structural te 35. structural testing applications sinfohost.nmt.edu/~ashok/1b_book_chapter.pdf ·...

985

Structural Tes35. Structural Testing Applications

Ashok Kumar Ghosh

This chapter addresses various aspects of testing ofa structural system. The importance of the man-agement approach to planning and performingstructural tests (ST) is emphasized. When resourcesare limited, this approach becomes critical to thesuccessful implementation of a testing program.The chapter starts with illustrations of some ofthe past structures that were built using con-cepts developed through testing. Most often, thesestructures were built even before the principlesof engineering mechanics were understood. Atpresent, due to the unprecedented expansion ofcomputing power, numerical and experimentaltechniques are interchangeably used in simulatingcomplex natural phenomena. Despite encouragingresults from simulation and predictive modeling,structural testing is still a very valuable tool inthe industrial development of product and pro-cess, and its success depends on judicious choice oftesting method, instrumentation, data acquisition,and allocation of resources. A generic descriptionof the current test equipment and types of meas-urements is included in this chapter. After carefulselection, three case studies are included. Thecomplexity involved with the modeling of struc-tural steel retrieved from the collapse site of theWorld Trade Center (WTC) under high-rate andhigh-temperature conditions is highlighted in thefirst case study. The second case study highlightsthe importance of the planning phase in providingthe basis for manageable and high-quality testing

35.1 Past, Present, and Futureof Structural Testing ............................. 987

35.2 Management Approachto Structural Testing ............................. 99035.2.1 Phase 1 – Planning and Control ..... 99035.2.2 Phase 2 – Test Preparation ............ 99335.2.3 Phase 3 – Execution

and Documentation ..................... 996

35.3 Case Studies ......................................... 99735.3.1 Models for High-Rate

and High-Temperature SteelBehavior in the NIST World TradeCenter Investigation ..................... 997

35.3.2 Testing of Concrete HighwayBridges – A World Bank Project ...... 1002

35.3.3 A New Design for a LightweightAutomotive Airbag ....................... 1009

35.4 Future Trends ....................................... 1012

References .................................................. 1013

of concrete highway bridges. The final case studydetails the development of a lightweight auto-mobile airbag from inception through innovation.This case study also illustrates the close ties be-tween structural testing and numerical simulation.The chapter closes with examples of a few futurestructural systems, highlighting the complexityinvolved in their testing.

A structure is an engineered system that must fulfillmultiple requirements. Testing of a structural systemhas a definite role to play in the overall process ofdevelopment, be it in the context of developing newproducts or assessing existing products. The design-by-testing methodology (Fig. 35.1) seems to work well inthis development process.

Testing expands the knowledge base regardingdifferent parameters that influence the design and devel-opment. As one learns more about a particular process,one tends to cut efforts on physical testing to reducetime and cost, resulting in greater emphasis on predic-tive modeling. Three primary reasons attribute to sucha shift in emphasis.

PartD

35

986 Part D Applications

First, the unprecedented expansion of computingpower in the last decade has enabled us to develop high-performance scientific computing hardware. Betweenthe 1990s and the present, this computing speed hasimproved 20 times from 2 tera-operations per second(OPS) to 40 tera OPS. All these computational capa-bilities are ideal for supporting highly complex andinterdisciplinary application-oriented developments.

Second, with these improved supercomputing capa-bilities, it is often possible to model a complex physicalphenomenon. For most of the conventional problems,there is extraordinary agreement between the simula-tion and the real situation, thus making the requirementfor experiments less important. The initiative Engi-neering Sciences for Modeling and Simulations-BasedLife-Cycle Engineering is a collaborative research pro-gram by the National Science Foundation (NSF) andthe Sandia National Laboratories (SNL) that focuses onadvancing the fundamental knowledge needed to sup-port advanced computer simulations. SNL is movingtowards engineering processes in which decisions arebased heavily on computational simulations, thus cap-italizing on the available high-performance computingplatforms. In the future, dominant use of analysis willbe a part of the design process.

Third, engineering design and development throughexperimentation will be used as a fact-finding tool thatallows one to understand the circumstances surround-ing a given problem and the main variables controllingit. Characteristically, complex systems are nonlinear,coupled, and exhibit multiple subsystems having mul-

Physicalprototyping

Analysis andoptimization

Identificationof the need

Definition ofthe problem Design Testing ProductionStop

Fig. 35.1 Design-by-testing methodology process flowchart

Identificationof the need

and definitionof the problem

Physicalprototyping and

testingProduction

Virtual prototyping

Parametric designand optimization

DesignSimulation-driven

product development

Fig. 35.2 Simulation-driven product development process flowchart

tiple lengths and time scales. Reliable predictions ofsuch systems need conscious coordination and integra-tion of experiment, theory, and computer simulation.The challenge is to increase predictability by creatinga robust design with reduced sensitivity to any influenc-ing variable. A characteristic of future development willbe simulation-driven product development (Fig. 35.2),where repeated tests are eliminated to cut the time andcost.

Numerical and experimental techniques are inter-changeably used in simulating complex natural phe-nomena. In the past, scientists such as Leonardo daVinci, Galileo, and many others adopted experimen-tal techniques to obtain solutions to complex problemswhereas continuum mechanics methods would lead topartial differential equations. Many argue that, in the fu-ture, numerical techniques will take over the disciplineof experimental techniques to solve so-called not-so-complex problems. This seems like the end of the roadfor a fading discipline. In the past, considerable ef-forts were made towards the development of knowledgebased on actual testing of structural systems. Most ofthese test results were used to validate predictive mod-els. Validating a predictive model based on experimen-tal test data is not an efficient use of resources. Instead,simulation-driven predictive models will be developedwith minimal involvement of experimentation. Struc-tural testing would be imperative for problems that arequalified as either an unknown or important situation.

Future structural systems will be active, intelligent,and adaptive, similar to live biological structures. Live

PartD

35

Structural Testing Applications 35.1 Past, Present, and Future of Structural Testing 987

in this context means that these structures will havecharacteristics like build-in sensors, being adaptive tothe prevailing environment, and self-diagnostics andself-repairing/healing. Understanding biological com-plexity requires sophisticated approaches to integratingscientists from a range of disciplines. These includebiology, physics, chemistry, geology, hydrology, socialsciences, statistics, mathematics, computer science, andengineering (including mechanics and materials). Thesecollaborations cannot be constrained by institutional,departmental or disciplinary boundaries. The NSF isin the process of initiating a model-based simulation(MBS), which integrates physical test equipment withsystem simulation software in a virtual test environmentaimed at dramatically reducing product developmenttime and cost [35.1]. MBS will involve combiningnumerical methods such as finite element and finitedifference methods, together with statistical methodsand reliability, heuristics, stochastic processes, etc., allcombined using supercomputer systems to enable sim-ulations, visualizations, and virtual testing. Expectedresults could be less physical testing or, at best, strategi-cally planned physical testing in the conduct of researchand development (R&D). The manufacturing of the pro-totype Boeing 777 aircraft, for example, was based oncomputer-aided design and simulation.

The training of engineers and technicians in test-ing and laboratory activities has decreased significantlysince the 1960s due to reduced funding. Funding avail-able for conducting tests, acquiring and maintainingequipment, and developing new techniques continuesto decrease, and yet more detailed information is ex-pected from the limited testing performed. Thus, it isabsolutely imperative to plan the test program so that itis possible to understand the system from the minimumnumber of tests performed.

This chapter deals with different aspects of modernST, where the material used is inert and the loadings

applied are conventional. This chapter will start witha discussion of the past, present, and future of structuraltesting in Sect. 35.1 with some historical illustrationson structural testing; salient features of structural test-ing that are characteristic of the past will be touchedupon. Section 35.2 covers the management approachto structural testing, describing guidelines for planningand performing tests. A generic description of the testequipment and types of measurements is included. Sec-tion 35.3 consists of three case studies.

Case study 1 describes modeling for high-strain-rate and high-temperature steel behavior in the NationalInstitute of Standard and Technology (NIST) WorldTrade Center (WTC) investigation. A major part ofthe investigation was the metallurgical analysis ofstructural steel recovered from the two WTC tow-ers. The case study highlights the challenges facedin terms of cataloging the recovered steel to charac-terize the failure modes, quantifying the temperatureexcursions that the recovered steel experienced, andcharacterizing and modeling its mechanical proper-ties.

The second case study highlights how managementplays a crucial role in the planning of a major testingprogram funded by the World Bank. The planning phaseprovides the basis for a manageable and high-qualitytesting process. The project involves testing of concretehighway bridges.

Case study 3 details the development of a lightweightautomobile airbag from inception through innovation toengineering development. The study demonstrates theengineering progression of the design and the role thattesting and analysis played in the development. New testand analysis methods were developed as required dur-ing the engineering phase as the design was eventuallycompleted and qualified. This case study also illustratesthe close ties between structural testing and numericalsimulation.

35.1 Past, Present, and Future of Structural Testing

A structure can be a large system (building, bridge, ship,or aircraft) or it can be something as small as a mechan-ical robotic actuator or an electronic component madeto perform multiple functions; for example, a build-ing structure has the primary function of withstandingloads (self-weight/dead, superimposed/live, wind, seis-mic, etc.), while providing the users with safety andcomfort from the effects of external loads, from shock

and vibration, acoustics, thermal, radiation, etc. Thebuilding is also required to be aesthetically pleasing.Through testing, the response of the structure underapplied loads (force, pressure, temperature, shock, vi-bration, and other loading conditions) is determined. Itis also common for a structure to be required to fulfillnew, different or extended functions after it has beendesigned and built. Most structural testing is considered

PartD

35.1


to be an art performed by experienced engineers whoare familiar with the structure, who will then evaluateand interpret the test results based on their knowledgeand experience as much as through prescribed proce-dures and formulae. In this chapter, an effort is made topresent this knowledge of structural testing in a system-atic manner.

The history of documented structural testing is asold as the time of Leonardo da Vinci’s tensile strengthtest. The test involved loading iron wires of differentlengths. Leonardo da Vinci used a setup similar to theone shown in Fig. 35.3. He used wires of similar diam-eter and different lengths to suspend the basket (b) froman unyielding support (a). The basket was filled slowlywith sand, fed from a hopper (c) with an arrangement sothat, when the wire breaks, a spring closes the hopperopening.

Da Vinci experimentally recorded the failure loadsfor different length wires. He observed that longer wireswere weaker than shorter wires. Early investigators ofthe time had no explanation for this observation basedon the concept of the classical mechanics of mater-ials. What was true for steel as observed by Leonardoda Vinci in the 15-th century is true today for any engi-neering material. The compressive strength of concretecylinder (length/diameter = 2), a heterogeneous mater-ial, shown in Fig. 35.4, clearly demonstrates the samephenomenon [35.2].

Galileo Galilei demonstrated how structural shapeand geometry can influence load-carrying capacity.He arrived at the correct conclusion that the bendingstrength of a rectangular cantilever beam is directly pro-portional to its width but proportional to the square of itsheight. After Galileo came the English physicist RobertHooke, who observed that the force with which a springattempts to regain its natural position is proportional tothe distance by which it has been displaced. Around thesame time, a group of mathematicians and physicistscontributed to concepts of force, mass, and formulationsof the principles of effect and counter effect. Towardsthe end of the 17-th century, the first theoretical investi-gations were carried out on the static behavior of vaults,which is among the most important elements of con-temporary engineering structures. The next noteworthywork was that by Leonard Euler, who published an ex-haustive treatise on plane elastic curves (bending lines).

