This action might not be possible to undo. Are you sure you want to continue?
(Part 2 of 5 Parts)
r i
Vo
THE _ SHOCK AND VIBRATION BULLETIN
Part 2 Ground Motion, Dynamic Analysis
JANUARY 1972
A Publication of THE SHOCK AND VIBRATION INFORMATION CENTER Naval Research Laboratory, Washington, D.C.
t
1
DDC
NATIONA TIC b INFORM4ATIOtJ SERVICE
Spdngfl.Id, Va. 22151
Office of The Director of Defense Research and Engineering
been approved for public release and sale; its distribution is unlimited. 67 document has /(~
This
Best
Available
Copy
2_ .
SYMPOSIUM MANAGEMENT
THE SHOCK AND VIBRATION INFORMATION CENTER William W. Mutch, Director Henry C. Pusey, Coordinator Rudolph H. Volin, Coordinator Edward H. Schell, Coordinator
Bulletin Production
Graphic Arts Branch, Technical Information Division, Naval Research Laboratory
.. II
for
N
............................
UST.
AVAIL &U/iu VECIAL.
" 
o
IF
BOultin 42
(Par o Prs
THE SHOCK . AND VIBRATION BULLETIN
JANUARY 1972
A Publication of
THE SHOCK AND VIBRATION INFORMATION CENTER Naval Research Laboratory, Washington, D.C.
The 42nd Symposium on Shock and Vibration was held at the U.S. Naval Station, Key West, Florida, on 24 November 4971. 'he U.S. Navy was host.
Office of

The Director of Defense
Research and Engineering

Fischer. Mechanics Research. Texas ... Johnson and R. Yang and H... ... Department of Mechanical Engineering.. . California ROCKING OF A RIGID....... . . Army Mobility Equipment Research and Development Center. R... L.. California 55 A SHOCKISOLATION SYSTEM FOR 22 FEET OF VERTICAL GROUND MOTION .... ... Kim and P.. Southwest Research Institute...S... . ... .. . Jr. B..... Ukrainetz... . Livermore... A...... P. Army Ballistic Research Laboratories.. 21 33 J. Galletly.. Columbus Laboratories. 75 87/ . . . Port Hueneme. Bernreuter.... . Naval Civil Engineering Laboratory.. .. ..... ..RESPONSE:SPECTRA E.. 111 SEI1MIC EVALUATION OF ELECTRICAL EQUIPMENT FOR NUCLEAR POWER STATIONS 3 H. . D.... Ohio .. Norris.. Livermore.. Aberdeen Proving Ground..S...... #. .. S... BATTELLE.. 109 Y.. . .. . ... NUCLEAR EXPLOS IONS D... ........ 97 CIRCULAR CANTILEVER BEAM ELASTIC RESPONSE TO AN EXPLOSION ... G.. M.. Saffell. M.... Miller and D.. University of California....CONTENTS. .... Port Hueneme.. 45 R. . Fort Belvoir... axd F..... Virginia iii ... Jackson.... R. L. U. C.... California THE COMPARISON OF THE RESPONSE OF A HIGHWAY BRIDGE TO UNIFORM GROUND SHOCK AND MOVING GROUND EXCITATION ..... Los Angeles... S.... The Ralph M.. E... ittjsburh.. PAPERS APPEARING IN PART 2 Ground Motion SINE BEAT VIBRATION TESTING RELATED TO EARTHQUAKE..... S. . .. Tokaz... G. Ferritto.. Westine. ..... Morris.... .... San Antonio. C...... .. Naval Civil Engineering Laboratory..... U... Saskatoon.. Pennsylvania  ....67 E... Canada MEASUREMENT OF IMPULSE FROM SCALED BURIED EXPLOSIVES .. D. Parsons Company. California DEVELOPMENT OF A WAVEFORM SYNTHESIS TECHNIQUEA SUPPLEMENT TO RESPONSE SPECTRUM AS A DEFINITION OF SHOCK ENVIRONMENT ... ..... .. Los Angeles. Inc.. Westinghouse ResearchLaboratories. ..... UNDERWATER BOTTOMFOUNDED STRUCTURE SUBJECTED TO SEISMIC SEAFLOOIR EXCITATIN . .. J. Lawrence Livermore Laborato'y. THE IMPULSE IMPARTED TO TARGETS BY THE DETONATION OF LAND MINES ..... EVENT DIAL PACK ..... ..... L.... J.... . ....... . 123 B... .... Lawrence Livermore Laboratory. . ..... .... Bernreuter... California DEFORMATION AND FRACTURE OF TANK BOTTOM HULL PLATES SUBJECTED TO MINE BLAST .. Ahlbeck. D... Haskell... Prause znd D... Hamm2r and H.... University of California. R........ . F... Vulnerability Laboratory... Zwlbel... California THE RESPONSE OF AN ISOLATED FLOOR SLABRESULTS OF AN EXPERIMENT IN .. Columbus........ SHOCK INPUT FOR EARTHQUAKE STUDIES USING GROUND MOTION FROM UNDERGROUND . . University of Saskatchewan. N. Md.
Now York DYNAMICINTERACTIOI BETWEEN VIBRATING CONIVEYORS AND .. New Jersey EFFECT OF CORRELATION IN HIGHINTENSITY NOISE TESTING AS INDICATED BY THE RESPONSE OF AN INFINITE STRIP. Civil EngineeringDepartmen.........J....... Columbus Laboratories.s and G....... F.. Day I and S. W. INTEGRATED DYNAMIC ANALYSIS OF A SPACE STATION WITH CONTROLLABLE SOLAR ARRYS....... Hartman.. The Johns Hopkins University.... C....... .... Naval Ship Research and Development Center. Hampton.. Columbus. Whippany... SUPPORTING STRUCTURE . W.'... CHARACTERISTICS OF LARGEFLEXIBLE STRUCTURES. .Dynamic Analysis THE EFFECTS OF MOMENTUM WHEELS ON THE FREQUENCY RESPONSE .. .... Virginla 137 .... H.. Tomer..... T.~. Commander.. and R.. Weinberger... Mozer.C.... M.. J. Pennsylvania FINITE AMPLITUDE SHOCK WAVES IN INTERVERTEBRAL DISCS ... M. North Eastern Research Associates. Bell Telephone Laboratories....... and O.... Advanced Technology Center... L. D..... . G.. E.... . 1.... Mason. NASA Langley Research Center.. of. Washington....C... Louisville... Philadelphia. R.....rA~W. rr. ...... New York. .....Louisville. J........ J....... Koch.. Kentucky. N... Inc. Ohio 181 195 VIBRATION ANALYSIS AND TEST OF THE EARTH RESOURCES TECHNOLOGY SATELLITE .. Professor. .. Rex Chainbelt Inc... ..... Hoboken. Maryland ACCELERATION RESPONSE OF A BLASTLOADED PLATE . T.. Inc. v2 <~. L. Kentucky 163 RESPONSE OF A SIMPLY SUPPORTED CIRCULAR PLATE EXPOSED TO THERMAL AND PRESSURE LOADING . L....... N.... Baltimore.. General Electric Company...... *.. and M. . . Upper Montclair........ North Eastern Research Associates..... Colorado '29 r ..... .... Washington. Fagel..... DYNAMIC RESPONSE OF STRUCTURES SUBJECTED TO TIMEDEPENDENT BOUNDARY CONDITIONS USING THE FINITE ELEMENT METHOD ...... Sardella.. Syracuse... 171 WHIRL FLUTTER ANALYSIS OF PROPELLERNACELLEPYLON SYSTEM ON LARGE STHE SURFACE EFFECT VEHICLES ....'V~rt ........ YuanNing Liu....PARAMETRICALLYEXCITEDCOLUMN WITH HYSTERETIC MATERIAL D. Maryland. Upper Montclair...... Inc.. A... C.. Fairchild Industries. Cokoni.. and Me D. Naval Ship Engineering Center.'Mathis..... Texas 203 213 221 o235 PAPERS APPEARING IN PART 1 Invited Papers SMALL SHIPSHIGH PERFORMANCE Rear Admiral H.. Dallas...... IBM Corporation. East Fishkill. .......... Design Engineer...J.. D.. Battelle.. Martin Marietta CrporationDenver.. Heinrichs and A......... University.. P".. and Stevens Institute of Technology..<fl~ . Louisville.... D. Rhodes. Morrow. Syracuse University.J........ Space Division.. Germantown..........CJ .... T.......... F.......... N.. Workman. EvanIwanowski..... Cohen.. Professor. .
Hyattsville. Naval Ship Engineering Center. Gens. Sandia Corporation. Virginia v I¢ . S. General American Research Division. Seattle. Samsury. Washington PREDICTION OF FORCE SPECTRA BY MECHANICAL IMPEDANCE AND ACOUSTIC MOBILITY MEASUREMENT TECHNIQUES R.Specifications SURVEY. C.D. Mitchell.ED PIPES . Taylor. B. V. and J. NASA/Marshall Space Flight Center. Albuquerque. Florida Research & Development Center. Jr.D.Otts. London. Maryland Transportation and Packaging A SURVEY OF THE TRANSPORTATION SHOCK AND VIBRATION INPUT TO CARGO F. Military Traffic Management and Terminal Service. Ph. L. England LIQUIDSTRUCTURE COUPLING IN CURI. Kao. The Boeing Company. WrightPitterson Air Force Base. C. Huntsville. SURVEY TESTING J. New Mexico HIGHWAY SHOCK INDEX R.D. Albuquerque. Maryland Measurement and Application of Mechanicil Impedance FORCE TRANSDUCER CALIBRATIONS RELATED TO MECHANICAL IMPEDANCE MEASUREMENTS E. West Palni Beach.. W. R.A PANEL SESSION SOME ADMINISTRATIVE FACTORS WHICH INFLUENCE TECHNICAL APPROACHES TO SHIP SHOCK'HARDENING D. Florida THE MEASUREMENT OF MECHANICAL IMPEDANCE AND ITS USE IN VIBRATION TESTING N.II L. N. U.. Hunter. 'F. and N. New Mexico A I TRANSIENT TEST TECHNIQUES FOR MECHANICAL IMPEDANCEAND MODA. Army Transportation Engineering Agency._Ewins and M. B. Air Force FlightDynamics Laboratory. Machineey Dynamica Division. R. Huntsville. Annapolis. General American Transportation Corporatic Niles. OF VIBRATION TEST PROCEDURES IN USE BY THE AIR FORCE W. Sandia Laboratories. M. L. illinois THE DYNAMIC ENVIRONMENT OF SELECTED MILITARY HELICOPTERS M. Lund. and J. C. Yarchb. Schock. Imperial College of Science and Technology. E. . Kennedy. Sainsbury. Favour. Wyle Laboratories. Alabama DYNAMIC DESIGN ANALYSIS VIA THE BUILDING BLOCK APPROACH A. Davidson and D. Assistant Project Engineer. M. Olson. Ohio MOBILITY MEASUREMENTS FOR THE VIBRATION ANALYSIS OF CONNECTED STRUCTURES D. Klosterman. Alabama and G. Pratt & Whitney Aircraft. Ohio SPECIFICATIONS .D. Ludwig. Lemon. J. Newport News. Senior Engineer. Ostrem. Structural Dynamics Research Corporation Cincinnati. Ph. G. F. Naval Ship Research and Development Center.
E. Minnesota vi 1' . California. Point Mugu. Lockheed Missiles & Space Company. D. Ground Equipment and Materials Directorate. The Boeing Company. H. Forkiois and E. S. Everett. Niles. Naval Research Laboratory.:Yound. Army TaikAu"omftive Command. Muth.DEVELOPMENT OF'A ROUGH'ROAD SIMULATOR AND SPECIFICATION FOR TESTING OF EQUIPMENT TRANSPORTED IN WHEELEDVEHICLES H. Shang and J. Minnesota. Lockheed Missiles & Space Company. M. Woods. Valley Forge. Minneapoll3. General American Research Division. E. Michigan State University. J. Palo Alto Research Laboratory. Fast Lansing. Culver City. Warren. Alabama DEVELOPMENT OF A PRODUCT PROTECTION SYSTEM D. General American Transportation Corporation. Illinois A STUDY OF IMPACT TEST EFFECTS UPON FOAMED PLASTIC CONTAINERS D. Rucker and R. Development. Directorate for Research.G.C. Industrial and Systems Engineering Department. Michigan MOTION OF FREELY SUSPENDED LOADS DUE TO HORIZONTAL . and S. Hughes Aircraft Company. Naval Missile Center. Army Missile Command Redstone Arsenal. Sunnyvale. Palo Alto. NASA Langley Research Center. TheUniversfty of Alabama In Huntsville. and A. Tuft. R. Rochester. Shoulberg and R. IBM General Systems Division. Huntsville. Eginegeriv an. Zwibel. Larson. Pierce. L. S. LABORATORY CONTROL OF DYNAMIC VEHICLE TESTING J.1lP MOTION IN RANDOM HEAD SEAS H. California PAPERS APPEARING IN PART 3 Test Control ON THE PERFORMANCE OF TDM AVERAGERS IN RANDOM VIBRATION TESTS A. V. California EQUIPMENT CONSIDERATIONS FOR ULTRA LOW FREQUENCY MODAL TESTS R. California A TECHNIQUE FOR CLOSEDLOOP COMPUTERCONTROLLED REVERSEDBENDING FATIGUE TESTS OF ACOUSTIC TREATMENT MATERIAL C. Port Huenem:. General Electric Company. M. Naval Civil Engineering Laboratory. U. Curtis. A. Washington.. Hampton. Grant. W. D. Clements. E. E. Smith. C. Trotter and D. W. Jr. S. Missile Systems Laborator. W. Aima. and R. Aerospace Group. Seattle. MTS Systems Corporation. California A MULTIPLE DRIVER ADMITTANCE TECHNIQUE FOR VIBRATION TESTING OF  COMPLEX STRUCTURES S. Virginia PROGRAMMING AND CONTROL OF LARGE VIBRATION TABLES IN UNIAXIAL [F AND BIAXIAL MOTIONS R. Wyskida. McDaniel. Pennsylvania COMBINEDAXIS VIBRATION TESTING OF THE SRAM MISSILE W. Grandle. Washington SHOCK TESTING UTILIZING A TIME SHARING DIGITAL COMPUTER R. Canon. Michigan IMPACT VULNERABILITY OF TANK CAR HEADS J. U.
