic Assessment of Deep-water Platforms." Journal, Structural

Division, American Society of Civil Engineers, Vol ST7, July 1976, pp 1447-1462

[12] Hart, W H and Lin, N K., A Proposed Stress History for Fatigue Testing Applicable to Offshore Structures, University of Florida, Gainesville, FL, 1985

[13] Olagnon, M., "Characterisation of Sea States for Fatigue Testing Purposes," Proceedings, Confer-

ence on Offshore Mechanics and Arctic Engineering, American Society of Mechanical Engineers, Houston, TX, 1988

[14] ACFM Crack Micro-Gauge Model UIO User Manual, Technical Software Consultants Ltd, Milton

Keynes, UK, 1990

Tiêu đề	Automation in fatigue and fracture: testing and analysis
Tác giả	Claude Amzallag
Trường học	University of Washington
Chuyên ngành	Materials Science
Thể loại	Publication
Năm xuất bản	1994
Thành phố	Philadelphia

Định dạng
Số trang	668
Dung lượng	14,3 MB