Knowledge of the properties of building materialshas gradually developed into an important branch ofengineering science. In the later part of the 18-th cen-tury, numerous strength tests, mainly compression ofstone and mortar and bending tests with iron-reinforced

c

a

b

Fig. 35.3 Testsetup usedby Leonardoda Vinci (seetext)

0 15 30 45 60 75 90 105

Strength (% of std 15×30 cm cylinder)

Diameter of cylinder (cm)

120

115

110

105

100

95

90

85

80

Fig. 35.4 Concrete cylinder under compression

stone beams, were carried out by a large number ofscientists. Thanks to the ceaseless research work of un-told, able engineers with theoretical backgrounds, themethods of structural analysis have since been steadilyperfected and extended to new tasks. The knowledgebase generated through these efforts has contributed tothe development of numerous predictive models. No-one contributed as much as Navier did in the later partof 18-th century and early part of the 19-th century.In his project for the suspension bridge over the riverSeine, Navier [35.3] carefully selected the material anddesigned the shape of the anchors based on actual cal-culations of mechanics Fig. 35.5.

In the past, structures were often developed beforethe principles of engineering mechanics were under-stood. The era of theoretical mechanics followed theera of applied mechanics, and personalities such asPoisson, Cauchy, Lamé, Saint-Venant, and many oth-ers contributed to the growth of the knowledge base onthe theory of structures.

PartD

35.1

Structural Testing Applications 35.1 Past, Present, and Future of Structural Testing 989

Anchor

Fig. 35.5 Navier’s concept for an-choring the cable (see text)

The Industrial Revolution in England transformedthe whole world with coal, steam engines, and railways.Great Britain was the first country to produce iron andsteel in large quantities. With improved quality (highstrength, elasticity and uniformity) came improvedeconomy, as engineers could determine the dimensionsof the structural elements very accurately. With the in-vention of the hydraulic binding agent called cement byJoseph Aspdin came the age of cement and reinforcedconcrete. During the 19-th century, the dimensions ofstructures were generally determined so that the admis-sible stress, calculated on the basis of Hooke’s law, waslimited to a certain fraction of the breaking stress (thesafety factor). It was found, however, that the safety fac-tor thus defined did not represent a reliable measure ofthe actual safety of the structure, especially for inde-terminate structures and structures made with concreteand masonry. With the statically indeterminate struc-tures, however, the local application of a stress beyondthe elastic limit will generally lead to a compensationof forces, inasmuch as overstressed parts are relievedby the action of less heavily stressed members. It is ob-viously only with the aid of extensive tests that thesehighly complicated conditions can be clarified. Thereare other problems connected with the strength of ma-terials, which are impossible or difficult to solve on thebasis of the classical methods of the mathematical the-

Test specimen

ScalePivot

Fig. 35.6 Capp’ssingle-lever ex-tensometer

ories. Among these problems is the concentration ofstresses, which occurs in complex structural elementsin the immediate vicinity of abrupt changes of cross-section, where the concept of stress optics can be ideallyapplied.

As far as measurement devices are concerned,for a long time experimentalists from all over theworld developed various methods independently with-out any standardization. Capp’s single-lever extensome-ter [35.4] was one of the early measurement devices ofsurface strains (Fig. 35.6). For more information on thetheory and application of various kinds of extensome-ters see Chap. 13.

The next generation of extensometers were mountedwith mirrors replacing the moving pointer of the rack-and-pinion system, resulting in greater accuracies. Theywere difficult to calibrate, sensitive to vibrations, bulky,and had restrictions on where they could be mountedon the structure or specimen. Other optical methods arethe moiré method (for surface displacement), photoe-lasticity (for complicated assemblies and parts), brittlecoating (for surface crack pattern), holography (forin-plane displacements), speckle pattern (for onlinemonitoring of surface deformation), and digital imagecorrelation.

Earlier descriptions of mechanical measurementsof displacement showed the use of mechanical gages.The first electrically based displacement measurementsystem was the linear variable differential transformer(LVDT). It was not until the invention of electric re-sistance strain gages and their application in variousdevices that experimental mechanics began to havea great impact on the design of structures.

Most technologies associated with structural testinghave been revised or improved in the past few decades;several new technologies have also been introduced.Portable equipment has been developed, permittingfield investigation of a wide range of potential prob-lems with capabilities to perform complicated tests

PartD

35.1


under adverse conditions. Most of the data-miningmethods/techniques are discussed in Parts B & C ofthis handbook. Despite encouraging results from simu-lation and predictive modeling, industry is still far from

reaching the point where it can discard experimentation.Testing remains an important activity, and its successdepends on judicious choice of testing method, instru-mentation, data acquisition, and allocation of resources.

35.2 Management Approach to Structural Testing

Test purposes and evaluation criteria are the necessaryguides in planning and performing tests as well as thebasis for interpreting the test results. For any type ofstructural testing the activities can be grouped into threephases

• Phase 1, planning and control• Phase 2, test preparation• Phase 3, execution and documentation

An important part of testing is the effort spent dur-ing the planning and preparation. As a rule of thumb,the distribution of efforts is 20% for planning, 40%preparation, and 40% for the actual execution and doc-umentation of the test program. This means that, duringthe creation of the functional specification, a master testplan is established. The master test plan describes whoperforms which type of test and when these tests areperformed. On the basis of the agreed test plan, moredetailed test plans are made. It is worth noting that theremay be legal or regulatory requirements that must beconsidered while planning for such tests.

35.2.1 Phase 1 – Planning and Control

The planning and control phase starts during the spec-ification of the functional requirements. The planningphase provides the basis for a manageable and high-quality testing process. After the test assignment hasbeen approved, the test team is to be assembledcarefully and the test strategy is defined. Manpowerselection is crucial for successful outcomes from thetesting program. Most testing requires specialist knowl-edge and skills and the accompanying education toderive high-quality results. A test team typically con-sists of personnel belonging to a large variety ofdisciplines. Adequate participation of testing specialistsis essential, both in the area of test management and inthe area of testing techniques. Test strategy is basicallya communication process with all parties involved, try-ing to define the parts and allotting the necessary time

frame. The aim is to have the most feasible coverage ofthe testing organization and to define the test infrastruc-ture. The objective of the next part of this phase is tomanage the progress of testing with regard to the timeand resources used. In accordance with the test plan,the testing process and the quality of the test objectare documented and reported. The most important de-liverable of a test program is a quality test report whichalso describes the accompanying risks. From the start ofthe testing process, testers develop a view of quality. Itis important that quality indicators are established dur-ing all phases of the testing process. Periodically, andwhen asked, management receives a quality report onthe status on the testing process to decide if the testingshould be continued or discarded. Major issues that in-fluence the decision-making process during this phaseare described in the following subsections.

Safety RequirementPotential hazards associated with conducting structuraltests can be introduced anywhere from a minor sourcesuch as electrical shock to a major source like the failureof a hoisting crane, the snapping of a prestressed ten-don during tensioning, or the inhalation of toxic fumes.Insights into safety-related issues can be very usefulduring the planning stage. There are many agencies andregulations such as the Occupational Safety and HealthAdministration (OSHA) [35.5]), the National Environ-mental Policy Act (NEPA) [35.6], and the ResourceConservation and Recovery Act (RCRA) [35.7]) thatenforce safety laws in the workplace.

Accuracy and ResolutionSome types of tests (e.g., modal analysis) often requiremany cycles of testing to develop statistically suffi-cient insights into structural behavior. Test quality isdefined through two parameters, obtaining sufficient in-formation to describe structural response, and verifyingthat the information obtained is valid. A well-executedstructural test provides objective evidence of structuralbehavior.

PartD

35.2

Structural Testing Applications 35.2 Management Approach to Structural Testing 991

Specifications (Requirements of Code)Codes and specifications are developed to bring stand-ardization to experimentalist all over the world. Struc-tural testing has been profoundly affected by thedevelopment of codes and specifications. The Ameri-can Society for Testing and Materials (ASTM [35.8])has developed test procedures for various kinds ofmaterials. There are organizations who develop testprocedures and specific requirements for structural test-ing. The American Railway Engineering Association(AREA [35.9]), the American Association of StateHighway Transportation Officials (AASHTO [35.10]),and the American Institute of Steel Construction(AISC [35.11]) are a few. The way a technique isapplied depends on the way the specifications arestructured. Based on the testing strategy and the de-velopment documentation, appropriate test specificationtechniques are selected and tailored during the prepa-ration phase. Chapter 17 of the International BuildingCode 2003 (IBC [35.12]) deals with structural testsand special inspections. Where proposed constructioncannot be designed by approved engineering analysis,or where the proposed construction does not com-ply with the applicable material design standard, thesystem of construction or the structural unit and theconnections shall be subjected to the tests prescribed insection 1714 [35.12].

Classification of Structural TestingThere are different ways in which structural testing canbe classified: full scale and models, calibrations, accep-tance or quality assurance types, proof tests, materialcharacterization, etc. These tests are aimed to deter-mine stiffness or load deformation, energy absorption,strength or failure load, vibration mode or other struc-tural characteristics.

Full-Scale Test. Characteristics of a structure areinfluenced by many parameters such as material (non-homogenous) properties, construction and fabricationtechniques, load environment, boundary conditions.The influence of these parameters is not linear, thusmaking prediction of scaled models difficult. When pos-sible, full-scale structural tests are recommended foraccurate prediction of the characteristics of the system.

Load Test. The purpose of conducting load tests is usu-ally to prove that a structure or its components canwithstand the anticipated loads. For example, a highwaybridge is designed to withstand a particular traffic load.

Through load tests on the actual structure, this value canbe determined. Section 35.3.2 deals with the load test ofa bridge structure.

Destructive/Failure Test. Often tests are carried outto understand the failure characteristics of a structure.Failure tests also indicate the reserve strength and de-formation capabilities in a structure. Destructive testingis performed to gain knowledge of the failure modes. Atpresent, the knowledge base of the structural response toa blast load is not well established. To develop designcriteria, destructive tests are performed to understandhow a structural system will respond to blast loads.Once this design basis is established, computer simula-tion will replace future destructive testing to save timeand cost.

Partial Load Test. Often, samples are taken from thetest structure in order to understand the characteristicsof the material used for the structure. This test is per-formed when it is necessary to predict the behavior ofthe structure, but the design and construction drawingsare not available or have changed. More on this type oftesting is presented in Sect. 35.3.2.

Nondestructive Test. Nondestructive tests are done tounderstand the elastic response of a structure and canbe classified into diagnostic and proof tests. In a diag-nostic load test, the selected load is placed at designatedlocations on the structure and the response of an individ-ual member of the structure is measured and analyzed.During a proof load test, a structure is carefully andincrementally loaded until the structure approaches itselastic limit. At this point, the load is removed and themaximum applied load (the proof load) and its positionon the structure are recorded.

Scaled/Model Test. Full-scale tests are time and re-source intensive. Often, scaled tests are done to predictthe behavior of a structure by extrapolating observa-tions based on scaled/model tests. There can be seriousconcerns with any scaled test, and the output shouldbe properly analyzed to arrive at a proper prediction.Sometimes these concerns can be addressed by con-ducting a series of tests with different scale factors todetermine the relation of the observed behavior withsize; for example, modal analysis (mode shapes andfrequencies) can predict how a structure will storedeformation energy under a given loading condition.Modal analysis judgment and experience give insight

PartD

35.2


No

Identified:• Scope of the test program is clearly defined and understood.• Resources are recognized, including information on the structure to be tested (new or old).• Specific requirements including the time frame are determined.• Deliverables are defined.

Is therea code of specification

in place?

Yes

Study code specifications on• test specimen requirement,• loading requirement,• data acquisition systems needed.

Evaluate potential hazards anddetermine the controlling measures.

Visualize the structure in service.Define:• loading environment,• boundary conditions,• simulate a test program fulfilling all the testing requirements.

• Prepare the test structure.• Simulate the boundary conditions.• Arrange the instrumentation and data acquisition system.• Check the connections carefully.• Check the safety requirements.• Prepare for the testing and documentation.

Carry out load sensitivity analysis anddetermine possible loading that can beaccounted for in the test program. Develop apredictive model to superimpose various loadcases.

YesIs it possible toaccount for all possible

loading?

No

Fig. 35.7 Flowchart of the test preparation during phase 2

into the dynamic response of a structure or componentand avoid instability.