. Deitrick. D. o. D. Radiation Incorporated. Honeywell Inc. NASA. F. Binder University of Southern California. K. California... .C. Greenbelt. Wolfe. Agrawal. Jamison. R. Hammer.... Olbert and T. H. Goddard Space Flight Center. Sunnyvale. Tedesco.. Air Force Flight Dynamics Laboratory.. N. Clarksburg. Department of Defense. . H. Herzing.. Florida THE EFFECTS OF VARIOUS PARAMETERS ON SPACECRAFT SEPARATION SHOCK W. C. TRW Systems Group. SINUSODAL VIBRATION OF POSEIDON SOLID PROPELLANT MOTORS L. Maryland Test Facilities and Technlques. E.. Maryland NONDESTRUCTIVE TESTING OF WEAPONS EFFECTS On COMBAT AND LOGISTICAL VEHICLES . THE EFFECT OF THE FINOPENING SHOCK ENVIRONMENT ON GUIDED MODULAR DISPENSER WEAPONS K.. . Melbourne. El Segundo. R. J. Minnesota DEVELOPMENT OF A FLUIDIC HIGHINTENSITY SOUND GENERATOR H... Research Specialist. Huntington Beach.. F. and A. Hughea Aircraft Company. Lockheed Missiles & Space Company. Washington.. and R. Jr. and J. COMSAT Laboratories. California WALL FLOW NOISE IN A SUBSONIC DIFFUSER E. Calfornia. Colorado TECHNIQUES FOR IMPULSE AND SHOCK TUBE TESTING OF SIMULATED REENTRY VEHICLES N. Government and Aeronautical Products Division Hopkins. Hammond. California vii .. Ohio DEVELOPMENT OF A LIGHTWEIGHT. S. D. WrightPatterson Air Force Base. California SIMULATION TECHNIQUES IN DEVELOPMENT TESTING A. Bangs. California VIBRATION FIXTURING .. Timpke. Huntington Beach. Los Angeles. Culver City. R. Weapons Laboratory... Illinois A ROTATIONAL SHOCK AND VIBRATION FACILITY RM T.. Long Beach.. Rock Island. California CONFIDENCE IN PRODUCTION UNITS BASED ON QUALIFICATION VIBRATION R.. SATURN AND ORBITAL WORKSHOP PROGRAMS R. Pendleton. Keegan and W. Colorado Springs.. Keeffe.NEW CELLULAR DESIGN. LINEAR MECHANICAL SPRING ELEMENT R. B..'1p A DATA AMPLIFIER GAINCODE RECORDING SYSTEM J. E.. L. U. A. McDonnell Douglas Astronautics Company.. Califorila STABILITY OF AN AUTOMATIC NOTCH CONTROL SYSTEM IN SPACECRAFT TESTING B. . Redondo Beach. Stafford. Denton and K. Fandrich. O'Keefe. L. McDonnell Douglas Astronautics Company. Kaman Sciences Corporation.. Space and Communications Group. Army Weapons Comna. Leete. Advanced Research Projects Agency. Jo_son. Huhes Aircraft Company. N. California State College. F.
Minnesota TRANSIENT RESPONSE OF PASSIVE PNEUMATIC ISOLATORS G. Army Electronics Command.'ado P =kGY ABSORPTION CAPACITY OF A SANDWICH PLATE WITH CR!Tft ABLE CORE D. U. C. Portsmouth. A. C. Ft. T. Tweedie. Bethesda. Medford. J. U.C. STAGNANT FLUIDS W.S. UnWfisity of Minnesota. Fox. Valley Forge. Space Division. Minnesota. Maryland OPTIMUM DAMPING DISTRIBUTION FOR STRUCTURAL VIBRATION R. Mass. Colo. University of Minnesota. Minneapolis. Minneapolis. iXxjcinovic. Chicago. Tufts University. E. Barrett and W. Minneapolis. Naval Ship Research and Development Center.. Ballistic Research Laborator. Pa. A. Minneapolis. Admiralty Surface Weapons Establishment. Nelson. Gainesville. V. L.K. L. University of Florida. Minnesota A LAYERED VISCOELASTIC EPOXY RIGID FOAM MATERIAL FOR VIBRATION CONTROL C. Steiner. Illinois SHOCK MOUNTING SYSTEM FOR ELECTRONIC CABINETS W. Minnesota DYNAMIC RESPONSE OF A RING SPRING R. Delany. J. Frohrib. Inc. D. Aberdeen Proving Ground. Plunkett. Zeidler.FREEFREE BEAMS IN DENSE. Department of Engineering. and E. Martin Marietta Corporation. lIT Research Institute. DAMPING OF A CIRCULAR RING SEGMENT BY A CONSTRAINED VISCOELASTIC LAYER Cpt. Barry Division of Barry Wright Corporation. Kilcullen. and F. Naval Ship Research and Development Center. Kacena. Washington. Medtronic. nlinois ON THE DAMPING'OF TRANSVERSE MOTION OF. Argonne. Plunkett. K. DYNAMIC ANALYSIS OF THE RUNAWAY ESCAPEMENT MECHANISM G. Stahle and Dr.PAPERS APPEARING IN PART 4 Isolation and Dmping TRANSIENTRERONSE OF REAL DISEPATIVE STRUCTURES "1. New Jersey.les. R. General Electric Company. and D. Almy. Argonne National Laboratory. Denver. Taylor. Florida Prediction and Experimental Technlqubs A METHOD FOR PREDICTING BLAST LOADS DURING THE DIFFRACTION PHASE W. Science and Mechanics. METHODS OF ATTENUATING PYROTECHNIC SHOCK S. Burbank. D. Blake. Eshleman. Hemp. Department of Mechanical Engineering. Maryland viii . Leibowitz and A. W. OPTIMIZATION OF A COMBINED RUZICKA AND SNOWDON VIBRATION ISOLATION SYSTEM D. California EXPERIMENTAL DETERMINATION OF STRUCTURAL AND STILL WATER DAMPING AND VIRTUAL MASS OF CONTROL SURFACES R. C. University of Minnesota. Monmouth.
White Oak. Virginia. G. Alabama SUPPRESSION OF FLOWINDUCED VIBRATIONS BY MEANS OF BODY SFIRFACE MODIFICATIONS D. FTields. Ralston. J. Joga Rao. Advanced Technology Center. Silver Spring. Naval Weapons Center.. Witt. Dallas. and H. Ohio RESPONSES OF A MULTILAYER PLATE TO RANDOM EXCITATION H. and W. Illinois BLAST FIELDS ABOUT ROCKETS AND RECOILLESS RIFLES W. Bell Laboratories. Canada DIAL PACK BLAST DIRECTING EXPERIMENT L. Parmenter. S. Air Force Flight Dynamics Laboratory. N.J. V. Ohio RESPONSE OF HELICOPTER ROTOR BLADES TO RANDOM LOADS NEAR HOVER C. Bandgren. Bessey. P. and R. Massachusetts INSTRUMENTATION TECHNIQUES AND THE APPLICATION OF SPECTRAL ANALYSIS AND LABORATORY SIMULATION TO GUN SHOCK PROBLEMS D. W. Bake]r. and V. G. Sallet and J. Naval Ordnance Laboratory. C.DRAG MEASUREMENTS ON CYLINDERS IN EVENT DIAL PACK S. Maryland AN EXPERIMENTAL TECHNIQUE FOR DETERMINING VIBRATION MODES OF STRUCTURES WITH A QUASISTATIONARY RANDOM FORCING FUNCTION R. Saunders. Butler. Alberta. WrightPatterson Air Force Base. Naval Ordnance Laboratory. Byrne. WrightPatterson Air Force Base. bitence Research Establishment Sdffield. National Aeronautics and Space Administration. J. F. E. F. Dahlgren. Merkle. J. Clncinnati. General American Research Division. PAPERS APPEARING IN PART 5 Shock and Vibration Analysis BANDWITHTIME CONSIDERATIONS IN AUTOMATIC EQUALIZATION C. F. W. Fgelo. L. General Electric Company. Inc. B. Watertown. N. Jr. Whippany. Ohio FACTOR ANALYSIS QF VIBRATION SPECTRAL DATA FROM MULTILOCATION MEASUREMENT R. California RESPONSE OF AIR FILTERS TO BLAST E. E. and W. Ungar. Texas A REGRESSION STUDY OF THE VIBRATION RESPONSE OF AN EXTERNAL STORE C. S.. Southwest Research Institute. China Lake. San Antonio. Marshall Space Flight Center Huntsville. Millis. Aircraft Engine Group. A. Naval Weapons Laboratory. W. Bolt Beranek and Newman Inc. Army Materials and Mechanics Research Center. T. Cambridge. Culbertson. Massachusetts. Texas TRANSONIC ROCKETSLED STUDY OF FLUCTUATING SURFACEPRESSURES AND PANEL RESPONSES E. Lakshmikantham and C. E. Morrow.. and R. Westine. DeVost. Geo'ge C. Christiansen and W. Air Force Flight Dynamics Laboratory. Berezow. Erwin. Niles. Golueke. Arroyo. Silver Spring. Maryland ix  .
F. Denver. B. National Aeronautics and Space Administration. P. Sierakowski. Naval Ship Research & Development Center. California DYNAMIC WAVE PROPAGATION IN TRANSVERSE LAYERED COMPOSITES C. Department of Aerospace Engineering and Engineering Physics.S. University of Virginia. S. NASA Langley Research Center. . W. Ross. Warren. Inc. Maryland ON THE US E OF FOURIER TRANSFORMS OF MECHANICAL MHOCK DATA H. Matula. St. Development and Engineering Directorate. Sunnyvale. Pllkey and 1o Ping Wang. A. A. Washington University. ACCEPTANCE J. USAARDC Ballistic R13sear. Southwest Research Institute. J. and W. D. Missouri STRUCTURAL DYNAMICS OF FLEXIBLE RIB DEPLOYABLE SPACECRAFT ANTENNAS B. and J. Park. Naval Civil Engineering Laboratory. L. J. Mains. California WAVE ANALYSIS OF SHOCK EFFECTS IN COMPOSITE ARMOR G. J. Case. T. Army. Aerospace Engineer. Gaberson and D. E. Goddard Space Flight Center. Jr.* * THE EFFECT OF "Q' VARIATIONS IN SHOCK SPECTRUM ANALYSIS M. Potsdam. Texas PERFORM: A COMPUTER PROGRAM TO DETERMINE THE LIMITING PERFORMANCE OF PHYSICAL SYSTEMS SUBJECT TO TRANSIENT INPUTS W. Jr. Martin Marietta Corporation. Raney. Maryland STATISTICAL LOADS ANALYSIS TECHNIQUE FOR SHOCK AND HIGHFREQUENCY EXCITED ELASTODYNAMIC CONFIGURATIONS K. G. MAINTENANCE. Bangs. Aberdeen Proving Ground. Parks. Louis. R. McGrath. B. Lockheed Missiles and Space Company. Filbey. Gainesville.. Hampton. Research. Coale. L. Pal. Kertesz. Weidman. San Antonio. TankAutomotive Command. Wrenn. L. G. Halle. Virginia x . U. F. Maryland RAPID FREQUENCY AND CORRELATION ANALYSIS USING AN ANALOG COMPUTER J. NASA. C. Port Hueneme.. Goddard Space 'Uight Center. Greenbelt. Hedges. Michigan U4VESTIGATION OF LAUNCH TOWER MOTION DURING AEROBEE 350 LAUNCH R. California INFLUENCE OF ASCENT HEATING ON THE SEPARATION DYNAMICS OF A SPACECRAFT FAIRING C. M. W.h Laboratories. and R. Lockheed Missiles & Space Company. Charlottesville.Westine. DYNAMICS APPLICATION. Head. Colorado. Aerospace Engineering Department University of Florida. Saczalski and K. Bethesda. Department of Civil and Environmental Engineering. Florida RW PLANE ANALYSIS FOR VULNERABILITY OF TARGETS TO AIR BLAST P. Clarkson College of Technology. New York Structural Analysis NASTRAN OVERVIEW: DEVELOPMENT. Cunningham. J. Virginia EXPERIENCE WITH NASTRAN AT THE NAVAL SHIP R&D CENTER AND OTHER NAVY LABORATORIES P. NASTRAN Systems Management Office and D. Sunnyvale. Kinsley and W. Maryland RESULTS OF COMPARATIVE STUDIES ON REDUCTION OF SIZE PROBLEM R.
Yang. F. Arcidlacona. Layher.. Gibbs & Cox. and J. Ni. Inc.C. R. and W. Waiston. Wada. Lockheed Missiles and Space'Company.STRUCTURAL DYNAMIC ANALYSIS AND TESTING OF A SPACECRAL T DUAL TRACKING ANTENNA D. University of Maryland. A. Wrenn. Jr. K. College Park. Bamford. Walters. New York. Washington. . Sunnyvale. Mechanical Engineering Department. Maryland SIMPLIFIED METHOD FOR THE EVALUATION OF STRUCTUREBORNE VIBRATION TRANSMISSION THROUGH COMPLEX SHIP STRUCTURES M. Jet Propulsion Laboratory. Naval Research Laboratory. D. New York ix . G. Chernjawski and C. California j Ship's Problems DETERMINATION OF FIXEDBASE NATURAL FREQUENCIES OF A COMPOSITE STRUCTURE OR SUBSTRUCTURES C. Heidenreich. A. and J. R. H.C. J. D. A. California LONGITUDINAL VIBRATION OF COMPOSITE BODIES OF VARYING AREA D.. Pasadena. EQUirVALENT SPRINGMASS SYSTEM: A PHYSICAL INTERPRETATION B. Guzy. Woods and B. Sccp. J. P.S. Garba.
sine beat vibration inputs during test can be related to foundation or building floor shock response spectra for computersimulated earthquakes. In particular.) In general. In addition. (Building floor seismic response records exhibit similar sine beat characteristics. the original seismic disturbance can be generated In terms of a nonstationary random process involving damped sinusoids with random frequencies as well as .GROUND MOTION SINE BEAT VIBRATION TESTINGRELATED TO EARTHQUAKE RESPONSE SPECTRA E. it becomes more practical to rely upon a conservative method of environmental testing. lb illustrates the corresponding narrowband vibration response of a simple (massspringdamper) oscillator wherein the frequency is restricted to the system natural frequency. INTRODUCTION Itiscustomary to develo. it can be described in statistical terms based upon "the analysis f noise in communication circuits" as pioneered by Rice[3]. Similarly. and such information can be used to authenticate parallel computeraided dynamic analyses. the safeguards equipment required for nuclear power plants).nathematical models of building structures l] and then make computeraided analyses of their dynamic response to seismic disturbances. Experience indicates that the inadvertent loss of principle function under shock can pose more of a design problem than an obvious strength failure. Fig. A series of sine beat vibrations applied at experimentally determined. This means that the equipment is conservatively tested by a procedure which makes it most vulnerable. G. earthquake vibration test criteria have been developed for evaluating and demonstrating the reliability of electrical control devices for Electric Utility Systems. natural frequencies of the equipment is potentially more damaging than the original seismic motion. a series of sine beat v 'rations are applied at experimentally determined natural frequencies of the equipment. RANDOM MOTION THEORY Random mudon is nonperiodic and. FISCHER WESTINGHOUSE RESEARCH LABORATORIES PITTSBURGH.. unpredictable with time. "although the amplitude fluctuations persist.) By way of contrast. As judged by the envelope of the peak magnitudes. PENNSYLVANIA Vibration test criteria are developed for evaluating the earthquake resistance and reliability of electrical switchgear. The test results provide data on natural frequencies and damping for typical nonlinear systems. Figure la illustrates a random vibration excitation which is characterized as broadband in the sense that it appears to'include all frequencies in addition to its large fluctuations in amplitude. therefore. . this moreorless harmonic motion is referred to as a random sine wave. although they become more complicated and difficult to model (2] . mechanical equipments located in such structures may require dynamic seismic analyses. (Earthquake accelerographs of freefield ground motions usually exhibit such characteristics. To obviate either difficult or questionable analyses. including sensitive control devices. In the special case of sensitive electrical control devices subject to possible malfunction during earthquakes (for example. The test table input can be related to the floor response spectra as calculated for a particular power plant structure and location in an active earthquake zone. However.
when a random excitation is put through a lightlydamped.Frequency spectra showing the output response of a simple oscillator for a random vibration excitation SINE BEAT VIBRATION In general. The vibration response wave (see Fig. On the other hand. 1b) tends to be skewed. Because resonant vibration buildup is an important engineering phenomenon. It is also called the twodimensional error distribution with reference to the "random walk" problem. it is essential to evaluate the frequency content of seismic vibration excitation. but they can work th.) Figure 2 shows the frequency spectra chart of PSD for a random broadband excitation and a narrowband response. the true sine beat vibration which can be expressed in terms of two rotating (acceleration) vectors as follows: (a)Random Vibration Excitation Fig.Narrowband Simple System Response InForm of a Random Sine Wave / TFrequency. narrowbandpass filterp the outputcresponse usually appears to be a harmonic signal with a slowly varying amplitude. (Earthquake grcind motion response spectra usually appear as broadband excitation. the latter demonstrating the filter action of a simple oscillator. the density distribution function of the amplitude variation of the envelope of the random sine wave (see Fig. the accelerationtime histories shown in Figs. A special case oftho fluctuating sine wave is) of course. hence the socalled sine beat vibration. can be expressed explicitly by means of the wellknown Rayleigh distribution curve. la. rial in the elastoplastic range of cumulative fatigue. the present discussioncan be simplified if limited to a "stationiry random process" consisting of an ensemble of motiontime histories. the latter considered to be stationary if the statistical properties are not affected by a translation of the origin of time. which was first solved by Lord Rayleigh(6]. u Fig.Envelope Curve of Peak Magnitudes : Time . For this purpose the quantity "power spectral density" (PSD) is used as a measure of frequency content of randomtype functions. Consequently. The energy is transmitted primarily in the neighborhood of the natural frequency of the system filter. whereas building floor response spectra correspond to narrowband quasiresonance buildup at one or more natural frequencies of the building. However. Strength failures caused by random ceismic disturbances involve a relatively ifw load cycles. (Important frequency effects in earthquake freefield accelerographs appear to be limited to a range from 1 to 25 Hz.) I 2 LIP . It. The probability density of the instantaneous values of the filtered response tends to be normal. 2. by means of the ergodic hypothesisp the required assemblyaveraging of random data can be replaced by the more simple task of timeaveraging over a single record of long duration. 2 .Accelerationtime history records of broadband excitation and narrowband response random phase relationships [4]. Hence. 1b) appears to be a sine wave at a single frequency but with pulsating amplitude. iiat. which is employd in studies of cumulative vibration fatigue. 1. b have now been characterized by means of a statistical analysis as plotted in Fig. or Gaussian (synmetrical)(5]. Response /Excitation (b).
novel feature introduced by this proposed sine beat vibration testing is that the test results and parallel design calculations for the equipment can be related to the peculiarities of the foundation or building seismic response spectra as supplied by the Electric Utility System fur their installations. Hz freq. HzA (2 beat pulses) test frequency. the motiontime history of the earthquake disturbance can be applied to each oscillator and a summation obtained of the total transient response of the building or ent etoipm equipment. hh::~ eisi F+F. the frequency of the sine beat vibration of the test machine imunting plate corresponds to each natural frequency in a continuous the equipment as determined by sweep frequency search from I to 25 Hz. In turn.F F where F Sw  natural (test) freq. . in turn. l: Fl. However.. specified o develop test 93.j•merit[8. or test ~2Hz Normal Mode Frequencies 6 Hz 22 H (a)Dynamic Model of Building (b)Equivalent Simple Oscillators Fig.[ArI Z ZZ. TRANSIENT ANALYSIS OF BUILDINGS AND EQUIPMENT The well known normalmode method can be applied specifically to mathematical models of buildings and equipment in order to determine a dynamically equivalent series of eiso dtrieadnmclyeuvln simple oscillators in terms of the uncoupled normalmode shapes and fixedbase natural frequencies.t. 3 ..1 . Hz 2 (no.. environmental testing techand several niques are wellestablished actual reproducing available for are machines earthquake motiontime histories [7 . of equipment.Computeraided dynamic analysis of building subject to seismic disturbance 3 .. spectra in the mounted equiphock response Also.g.3. steadystate vibration testing at equipment natural frequencies is wellknown in terms of the Navy MILS167 for shipboard environmient. for seismic test criteria the sine beat moion input is much preferred because it pro 'uces only a limited quasiresonance magnification and less cumulative fatigue. 0'. the equipment's capabilities are evaluIn general. Hence. of envelope. table motions which... of cyc/beat) = p c 3 ) In actual testing.
a distinction~must be made between the broadband seismic iexcitation at ground level and the narrowltand floor motion at the various building elevationsthe latter resultk. this situation is usually covered by specifying an envelopetype of building floor response spectra. the magnification of the' floor motion inthe equipment amounts to 5.) It should be recognized that when a S. equivalent sine beat vibratlon ciabe applied to the test machine mounting~plate in order to'evaluate tfie seismic capabilities of the equipment.139 Peak Acceleration . which occuriswhen a natural frequency of the equipment coincides with a natural frequency of the building structure.the broadband seismic excitation at the base of the buil dtigcan be' (b) flteredW'andgniied by the building structure. It is information derived from a transient analysis of a pertinent building structure model with a base acceleration input corresponding to a Fig. 4b conslsts of various harmonic oscillations depending upon the different paths along which the ground disturbancehas been propagated.itest using 5 cycles/beat at 6 Hz will produce this same magnification in equipment having 5 percent damping. anl(c) amplified by the equipment response at the coincident building natural frequency of 6 Hz. In other words. The resulting floor motion shcwn in Fig. a 69 with . 2 9Peak . (In turn. the test machine will duplicate (or exceed) the resulting equipment response motion shown'in Fig 4c.5 p'rcenT damping.) FILTER ACTION OF THE BUILDING STRIICTURE Figure 4 illustrates how (a). 6 and 22Hz needbe considered.Typical accelerogranis of OBE horizontal motion as transmitted by dymanic model 4 .5pek Respon 1structure 19' " natural frequency of the equipment does not coincide with a natural frequency of the building then the quasiresonance motion buildup in the equipment will be much less evident than as slzwn by Fig.However. At 6 Hz with. However. which no longer coincides with a building structure natural frequency. then a more severe response condition might have to be recognized. the input the amplitude should be reduced to produce floor response specified at the test frequency. '  BUILDING EQUIPMENT "DESIGN" lime 0~g (a) Rjndm'type Ground Motion Input toComputer Modl of StcureEqutpent I PROCEDURE Figure 5 shows a semilog plot of typical floor response spectra as supplied by the ArchitectEngineer of the building.29 0 O. when there is interaction of two or more equipment modes.) Mf Uresult 2x byBuIlding Strulure 0.only Time applied at the equipment natural frequency. Alsol i 'the following discussion an essential distinction must be made between the customary shock response spectra used in design. (Similarly. (Similar frequencies found'in the equipment. 1b. and inthe form of an. 0 times by the building structure. JIn turn. although a sine beat vibration test would be 0 0. and (b) thedynamically equivalent seriis of simple oscillators. it now approximatesfa random sine wave as shown in Fig. Figure 3 shows (a) a typical mathematical model of a nuclear power plant building. 4 .and the original motiontime historyused as the basis for the equivalenitsine beat vibration testing. Beat 416 Ii SneBealal .} 1 o 0. this motion under the equipment has been filtered at 6'Hz and magnified 2. Essentially. The net Is a somewhat conservative sine beat test compared to using actual floor motion inputs. (An equivalent sine beat vibratio. 5 times. 4c. Only the three lowest modes of vibration with natuial frequencies of 2.) It Is appreciated that in complex structures there may be important crosscoupling effects and fluctuating response motions when natural frequencies are close together.5 (c Manifiecd qulpinenio of 5 5 Sine . 'since'they are the only ones calctated within the most imp6tant earthquake hazard range ibr vibration buildup from about 1 to25 Hz..ng from the filtering action of'the mechanicalsoilstructure system.model are not necessarily coincident with those in the building.
5 Cycles/Beat 0. which is the coincident buildingequipment natural frequency. f V2 I * 00 1 13 to 3 1 0.Sine beat motion test for DBE 2 Second Pause 1.OBE horizontal floor response spectra Fig. 2. 13 g. The filtering action of the building. and the OBE has been established on the basis of the seismic history of a specific power plant site. the input motion has been magnffied times to give apeak value of 0. which can be determined for the equipment from the motion buildup during the continuous sweep frequency test.8 x OBE 5 . 5.27 g corresponding to the horizontal asymptote for frequencies above 30 Hz. the ground motion is with a peak value of 0.8 'random . At an upper floor level where the equipment might be located.0 Hz.. for 5 percent damping in the equipment at 6 Hz. Free Vibration 29 (b) Test Equipment Acceleration Ig  .5 x (a) Test Table Motion Under Equipment Fig.2:0 _ Fmodel Damping _ 1 .i25% L9 1.56 horizontal motiontime history for anOperltting Basis Earthquake (OBE). In other words.iI . 5). the maximum equipment response acceleration becomes 1.0 Hz. Q =5. Presumably the includes a soilstructure interaction factor. The 2. 4c with Fig.7 g 11 5 H __ _ W 1.7 g Peak Respon se 2g =5%.5 Equlpmenthf'erequency. I 11 0.30. Hz 30 5. (Compare Fig. 5response at equipment location in building S6.5 g. 6 . 4b with Fig. 5&) Figure 5 also gives floor motion spectra (actually the peak acceleration measured on the equipment at each of its natural frequencies) for several values of percent critical damping. 4.2 0. (Compare Fig. Finally. was based upon the data to be found and interpreted from Fig.49g Peak Input. as previously illustrated by Fig.