To perform a modal test, the test item must be ex-cited (naturally or artificially) in the same manner asthe structure would vibrate in different modes. Theresponse to the excitation can be measured by the con-ventional transducers, which were discussed in Parts B& C of this Handbook. A typical modal test includes anarray of accelerometers placed on a structure and con-nected to amplifiers and a data recording system. Anexcitation can also be fed to the recording system to col-lect simultaneous data. Thus a modal analysis is oftenused to troubleshoot a vibrating system.

Material Test. Structural testing and materials testingmust both be considered to construct an accurate pre-dictive model. For material testing, specimens shouldbe extracted from similar structures to be analyzed.Chapters from part B & C of this Handbook containsdetailed descriptions of various conventional tests thatare performed to characterize materials used to build thestructural system.

Forensic Tests. Forensic tests are performed to un-derstand the real cause of any failure of a structure.Failure is a quick phenomenon to which a number offactors contribute. Analysis of such a phenomenon isvery challenging. Forensic tests that determined the se-quences of various phenomena that contributed to theeventual collapse of the Twin Towers are highlighted inSect. 35.3.1.

The output of the planning and control phase isa statement or description of the work to be per-formed based on a feasibility study. Feasible meansthat a test is possible on the basis of availability ofresources, such as manpower, funding, equipment, man-agement tools, and time. This phase will also clearlydefine

• the tests to be performed• the available resources• the specifications (administrative, legal, codal andsafety) to be met• the output and information on the quality and accu-racy of the results

PartD

35.2


35.2.2 Phase 2 – Test Preparation

After the planning and control phase, preparation foractual tests will start. In this phase a systematic studyis carried out to understand the test program and deliv-erables to ensure preparation for a specific test will bethorough and adequate. The structure to be tested is tobe designed and developed as if it was a new structure.In the case of an existing structure, the test structure willbe identified. Figure 35.7 shows the flowchart of the testpreparation during this phase.

Loading on StructureThe loading conditions used in the laboratory and fieldtests should simulate the service or use conditions.Often, equivalent static loads are used to adequatelyrepresent and simplify some complex dynamic environ-ment. In addition to the magnitude of loading, the pointof application and the directions are factors that influ-ence any testing program. When the structural systembehaves linearly, it is possible to apply loading sequen-tially. Again, for a linear-elastic system, it is possibleto determine the forcing functions mathematically fromthe measured responses.

Dead LoadDead loads do not typically change with time, or theychange so slowly with time that they produce only static

Line drive supply

Vibration meter

AccelerometerTarget beam

Impact hammer

Heavy support

Spectrum analyzer

Fig. 35.8 Schematic diagram of vibration test setup

responses. Dead loads include the weight of the struc-ture and loads arising from any permanently attachedelements. Dead loads can be accurately estimated know-ing the dimensions and density of the elements.

Superimposed LoadLoads that are temporary in nature come under the su-perimposed load category, like the weight of people ina building. Loads due to wind, earthquake, vibration ofmoving vehicles, etc. are treated as live or superimposedloads. Often transient loads that arise from impacts,gusts, and other rapidly varying loads are accounted forby additional multiplying factors. Based on safety andserviceability requirements, multiplying factors for thebuilding structure are recommended in the InternationalBuilding Code (IBC) 2003 [35.12]. When these valuesare not available, laboratory tests can be conducted todetermine them.

Thermal LoadSimulations of thermal loading are rather complex,especially when full-scale tests are conducted. Casestudy 1 illustrates the challenges encountered whilemodeling the behavior of structural steel under hightemperatures due to the combustion of the aircraftfuel inside the Twin Towers before their collapse. Thetowers were subjected to dead loads, imposed loads,thermal loads, and impact loads all at the same time.

PartD

35.2


Chapter 7 of the IBC 2003 deals with construction re-quirements for fire-resistance design for materials usedin structures. ASTM E119 contains the necessary testprocedures for fire testing.

Dynamic/Vibration LoadDynamic testing is a widely accepted test method tocharacterize a structure and its components subjectedto time-varying loads. The structure is excited at var-ied frequencies to determine the proof load, dampingcharacteristics, dynamic modulus, mode shape, etc. Fig-ure 35.8 shows a schematic of a typical experimentalsetup used to obtain dynamic characteristics of lami-nated composite beams under an impact load. Vibrationof the beam causes the support to vibrate, which in turninfluences the vibration of the beam. Thus, for properinterpretation of the test observation, the coupling effectbetween the support motion and the structure should becarefully accounted for.

Blast and Explosive LoadAfter the 9/11 terrorist attacks on the Twin Towers,the importance of investigating structures under blastloads has intensified. Blast loads create waves in theair which impact on exposed structures. Loading sys-tems for blast waves and explosive detonations usuallyrequire special facilities. Even for very small quantitiesof explosives, rugged facilities are needed to contain,or at least focus or direct, the released energies. Theoverpressure/duration relationship for a high explo-sive or nuclear explosion in air is shown in Fig. 35.9,where p is overpressure (or air blast pressure) and t istime.

0 0.2 0.4

Positive phase

Negative phase

0.6 0.8 1 1.2 1.4 1.6 1.8

p/p0

p/p0 = (1–t/t0) e–t/t0

t/t0

1

0.8

0.6

0.4

0.2

0

–0.2

Fig. 35.9 Blast load–duration relationship

In the figure, the decay of pressure after the first in-stantaneous rise is expressed exponentially. The valueof the peak instantaneous overpressure p0 will dependon the distance of the point of measurement from thecenter of the explosion. In a TNT (trinitrotoluene) ex-plosion, p0 might be 200 psi (1.38 MPa) or 300 psi(2.07 MPa) at the point of burst of the explosion, butwould rapidly diminish with distance. In a nuclear ex-plosion equivalent to 1000 tonnes of TNT, the peakoverpressure would be 2000 psi (13.78 MPa) at 30 mfrom the center of the explosion.

When an explosive charge is detonated in contactwith or very close to a structure in air, the loading can nolonger be considered uniform over the area of compo-nent faces. The accurate measurement of dynamic loadson structures is a demanding subject. The physical prop-erties of blast waves as they strike the structure are mostcommonly recorded in terms of pressure. When an ex-plosion occurs within an enclosed room, the impulsethat loads the structure may be broken into two dis-tinct parts, the shock front and a quasistatic gas pressureload. This initial blast load is due to the primary shockfront and secondary reflected shock fronts. Air shockloadings typically have very short time durations, butmay have peak pressures of several thousand pounds persquare inch (1 psi = 689.47 Pa) or more. At later times,the structure is loaded primarily by a low-intensity long-duration gas impulse that is due to the expansion of theexplosive byproduct gases within the structure.

Recording the pressure and acceleration data isa challenge by itself. High-speed photography (typically2000–3000 frames per second or higher) is also used tomeasure structural response.

Wind LoadThe design and development of a structure that will beexposed to wind loads for most of its service life willusually be tested in a wind tunnel to understand theresponse of the structure. Aerodynamics plays an im-portant role in the design of an aircraft, automobile,or a high-rise building. The Tacoma Narrows suspen-sion bridge collapsed due to loss of rigidity becauseof a 42 mph (18.78 m/s) wind, even though the struc-ture was designed to withstand winds of up to 120 mph(53.64 m/s).

Probabilistic LoadsIt is well known that structures and material char-acteristics are seldom known deterministically. Thisphenomenon was observed by Leonardo da Vinci inthe 15-th century, and is still true today. Degrees of

PartD

35.2


uncertainty in determining material characteristics callfor probabilistic approaches to determine the strength ofmaterial (as illustrated in case study 2).

Loading SystemsLoading systems are used to apply loads during struc-tural testing. The objective is to replicate service loadas nearly as possible. It is very difficult to create exactloading situations in the laboratory, and thus approxi-mations must be made.

SimulationConsider the simplest situation of applying a uni-formly distributed load (UDL) on a plate structure.Figure 35.10 shows at least three ways one can achievea UDL on a plate structure. The UDL remains uniformas long as the plate does not deform excessively. Assoon as the plate deforms, the curvature effect will makethe load nonuniform. While trying to simulate UDL,careful consideration is needed to make sure that theUDL is accurately represented throughout the loadinghistory. One way to account for this difference betweentrue UDL and simulated UDL is by adjusting for cur-vature effects in the response. Thus, it is important tocheck the method used to apply the loads. This checkincludes the locations, directions, and manner of appli-cations (point, line, distributed, or inertial). It shouldalso include loading rates and other time-dependent de-scriptions of the loading, such as magnitudes as the testprogresses.

Control signal

Load signal

Stroke signal

Closed loop

Readout equipment anddata/control processor

Feedbackselector

Valve drive orcontroller

Loadcell

Specimen

Servovalve

LVDT connector

Fig. 35.11 Components with connections for a closed-loop control system

Rows of hydraulic jacks

Pneumatically Using sand bags

Fig. 35.10 Application of uniformly distributed loads can be donedifferent ways, all of which are different when the beam starts tobend excessively

Loading EquipmentTesting machines are mechanical systems with vari-ous controls, including manual, mechanical/hydraulic,and electrical control with feedback systems. Two ba-sic types of controls have the ability to impose eitherloads or displacements. An excellent description of testcontrol systems is provided in [35.4]. Three main typesof force testing machines have been developed and arein use – the dead weight, screw-driven, and hydrauli-cally actuated. Primarily, there are two ways to controlloading conditions (interaction between diagnostic andfeedback).

1. An open system: performing tests in which the op-erator retains all the control on loading/unloadingand the loading rate

2. Closed-loop systems: performing tests under an en-vironment where the machine controls loading/

PartD

35.2


unloading and the loading rate. However, the opera-tor retains a manual override option. A simple blockdiagram of a closed-loop actuator with a servo–valve combination system is shown in Fig. 35.11.

CalibrationMeasurement processes depend on the calibrations ofthe instruments used. Inconsistent calibrations are com-mon and should be determined in any testing programand the proper corrections should be applied in the re-sponse data. By performing periodic checks, one candetermine inconsistency in instruments behavior. Of-ten, the instruments used for measurements could bethe source of undesirable responses in a testing pro-gram. Calibration activities are essential in structuraltesting because they give credibility to the data gener-ated.

Test ControlThere are many ways to apply control and feedback sys-tems in structural testing. It is important to select theproper test equipment to match the requirements.

35.2.3 Phase 3 – Executionand Documentation

In order to satisfy the agreed delivery schedule, the ex-ecution phase should start as soon as the preparation forthe test program is complete. Sometimes, it may be nec-essary to conduct exploratory tests to determine the testdomain. There are other times when part of the data isused to support a later part of the testing process. Thedifference between the actual test results and the ex-pected results can indicate a problem area. This can bedue to some error with the infrastructure, an invalid testcase, or poor judgment. During the entire test executionstage, one should allow for quick and reliable reporting.Possible questions to be addressed are what percentageof the program is complete, how do these tests comparewith the prediction, what are the trends, should testingbe continued, etc. The infrastructure for testing includesall facilities and resources needed for structural testing.The selection of the correct infrastructure is strongly de-pendent on the type of resources and hardware platformavailable.

Obtaining DataThere are wide ranges of transducers, gages, sensors,and other measuring equipment available, as describedin Parts B & C of this Handbook. The purpose of a trans-ducer is to convert the physical changes occurring in

a structure due to applied loading into an electricalchange, usually by altering voltage or current. Thereare transducers that provide a point response of thestructure and there are others that represent full-fieldor whole-field results. Judicious selection of the test-ing machine, the right kind of sensors to collect data,and proper processing tools will yield the expected re-sults. Often it is expected to combine both point andfull-field techniques to better understand the responseof the structure.

Point Techniques. Measurement devices such as straingages, capacitance gages, load cells, accelerometers,and thermocouples are used to collect structural re-sponse in terms of accelerations, displacements, forces,velocities, temperatures, and strains. Response data ob-tained through the point technique are then fed into ananalytical model to analyze the response of the wholestructure. For more information on these techniques,refer to Chap. 12 for strain gages.