This elementary computeraided analysis is equivalent to the normalmode evaluation of the equipment response (see Fig.Vibration magnification at quasiresonance equipment natural frequency. there is no further magniBUILDING EQUIPMENT "TEST" PROCEDURE For seismic testing of equipment. influential. Howeverj for test purposes it . if the equipment natural frequency is 30 Hz or greater.5 times. NorthSouth and vertical). the Electric Utility Company and the ArchitectEngineer supply the foundation and floor response spectra with a statement regarding ODE and DBE. SINE BEAT VIBRATION TEST METHOD Figure 7 provides the theoretical data needed for the correlation of floor response spectra (Fig. In view of the cumulative fatigue effects associated with testing the equipment in three directions (EastWest. I .. whereby DBE = 1. 6a).8 (OBE). (In practice. the vibration machine input is intended to simulate the worst aspects ofthe motion of the building floor at the attachment of the equipment. L El Centro earthquake at 5 percent dampingthe shock response spectra magnification is only The floor response spectra shown in Fig. In summary. since 1. 5 can be applied directiy. Fig. the Intensity multiplier is used to achieve a more safe Design Basis Earthquake. it becomes apparent that sine beat vibration testing can be made very conservative with regard to probable local seismic activity. (For the 25 Il Steady Stale. percent of critical). or else apply the conservative substitute shown by Fig. 4c and 6b). & 5 2 4 I. which shows that the selective filtering action of the building structure can be quite. 27g) = 0. 8 (0. 6a s huws that a sine beat vibration input consisting of 5 cycles/beat at 6 Hz and a peak amplitude of 0.5 g at 6 Hz and 5 percent damping. Also. percent of critical damping Fig. 49 p is the appropriate test machine input to represent the damaging effects of the floor motion under the equipment as shown in Fig. The peak horizontal response acceleration at the equipment cg will be 1.13 g = 11. 7 . 5 can be used for equipment design purposes as follows. As many as 5 successive sine beats have been used in actual testing at each 6 E L :/j 5 \  >  I _"  fication ofrthe basic floor motion which is'onlyt 0. Based upon the transient analysis of a linear massspringdamper model with a sine beat base motion excitation. 6 7 I ' 10 8 E. it is Important to recognize that for design purposes the floor response spectra of Fig.) Finally. 7 g and then gradually decays as a free vibration at C Hz before the next sine beat vibration is applied.. hence the time duration of the corresponding strong motion accelerograph. The total number of successive beat effects applied to the equipment depends upon the earthquake magnitude. In addition. 4b. This means that 1. the vibration magnification curves (socalled Qfactor at resonance) are calculated for both 5 cycles/beat and 10 cycles/beat over a range of damping values (i. 6a..maximum overall magnification of the groundmotion into the equipment at 5 percent damping is 1.CNles/Beat ! esIRandom.5 times the weight of the equip1C • 10 Cycles/Bea g Y 10 / I 7C\ . ment should be applied as an equivalent static horizontal load at the cg in order to verify that stressed members are below the yield point.49 g. 5) and test table input (Fig. since the probability of an earthquake worse that the OBE is difficult to agree upon. Fig. . necessary to apply the actual floor motion shown by Fig. A conservative approximation of the earthquakeinduced floor motion is applied by means of the sine beat vibration motion. .5 g/0. 6b shows that the equipment response amplitude reaches a peak of 2. 4b.27g.
" peak acceleration of 2. compared to steadystate vibration testing. indicating that sine beat vibration testing can be made quite severe. Notice that at equipment natural frequencies of approximately 4. Also. At each experimentally determined natural frequency of the equipment. Hz 25 should be increased by the DBE factor (1. 75 g instead of 1. The previous example for selecting a sine beat vibration test corresponding to building floor response spectra can be summarized as follows. or the input "cycles/beat" selected on the low sidej the net result in both cases would be a smaller Qfactor and a larger test machine acceleration. 49 g at 6 Hz. the sine beat input acceleration peak should be only 0. 49 g at 6 Hz). the peak acceleration of 0. 8 . The upper and lower sideband frequency ordinates correspond to the intersections of a vertical 'line at a resonant (test) frequency abscissa with the two radial lines drawn for some value of percent critical damping. 4Hz instead of the 0. In addition. 5 times at 5 percent damping) in order to get the peak value of the sine beat input acceleration (0. the Qfactor near zero damping means only a limited resonance buildup. ordinary strucHowever. (Similar curves can be plotted at 10 cycles/beat and steadystate.Figure 7 also shows the Qfactors for steadystate vibration resonance (Q = 100/2 g ) and for random (white noise) excitation equal to the square root of Q. the response at 5 percent equipment damping is only 0. This example shows the need for adjusting both building and equipment designs to avoid coincident natural frequencies. 2 and 8. the corresponding (2) sideband frequencies can be determined along the abscissa for resonance curves at various values of damping in the simple oscillator system. seismic.7 g in the power plant equipment under test amounts to an overall magnification of 21 times. 13 g corresponds to an earthquake characterized as VII on the Modified Mercalli Intensity Scale (6.) The stairstep construction shown by the dashedlines gives the maximum spacing between "offresonance" test frequencies. Figure 8 shows the above information plotted as a family of fanshaped curves for a sine beat excitation at 5 cycles/beat. 245 g at 4. together with the assurance that at least 1/2 peak amplitude will be excited. when testing unknown systems there will be no destructive buildup produced inadvertently. 2 and 8. This proposed test method is conservative in that should either the damping of the equipment be evaluated on the high side.5 Damping 6 OIl fResConace Tests S Cri/ / Damplin Sle / 15  2 N 2_ 1 15 1 3: 0 5 3 5 4 o 2 nates for the OBE response spectra curve 0 10 1 Resomant Frequency. The most damaging sections of earthquake oscillographs usually correspond to a sine beat vibration excitation of about 3 cycles/beat at various preferred frequencies from about 1 to 10 Hz. as shown by Fig. in this example for 5 cycles/ beat. which are noncoincident with the building structure natural frequency at 6 Hz. 4 7I .Chart for determining number of offresonance tests at 1/2 peak amplitude usually evaluated by testing at successively larger glevelsp the danger of fatigue failures is reduced because of lower average stressing. since equipment fragility levels are Fig. theordi 40 . meaning damage "slight to moderate in well built.8 times) and then reduced by the Qfactor (5. wherein the sideband widths becomes increasingly smaller. Hence. In other words.motion response magnifactors fall soniewhat below the random fication excitation values. comparison. OFFRESONANCE SINE BEAT TESTS The computeraided analysis used to obtain the Qfactor curves of Fig. 4a. Somewhat arbitrarily at the 1/2 peak amplitude ordinate. It should be emphasized that in the example of Fig.4 Hz. By 50 . 7 has also supplied data for plotting resonance curves of magnification versus frequency from 1 to 25 Hz. 6b the tures. The quasiresonance buildups produced by the sine beat vibrations fall'in between the latter two extremes. 5 Richter Scale).5 g maximum at 6 Hz.
and Huntsville. sine beat test inputs and subsequent equipment resp. all emphasize the importance of quasiresonance phenomena as the source of damage or malfunction. On the otherhand. L. 2. Blume. 8. but persist over a relatively wide frequency range and they can be excited in 411 three directions of testing. Shock & Vibration Handbook. "Mathematical Model Analysis for the Dynamic Design of Machinery".Of course." 28th Shock a. Henc% sine beat vibration testing appears appropriate for evaluating complex equipment where it is most vulnerableat its natural frequencies as determined by a vibration sweep at a reduced acceleration input from I to 25 Hz. 1968. 7. Miles and W. of Defense.d Vibration Symposium. to establish "an acceptable seismic risk". McGrawHill. 1961. there are insufficient strongmotion earthquake accelerograms available for power plant structures. Shipway. Norco. the severity of the test can be simply adjusted and the test results can be made conservative with respect to any specified seismic environment. (As a practical matter. Portland Cement Assoc. 8 REFERENCES 1. S. E. In switchgear equipment. 51. typical complex electrical switchgear systems usually introduce nonlinear effects in terms of snubber springs. Also. Bogdanoff.ses can be directly related to the foundation or building floor seismic response spectra as supplied by the Electric Utility Company. Bell Sys. T. University of California in Berkeley. 9. The use of computeraided analyses of buildings and equipment. S. both of which conditions appev' typical of seismic disturbances. October 1967. S." Chapter 11. 2 April 1961. then they can be used to start the stairstep construction in both directions to establish additional offresonatice test points. 23. Thompson: "Statistical Concepts in Vibration. Mustin. *avoids . Sine beat vibration testing produces only a limited quasiresonance buildup and excessive fatigue in the equipment. Experimental Mechanics. Penzien. July 1960. G. when there are many equipment natural frequencies distributed over the range from 1 to 25 Hz. the stairstep construction may indicate that additional offresonance tests are unwarranted. "Design of Equipment to Withstand Underground (Nuclear Weapon) Shock Environment. Fischer. Cal. 46 (1945). 3. The primary purpose of the previous development is to avoid unwarranted cumulative fatigue and wear of equipment being tested for earthquake resistance. In general. 1961. Rice. New York. G. "Response of a Simple Structure to a Random EarthquakeType Disturbance. Equipment resonances appeai heavily damped. G. W. there does not appear to be any obvious correlation with field service reports. SSA. 6. friction. U. OA New Technique for Seismic Shock Simulation. "Design and Research Potential of Two Earthquake Simulator Facilities. etc. when there are only one or two obvious equipment natural frequencies. crosscoupling. J.) CONCLUSIONS In general." Bull. and the specification of earthquake response spectra. Tech. 5. J.S." Shock & Vibration Info. SESA. et al. Center. E. where possible loss of principle function is a more important consideraton. clearances. J. or only building structure natural frequenciei to consider. Design of Multistory Reinforced Concrete Buildings for Earthquake Motions.. U. 282 (1944) and 24. 0. let alone specific types of equipment. it is generally agreed that an isolated peak acceleration response is not a reliable indication of damage. et al. 4. G. Also. Fischer. Dept. of Defense. Dept. et al. J.)) Wyle Labs. Ala. "Theory and Practice of Cushion Design.." Richmond Fie!d Station. John A.
Frankly. I think that random transients would be v. Gaynes (Gaynes Testing Laboratories): How did youmonitor the relays andswltches todetermine wheter they were functioning or not functioning? Mr. When you use three dimensional testing you never test nearly as severely as we are suggesting with sine beat testing.. 9 * . Mr. but as far as malfunctioning.  .aperkir i.y to go. With actual earthquake records. Haag.(MTh Systems Corporation): We have been performing several tests in the seismic shock area including the sine beat. I will take testing with simultaninput anytime. Of course ifthe circult breaker pops open it is pretty obvious that you are Introuble. it seems to be the electrical operation that is the most significant. but I started out by saying: "In the interests of simplicity we pretty much stuck to the sine beat. you just think you are pushing something In acertain direction..DISCUSSION Mr. and determine the resonance modes by moitoiligl the dtaplaceiment of the specimen. We also have gone quite a bit further. To find the resonances. Usbg the sine beat seems tobe a more severe test than using random transients. There are accelerometers located on the structure etc. . depending on the application. The definition of realfunction is quite a touchy point. Fischer: Usually we had an electrical hookup to an oscillograph element. Again. We feel it is a severe test. approach by usigrandom trnsients In determining the ability of the product to withstand vibration. we do pretty much the same as required in MILSTD167. somebody gives you a definition of what constitutes a malfunction.. and I think you will find that your feelings are correct. We have gone into the shock specters". such as computers for the Safeguard system. where you sweep the irequencies. It can always escape at right angles. and we pretty much confirm your approach. Piunkett pointed out tids morning. We will be presenting a paper in the near future on that particular comparison. We Justify the simplicity bysaying it is a conservative test. but If the shock spectrum can be adequately defned. I agree. as I believe Dr. In some applications. Fischer: Well it could be more authentic. so one practically has to use solid state circuitry. any kind of relay flutter is not allowed at all. Heous Mr.
' 4. a laboratory facility for the seismic evaluation of Class I electrical equiptrent was designed and installed early in 1971 at the BattelleColumbus Laboratories. Columbus Laboratories Columbus. Prause and Donald R.o=' A f. These must be designed to remain functional when subjected to the Design Basis Earthquake (DBE). nearly all locations in the United States have a history of earthquakes. often consists of a light frame strcture supporting flexible racks and panels to which a variety of components such as transformers. . Significant results from this work are included to provide equipment designers with some typical experience. it is considered impractical. however. Ohio A review of current seismic specifications for electrxcal equipment and the different types of vibration excitation that might be used to satisfy these specifications led to the design and construction of a seismic vibration facility with the capability of reproducing typical building floor earthquake acceleration time histories.'. $. A comparison of the acceleration response spectra for the table motion and a typical command signal demonstrates the feasibility of using this type of realistic simulation of earthquake motions for evaluating electrical equipment performance. although the frequency of occurrence and severity vary considerably.1$. As a result of this conclusion. if not impossible. and these are subject to less severe seismic design requirements.W ' V ° ' ii . t t.. piping systems. Analytical techniques can be used to determine the adequacy of the basic structure of an electrical equipment cabinet to survive an earthquake. INTRODUT1ION The design of nuclear power generating stations requires the consideration of many types of possible accident situations in order to insure that the public is protected from potential exposure to nuclear radiation. SEISMIC EVALUATION OF ELECTRICAL EQUIPIENT FOR NUCLEAR POWER STATIONS Robert H. switches and meters are mounted. As a result. . The considaration oi earthquake effects is particulLrly important because the forces can be extremely large and all pacts of a nuclear station could be affected simultaneously. ail nuclear power generating stations being built in this country are designed to withstand earthquake motions even though they may be located in regions where earthquakes are ignored for other types of structures.ponse to earthquake excitation of building structures. to determine by analysis whether the electrical components will perform reliably without damage or temporary interruption of operation during a seismic disturbance. that has been selected for the particular plant location. relays. H  o *. and many types of mechanical equipment [1]. Electrical equipmerit. SEISMIC RDUIRENTS FOR EECTRICAL EQUIPM T The components and structures of a nuclear plant that are critical to the hutdown and maintenance of the reactor in a safe condition are designated as "Class I". sometimes referred to as the Maximum Credible Earthquake. Modern dynamic analysis techniques have proved to be valuable tools for predicting Lhe re . Manufacturers of Class I electricali equipment are now being required to verify that their equipment is capable of acceptable performance during a seismic disturbance. The design of the vibration table and control system and the development of the acceleration timehistory command signal are discussed.  . Preceding page blank 11 _ _ J . The design of this facility was based on the desire to achieve a realistic simulation of the earthquake vibration environment. Contrary to popular opinion. however. Ahlbeck BATTLLE. Structures and components whose failure would not interfere with a controlled shutdown of the reactor or contribute to an excessive release of radiation ar generally designated as "Class II".
. 4s . 0. during the earthquake of May 18.. simple structure will respond to that vibration. Therefore. In particular. the most severe part of the earthquake occurred during an interval of only The maximum acceleration of 1015 seconds. the earthquikeresponse spectrum for a building floor ie the maximum response of a series of 3ingledegreeof' freedom oscillators that are excited by the floor motion. California Earthquake. stresses. Whileit is expected that the severity of the DBE wiii vary for different locations in the country..o3'o . Each oscillator has a fixed percent of critical damping. so the series of maximum responses gives a good representation of the frequency content of the floor motion. 0 about equal intensity and a vertical component that usually has lower peak accelerations. but a different natural frequency.Fourier 1 spectrum or Fourier series to describe cotp .'$' ' 0 4  4 32 8 20 16 1 T*X.) to a typical earthquake ground mction selected for that location. The floor response spectrum does not describe the acceleration versus frequency characteristic of the floor motion directly... since experience has shown that thistype of description often confuses even specialists in other areas of structural dynamics and vibrations.ex signals. The earthquake ground motion will have acceleration components in two orthogonal horizontal directions of the acceleration time histories of the floor response could be included directly in equipment specifications. . there is also considerable difference in the type of specifications being used to assure compliance with seismic requirements. eta. California. Adding to this proble is the fact that there are no standard requirements for seismic evaluation.+. The predictions of the building floor motions in response to the DBE ground motion are then used to establish the seismic requirements for equipment that will be located in that building. The use of a response spectrumhas & methbdof charparticular signLfcance:as acteriziing a complex ti. although the energy is concentrated at higher frequencies. Fig. it has been customary in earthquake eniineering to describe the equipment vibration environment by a response spectrum. . I Ground acceleration record of NorthSouth component of El Centro. accelerations. The greatest source of confusion from nearly all types of seismic specifications is5the use of response spectra as the principal means of describing the earthquake vibration environment. so a manufacturer receivesadifferent set of ieismic specifications for each new nuclear power jenerating station. May 18.33 g measured at El Centro is the strongest ground motion that has been recorded (the authors have no information on very recent earthquakes such as the one in Los Angeles in 1971).. Fig.SECONDS Fig. The earthquake analysis of a Class 1 p 0 Z 2i _ .o. While y 0 o I . . building requires a detailed mathematical model of the building's stiffness and mass distributions.ansient vibration in a way that is useful lor determiniag h. shows a typical 5second record from the most severe part of the acceleration time history preulcted for the floor of tie auxiliary building of a nuclear power station. Data of this type are often used to establish the DBE ground motion at nuclear plant sites by retaining the time history and by linearly scaling the amplitudes to achieve a desired maximum acceleration. . 2  Typical horizontal building floor acceleration for DBE 12 .A basic problem faced by these manufacturers is that only rarely are those persons responsible for the design of this type of electrical equipment also knowledgeable in earthquake engineering technology. a brief review of the development and implications of earthquake response spectra for earthquake engineering seems justified. 1940 0411 0711.0  0. o T. Although the ground motion lasted for about 45 seconds. for example. 1 shows the component of ground acceleration recorded at El Centro. 1940.and this is the principal sourieof misun4derstandingby those who are familiar with theuse of "the. Fig. EARTHQUAKE'RESPONSE SPECTRA The design of each Class I building for a nuclear power generating station requires a detailed analysis of the response of that building (deflections. 2.
3  Series of sechanical oscillators used to determine floor moLion response spectrum 13 . 4 . and constant amplitude sinusoidal excitation. and the maximum absolute value the response. X (1) X3 (t) ftWthe Le_ Floor Fig. rvency. However. sine beat excitation. the most severe vibrations will be experienced by floorwounted equipment with natural frequencies close to those of the building. Tie process can be repeated to determine a family of curves of 1z I ma versus natural frequency. Equipment which is quite rigid (natural frequencies above about 30 Hz) will follow the floor motion exactly. It is a matter of choice whether the maxicul response parameter of relative displacement tz.. Some of the different types of vibration currently being used to evaluate electrical equipment are decaying sinusoidal excitation. 1(t). because they are relAted for zero damping by the following equations: 'n max' or absolute acceleration 1"n + max lu o3 0 /I op 0 03 0 I 1. 4 shows an acceleration response spectrum for the horizontal floor acceleration time history shown in Fig. and this is the displacement response spectrum for the floor acceleration. "erttz NOalwaO I 2 n+ I mAx  . is the relative displacement between the mass and the floor. Iz Imax .The diagram shown in Fig. This makes it possible to devise any number of vibratory motions that will satisfy many spectrum of whilch for will little resemdamping. _ • " X 1 0.crum is identical to the maximum floor acceleration that would be observed from the floor" acceleration time history (see Fig. 2). can be obtained for of . is the undamped natural frequency of the nth oscillator. 2.X(t) n+  2nzn(t) 2 nn y(c) (2) Fig.Floor horizontal acceleration response spectrum for DBE SEISMIC VIBRATION STUDIES A detailed discussion of the approximations involved for small damping can be found'in Reference (2). value of a have selected response blance to a typical earthquake floor acceleration time history. The use of the response spectrum has some important advantages for designing structures to withstand earthquakes. The floor anceleration time history. wn' . Most of the horizontal vibratory eitergy fron earthquakes occurs at frequencies below 10 to 15 Hz.1 (4) max wunIZnI Fig. '(t).eachoscillator.or diferel values of damping. The equation of motion for any one of the michanical oscillatorn is n where zn(t) . All of these employ single frequency excitation so that a number of evaluations at different frequencies are required to cover the response spectrum. The'ire are many techniques available for calculating the displacement response time history. Therefore. its use as a specification of the earthquake vibration environment for equipment evaluation has some significant limitations resulting from the absence of all phase angle information. and C is a selected value of critical damping ratio. Therefore.% c 1% Jc oGO 04 0 _% [in is plotted. would be predicted from a structural dynamics analysis of the bilding response to the DBE..I x. z(t). and the largest response of the building occurs at its natural frequencies. 3 of a Sseries of mechanical oscillators resting on a building floor ha proved useful for explaining the response spectirum. the acceleration amplitude which is approached asymptotically at high natural frequencies on the response spec.relative velocity 60 4 30 20 .1 I 1 O 0 1 1 20 03 1 1 1 I 111 O0 1 20 1 30 40 I 0 max ax 30 40 r.
the peak response spectrum amplitude of 1.10).11 g. The coupling effects of simultaneous vertical and iorizontal floor motions during an actual earthquake ure reduced in importance by their considerable difference in both amplitude and frequency content. 2. Furthermore. However. Since the type of equipment to be evaluated by this seismic vibration simulator was not expected to respond to large displacement components at low frequencies (below I Hz). 2. while desirable for achieving the most realistic simulation. but it is still different from that of an earthquake motion.sin about weight with maximum base dimensions it 80 by 48 inches.5. To provide a table wit's minimum weight and maximum stiffness. and the actuator can be rotated 90 degrees to provide motion in either vertical or horizontal directions. Normal * !4 .5 Hz. thereby keeping the oilcolumn stiffness high and maximizing the highfrequency response capability. Large peaks were found at 0.9 Ilz in the spectrum of horizontal acceleration. The principal goal was to be able to evaluate equipment efficiently while reducing the possibities of subjecting it to vibration which might be considerably less severe or more severe than necessary. with minor peaks at 1.Also. with lower amplitude peaks from 7 to 11 Hz and from 17 to 21 Hz. For these reasons. The single hydraulic actuat. Simultaneous vibration in vertical and horizontal directions. if the equipment being evaluated has its lowest resonance at 8 Hz with only 2 percent damping (Q = 25)9 the sinusoidal excitation would produce a peak acceleration of 2.25 g in the vertical direction were recorded during five seconds of the floor vibration response data for the DBE. with a predominant 0. while overtesting might result in an unnecessary increase in the equipment cost. 5. was not incorporated because of its considerably greater cost and complexity. equipment with several resonant frequencies below 30 Hz will not respond to single frequency excitation in the same way it would respond to an earthquake motion. It is quite unlikely that the maximum accelerations will occur simmltaneously in time for the two directions. The vertical acceleration power spectrum showed a predominant peak at 1. T Is limit was chosen to beobout 3000 pour . the response spectrum amplitudes for these motions vary more than the earthquake response spectrum for changes in damping. 4 can be satisfied by using a 0.37 g in the horizorital and 0. This requires making conservative estimates of the expected equipment damping or measuring equipment damping at resonances. For example. The horizontal displacement reached a peak amplitude of 12 inches. SEISMIC VIBRATION FACILITY DESIGN The first task in designing the BattelleColumbus seismic vibration facility was to deffne the amplitude and frequency response requireents needed to reproduce the typical floor acceleration time history shown in Fig. Acceleration power spectral density curves for the 5second signals were generated by Fast Fourier Transform techniques to provide some idea of frequency content.r is attached below the table centroid.75 g compared with about 1. Peak accelerations of 0. constantamplitude (Q . Analog acceleration signals were recorded on FM tape by digitaltoanalog conversion of the original digital computer data for use as simulator command signals. Preliminary procesaing of the acceleration data consisted of numerical integration to obtain velocity and displacement signals. Physical Nize of the simulator was based on an estimate of equipment sizes to be evaluated in the foreseeable future. Command accelerations and the resulting filtcred displaciments used as a basis for the facility design are shown in Fig.a procedure which is both expensive and possibly inaccurate because of nonlinearities. one of the first design decisions was to limit the total actuator stroke to 6 inches (peaktopeak). The use of sine beat excitation is preferable to constantamplitude sinusoidal excitation because the variation in the response spectrum amplitude with damping is less. The design of the seismic vibration facility included consideration of several tradeoffs necessary to achieve a "costeffective" facility. Thecs amplitude and frequency reuirements led to the selection of a servovalvecontrolled hydraulic actuator to power the vibration table.3 and 0. It would be quite dangerous to "undertest" the equipment.1 g at 8 Hz for 5 percent damping shoin in Fig. it seemed desirable to develop a more realistic simulation of the earthquake vibration environment. o equilateral triangle of steel Ibears with three anglebisccting beams In the center was designed. This was done in order to minimize the entrapped oil in the hydraulic actuator. and the large number of evaluations required might produce fatigue failures that would never occur during an earthquake. Negligible response was found above 10 Hz.311z frequency plus some additional lower frequency component (. The alternative of using the maximum flooracceleration amplitude to test at all frequencies may also severely overtest equipment. and 6.1/2C .0 Hz. To keep the commanded displacement within this 6inch limit.2. a highpass filter was used In series with the acceleration comand signal to attenuate frequency components below 1.5Hz.6 g'Sa for the floor earthquake motionan "overtest" of 70 p'rcent. sinusoidal vibration at the equipment base.he displacement did not return to the initial position within 5 seconds).
with short lengths of hose supplying fluid to an actuatormounted manifold. Floor msvoement d veW. and solenoid shutoff valve are located at the of the vibration table. the oilcolumn resonance adds a pair of lightly damped cnaplex poles to the root locus plot. . This base is fastened to a heavy "strongback" formed of 36inchdeep Ibeans located beneath the floor level of the laboratory. 6 shris the seismic vibration table pos.n . while in the vertical mode a combination of Roundwtys and linear ball [ Rqproduced 'fn lvailable O COPY  bushings is used for guidance at the three corners of the table.cot Fig. * . a.: modes of vibration. 10micron fi!ter. piston red. H on the Safety fine tions (overtravel limit switches and servo overcoand voltebe level) deenergize the shutoff valve in the event of a system failure and stop the table with a maximum qcceloration of about 0. results from the mass of the table and equipment oscillating on the effective stiffness of the entrapped oil. .base j 47.ll natural frequencies were predictad to exceed 50 Hz. a 1gallon bladder accumulator. 25gpm hydraulic supply and loog 76104 (15gp) flow control servovalve. .75 g. •Since estimated based on an assumed effective oil bulk modulus of 100.4 . Without compensation. Inhrd V r . _. fastened to a heavy concrete slab.z was oilcolu=n resonant frequency of 22 11z 4 ' IO I • ' m &" " . + 15 L * . and . Support and guidance of the table in the horizontal mode are providcd by Thomson "Roundways". Three Firestone "Airlde' springs connected through a pressure regulatev to shop air are used to support the static weight of the table and equipment for excit. The servovalve and crossover relief valve are attached directly to this manifold I 9. 5 .qo . .000 psi (in tests this the hydraulic power supply is located some distance from the simulator. Several methods can be erployed to control the An Ibeam frae is used to provide a base for accurate alignment of the guideways and actuator. 25Gmnomum 414" . I rate at peak vclocity dictated the choice o actuator size.tioned for vertical motion. Hofqlato Floor Acceletolson F!g.. Design of the servo controller was based on the salient fact of life of a hydraulically actuated inertial load: the This resonance oilcolumn resonance (3). j j :.tion in the vertical direction. a miim. The strongback. of the table when actuateit at the centroid vere calculated by means of a finiteelement digital computer program. b Horizontol FWoo Oocement double rodend cylinder was chosen to rate with an available 3000psi. 6  Battelle Seismic Vibration Table In vertical mozde Conflicting requirements of stiffnes (oilcolumn resonance) and maximu oil flow t 93 on. For this size of actuator.Acceleration and filtered displacement time histories of bilding floor response toi prvd side minu enrpe i "active" of the servovalve. and actuator support structure. provides the necessary seismic mass against which the simulator can react. A 1/2inch bore. c Verhicol Floor Acceleroton . Fig.