Full-Field Techniques. This technique examines a por-tion of the structure or at times, the whole structure.Most of the optical methods provide full-field struc-tural response. The use of such methods (moiré,holography, speckle, photoelasticity, brittle coating,optical-fiber strain gages, thermoelastic stress anal-ysis, or x-ray analysis) prevent the use of anothermethod in the same test. The theories behind most ofthe full-field techniques are provided in Chaps. Chap-ter 29 (optical methods), Chapter 20 (digital imagecorrelation), Chapter 21 (geometric moiré), Chapter 22(moiré interferometry), Chapter 24 (holography), Chap-ter 23 (speckle methods), Chapter 25 (photoelasticity),Chapter 14 (optical fibers), Chapter 26 (thermoelasticstress analysis), and Chapter 28 (x-ray stress analy-sis).

Instrumentation and Data Acquisitionand Processing

The term instrumentation encompasses all the devicesused to sense, connect, condition, display, and recordthe data. Selection of the instrumentation system is dic-tated by the requirements of the loading environment.For more information, refer to [35.4, Chap. 10].

Data Transmission. To avoid hazards in many testsituations, sensitive instruments and personnel can belocated remotely. Thus the cables that connect trans-ducers with the signal conditioner and recorder shouldbe such that their signal attenuation is minimized. This

PartD

35.2

Structural Testing Applications 35.3 Case Studies 997

attenuation is more pronounced in the case of high-frequency signals.

Signal Conditioning. Once a mechanical response issensed, the output of the sensor needs to be amplifiedfor use in the recorders. This is known as signal condi-tioning, and is done after the response is converted intoan electrical signal.

Computer Data Acquisition. This topic is covered ex-clusively in ([35.4, Chap. 10]).

Completion PhaseThe completion phase is launched after the executionphase is completed. These activities are generally lessstructured or often forgotten, and when under highpressure, concessions are made. There should be someprovisions on how to carry out additional tests (if sowarranted) after evaluating test results based on themain test program.

Experimental Methods – Evaluation Criteria. Testevaluation criteria consist of methods according towhich test results can be interpreted and understood.Evaluation criteria are usually specified through meth-ods and techniques of engineering mechanics and

material science. The data obtained from the associatedmeasurement devices used in testing will make corre-lations and comparisons of theory and test data moredirect and easier to understand. Rarely do test resultsand theoretical predictions agree completely, even moreso in the case of a complicated system. Much of the timeneeded for a test setup is used to ensure that the load-ing conditions are realistic, while attempting to complywith the test purposes and evaluation criteria.

On rare occasions, when large discrepancies be-tween theoretical predictions and test results occur, theexperiment is systematically debugged in the followingorder:

1. Questions will arise about the validity of the cali-bration of the instruments, i. e., the relationshipsbetween known inputs and observed outputs and theuncertainties associated with the two measurementprocesses are quantified and defined;

2. Questions will arise on the signal conditioning anddata acquisition;

3. Questions will arise about the capabilities of the testpersonnel.

It is better for everyone to get involved and discoverthe problem as early as possible in the laboratory orunder controlled conditions.

35.3 Case Studies

After careful selection, three case studies have been in-cluded in this chapter on structural testing. The firststudy is based on modeling aspects of high-rate andhigh-temperature tests on steel retrieved from the col-lapse of the World Trade Center disaster. This studyonly highlights the issues that were not addressed inearlier chapters. This investigation falls under forensicstudy. The next case study deals with the challengesfaced while conducting nondestructive and load tests onmultiple concrete highway bridges. The third case studyillustrates how structural tests can innovate a productdesign.

35.3.1 Models for High-Rateand High-Temperature SteelBehavior in the NIST World TradeCenter Investigation

DescriptionIn September 2002, the NIST became the lead agencyin the investigation of the World Trade Center (WTC)

disaster of September 11, 2001. That investigation ad-dressed many aspects of the catastrophe, from occupantegress to the factors that affected how long the TwinTowers stood after being hit by the airplanes [35.13,14].A major part of the investigation was the metallurgi-cal analysis of structural steel recovered from the twoWTC towers. The analysis included tasks to catalogthe recovered steel, characterize the failure modes ofthe columns, quantify the temperature excursions thatthe recovered steel experienced, and characterize andmodel its mechanical properties. The last task consistedof two major subtasks:

1. Assess the quality of the steel recovered from thecollapse site to ascertain whether it had the yieldstrength called for in the design drawings, and de-termine whether its properties were consistent withthose expected of construction steels from the WTCconstruction era.

2. Provide material models for room-temperaturestress–strain behavior, high-strain-rate stress sen-

PartD

35.3


sitivity, and high-temperature plasticity and creepbehavior to the groups that conducted the finite ele-ment modeling of the collapse sequence [35.15].

This case study summarizes the findings andmethodologies of the second subtask.

Challenges and StrategyChallenges. The team faced five challenges in complet-ing the subtask of the metallurgical analysis:

1. Very little of the structural steel from the fire and im-pact floors remained when NIST assumed control ofthe investigation. Although the salvage teams recov-ered many important perimeter columns from nearthe area of impact in WTC 1, most of the structuralsteel was recycled very soon after the excavation ofthe site.

2. Identifying and cataloging the recovered steels wasextremely time consuming. After the recoveredcolumns arrived at NIST, team members attemptedto identify their locations in the building using theoriginal structural plans and various remaining iden-tifying marks.

3. The WTC towers were constructed from a muchwider variety of steels than is common in high-risebuilding construction. Unlike many tall buildings,which might use only three or four grades of steelsupplied to ASTM specifications, the WTC em-ployed 12 yield strength levels, many of whichwere proprietary grades from foreign steel mills.Four different fabricators, who purchased steel fromboth domestic and foreign mills, worked on the fireand impact floors alone. The wide variety of steelsmade it impossible to completely characterize eachone.

4. Locating suitable areas to harvest test coupons wasfrequently difficult. The collapse and subsequent re-covery efforts damaged many of the columns byplastically deforming them. Truss elements provedextremely vexing in this regard. Not only were theiroriginal locations in the building unidentifiable, butthey were compressed into tight balls for removalfrom the collapse site during the recovery. Thishandling rarely left any material suitable for me-chanical testing.

5. The accelerated time schedule necessary to deliverevaluated properties to the modeling groups fre-quently required the team to evaluate propertiesusing literature values, even when recovered steelexisted in the NIST inventory.

Strategy. To evaluate the quality of the steel and to es-tablish baseline mechanical properties, the group testedat least one tensile specimen from each yield strengthlevel. Additional specimens from relevant steels fromthe fire and impact zones were also characterized. Tocharacterize the high-rate stress sensitivity, the focuswas on perimeter and core column steels from near theimpact zone. These columns were most relevant for de-termining the severity of damage to the buildings causedby the impact of the aircraft. To characterize the high-temperature deformation behavior, the focus was onsteels from core columns and the floor trusses. Creepcharacterization focused on the floor truss steels only. Inparallel to the experimental characterization, the teamdeveloped methods to estimate properties of other steelsnot characterized, using literature data and models aswell as test results on WTC steels.

Test MethodsRoom-Temperature Tensile Tests. Evaluating the qual-ity and establishing the baseline tensile behavior of thesteels from the fire and impact zones required severalhundred room-temperature tensile tests that encom-passed all the relevant strength levels and forms, suchas plates, rolled shapes, and truss components. The testprotocol generally followed ASTM E 8 (and in somecases ASTM A 370) for comparison to mill-test reportdata. All tests retained the extensometer on the speci-men past the point where the specimen began to neck tocapture the full stress–strain behavior. Results of thesetests were used to assess the quality of the steel, as wellas to form the basis of the stress–strain models for thehigh-rate and high-temperature behavior.

High-Strain-Rate Tests. Relevant steels from the im-pact zones were tested at rates up to 500 s−1 to establishthe strain rate sensitivities of their strengths. The testsuite comprised eight perimeter column steels with50 ksi ≤ Fy ≤ 100 ksi (345 MPa ≤ Fy ≤ 689 MPa) andfive core column steels with 36 ksi ≤ Fy ≤ 42 ksi(248 MPa ≤ Fy ≤ 290 MPa). The core column groupincluded examples of both plates from built-up boxcolumns and wide-flange shapes. Specimens from bothshapes were machined from sections flame-cut from theoriginal columns. Specimen gage sections were locatedat the one-half or one-quarter depth positions (as re-quired by E 8 and A 370) and well away from anyflame-cut edge. Tension tests in the range 50 s−1 ≤ ε̇ ≤500 s−1 employed a servohydraulic test machine witha special slack adapter grip that allowed the actuator toreach full speed before loading the specimen [35.16].

PartD

35.3


The gage section of the typical test specimen was32 mm long; the cross section was 6.35 mm × 3.17 mm.This specimen was geometrically similar to the larger,standard tensile specimen used in the quasistatic tests,which allowed pooling of the data for evaluating theeffect of strain rate on ductility. In the high-strain-rate tests, the specimen stress was measured by usinga strain gage bonded to the nondeforming grip end ofthe specimen, while the specimen strain was measuredfrom a high-elongation gage bonded to the gage sec-tion. The 1% offset yield strengths YS01 for all testswere evaluated using the European Structural IntegritySociety (ESIS) procedure, which is described graphi-cally in the inset to Fig. 35.12 [35.17]. Choosing the 1%offset yield strength YS01 instead of the more common0.2% offset YS002 removes most of the spurious effectsfrom load cell ringing, which can be misinterpreted asincreased strength.

High-Temperature Tests. Two types of tensile testscharacterized the high-temperature behavior of repre-sentative specimens of steels from perimeter columns,core columns, floor trusses, and truss seats. A firstgroup of standard, high-temperature tensile tests, whichfollowed ASTM E 21, complemented the room-temperature tensile tests and formed the basis forthe high-temperature plasticity models. In general, thehigh-temperature tensile tests employed round spec-

0.40 0.1

Perimeter column flangeWTC1 column 130 floors 90–93

(1) dε/dt = 417 1/s(2) dε/dt = 260 1/s(3) dε/dt = 100 1/s(4) dε/dt = 65 1/s(5) dε/dt = 6.1×10–5 1/s

0.2 0.3

Engineering stress (MPa) Engineering stress (ksi)

Engineering strain

0 0.01 0.02 0.03

ESIS procedure for YS01

1% offset

YS01

(1)

3rd orderpolynominal fit

for 0.01 <ε<0.05

0.04 0.05 0.06True strain

True stress (ksi)True stress (MPa)

(5) (4) (2) (3)(1)

800

600

400

200

0

800

600

400

200

0

120

100

80

60

40

20

0

120

100

80

60

40

20

0

Fig. 35.12 Example stress–straincurves for high-rate tensile tests. Theinset describes the procedure for es-timating the 1% offset yield strengthYS01

imens with d = 12.7 mm. Strains were measured ona 25.4 mm gage length. Test temperatures were 400 ◦C,500 ◦C, 600 ◦C, and 650 ◦C, which spans the temper-ature region where the strength changes most rapidly.To maximize the number of different steels tested anddefine the temperature dependence of the strength, gen-erally only a single test was made at each temperature.A second group of short-time creep tests characterizedthe time-dependent deformation behavior of steels fromthe floor trusses. These tests employed smaller speci-mens, with a 32 mm-long uniform cross-section that is6.0 mm by 3.17 mm. Characterizing the creep behaviorrequired about 20 tests per steel, at the same tempera-tures used in the high-temperature tensile tests.

Deformation ModelsHigh Strain Rate. High-strain-rate tensile tests, as dis-played in Fig. 35.12, show the strain-rate sensitivity ofthe steels. Although several models exist for describ-ing the change in strength with strain rate, a particularlysimple one is

σ = σ0 K

(ε̇

ε̇0

)m

, (35.1)

where the strength σ can represent the 1% offset yieldstrength YS01, for example. Figure 35.13 plots the cal-culated strain rate sensitivity m of the 1% offset yieldstrength YS01 for the perimeter and core column steels.