'al testing. floor seldom exceed 0. for example. Since the maximn. and force limitation is the most significant parameter for any type of sinusoi. Lower frequencies are of little practical interest because the lowest equipm~ent resonant frequencies are seldom below 5 Itz. 10 shows the maximum table accelerations that have been measured for the frequency range of maxiatum 1 to '5 lci . Rod Sec . One disadvantage of bypass flow is the lower frequency of the resonant peak for desirable levels of damping.5 for a DBE.2 g. Adjustable bypass flow using a needie valve and small tube acrossthe actuator is an effective.. method of c~ntrolling the resonance.7 efec ofactato Te byassadequate flow on root loci of the lowerfrequency tormst hydraulic eyetew With Inertial load 16 * . and the resulting table acceleration was monitored using a QuanTech Model 304TDwave analyzer to track the fun "uental component of acceleration with a 1lz bandwidth filter. A constantamplitude voltage command to the double integration circuit wgas varied from 1 to 40 Hz.. The results of the sinusoidal closedloop frequency response evaluation are shown in Fig. 9. Accelerations are limited by the flow rate of the at frequencie from abo t 1 to 10 and by the taximum actuator force of about pounds ataccelerations higher frequencies.. root. . i Fi. Primary servosystem feedback is derived from a posttion transducer (DCDT) mounted on the actuator. the system with bypass flow is far more stablethis is shown by the reduced angle of a vector from the origin to the gaindependent Fig. . A &odifiedlacceleration feedback signal from a straingage accelerometer mounted on the table is also summed with the error signal to improve system response inthe 10 to 20Hz range. oo . 8 . flow rate. The positin feedback signal is subtracted from the filtered.0 1200 Jos 20/ OO X×z vibration facility Fig. For comparable values of forwardloop gain (the parameter Ka represents servo amplifier gain in amps per volt). doubleintegrated acceleration command signal to provide the errorsignal to the servovalve.effects of oilcolumn resonance [4]: acceleration feedback. 8 shows the control system in blockdiagramform with linearizedtransfer functions for the major cimponents.Seismic control system SEISMIC VIBRATION FACILITY PERFORANCE Closedloop frequency respons: evaluations were conducted with a '1500pound dummy equipment load on the table at normal acceleration amplitudes.1"' iloo No bypol flow 0 •Rod See Zo 2oo 1amplitude 0While o) Ht. the acceleration capability is for frequencies above 2 liz. or controlled bypass flow. and the effective system bandwidth (3 dB point) is about 14 Hz for the command signal of 0. 7. The effects of introducing bypass flow (based on a linear estimate of system response) are sketched in the root loci of Fig. iU . S lservovalve hydruicsly ld7500 7 Th effct f acuatr byassg Fig. A secondorder resonance with a + 3dB peak ic evident at 10 1100 at 5 05 =Rod 0o sec 0 'oo . Fig. pressure feedback (inconjunction vith a pressure control servovalve). "a e o o0oo .'. the maximum acceleration determined by the actuator stroke.o bctuntor byaw X r . yet inexpensive. the closedloop frequency response of of the control system determines the reproducing a complex accurately Icapability acceleration time history.0 0 o2 X$o.
0 06 0.. end seismic evaluation programs have been completed for a variety of Class I electrical equipment. 01 2 0 comand . Hertz I I I I 30 40 60 Fig.0the o* 04.6 O b.. Fig. ance by comparing the time histories of the comand and output signals. The increase in amp othe 0 t 3 01 oacteristics 0001 00 litude required to compensate for filter charcan be reduced by using an increased rate of filter attenuation (sharper filter) in the control system or bypreprocessMtories Equgment Weight. 10 Maximum acceleration capability of seismic vibration table for sinusoidal excitation Figures 11 and 12 show the capability of the vibration table to reproduce a flooracceleration time history. Fig. ~~~p 6. .. constant amplitude acceleratiorn comoand signal . 11 show that the use of a highpass filter to attenuate lowfrequency components and the presence of some highfrequency noise makes it difficult to evaluate control system perform 17 LI .1500:b 05. However. The results In Fig. a comparison of the response spectra. Highactuatori frequency noise can be reduced by using an actuator with lowfriction seals and extending the control system bandwidth by improved compensation for the oilcolumn resonance. Horizontol Floor Accsleration (Al 2. . . i ing the original flooracceleration time hiswith digital computer filter algorithms ! ' I ~03 o 04 S3 requency. Horizontal Toble Accelerotion Fig. prior to obtaining an analog signal. 11 . and electrical performance of this type of equipment.. cc . 2  I 3 4 111 I 6 a 10 20 Frequency. 3. lb 1500 3000 I 0I 1. damping factors. The amplitude of the tabl e. . 13 shows a typical electrical equipment cabinet mounted to the seismic vibration table and being subjected to earthquake excitation in the fronttoback horizontal direction. . for flooracceleration command signal and the resulting table acceleration indicates acceptable agreement. above 4 to 5 Hz). Fig.0 [ 0.0 .e.al acceleration can be increiiid above the 0. . orti. 1971. 12.37g maximum accelerution for the DBE so that. The equipment is energized electrically during the seismic evaluation and all critical electrical performance parameters are monitored continuously.Comparison of horizontal table acceleration with DBE floor acceleration comnand signal SEISMIC EVALUATION RESULTS The seismic vibration facility hap been in operation at BattelleColumbus since February. the response spectrum of the table motion exceeds required spectrum at all frequencies where any type of equipment resonance might occur (i. . or ri. 9 Table acceleration response to sinusoidal. Results from this work are cuumrrized briefly to provide equipment designers with some typical data on the natural frequencies. o o Curve A Equipment Weight. ..
Equipment Size. i base 03o 02 02 0376 0 02O 0 1 b 1 . . 6 36 36 60 36 30 60 24 24 20 24 30 30 90 qO 90 56 75 75 2350 700 3500 1200 800 2000 4. However. 13  Time exposure of typical equipment during horizontal vibrations The data in Table 1 indicate that it is unrealistic to expect equipment of this type to be designed to have horizontal natural frequencies above the maximum excitation frequency for earthquakes (about 20 l1z).. Therefore. 12  Comparison of vibration table horizontal acceleration response spectrum with floor response spectrum for DBE i  : •.r3l 40 60 litudes from accelerometers located at the and near the top of the equipment cabinet.. .5 10. .. .. 11. 12.requires 20 30 2 3 4 6 Hes 6 0 SI 0at.. >25 >25 10 7.1 . in.ed by comparing the acceleration amp0540esonso046 20% rCo. . ::< _. 5_ Fig. lb Natural Frequencies..i 1100 o eqpmtft e.5. .0 9. :. Wt. Natural frequencies shoikld be increased as much as possible during design by locating heavy compor. evaluation procedure which is based on excitingthe equipment cabinet with sinusoidal vibration only at natural frequencies has the risk of missing a resonance of some small component that could cause an electrical failure.5 20 >25 >25 .' ?s. While this procedure is satisfactory for determining the major structural natural 4 frequencies of the cel. .. .VIBRATION CHARACTERISTICS It is a general practice to conduct a brief vibration survey of each equipment in order to identify major natural frequencies pr'or to evaluating its electrical purformance inder specified seismic conditions. .. . 2 3 4 5 . 10. 5.. Fig.. . 1 lists the natural frequencies which have been measured for a variety of electrical These natural frequencieswere determi..5. . . TABLE I Typical Natural Frequencies of Electrical Equipment Cabinets Item No. Hz Horiz..eonse requipment. Equip. experience indicates that it is even more important for designers to consider the details of the mountings used for the components inside the cabinets.. . I F040i0tnc.. the identification of local panel resonances and resonances of components mounted inside the cabinet more extensive an measurements than are practical. . .5 11. ..5. d. .en mounting bracket for a transformer weighing about 100 pounds is the only structural failure of any type that 18 4 I.qhl • oTable . 7..5..5. . A brv.5 9. ... Vert.10 40: ~unit 1° jO To*I. ..ents near the cabinet base and sizing the structural members at the cabinet base to transmit the inertia: loads to the mounting bolts without exceesive deflections.
has been observed. However, even this failure occurred only.after extensive exposure to sinusoidal excitation at resonance caused a fatigue condition that would not be expected froman earthquake, The data in :able 2 show the apparent magnification factors and corresponding critical damping ratios measured while a typical equipment cabinet was excited at resonance at different acceleration amplitudes. The selection of an appropriate damping factor is oue of the most difficult and important judgments that must be made for any type of dynamics study. This type of equipment usually has damping factors no lower than about 3 percent, and in most cases the damping is in the range of 5 to 10 percent of critical. However, the increase in damping with vibration amplitude shown in Table 2 can be expected for this type of equipment, indicating that nonlinear effects are quite significant,
CONCLUSIONS The resultw'which.have been achieved to date indicate the'feasibility of using the actual flooracceleration timehistoriespredicted for a DBE to evaluate electrical.equipment performance. The useof tlhstype of realistic excitation substantially reduces the possibility of overtesting or undertesting the equipment. Furthermore, the vibration evaluation portion of a sismic evaluation program can be reduced substantially, since a preliminary survey for natural frequencies is unnecessary and the total exposure to the earthquake environment can be limited to about 60 seconds in each direction. It is expected that the appropriate acceleration time histories can be supplied by the nuclear utilities once the advantages of using a more realistic earthquake simulation are recognized. However, there aresome cases in which the use of an artificial earthquake signal has some important advantages. This type of signal can be generated using digital or analog techniques to modify wideband noise with appropriate filters to produce a timehistory signal having any desired response spectrum characteristic. Battelle's digital computer facility has been used to generate an artificial signal having a relatively flat (constant amplitude) acceleration response spectrum over the frequency range of 2 to 20 Hx. This type of excitation is desirable for evaluating equipment that will be installed in several different nuclear power generating where the seismic requirements are similar but where the particular time histories and response spectra depend on the different building natural frequencies and location of the equipment in the building.
TABLE 2 Effect of Acceleration Amplitude On Damping or ATpcraloElctritce Damping For A ypical Electrical E Equipment Performance Equip. Equip. Base Accel., Acce.* Top Accel., cc8 Magnification Factor F 11 8.4 6.7 Damping, Percent Critical C 4.5 6stations 6.0 7.5
0.11 028 0.25 0.39
1.2 2.1 2.6
0.67
2.9
4.3
11.6
ACKNOWLEDGMNT
*  Constantamplitude horizontal
sinusoidal vibration at 8 1z natural frequency.
EQUIPMNT PERFORMANCE It is encouraging to report that most of the electrical equipment that has been subjected to vibration simulating the DBE for nuclear power generating stations has performed without any significant design changes. No examples of cabinet structural to the main structural members or holddown bolts have been observed. The few electrical failures that have occurred were usually caused by highvoltage arcing or relays which malfunction. Failures have been observed in some meters which are often mounted on a flexible door panel, but a failure in this type of component seldom effects the primary functional performance of the equipment
The authors are grateful to the assistance of Mr. Charles Rodman and the other members of the Mechanical Dynamics Division at Battelle who have contributed to the design and operation of the seismic vibration facility. They also wish to express their appreciation for the cooperation and encouragement of Mr. Julius Tangel of the Public Service Electric and Gas Company.
(satisfactorily
REFERENCES 1. John A. Blume & Associates, "Summary of Current Seismic Design Practice for Nuclear Reactor Facilitios", United States Atomic Energy Report TID25021, Sept. 1967.
1'damage
2. R.D. Kelly and C. Richman, "Principles and Techniques of Shock Data Analysis", SVM5, The Shock and Vibration Information Center, Naval Research Laboratory, Washington, D.C., 1969.
19
3.
V.H. Larson, "The Control of Acceleration by Electrohydraulic Shaker Systems", 145 Systems Corp., Technical Bulletin 840.001.
L.H. Geyer, "Controlled Dampin Through Dynamic Pressure Feedback",, Moog, Inc. Technical Bulletin 101.
4.
20
S..'
V
t
J%<AC
~I
A.m
4
~~
4,~
"~
Y
.