PartD

35.3


The plot also includes the strain-rate sensitivitiescalculated from the reported yield strengths for someother construction and low-alloy steels from the WTCconstruction era [35.18–24]. The strain-rate sensitivi-ties of the yield strength of the WTC steels are similar.The accelerated delivery schedule for the strain-rate

200 400

1% offset yield strengthYS01 α (dε/dt)m

600 800

20 40 60 80 100 120

Strain rate sensitivity (m)

Fy (MPa)

Fy (ksi)

0.07

0.06

0.05

0.04

0.03

0.02

0.01

0

Perimeter columnsCore columnsLiterature

Fig. 35.13 Strain-rate sensitivity m of the 1% offset yieldstrength YS01 for WTC steels and structural steels of simi-lar strength and composition

8000 200

Perimeter columnsCore columnsTruss seatsTruss componentsLiterature dataModelEurocode ε = 2% yield strength = fmaxEurocode proportional limit fp

A2 = 0.074m1 = 8.07m2 = 1s1 = 635°Cs2 = 539°C

f = A2+(1–A2)exp(–0.5[(T/s1)m1+(T/s2)m2]

400 600

f = YS002/YS002 (T =20 °C)

T (°C)

1

0.8

0.6

0.4

0.2

0

Fig. 35.14 High-temperature yield strength YS002 for steels recov-ered from the WTC, structural steels from the literature and themodel for the normalized yield strength f

sensitivity data required that the team supply a singlestrain-rate sensitivity for all steels, which was evaluatedusing a subset of the steels in Fig. 35.13. Because thatstrain-rate sensitivity was based primarily on data fromthe higher-strength perimeter column steels, it slightlyoverpredicts the strength increase of the lower-strengthcore column steels. Significantly, none of the perime-ter column steels suffered from brittle failure at the highdeformation rates expected from the aircraft impact. Atrates of up to 500 s−1, the total elongation to failure Eltwas never less than 20% for any of the steels character-ized.

High Temperature. NIST supplied three models for cal-culating the high-temperature behavior of the relevantsteels from the fire zone:

1. normalized yield and tensile strength as a functionof temperature

2. tensile stress–strain (plastic) behavior as a functionof temperature

3. creep strain as a function of temperature, stress, andtime

Figure 35.14 compares the measured 0.2% offsetyield strength YS002 of the recovered steels to a suiteof literature data for structural steel [35.25–31].

The solid line represents the model for the tem-perature dependence of the yield strength, which wasdeveloped from the literature data in Fig. 35.14 beforesignificant testing was complete. The data for the WTCsteels generally lie slightly below the bulk of the liter-ature data for T > 500 ◦C, probably because the WTCtests employed a slower testing rate.

In addition to the generic yield and tensile strengthbehavior model of Fig. 35.14, NIST developed a secondset of models to predict the stress σ as a function ofstrain ε and temperature T for all steels in the fire zone

σ = RTS K (T )εn(T ) . (35.2)

The temperature-dependent terms in this stress–strainmodel account for the decreasing work hardening withincreasing temperature. The form of the functions K (T )and n(T ) is the same as the master curve for high-temperature normalized yield strength in Fig. 35.14.The accelerated delivery schedule required NIST to de-velop models for all relevant steels based on test datafrom only two steels. The behavior of A 36, Fy = 36 ksi(248 MPa) steels was estimated from literature stress–strain curves [35.25]. The behavior for all other steels,including the wide-flange core column in Fig. 35.15, is

PartD

35.3


based on the stress–strain response of nominally ASTMA 242, Fy = 50 ksi (345 MPa) micro-alloyed steel re-covered from the WTC floor trusses.

To calculate the stress–strain behavior of an un-characterized steel, the stress is scaled by the ra-tio, RTS = TSref/TS, of the room-temperature tensilestrength of the reference steel on which the model wasdeveloped to the tensile strength of the uncharacterizedsteel. Scaling by the room-temperature tensile strengthTS produced greater fidelity than scaling by the ra-tio of room-temperature yield strengths, as previousstudies have done [35.32]. Figure 35.15 compares thestress–strain curves at different temperatures for speci-mens from the flange of a wide-flange core column tothe prediction of the global model of (35.2). Note thatthe parameters used to predict the stress–strain curveswere developed using data from the ASTM A 242,Fy = 50 ksi (345 MPa), micro-alloyed truss steel, andnot the plotted wide-flange core column steel. The qual-ity of the prediction confirms the choice to scale theresults by the ratio of the room-temperature tensilestrengths, RTS, in (35.2).

The third set of models NIST supplied includedtwo models to represent the creep behavior of all thedifferent steels exposed to the fires. The first, for thelow-strength A 36 steels, came directly from previousNIST research, and was based on literature data for thecreep of low-strength structural steel [35.32, 33]. Thesecond model, used for all other higher-strength steels,was based on the creep data from the ASTM A 242,Fy = 50 ksi (345 MPa) micro-alloyed steel recoveredfrom the WTC floor trusses. Both models broke the tem-perature T , stress σ , and time t dependence of the creepstrain εc into separate functions

εc = A(T ) (RTSσ)C(T ) t B(T ) . (35.3)

The exact forms of B(T ) and A(T ) differ slightly be-tween the two models.

Like the high-temperature stress–strain models, thedifference in creep response of the various steels wascaptured by scaling the stress by the same tensilestrength ratio RTS. An extensive comparison of the re-sults of this method on a suite of literature data setsfor creep of structural steel established the superiorityof this method over two other possible scaling ratios,no scaling and room-temperature yield-strength scal-ing [35.34, 35]. No scaling of the applied stress (i. e.,setting RTS = 1) is equivalent to assuming that allsteels have identical creep behavior. Scaling the stressby the room-temperature yield strength ratio provedto be particularly unsuitable. Creep curves from this

0.250 0.05 0.1 0.15

403 °C

400 °C

20 °C

k0 = +183 MPak4 = +845 MPak1 = +6.576k2 = +6.597tk1 = +512 °Ctk2 = +512 °C

σ = RTSK(T)ε n(T )

K(T ) = (k4–k0) exp{–1/2[(T/tk1)k1+(T/tk2)k2]}+k0

n(T ) = (n4–n0) exp{–1/2[(T/tn1)n1+(T/tn2)n2]}+n0

n0 = +0.034n4 = +0.151n1 = +10n2 = +10tn1 = +520 °Ctn2 = +500 °C

500 °C

12WF92 Fy = 42 ksi core columnWTC1 column 605 floors 98–101

Experiment: dε/dt = 1×10–3 1/sModel

600 °C600 °C650 °C

0.2

Engineering stress (MPa) Engineering stress (ksi)

Engineering strain

700

600

500

400

300

200

100

0

100

80

60

40

20

0

Fig. 35.15 High-temperature stress–strain curves and the predictionof the global model (35.2) for high-strength steel

0 2000 4000 6000

0 0.5

(1) 297 MPa(2) 278 MPa(3) 252 MPa(4) 200 MPa

a0 = –55.45a1 = +9.476×10–3 °Ca2 = –3.52×10–5 °C2

b0 = +0.398b1 = +3.553×10–11

b2 = +3.698c0 = +3.233c1 = +1.167×10–2 °C–1

RTS = +1

ExperimentGlobal model

X(1) X(2) X(3)

(4)

εc = A(T) (RTSσ)C(T )tB(T )

A(T ) = exp(a0+a1T+a2T 2)B(T ) = b0+b1(T/1°C)b2

C(T ) = c0+c1T

Truss angle ASTM A242500 °C

1 1.5 2Engineering creep strain (ε0) t (h)

t (s)

0.06

0.04

0.02

0

Fig. 35.16 Creep curves and the prediction of the global model(35.3) for creep of floor truss steels that conformed to ASTM A 242

method deviated the most from the literature experi-mental data.

Figure 35.16 compares the experimentally deter-mined creep curves for the Fy = 50 ksi (345 MPa)micro-alloyed floor-truss angle steel to the prediction ofthe global model in (35.3).

PartD

35.3


This shows the excellent fidelity of the predictionof the global strain–time model for that steel. However,because NIST did not characterize the creep responseof any other WTC steels, it is not possible to assess thefidelity of the prediction on the WTC steels.

SummaryNIST conducted more than 500 mechanical propertytests to assess the quality of the steels and developmaterial models in support of the World Trade Cen-ter collapse investigation. These data supported modelsfor room-temperature stress–strain behavior, strain-ratesensitivity, and high-temperature deformation. The casestudy highlights the challenges faced in terms of cat-aloging the recovered steel to characterize the failuremodes, quantifying the temperature excursions that therecovered steel experienced, and characterizing andmodeling its mechanical properties.

35.3.2 Testing of Concrete Highway Bridges– A World Bank Project

IntroductionThe National Highways Development Project (NHDP)launched in 1999 was India’s largest ever high-ways project, covering a length of nearly 24 000 km(15 000 mile). The project cost (estimated to beUS $25 billion) was funded by the World Bank, theAsian Development Bank, and the Indian Govern-ment [35.36]. The objective of this project was todevelop a world-class highway network with uninter-rupted traffic flow. A large part of this network involvesimproving the existing system, which was constructedduring the 1950s. Most of the bridges in these highwayshad signs of distress due to lack of regular mainte-nance, and in most cases, records on their design and

13 m

6 m

Transverse girders

Telescopic working platform

Test truck

Bridge deck slab Railing

Longitudinal girder

Fig. 35.17 Test truck with telescopicworking platform

construction details were nonexistent. The traffic onthese structures had increased dramatically since theiroriginal construction. This situation called for tests toestimate the load-carrying capacity, particularly to de-termine if they had developed material deteriorationand loss of section strength over the years of usage.Due to budget and time constraints, only a few bridgescould be completely tested and assessed. This casestudy describes how the management approach, ex-plained in Sect. 35.2, was implemented to test a largenumber of bridges within the constraints of time andcost.

Challenges InvolvedThe investigating team came across three major chal-lenges in completing this project.

1. Test structures were located on National Highwaysthat connect different states. The traffic volume onthese highways is very high. Even for testing pur-poses, highway closure for more than 15 min wasnot permitted. Thus a testing method that allowedintermittent loading, allowing normal flow of trafficbetween two successive tests, was adopted. Inter-mittent loading is possible only by using articulatedtrucks. A market survey demonstrated that no suchtest truck with a telescopic working platform existedin the country (Fig. 35.17). Importing such a truckwas not feasible from the perspective of cost andtime.

2. Most bridges had visible distress marks and hadno design/construction documentation available.As a result, a number of additional tests (mostlynondestructive and partial destructive) had to beperformed to determine the characteristics of theconcrete. It is a well-established fact that no single

PartD

35.3


test would be sufficient to assess the condition ofany of these structures.

3. The greatest challenge was developing a theoreti-cal model to predict the load–deflection behaviorof a distressed structure. This model was neededduring load testing of the highways.

Planning and Control PhaseIn order to meet the objectives of the test program,it was necessary to test these structures in threephases.

Phase 1 – Visual Assessment. Visual assessment wascarried out in order to detect all symptoms of dam-age and defects as per specifications given in a specialpublication (SP) of the Indian Road Congress (IRC:SP-35) [35.36]. A carefully selected assessment team,represented by experts from each of the contributingagencies (i. e., funding agencies, users, consulting firms,and the state highway department) were involved dur-ing this phase. On the basis of visual assessment, allof the bridge structures were categorized into threegroups

• Group 1 – bridges with little or no distress at all,which were not recommended for testing• Group 2 – bridges with minor distress, which wererecommended for testing to determine the severityof the distress• Group 3 – bridges having considerable distress, forwhich tests were not needed as their replacementcan be clearly justified

Bridges that fell into group 2 were identified andgrouped, and only representatives from each group wererecommended for testing. The objective of this test pro-gram was to

• evaluate safe load-carrying capacity and• establish a common procedure for posting a struc-turally deficient structure.

Phase 2 – Generation of a Scientific Database. Sincethere were no design/construction related data avail-able, this phase was adopted to generate followinginformation:

• physical/geometrical information of various ele-ments;• a map of crack (structural or nonstructural) patternsexisting in the structure;

• characteristics of the concrete used in variouselements of the structures – this was done by con-ducting nondestructive and partial-destructive tests;• A theoretical model to predict the response of thestructure.

Phase 3 – Load Testing. Load testing of the actualbridge structure was performed to ascertain the liveload-carrying capacity. The procedure for rating the liveload capacity requires knowledge of both the actualphysical condition of the bridge and the actually appliedloads. This phase involves the following tasks:

• Task 1 – Based on the traffic survey reportand following the specification given in IRC:SP-37 [35.37], the test load was determined. Load testwas done at four different stages of loading as givenin the procedure. The chosen loading arrangementshould be such that these values can be achievedquickly. Sand bags, which can be added or re-moved easily, were used to achieve the requiredloading.• Task 2 – Fig. 35.18 shows the distribution of sandbags to achieve the code-specified axle-wise loaddistribution.• Task 3 – Before conducting the actual load test,a theoretical model was developed to predict the de-

Sand bags to load

Fig. 35.18 Sand bags used to achieve code-specified loadwith the proper distribution

PartD

35.3


flection of the structure at various stages of loading,assuming– dead load of the transverse girders as concen-

trated loads– twisting of the transverse girders due to

eccentrically-applied live loads (in the form ofthe influence factor of the girder)

– railing dead load, as a uniformly distributed load(UDL)

– Deck slab and girders constructed monolith-ically (i. e., longitudinal girders worked likeT-beams)

– Prestressed concrete longitudinal girders, orig-inally designed as balanced sections with thesoffit stress due to self-load as −5 kg/cm2, andthe profile of the tendon was assumed to beparabolic.• Task 4 – Actual load test.

After the field tests, a rating analysis was performed andthe safety of the bridge structures was determined andreported, after proper documentation.

PreparationPreparations for the testing program included the fol-lowing tasks:

• Task 1 – On the basis of the traffic survey, the heav-iest vehicle plying on the bridges was determinedto be 25 tonnes. The code specifies to use the nextheavier vehicle, i. e., an articulated truck trailer of35.2 tonnes. Figure 35.19 shows the geometry of thetest vehicle, along with the axle-wise load distribu-tion to meet specifications of the code. The spatialdistribution of the sand bags to achieve differentstages of loading was ascertained through propercalibration.

5.6 t 10.6 t 9.5 t 9.5 t

16 %

1.18 m

30 %

3.2m

27% 27%

4.05m 1.36m 1.35m

Fig. 35.19 Geometric of the test vehicle with axle-wiseload distribution

• Task 2 – The working platform was designed, fab-ricated, and tested in the laboratory before beingtransported into the field. The platform was madeusing rectangular steel tubes in a telescopic ar-rangement to be easily transported in pieces andassembled at site quickly. Figure 35.20 shows thesimple but elegant arrangement of the suspensionsystem for the working platform. Using a propercounterweight, the requirement for a cable at the sitewas avoided.• Task 3 – The instruments intended to be used fornondestructive, partially destructive, and load testswere properly calibrated in the laboratory.

Execution and DocumentationThe test program constitutes three different kinds oftests:

• Test 1 – Nondestructive tests• Test 2 – Partially destructive tests• Test 3 – Load tests

Brief descriptions of each of these tests are given below.

Test 1 – Nondestructive Test. Four subtests were per-formed under nondestructive tests

1. subtest 1 – to determine compressive strength of theconcrete by impact hammer

2. subtest 2 – to determine the quality of concrete usingultrasonics

3. subtest 3 – to determine the reinforcement cover bycover meter

4. subtest 4 – to measure the corrosion of reinforcingsteel using a half-cell meter

Subtest 1 – Impact Hammer Test (Also Known asSchmidt Hammer Test or Rebound Hammer). Thistest indicates the hardness of the concrete surface andindirectly reflects the strength of the concrete. ASTMC 805 provides the necessary specification for this test.Characteristics of this test method are

• a large number of readings can be taken in shortduration,• a large number of locations on the structure can beselected,• a statistical approach with a high degree of confi-dence can be applied,• the actual structure is tested, therefore the test resultreflects the totality of the final product,

PartD

35.3


Rectangular hollow tube in tube (telescopic)

String pull

Pin connection

Counterweight

Reactionfrom siderail

Wheel

Deck reaction

StringFig. 35.20 A single arrangement onhow the working platform was an-alyzed, designed, fabricated in thelaboratory before being transported tothe site for field test

• the strength of concrete in a structure can be de-termined with an accuracy of ±15%. When littleinformation is available about the concrete quality,the error can be up to ±25%.

ASTM C 805 specifies that ten rebound numbers berecorded for each location. The minimum number oflocations where tests should be performed is ten per ele-ment. Thus for any element, there are 100 test readings.These are analyzed statistically as: fck (most proba-ble characteristics strength, MPCS, of concrete) = fm −1.64 S, where fm is the mean value of the strength dataand S is the standard deviation (SD) of the data. Fig-ure 35.21 shows the frequency distribution of concretein the deck slab of the bridge with identification number38/1.

Subtest 2 – Ultrasonic Test. In an ultrasonic test,compressive strength estimates are derived from time-domain signals. The density of a material affects bothpulse velocity and strength. Thus there is a unique cor-relation between the longitudinal (compressive) wavevelocity and material strength. ASTM C 597 outlinesthis test procedure.

The operational principle of modern testing equip-ment includes the measurement of longitudinal, trans-verse, and surface waves. The receiving transducerdetects the onset of the longitudinal waves and deter-mines which is the fastest. From the velocity of thispulse, estimation can be made on strength. Table 35.1gives the classification of concrete on the basis of pulsevelocity.

The zero setting of the equipment is done duringthe calibration of the device. Characteristics of this testmethod are

• surface condition and moisture content can influ-ence pulse velocity by up to 20%• existence of cracks in the material has a strong in-fluence• the presence of steel reinforcement can increasepulse velocity by about 40%• age, aggregate type, and size of the concrete all in-fluence pulse velocity

Subtest 3 – Cover Test. Major causes for the deteri-oration of reinforced cement concrete (RCC) structureshave been the corrosion of the embedded reinforcement.The main cause for the occurrence of corrosion is iden-tified as the absence of proper cover, which providesthe protective shield. When reinforcement details are

0 10

Samples = 10Data size = 100Max. value = 40MPaMin. value = 26 MPaMean = 33.96 MPaSd = 3.1 MPaMpcs = 28.85 MPa

20 30 40 50

Frequency

Strength (MPa)

40

35

30

25

20

15

10

5

0

Fig. 35.21 Frequency distribution of concrete strength

PartD

35.3


Table 35.1 Classification of the quality of concrete on thebasis of pulse velocity

Longitudinal pulse velocity Quality of concretekm/s 103 ft/s

> 4.5 > 15 Excellent

3.5 – 4.5 12 – 15 Good

3.0 – 3.5 10 – 12 Fair

2.0 – 3.0 7 – 10 Poor

< 2.0 < 7 Very poor

Table 35.2 Assessment of rate of corrosion

Measured electrode potential Corrosion rate

< 250 mV Insignificant

> 250 mV and < 350 mV Corrosion initiated

350 mV Corrosion is certain

> 350 mV High to severe

not known, the position of the reinforcement close tothe surface can be determined by a cover meter. A covermeter has a probe and an indicator on the front panel.The position and direction of the reinforcement embed-ded in the concrete were determined by sweeping theprobe symmetrically over the surface. The needle of theindicating instrument will deflect if the probe nears a re-inforcement bar. This is measured by placing the probein different positions until a maximum needle deflectionis obtained.

Subtest 4 – Corrosion Test. When reinforcing steel isembedded in concrete, it does not normally corrode.The inherent cement-alkaline environment acts as a pro-tective passive layer on its surface. However, if thedepth of cover of the concrete is insufficient, then thepassive layer can break in the presence of excessiveamount of chloride ions. The chloride can originatefrom sodium chloride (common salt) in marine loca-tions or from de-icing applications, or from the use ofa particular admixture, e.g., calcium chloride (acceler-ator), or from surrounding soil, from the contaminatedunwanted aggregates themselves or even from the mixwater or curing water. The breakdown of the passivelayer will force the steel to rust and expand in volume.As a result, the concrete cracks.

In order to assess the rate of corrosion of embeddedsteel in concrete qualitatively, half-cell potential meas-urement, an electrochemical nondestructive technique,was adopted for the detection of the state of re-barswithin the concrete structure. The procedure for this testis to divide the surface under test into a grid system ofsuitable dimensions as per site conditions. Intersections

of these grid lines are marked numerically as shown inTable 35.3. Assessment of the rate of corrosion on thebasis of half-cell potential is given in Table 35.2. ASTMC 876 has the necessary specifications for this test.

Test 2 – Partially Destructive Test. Of all the testsavailable for the determination of compressive strengthof concrete, a core test may be the most direct test ofin situ concrete. In this test method, cylindrical coresare cut using a rotary diamond drill (ASTM C 42) fromthe candidate structure and then polished and tested un-der uniaxial compression. While identifying a locationfor core cutting, regions where embedded reinforcementsteel is present should be carefully avoided.

Cores can be used to determine not only the com-pressive strength of concrete but also some of the othercharacteristics of the structure, i. e., carbonation depth,cement content, chloride content, and sulfate content inthe mortar (ASTM C 1084). The presence of excessivechloride and sulfate in the concrete can cause corro-sion of the embedded reinforcement. The process ofdrilling, including the vibration and impact, can weakenthe interface between the cement paste and the aggre-gates. Core testing has an accuracy of approximately±12% if a single core is tested. If multiple cores aretested at the same location, the mean core strength willhave an accuracy of ±√

(23/n)%, where n is the num-ber of cores [35.38]. Table 35.3 shows typical core testobservations.

Test 3 – Load Test: Deflection Prediction. From theavailable geometrical and material data, a theoreticalmodel is developed. Theoretical deflection is due toboth the self-weight of the structure and the superim-posed load due to the truck trailer loaded with sandbags.

Equation (35.4) gives the deflection due to a uni-formly distributed dead load of intensity w.

y =(

1

EI

) (−wLx3 −wx4 −0.0416wL3x).

(35.4)

As per IRC SP 37 [35.39], the position of the truck-trailer should be such that it causes the absolutemaximum bending moment in the girder. The bend-ing moment under the load will be at the absolutemaximum when a load and the resultant of the set ofloads are equidistant from the center of span for thevarious possible combinations of loads. When morethan one truck-trailer are used, as per the code spec-

PartD

35.3


ification, the clear distance between vehicles CD is18.5 m. Figure 35.22 shows the longitudinal placementof two truck-trailers, one completely inside the span,i. e., DEFG and the other partially, i. e., ABC on thespan.

0 10 20 30

10.6 t

5.6 t

9.5t9.5t

A B C D E F G

Deflection due tosuperimposed load

Deflection due toself-weight of thebridge

40 50 60

Truck 2 Truck 1

Span (m)

Deflections (mm)

0

5

10

15

20

25

30

Fig. 35.22 The longitudinal placement of two truck-trailers as per code specification

For this position of live loading, deflection can befound by the method of superposition given by equa-tions (35.5,35.6) as

y =(

1

6EI

)[−P(L −a)x3 + Pa(L −a)(2L −a)x

],

for x < a , (35.5)

y =(−Pa

6EI

)[(

(L − x)3

L

)+

((L2 −a2)x

L

)− (

L2 −a2)] ,

for x > a . (35.6)

xa

y

L

P

Fig. 35.23 Deflection of a simply supported beam due toa concentrated load P at mid span

Total deflection can be obtained after combining the liveload and dead load deflections.