SHOCK INPUT FOR EARTHQUAKE STUDIES USING GROUND MOTION FROM UNDERGROUND NUCLEAR EXPLOSIONS* D. L. Bernreuter, D. M. Norris, Jr., and F. J. Tokarz Lawrence Livermore Laboratory, University of California Livermore, California Comparisons are made between the ground motion from earthquakes (EQs) and underground nuclear explosions (UNEs). It is shown that peak glevels and response spectra form a reasonable basis for comparison. Several approaches which attempt to characterize the time history are also discussed. It is shown that (1) the peak glevelsfrom UNEs have magnitudes comparable to those estimated for the strongest EQs, (2) the spectra from UNEs are similar to those from EQs, and (3) a time history comparison shows that both the duration of strong motion and the number of nearpeak glevel cycles for UNEs fall within the'range established for strong EQs. Based on these results it is concluded that ground motion from UNEs can provide an EQlike environment for testing fullscale structures. INTRODUCTION The recent San Fernando earthquake (M = 6.3) provided the first real test of earthquakeresistant design as practiced in California. Many structures failed the test. This earthquake caused 64 deaths and an estimated $553 million damage. The major cause of death was catastrophic collapse of multistory buildings. Earthquakeresistant design criteria used in California building codes are based on static approximation of the dynamic loads produced by earthquakes. These criteria require that the structure be capable of carrying a set of static lateral loads whose magnitudes and distribution are chosen so as to approximate the effects of the dynamic loads an earthquake might be expected to produce. In view of the damage caused by the San Fernando earthquake, it is clear that this static approximation method isinadequate, The design of structures to withstand EQ (earthquake) forces requires a more detailed theoretical approachone involving thorough dynamic analysisto guarantee basic survivabTlity. Significant capability for such analysis now exists with the computers and software presently available. Computer methods produce a dynamic analysis by idealizing the real structure into a mathematical model and then determining the response of the model to some prescribed ground motion. However, due to the lack of sufficient experimental data, current methods of formulating mathematical models of real structures are not yet developed to the point where they can readily produce accurate models for most structures. This is true even for very low level seismic input, as shown by Tokarz and Bernreuter [1). It has been very difficult to correlate results obtained from analysis with actual dynamic behavior of fullscale structures, principally because not much detailed data exists on actual dynamic behavior of structures during earthquakes. Dynamic tests have been conducted on some largescale structures in order to develop more accurate mathematical models [2J. However, because of the power and mass limitations of existing vibration generators, the amplitudes of vibration in these tests have not been large enough to cause substantial inelastic deformation. The dynamic response information is therefore valid only for small amplitudes of vibration. Since substantial inelastic deformation usually occurs prior to catastrophic failure, it is most important to treat the vibration problem at the relevant amplitudes. This highly nonlinear behavior makes it extremely difficult to apply the data that can be obtained from model tests [21. The cost of building a shaker facility large enough to test fullscale structures has been estimated to be approximately $20 million [31. Even then the size and type of structure would be severely limited. An alternative approach would be to use the ground motion from an underground nuclear
Work performed under the auspices of the U. S. Atomic Energy Commission. 21
?4
explosion (UNE) to excite properly located structures to EQlike ground motion. Several
major advantages of using the ground motion (2),there is no limit on the size or type of
In addition, no clear relation between EQ magnitude and peak glevel has been established.
Ibis is shown in Figs. 1 and 2.
from UNEs are: (1) the ground motion is available free as a byproduct from nucleartests, structure that could be tested, and (3) true soilstructure interaction would be achieved. The ability to test any type of structure and the achievement of true soilstructure .
Figure I gives Itshould be
a correlationof glevel vs distance from the closest point of observed faulting. The EQ
magnitudes range from 5.5 to 8.3.
noted that the peak glevel from the San Fernando EQwas around 1lg. Figure 2 gives
a correlation by Houcner (111 relating EQ m agnitude to peak glevel and range. Also shown in this figure are the Parkfield EQ (M = 5.7)
interaction are important. For example, underground nuclear reactors and storage containers can be tested. In fact, for marr types or large structures this technique of subjecting the full. scale structure to ground motion from a UNE seems to be the only practical way to investigate the structure's response to an EQ. CHARACTERIZATION OF EQ GROUND MOTION The feasibility of simulating EQ ground motion with UNE induced ground motion depends on the similarity of the two phenomena. The mechanism of energy release is much different in a UNE than in an EQ. There are considerable similarities and also considerable differences between the induced ground motions depending on the criteria of comparison. The choice of criteria will depend upon the purpose for making the comparison. For example, criteria used by seismologists to compare UNE and EQ ground motion from the viewpoint of seismic detection of underground testing [4] are quite different from criteria used by structural engineers whose basic concern is the response of structures. In order to show that UNEinduced ground motion is similar to that from EQs it is necessary to establish a means of properly characterizing EQs from the point of view of structural response. Many studies have been made to develop the best means of characterizing EQ ground motion [5101. However, this has proved difficult to accomplish, one reason being that only a few records of strongmotion EQs exist (here strong motion is arbitrarily taken to be a peak glevel greater than 0.1 g). These records show a large variation in pulse shape, time of due'ation, peak glevel, and number of cycles. Some of the complex causes of these variations are discussed in a paper by Trifunac 181. The most generally used criteria to characterize EQs for structural studies are (1) peak glevel, (2) response spectra, (3) Fourier spectra, (4) power spectral density, and (5) time history comparisons. None of these criteria are completely satisfactory. Peak glevel has the advantage of being extremely simple to compare. It is also physicclly meaningful. Nevertheless, peak glevel alone is not adequate to characterize EQs (7].
and the San Fernando EQ (M
= 6.3).
seen, these clearly do not fit Housner's projected correlation.
As can be
The response spectrum has been preferred for structural engineering studies of strongmotion earthquakes, because it combines both the representation of the exciting force and the response calculations. It thus lumps together under one representation the major parameters of interest to the structural engineer. The major disadvantage of the response spectrum is that only peak response is determined. The numbei of nearpeak response cycles, which are important for studying postyield behavior of structures, is lost. The Fourier spectrum of an input function shows directly the significant frequency characteristics of the function, and from itthe timehistcry response of the system can be computed. However, in the study by Jenschke et al. [51 it was found that results obtained using the Fourier spectral method were not satisfactory: the sine and cosine Fourier transforms showed high irregularities of the same order as the ground acceleration function, and no correlated characteristics could be found. These high irregularities make it difficult to use the sine and cosine Fourier spectra for comparison purposes. Hudson [6] showed that the Fourier amplitude spectrum is much more regular and is very closely related to the relative velocity spectrum. It has no advantage over the relative velocity spectrum and in fact is inferior in that the peak levels may be lost; timehistory aspects are also lost. The power spectral density apprcach is attractive in that it allows a probabilistic approach to be used. However,. Jenschke et al. [5] found that this approach was inadequate since ground motions produced by EQs and UNEs are essentially nonstationary phenomena, even for dynamic systems having natural periods considerably shorter than the duration of the ground motion. This is a serious problem when dealing with the ground motion from UNEs because the duration of shaking is quite shortabt.ut 5 to 15 sec as compared to 30 sec for the El Centro EQ. Several investigators have made attempts to generate artificial accelerograms using various nonstationary processes. But as pointed out by Trifunac 181 the models generally
22
L kp..
Q
'AAA<C
,t.

.~..

IS. Both the this behvio ofstrctues Nosuc chracerzasuggestion and the previous one (total time above a certain glevel) give some useful information. No such characterization exi5ts.1). CasioRica lB8Nov. 196?(M * 4.. Perez 010] From Cloud and i Sfr used are too simple.ld Calif. o. Calif. Puget Sound. Only some accelerograms from the 1040 El Centro EQ have been studied in this way. 6.) 1 Distance miles 00 Fig. Direct comparisons are usually not madetimehistory except to obtain qualitative insight into the nature of the ground motion. Utah 30 As.7). hence at the present time this approachwhile interestingdoes not offer a meaningful characterization of EQs. Some generalized timehistory characterization of EQ ground motion is needed to study the postyield behavior of structures. et. 1 1933(M 6. but they do not properly characterize exst. 27J. Ltuyo Say.\ I8. Calif. 1964 (M c 8.5000i £ I I a 1 ! 1 1 1 1 1lol Pacoima Dam (San Fernando epicenter distance) a Boxcar Mississippi (a) Lo (a/g) 3.1 1I Pld. Sishop.6).. 1945 (M 7.g.C 10 • iI. Oregon 6 Nov.4).8). Portland. Hegnloire.7). It is clear that no adequate single characterization exists. 21 July 1952(M 7. Aloslr 10 July 1958 (M *7. Kern Sounty.02og (D +43) 8 \ . \.one 19(6M. 7. 1940 2.\ Jbt. 30 D¢c. 0. Cloud and Perez 1101 suggested comparing total time the acceleration y1he was above a certain glevel Figure 3 shows such a cmparison for the Parkfield and El Centro EQs. Logan.1 27. 1949(M "7. I8 Maky 3. It is not at all evident from this figure that the El Centro EQ was by far the more damaging. IMar.5). 6. Limo. Trifunac (81 suggests using the response envelope spectrum which is a threedimensional plot of magnitude of response of onedegreeoffreedom oscillators vs time and frequency. 19. El Contre. 1950 22.0) 16. 5. Other investigators suggest counting number of cycles of a given glevel. 1962 (M * 5.sontlag.8). ugese lod tion~~~~ copain toa ieteaclrto a bv the time histories relative to postyield behavior of structures. Based on this discussion we plan to establish a similarity between EQ and peak UNE glevels ground motion by showing that () the from UNEs are within the Izaionof23s range estimated for the strongest EQs. 1945 (\ 0.1). Son 23. Longkoch.3). Wash.7. 195(M. This results from a lack of motion knowledge of the actualAt character of strong accelerograms. •6 20.). 1966 (M *7. Mjsxco I I May 1960 (M . 9.01 oi . i .\ (M 7. and (3) the duration of ndPeez11] h bv icsinsmaie h p gd . (M 7. SonJose. \ 6 6. Moto 17A 09 . Costa Rica 5sOct.7.. El Centro. 7). PrinceWliam Soun. 1  Plots of glevel vs distance from closest point of observed faulting. . Peru 17 Ocr.5). 947 (M. (2) the response spectra from typical UNEs aresimilar to those of strong EQs. Me. Alaka 27 Mar. 13Apr. 1939(Me 6. 8. Calif. the present time these investigations have not progressed enough to make use of here. Calif..2). lOApr'.ico City.Calif.52 log (D + 80)_ 00Lg 10 (a/g) =3.2). above discussion summarizes the approaches generally suggested to characterize EQ ground motion for structural response studies.
01 1 0.I 1 .1 0 0. 10 sec Duration of acceleration Fig. For ranges less than I DOB the valke plotted is the initial acceleration.0. relative velocity. 0. These izclude relative displacement.14]. (3) distance from surface ground zero (range).*% 0 X7 Nv 7 Various types of respobse spectra can be generated. California " Station 2 by glevels fall within the range established major EQs. [12]. The most comprehensive dircusslon of •0.. I 0. It would be undesirable to locate test structures within this spall regime.0 NI Pokfield (M=5. See Fig. Response Spectra romparisons 0.is a plot of peak glevel versus range for "typical" nuclear explosive yields of I0.2 * 1._.. Since all of the 0 L. It is clear from this figure that the peak glevel of the strongest EQ can be easily duor exceeded (if desired). For a given yield the peak surface ground motion can vary by a factor of 20.4 =8 0 % .2 0. . For low assumed values of viscous damping (less than 10% critical) all of the " above spectra are related 16]. Hous ] (IlV From IN 650 0.2 " u 0. 100.2 " surfacc ground motion from UNEs Inthe strong motion regime is that by Bernreuter et al. .. (4) geology around the explosive. (2) the factors [121. To extrapolate this data to closein distances leads to gross errors.. goo spectra to be considered for comparison have low values of damping.. 20 ~ 40 ~No 60 so standard EQ spectrum has been pnerally accepted for comparison purposes [71 Distance to fault .1.the lower layers. .4N N S r a 0.. and absolute acceleration spectra.. . I for estimated EQ peak glevels.. But since devices of a given yield are usually buried in similar geologies (and at similar depths) it becomes more meaningful to talk about general curves of peak glevel versus yield and range. 3  Comparison of Parkfield and I Centro earthquakes 24 .7) a San Fernando (M=6. only the velocity spectrum will be used.sec () 10 * 0.1 Duration of acceleratlion . These include (1) depth of burial (DOB). and (5) geology through which the wave Is transmitted. and1000kt(kilotons). 0.miles Fig.1 0 .6 E "r 0. In this regime.8 _ *plicated.. .. El Centro. the motion is such that the top several hundred feet of earth spalls away from.3) Figure 4.Plots cf earthquake magnitude and peak glevels vs range. 2 ...4 N (a) Parkfield.5 0. by a UNE.3 Vt ._ . often causIng a larger peak'glevel when the spall gap closes. COMPARISONS OF UNE AND EQ GROUND MOTION Peak gLevel Comparison Most of the published correlations of UNE ground motion data deal with peak glevels below those of interest [13. California . The magnitude of the peak glevel at the ground surface from a UNE depends on several yield.01 0. 0 .. .
However. onestory. these are the spectra most used for design studies. 5  2 sec Comparison of E1 Centro. 5 andi 6 shows even greater differences between the' same EQ at different stations than between different EQs. Spectra of the closein ground motion from UNEs have not been extensively computed or studied. t (although several have been proposed).09 m % % Parkfield i % t 10 o •! 0. Figure 6 shows two stations. There is substantlal variation in spectra between different EQs and between the same EL recorded at different stations. N21E) spectra. for closein survivability.28 El Centro EQ (48 km) 70. NS).000 Range from surface ground zero  100. This limits (at this time) our choice of UNE ground motion for comparison. N8OE). and Taft EQs 25 . This is because most permanent NTS structures are qimple. Figure 5 shows a comparison of the abovementioned EQs.000 ft Fig. A comparison of Figs. Spectra of the UNE ground motion recorded off NTS have been extensively studied [15. Olympia (1949. the peak glevel and change in kinetic energy are the important parameters. extrapolation of these studies to locations much closer to ground zero Is not valid.5 g.16]. Olympia. 3.n 10.1 0kt 0. Comparison herein will be made relative to the El Centro (1950.10 .Peak radial acceleration for typical UNEs as a function of range for several yields 6. both having a peak acceleration level of 0. strongly built fieldtype structures located out of the very strong motion regime."•. and Taft (1952. 4 . Olympia EQ (72 km) e 60 *6 40 0 20 " 10/ 0 "Taft EQ (64 kin) 2 3 0 I Natural period Fig. Also. from the EQ.14 2.