Procedure for Load Testing. As per the requirement ofthe code:

• observations are made to find the existence of anycracks. If found, they should be marked and theirwidth should be measured;• for ease of observation of the behavior of cracksand their new formations during the test, a limewhitewash was applied the day before testing at thecritical section;• the load test was done during the morning hours ofthe day when the variation in temperature was low.• the test load was applied in stages of 0.5W , 0.75W ,0.9W , and W , where W was the gross-laden weightof the test vehicle (i. e., 35.2 tonnes);• for each stage, the corresponding loaded vehicle wasbrought to the marked position and the observationsof deflections was made instantaneously, and againafter 5 min;• after the placement of the load, the development ofnew crack and the widening of the existing crackswere observed;• prior to the start of testing, the theoretical deflec-tions at various stages of loading were calculatedand plotted. If the in situ deflections exceeded thesevalues by more than 10%, the test was discontinued;• for testing with multiple test vehicles, the individualvehicles were gradually brought into position andthe resulting deflections were continuously moni-tored;• the test vehicle was taken off the bridge, and in-stantaneous deflection recovery and the deflectionrecovery after 5 min were noted.

Figure 35.22 shows that the deflection caused by self-weight (uniformly distributed over the span) was aroundten times that for the superimposed load. Deflection wasmeasured at mid span using a dial gage with a resolutionof 0.01 mm.

Rating Analysis. After the load test, a load versusdeflection graph is drawn. The load corresponding toa deflection of 1/1500 of the span is determined af-ter extrapolating the load–deflection curve to obtainthe acceptance load. Most of the bridges demonstratednearly 100% recovery of deflection with unloading ofthe test vehicle. Test observations are summarized inTable 35.3.

PartD

35.3


Table 35.3 Summary of tests conducted on the bridge identified as no. 38/1 over the River Subarnarekha on NationalHighway 60

Test Location of test Observations Inference

Test 1, Deck slab MPCS ;28.85 MPa;

SD = 3.1 MPa Good

Subtest 1 Pier MPCS ;19.37 MPa;

SD = 9.12 MPa Average

Hammer test Longitudinal girder MPCS ;35.81 MPa;

SD = 5.80 MPa Good

(Strength in Transverse diaphragm MPCS ;30.85 MPa;

SD = 4.14 MPa Good

MPa) Pier cap MPCS ;24.88 MPa;

SD = 8.01 MPa Average

Test 1, Deck slab GoodSubtest 2 Web of girder GoodUltrasonic Flange of girder Goodpulse test Diaphragm FairTest 1, Soffit of central girder from

upstream (U/s) face30 Adequate as per code

Subtest 3 Web of central girder fromU/s

40 Adequate as per code

Cover (mm) Central diaphragm betweencentral and U/s girder


Soffit of deck slab betweencentral and U/s girder


Pier no. 2 from west side 55 Adequate as per codePier cap no. 2 west face 60 Adequate as per code

Test 1, Abutment (2.8 m from top) 1 – 580 mV Rate of corrosion is very highSubtest 4 Grid area = 0.09 square me-

ter around the exposed bar asshown on the rightmost cell inFig. 35.25.

2 – 630 mV

Corrosion Distance from 1 to 8 = 600 mm 3 – 640 mVintensity Distance from 1 to 2 = 150 mm 4 – 690 mVaround an 5 – 688 mVexposed bar 6 – 750 mV

7 – 594 mV8 – 650 mV9 – 620 mV

Test 2 Structuralelement

Strength(MPa)

Carbonation depth(mm)

Cement (%) Chloride (%) Sulfate (%)

Partially Top slab 30.1 Nil 15.41 0.033 0.045destructive Abutment

cap28 Nil 10.83 0.009 0.033

test Kerb 26 Nil 13.33 0.013 0.058(core test) Girder Weak 2 17.91 0.04 0.14

Upper abut-ment

Weak 2 11.25 0.1 0.033

Lowerabutment

24 Nil 10.83 0.015 0.055

Pier 16 1 10.83 0.007 0.065Test 3 Recovery of deflection Full recovery Loading is within the elastic

limitLoad test Rating of the bridge on the

basis of load testThe bridge is safe for a class-Atrain of vehicle, 70R tracked, andwheeled vehicles

PartD

35.3


Bridge characteristics:

• prestressed concrete girder bridge with 13 spans (1×21.3 + 11 × 48.8 + 1 × 12.5 = 570.6 m long) witha cross section as shown in Fig. 35.24. N is thenumber of transverse girder.

Largest span = 48.8 m; N=7

Fig. 35.24 Crosssection of thebridge identifiedas no. 38/1

• visual distress includes spalling, reinforcement barexposure, and cracks in the deck slab and on sup-porting structures

Summary of the test observations is given in Table 35.3.

35.3.3 A New Designfor a Lightweight Automotive Airbag

DescriptionIn 1991, Sandia National Laboratories and a fabricmanufacturing company initiated a project to developa lightweight fabric automotive airbag. The objectiveof the project was to address performance issues inher-ent with incumbent, heavyweight automotive airbags.For example, heavyweight fabric airbags, constrainedby imposed pack volume, must be small, necessi-tating high pressures in the airbag to decelerate theoccupant during an accident. Conversely, lightweightfabric airbags could be considerably larger, resultingin lower pressures exerted on the occupant duringan accident. Also, the lightweight fabric is soft andsmooth, which eliminates the facial abrasion experi-enced with heavyweight fabric airbags. Finally, themomentum of airbag deployment imparts a large im-pulse to the occupant’s face, a phenomenon termedface-slap, which is mitigated by the smaller mass oflightweight fabrics.

1

2

1

2

3

4

5 6

7

8

9

Fig. 35.25 Gridpoints aroundthe exposed baras marked nu-merically from 1to 9

It was anticipated that a lightweight fabric airbagwould require a new design to pass the rigorous testingof an automotive industry component. The incumbentdriver-side airbag, a two-piece circular design, is notvery structurally efficient. To check this notion, an in-cumbent two-piece, circular airbag was fabricated usingthe lightweight fabric and subjected to a burst-pressuretest. The airbag industry employs burst-pressure test-ing to assist in the design qualification of prototypeairbags. During this test, the airbag is inflated withambient air over several seconds until the airbag failsstructurally. While each automotive manufacturer re-quires different qualification standards, the result ofthe lightweight-fabric, two-piece circular design burst-pressure test conclusively demonstrated that lightweightfabrics would require a new design to produce a viableairbag.

Challenges InvolvedEarly lightweight fabric prototype designs convergedon gore-shaped structures similar to solid-canopyparachutes and hot-air balloons, as depicted inFig. 35.26. An initial design review concluded thatthe proposed prototypes might satisfy the engineeringspecifications, but certainly would not represent a cost-effective, marketable product.

Using cost as the exclusive constraint, a subse-quent design session produced a radically different yeteconomically viable airbag. The simple design useda single square piece of fabric and the enclosure wascompleted by folding the corners toward the centerof the square, as shown in Fig. 35.27. Seams wereconstructed along the diagonals where the free edgesformed, and a hole was cut in the center to accept theattachment hardware. This pattern ensures with its verysimple design that the airbag produced the strongestseams and no parent fabric was wasted.

Planning and Control PhaseTo prove this design, a significant quasistatic pressur-ization and dynamic-deployment cold-air inflation testseries was initiated. The experimental test lab con-sisted of a high-pressure source and serially plumbedsolenoid valves that could supply a dynamic burstof air simulating an inflator deployment. High-speedvideo cameras (1000 fps) were the primary diagnostictools that assisted the post-test evaluation of the earlyprototypes. Simple pressure-transducer measurementsmonitored the progress of the evolving prototypes. Thisenabled the airbag deformation pressure–time history tobe quantified during these inflations.

PartD

35.3


Quasistatic tensile tests of the fabric were performedto determine the ultimate capacity of the fabrics, alongwith the stiffness, or modulus of elasticity. To cor-rectly determine the modulus, a uniaxial straining ofthe material is required. Wrap grips were used in thesetests to wrap the fabric over a cylindrical fixture toavoid the hourglass deformation patterns common withother test fixtures. The crosshead of the test machinein these quasistatic tests moves at a maximum rate of10 in/min (0.0042 m/s) while the load and deforma-tion are recorded. The modulus was measured alongthe original fabric roll direction (the warp direction),across the fabric roll direction (the fill direction), andin-between or at 45◦ to each of these directions (thebias direction). This modulus data is used as the basisof the structural analysis model definitions. Prototype

Conceptual gore-shaped airbag designSandiaNationalLabaratories

Fig. 35.26 Early prototype design appears to be goreshaped

Structurally efficient airbags

Attachment hardware

SeamsFold lines

AirbagInnovationsResearch

Fig. 35.27 A simple design using square fabric to reducewastage

seams were also evaluated using the tensile tests, withconstruction variables such as seam type, thread size,needle size, stitches per inch, type of stitch, and rein-forcement materials.

SimulationA structural analysis investigation of the new designwas started to understand how load is distributed inthe airbag. The chosen approach called for finite ele-ment analyses using the nonlinear code Abaqus [35.40].This code allows for gross deformation of the airbagduring inflation using membrane elements that incorpo-rate the orthotropic behavior of the fabric. A simulationof the deployments from the cold-air inflation testingwas chosen to allow direct comparisons with a knownloading.

High-speed video coverage of the airbag deploy-ments was compared with the deformation from theanalyses, showing that the curvatures of the airbagswere dissimilar. These analyses used the loading meas-ured as pressure–time histories and the quasistaticmaterial properties described above. The testers con-jectured that the inflation dynamics (which occurs inthe time frame of 10–20 ms) were altering the effectivestiffness of the fabric. Dynamic tests were then justifiedto determine the strain-rate sensitivity of the fabric.

Postprocessing of the analysis strain showed a strainrate of 5–10 /s in the fabric during the pressurization(analysis is a very good tool for determining the strainrates of loading, and is generally not intuitive). Usingthis information, tensile tests were designed utilizinga high-speed MTS load frame. This test machine is ca-pable of producing a velocity of 200 in/s (5.1 m/s) inthe crosshead. A photograph of this test setup is shownin Fig. 35.28, where a seam test sample is shown ina temperature-conditioning fixture. The gage length ofthe sample was adjusted until the desired strain rate wasachieved, and verified using high-speed video coverageof the test. This was accomplished by running the testmachine at its maximum speed and adjusting the testsample length until the desired strain rate was measuredusing the video coverage. The correct specimen lengthwas iteratively determined by repeating the test untilthe strain rate of interest was finally achieved. The de-sired strain rates were bracketed to ensure consistencyof the measurements and the loading technique for thetest specimen.

Testing at the loading rate of the airbag deploy-ment showed an apparent stiffening of the fabric matrix,and produced as much as a 30% increase in the elas-tic modulus of the fabric. The strain-to-failure was also

PartD

35.3


frequently reduced depending on the amount of crimpin the fabric weave (essentially an expression of howtightly the fabric is woven).

There was also a large modulus discrepancy be-tween the fabric construction directions, warp and fill.Warp fibers are the yarns in the direction of the fabricroll, and fill fibers are passed in between and normal tothe warp yarns, to make up the fabric matrix. Typicaltensile test results of nonoptimized fabrics are shownin Fig. 35.29, where the load–displacement response ofthe fabric is shown for both the warp and fill directionsduring high-speed tests.

The complete loading history of the same fabric inthe warp and fill direction is shown in the figure (ma-chine startup dynamics occur early in the load trace, asnoted) with the actual tensile loading being representedby the second half of the load trace. Note that, not onlyis the slope or modulus of the material not the same,but the strain-to-failure of the material is different byalmost a factor of three. The strength of the fabric issimilar in the warp and fill directions (the normal wayfabric is designed or ordered). Most importantly, theelastic properties of this material cannot produce a bal-anced state of strain in the resulting airbag deploymentwith this mismatch in moduli (the two slopes of the loadtraces approximated by the lines in the same figure). Fu-ture fabric construction for this airbag was designed tobalance out the moduli in the warp and fill directions.

Validation of the ModelWhen this new understanding of the fabric propertieswas incorporated into the analyses, the curvature of

Fig. 35.28 High-speed MTS load frame

0 0.01 0.02 0.03 0.04 0.05

WarpFill

Machine startup dynamics

Fill modulus/warp modulus ~ 2

0.06

Load (lbs for 2 in. sample)

Displacement of crosshead (in.)

250

200

150

100

50

0

Fig. 35.29 Load–displacement history of fabric in the wrap and filldirections (1 lbs = 4.4 N; 1 in = 2.54 cm)

the airbag’s high-speed deployments improved greatly.A still image during deployment is shown in Fig. 35.30.In this figure, the photo of the airbag is overlaid withthe deformed mesh lines (shown in black) of the finiteelement membrane model. The curvature of the frontand rear panels of the airbag now match very well.This proves graphically that the dynamic modulus infor-

Fig. 35.30 Photograph and analysis overlay of inflatedairbag

PartD

35.3


Fig. 35.31 Warp stress distribution of the mounting area ofthe airbag

mation, replacing the earlier quasistatic tests, producesthe correct state of stress and deformation in the fabricstructure.

The airbag seams have warp- and fill-directed fab-ric on opposite sides of the seam due to the geometryand simplicity of the design. Any mismatch in the prop-erties of the warp and fill will cause shear stress todevelop at the seam interface. Conversely, shear stresscontrol down the seam line is made easier by the bal-ancing of modulus of the dissimilar directions. Anyimbalance in the moduli across the seam must be bal-anced by producing shear across the seam, which leadsto tearing or premature failure. The analyses showedthese effects of airbag construction. The warp stress

contours around the mounting area of the airbag areshown in Fig. 35.31.

Developing seams to move with the deformingpanels of fabric was the next step in project optimiza-tion. It is clear that different levels of bunching occurduring the loading process due to the different amountsof fabric adjacent to the stitching. This relates directlyto the shear stress formation down the line of the seams.This seam behavior was also optimized using the high-speed test frame so that the radii of the stitched materialincreased the ability of the material to handle the dy-namics of the inflation. This seam optimization is an ex-ample of an analysis that has identified an issue and itsroot cause, which was then addressed through testing.

Once the state of stress in the airbag is understood,it is possible to determined how this stress can bebalanced across the seams. Moduli balancing was ac-complished by literally designing the fabric weave to actin a balanced fashion. Yarns in the fabric weave were re-moved to achieve a stronger fabric. This was a radicalconcept to achieve a dynamically balanced fabric re-sponse. This concept led to altering the yarn propertiesin conjunction with the weave design to dynamicallyand statically optimize the fabric action [35.41].

A case study that details the project from incep-tion through innovation and engineering developmenthas been presented. It was noted that the implemen-tation of different constraints led to radically differentdesign proposals during the innovation phase. The studyalso demonstrated the engineering progression of thedesign and the role that testing and analysis played inthe development. New test and analysis methods weredeveloped during the engineering phase as the designwas eventually completed and qualified.

35.4 Future Trends

In nature, shape and structure spring from the strug-gle for better performance, thus they are adaptive andmultifunctional. Exciting progress is being made in de-veloping engineered systems with ideas taken from thebiological world. These systems are complex and callfor an interdisciplinary effort encompassing a breadthof science and engineering disciplines, including bio-logical sciences. Living organisms produce a strikingarray of structural concepts with a wide variety of bio-logical functions. For example, consider the case ofa daffodil stem, a biological beam with a high bend-to-twist ratio of ∼ 13, while that of an isotropic circular

beam would be 1.0–1.5. This natural beam will twistand bend in response to wind loads to reduce drag byup to 30%. Thus, structural testing of a biological beamshould incorporate loading under bending and twist-ing simultaneously. The particular combination of theseloading cases will be function of the characteristics ofbiological structures or materials.

Generating engineering concepts with ideas fromthe biological world will be the emphasis of future re-search efforts. The growth of the field (biomechanics,biomimetics, biomaterials, bioengineering, biologicalsoft tissue, etc.) can be seen by its maiden presence in

PartD

35.4

Structural Testing Applications References 1013

Chaps. 7, 31, and 32 of this Handbook. More than halfa dozen journals specific to biological materials havebeen started in the last decade. The number of pub-lications in this area during last 5 years exceeds thosepublished during the previous 50 years.

Advances in medical imaging (e.g., 3-D Dopplerecho and magnetic resonance imaging), medical im-age reconstruction techniques, and computer simulationmethods are playing important roles in our health caresystem, both for the design of new medical devicesand for the rehabilitation simulators that provide real-time feedback for health-care planning and training. Forproper validation of these structural systems, (medical

devices and simulators), state-of-the-art measurement,control, and analysis techniques will be needed.

Nanoscale technology will influence the futurein many ways. Multifunction nanotubes are wondermaterial with unusual combination of characteris-tics: strength (more than 460 MPa), creep resistance,electrical conductivity, damping, and many other de-sirable characteristics at high temperature (450 ◦C)through to cryogenic environments. Naturally, a struc-ture developed using this wonder material should betested in coupled modes. Chapters 16 and 30 ofthis Handbook contain more information on carbonnanotubes.

References

35.1 R. Sack: Model Based Simulation (National ScienceFoundation, Arlington 1999), White paper, See also:[www.eng.nsf.gov/cms/] for update

35.2 W.H. Price: Factors influencing concrete strength,Proc. J. Am. Concr. Inst. 47(6), 417–432 (1951)

35.3 H. Straub: A History of Civil Engineering – AnOutline from Ancient to Modern Times (MIT Press,Cambridge 1964)

35.4 R. Reese, W.A. Kawahara: Handbook of StructuralTesting (Society of Experimental Mechanics, Bethel1993)

35.5 H. Erwin: OSHA 1910 General Industry StandardsMade Easy, 2nd edn. (Erwin Training Institute, Ed-monton 1989)

35.6 National Environmental Policy Act under US DOE athttp://www.eh.doe.gov/nepa/

35.7 http://www.epa.gov/region5/defs/html/rcra.htm35.8 American Society for Testing and Materials:

http://www.astm.org/35.9 The American Railway Engineering and Mainte-

nance Association: http://www.arema.org/ (2008)35.10 http://www.aashto.org/ and http://www.transpor-

tation.org/35.11 American Institute of Steel Construction http://

www.aisc.org/35.12 International Code Council: International Building

Code (International Code Council, Country Club Hills2003)

35.13 National Institute of Standards and Technology:Federal Building and Fire Safety Investigation ofthe World Trade Center Disaster: Final Report of theNational Construction Safety Team on the Collapsesof the World Trade Center Towers (NIST, Gaithers-burg 2005), available at http://wtc.nist.gov

35.14 F.W. Gayle, R.J. Fields, W.E. Luecke, S.W. Banovic,T. Foecke, C.N. McCowan, T.A. Siewert, J.D. Mc-Colskey: Federal Building and Fire Safety In-vestigation of the World Trade Center Disaster:

Mechanical and Metallurgical Analysis of Struc-tural Steel (NIST, Gaithersburg 2005), available athttp://wtc.nist.gov

35.15 W.E. Luecke, J.D. McColskey, C.N. McCowan,S.W. Banovic, R.J. Fields, T. Foecke, T.A. Siew-ert, F.W. Gayle: Federal Building and FireSafety Investigation of the World Trade Cen-ter Disaster: Mechanical Properties of StructuralSteels (NIST, Gaithersburg 2005), available athttp://wtc.nist.gov

35.16 D.M. Bruce, D.K. Matlock, J.G. Speer, A.K. De:Assessment of the strain-rate dependent tensileproperties of automotive sheet steels, SAE Technicalpaper 2004-01-0507 (SAE, 2004)

35.17 European Structural Integrity Society: Procedure forDynamic Tensile Tests. In: ESIS Designation P7-00,ed. by K.H. Schwalbe (ESIS, Delft 2000)

35.18 D. Chatfield, R. Rote: Strain Rate Effects on Proper-ties of High Strength, Low Alloy Steels, PublicationNo. 740177 (Society of Automotive Engineers, War-rendale 1974)

35.19 H. Couque, R.J. Asaro, J. Duffy, S.H. Lee: Cor-relations of microstructure with dynamic andquasi-static fracture in a plain carbon steel, Met.Trans. A 19(A), 2179–2206 (1988)

35.20 R. Davies, C. Magee: The effect of strain-rate uponthe tensile deformation of materials, J. Eng. Mater.Tech. 97(2), 151–155 (1975)

35.21 A. Gilat, X. Wu: Plastic deformation of 1020 steelover a wide range of strain rates and temperatures,Int. J. Plast. 13(6-7), 611–632 (1997)

35.22 J.M. Krafft, A.M. Sullivan: On effects of carbon andmanganese content and of grain size on dynamicstrength properties of mild steel, Trans. ASM 55,101–118 (1962)

35.23 M. Langseth, U.S. Lindholm, P.K. Larsen, B. Lian:Strain rate sensitivity of mild steel grade ST-52-3N,J. Eng. Mech. 117(4), 719–731 (1991)

PartD

35


35.24 M.J. Manjoine: Influence of rate of strain and tem-perature on yield stresses of mild steel, Trans. ASM66(A), 211–218 (1944)

35.25 T.Z. Harmathy, W.W. Stanzak: Elevated-tempera-ture tensile and creep properties of some structuraland prestressing steels. In: Fire Test Performance,ASTM STP 464, ed. by ASTM (American Society forTesting and Materials, Philadelphia 1970), pp. 186–208

35.26 R. Chijiiwa, Y. Yoshida: Development and Practi-cal Application of Fire-Resistant Steel for Buildings,Nippon Steel Techn. Rep. 58, 47–55 (1993)

35.27 S. Goda, T. Ito, H. Gondo, I. Kimura, J. Okamoto:Present status of weldable high strength steel,Yawata Techn. Rep. 248, 5086–5163 (1964), inJapanese

35.28 J.M. Holt: Short-time elevated-temperature ten-sile properties of USS “T-1” and USS “T-1” Type Aconstructional alloy steels. In: United States SteelReport (Applied Research Laboratory, Monroeville1963)

35.29 J.M. Holt: Short-time elevated-temperature ten-sile properties of USS Cor-Ten and USS Tri-Tenhigh-strength low-alloy steels, USS Man-Ten(A 440) high-strength steel, and ASTM A 36 steel,US Steel Corp. Rep. 57, 19–901 (1964)

35.30 G.F. Melloy, J.D. Dennison: Short-time elevated-temperature properties of A7, A440, A441, V50, andV65 grades. Internal memorandum to J.W. Frame(Bethlehem Steel, Bethlehem 1963)

35.31 United States Steel Corp.: Steels for Elevated Tem-perature Service (United States Steel Corporation,Pittsburgh 1972)

35.32 B.A. Fields, R.J. Fields: Elevated Temperature De-formation of Structural Steel. NISTIR 88-3899(National Institute of Standards and Technology,Gaithersburg 1989)

35.33 D. Knight, D.H. Skinner, M.G. Lay: Short Term CreepData on Four Structural Steels. Report MRL 6/3Melbourne Research Laboratories (The Broken HillProprietary Company, Clayton 1971)

35.34 G. Williams-Leir: Creep of structural steel in fire:analytical expressions, Fire Mater. 7(2), 73–78(1983)

35.35 M. Fujimoto, F. Furumura, T. Abe, Y. Shinohara:Primary creep of structural steel (SS41) at high tem-peratures, J. Struct. Construct. Eng. 296, 145–157(1980)

35.36 National Highways Authority of India: http://www.nhai.org/

35.37 IRC: SP:35-1990, Guidelines for Inspection andMaintenance of Bridges, Indian Roads Congress(R. K. Puram, New Delhi 1991), http://www.irc.org.in/

35.38 S.S. Seehra: Specimen Testing vis-á-vis in-situ Testing, Nondestructive Testing of ConcreteStructures (Indo-US Workshop on NDT, Roorkee1996)

35.39 IRC: SP:37-1991, Guidelines for Evaluation of LoadCarrying Capacity, Indian Roads Congress (R K Pu-ram, New Delhi 1991), http://www.irc.org.in/

35.40 Abaqus Inc.: Abaqus Users Manual (Abaqus, Prov-idence 2005)

35.41 K. Gwinn, J. Holder: Fabric with Balanced Modulusand Unbalanced Construction, Patent 0930986B1(2004)

PartD

35

structural te 35. structural testing applications sinfohost.nmt.edu/~ashok/1b_book_chapter.pdf ·...

Documents