Cholame Shandon Array. and 1.74 (5] 213 . The accelerograms of the EQs and UNEs shown are quite different. .98 0.40 30" : 20 _ 1 1 Table I lists four events for comparison Parkflstd EQ. A Study of Figs. 6 .The duration of strong shaking for these EQs was Z0. The number of nearpeak glevel cycles of ground motion ranged from 3 for the Parkfield EQ to over 10 for El Centro.. [5] [51 (19] (171 Saturated tuff overlaid with dry tuff Tuff Tuff overlaid with deep alluvium 0. Spectra Event Mississippi (a) Mississippi (b) lialfbeak .46 purposes. 3. 2 N65E 270 ft from fault Peak glvel. Nevertheless. 740shows that the response spectra from typical UNEs are similar to those from strong EQs. 60 50 . and Fig.3 miles from fault Peak glevel. 10..Response spectrum from Parkfield EQ (June 1966) at two stations TABLE 1 Sources of UNJ. the spectrum was computed from a velocity transducer.Tlanca Aardvark 300 19 38 Yield (kt) 20200 Geology 'ruff overlaid with deep alluvium  Recording Rango Peak (it) gLevel 3280 6560 7000 5310 3600 0. and Parkfield EQs. 5 N85E A Sconditions. (No accelerogram was available for Halfbeak.5 .5 sec.*. *7 4.  ~ "L  777. Station No.) The duration of strong phase motion (peak glevel greater than 0. Figure 12 shows the accelerograms from the El Centro.' Station 1o. Mississippi. Carder and Cloud [17] also noted this similarity between EQ spectra and UNE spectra. and Boxcar (1.. These events were chosen since they cover a wide range in yield and geological Figure 7 compares Mississippi (a) and El Centro.45  Ref.1 g) is approximately 2 sec for Aardvark and over 7 'sec for Boxcar. and Olympia. TimeHistory Comparisons Figure 11 shows accelerograms from three. Furthermore. ~. Figure 8 compares Mississippi (b). k00 E m Parkfield EQ. respectively.. Cholame Shandon Array.. 10 compares liallbeak and El Centro.4. 0. San Fernando.5  J 0 1 Period 2  4 ec Fig. Aardvark. the variations between the UNEs and EQs are no greater than the variations befween the EQs themselves. *  .2 Mt). namely Aardvark. as shown in Figs.UNEs. Figure 9 compares Blanca and Taft. 5 and 6. the duration of strong motion and the number of cycles of nearpeak glevel for the UNEs fall 0 1 2 3 4  [ 80 C 60 60"* 4020. This figure shows in addition that the duration of strong motion from a UNE increases with larger yield and also with increased distance from ground zero. 0.
5 R "10 .Mississippi and Aardvark UNEs vs Olympia EQ. 8 . 0. Olympia EQ (72 km) Aardvark UNE (I...I' Mississippi UNE (Ikin) . Relative velocities normalized by dividing original velocities by 0. 0 0 2.1 g 1.5 10 .. ~ El Centro EQ (48 km) 0 Fig. 80 t20 0 20 40 60  80 rod/sec 100 120 Undamped natural frequency r'ig..u.' 40 6 \. I km). A • Ij. 7 20 so 60 40 rad/sec Undamped natural frequency  100 12 Mississippi UNE v El Centro EQ. original velocities by 0.1 g Relative velocities normalized by dividing 27 . .ii 2 ~ ' iv .
and Boxcar UNEs. .I~~~~ _ _ _ _ _ _ _ _ _I_ 0 80 1. I I 3 4 f Fig. ground motion with longer pulse durations and also more cycles of nearpeak glevel.4 Cs. ? to tho~se estimated for the strongest test structure). a test structure can be excited HQs. 4 2 Fig. the ground motion from any of these UNEs could be considered as being representative of a future major EQ. closein ground motions from UNEs are simFrom these results it is concluded that liar to those produced by major EQs. El Centro EQ .2 0.. SUMMARY 5 6 7 The. 9   sec O Blanca UNE vs Taft EQ 0. The oreoingcomarisns how hatnearpeak glevel cycles for UNEs fall within 7be oreoingcomarisns how hatthe range established by strong EQs. by grouand motion comparable to that of a future major EQ. the envelope However.e.2  ~80 * ~90 Halfieck UNE0 c7 0 k5 840> 0 1~0 0 ~ I Af ~ .6. (2) The upectra !rom typical UNEs are slnr~ir to tliis.2~r EQs (i. This (3) A time history comparison of ground can be accomplished by sequentially firing motion from typical IJNEs shows that both the IJINEs or by subjecting test structures to ground duration of strong motion and the number of motions from UNEs fired over a large time . It is therefore noncluded that from the standpoint of a time history comparison.. Taft EQ (N21 E) 00 2Og4erlanca UNE 0. Mississippi. to study postyield behavior of UCVelOpcir ypical UNE spectra would structures it will be necessary to generate UNE match an envelope from strong EQs).. issip N *J0. by predetermining the yield and (1) T7he peak gleveis from UNF~s are location of UNEs (relative to the location of a compar. More typical UNEs ground specifically: lent to that of generate a possible futuremotion major equivaEQ. Therefore.60 * ..41 I1 I 1 2 3 4 5 6 1 7Natural 2 period sec UNE vs El Centro EQ 3 0. Natural period Fig. 28 .see 8 9 10  Accelerograms from Aardvark.6 0 0.224. from the structural engineer's viewpoint.2 __________ Aardvark UNE .<2 V' 0.0 0 0.. 10 lialfbeak within the range establisHed by the EQs.
51840.allysevere damage. 2 0 0.4 0. One important fact to note is that lower yield explosives can be used. 2026 PSTCholme Shandon No. In order to properly locate structures a study must be made to correlate the ground nuction spectra from with yield. span.80. This list indicates that more than a sufficient number of UNEs are exploded every year to supply the required ground motion for any type of test program. 29 .sec Fig. and San Fernando EQs.0. This is of some importance since a large number of lcwer yield explosives are detonated as compared to the very highyield explosives.2 San *~1. Once this is accomplished it would be possible to subject test structures to increasingly strong motion ranging from elastic response to fi. Both of these possibilities need much further study. 2037 PST Strong motion 24 26 2830 I 0 4 6 8 10 12 14 16 18 20 22 24 26 28 30 I 1. The only reason for using very large yield explosives is to increase the duration of shaking. _V PA. Springer a' Kinnamannj 181 give a list of U. El Centro.6 Fernando EQ Pacoima Dam S160E 4 0.4 0. geology.2  0. % 2 6 8 10 12 14 16 18 20 22 El Centro EQ. UNEs.2  J 0.2 Parkfield EQ.00. 1 &A% .2 .8 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 Time . 12  Accelerograms from Parkfield. 62766. and depth of burial ofUNEs the explosive. S.4 0.
and D. "Response Envelope Spectrum and Interpretation of Strong Earthquake Ground Motion." Bull. 2. Kinnamann. Perez. 16091632. Am. 1969 12. Vol. Murphy.. on Eng. A2119A2132.J. 4." Environmental Research Corp. 3rd World Conf. 111125111142. Soc. I.. 1970). E. 181194. R. IIN201039. 9. 417430. Clough. 15171534. Earthquake Eng. Stevenson. Trifunac. R. EERC671. 4th World Conf. pp. W." in Proc. E. Atomic Energy Commission Rept. "Some Problems in the m robems to 6. "Dynamic Tests of Full Scale Structures. Underground Nuclear Explosions 19611970... Am. P. 1971 S. Liu and D. Vol. 30 . Earthquake Eng. "Control of the Dynamic Environment Produced by Underground Nuclear Explosives. Am. III. N. Murphy and J. K. Wiegel. "Seismic Wave Propagation. Miller. pp. 3. 1969 R. Earthquake Eng. Sept. R. Earthquake Eng.' Earthquake Engineering Research Center. Selsm.." Proc. 52. 1969 W. pp. Vol. Hudon o Application of Spectrum Techniques to StrongMotion Earthquake Analysis. "Feasibility Study of LargeScale Earthquake Simulator Facility." Bull. "Intensity of Earthquake3 Ground Shaking Near the Causative Fault." pp. No. and A. D.S. Evernden. Penzien. pp. 3rd World Conf. Cloud and V.J. 19. 3rd Plowshare Symposium (Davis. 1960 D. A. Atomic Energy Commission Rept. Am. Cloud and D. 59. F. fludson. Inc. W. F. California. J. 1964 R. 93106 in Earthquake Engineering Prenticeflall.. 1969 G. Jackson. 38283856. February 1971 If. Bernreuter.. 1970 J. Lynch. Jhaveri. U. Englewood Cliffs. and J. Vol. Reg. Seism.. 1. 13. 11194111115. E. K. pp. A. Tokarz and D.... "Ground Motions Generated by Underground Nuclear Explosions. 59. 14. Soc. Bouwkamp. pp. (R. UCRL50977. 343356. G. Rea. Rept. PrenticeHall." in Proc.." Proc. U. Lahoud. Vol. 15.. C. 979993. I. Housner. V. 1964). University of California. Am. Dec.. 6. With Nuclear Explosives (Las Vegas. May 1970 J." Bull. Geophys. "Identification of Earthquakes and Exlosions by Use of Teleseismic Data. W. Berkeley. "Analysis of Seismic Peak Amplitudes from Underground Nuciear Explosions. Soc. 1967 J. C. Vol." pp. S. "Seismic Source Summary for U. Nev. pp. 1969 15. TID7695. pp. Mueller and J. W. 2. "Structural Response to Close. Seism. editor). Wiegel. Vol. "Response Spectrum for Pahute Mesa Nuclear Events. Vol. 2nd World Conf. 1970 D." Bull. Am." to be published in Bull. L. Springer and R. N. D. Seism. 1969 11. 1970 M. UCRL73036. 23.in Horizontal Ground Motion from Underground Nuclear Tests at Pahute Mesa. W. E. L. Hudson. 1965 D. S. S. pp. 1965 . pp. B." Holmes and Narver. Clough." Bull. Penzien. editor). 17. Vol. R. Vol. 127149 in Earthquake Engineering (R.udson. R. Seism. Nev. 16. 23252341. Rept. "Characteristics of Strong Ground Motions. D. Carder. Mickey. pp. 22952309.' J. Las Vegas." Lawrence Livermore Laboratory Rept. "Comparison of Calculated and Measured Response of a HighRise Building to Ground Motions Produced by Underground Nuclear Detonations. "Design Spectrum. "Seismic Spectrum Scaling of Underground Detonaop tos"EvrnetlRsac tions. 59. Englewood Cliffs. 10. "Spectral and Correlation Analysis of GrouredMotion Accelerograms." Proc." in Proc. V. CONF700101. D. Soc.. pp. Seism. 1962 7.REFERENCES I. Symp. "StrongMotion Records and Acceleration. 5. 8. Vol. 18. Soc. 74. Jenschke. Soc. Bernreuter. A. See also Lawrence Livermore Laboratory Rept.
Bernreuter: That was from a paper by Cloud and a coauthor in the fourth world earthquake conference. the acceleration levels in a particular earthquake may vary from place to place depending on the type of soil or rock encountered. University of Southern California has made studies tryiny to relate ampliflcations of base rock motions to soil structure. what was the origin of that. soil conditions in the surrounding areas. Haag (MTS Systems Corporation): I am intcrcsted in knowing if any one has done any work In relating the earthquake accelerations to the type of Mr. He was looking at how long the level remained above a certain value. Haag: Have they been conclusive inany way? Mr.how many seconds it remained above 1/2 g or 1/4 g for different earthquakes. Are you aware of any studies that have been done? Mr Berreuter: Yes. Mr. 31 . So there is some hope that this may be clarified. Voice Was it the time after the initiation of the event? Mr. for example. Bernreuter: Yes. Harry Seed. lUernreuter: No.DISCUSSION Voice In terms of the duration at certain amplitudes in those plots that you showed. There are also several other papers on the subject. Bernreuter: I do not really feel that they have been conclusive. Once the San Fernando records are studied in detail. This is the first time wbhave had so many recording stations for a given earthqtake over varied geologies. there have been quite a few studies. *yjjThen this was not the actual time he dwelled at any particular level? Mr. For instance. One main reason being simply the lack of recorded motion. one might be able to understand this better.data? Mr.
buoyant spherical pressure hull supported by framed legs. pressure Frictionhull between the pads and the underlying the shear resistance of the soil prevent sliding of the structure.ROCKING OF A RIGID. multiplied by appropriate envelope factors. The structure considered is an earlygeneration type that would be constructed out of the water and lowered into position on th( ocean floor. G. 1. and foundation reaction are considered. This paper describes a study and analysis of the behavior of a particular kind of structure under such loading. h Fig. The structure could look like the simplified model shown in Fig. but the structure tends to rock rigidly in a manner that compresses the soil beneath it and may cause lifting off. One of the most severe conditions. cr it could be a manned station complete with life support systems and means for ingressegress. added mass. California This paper describes in analytical study of an earlygeneration seafloor structure subjected to severe seismic excitation. would occur when there were heavy seismic disturbances in the ocean floor. UNDSRWATER BOTTOMFOUNDED STRUCTURE SUBJECTED TO SEISMIC SEAFLOOR EXCITATION J. Effects of drag. S. which in turn soil rest and on pads on of the asea floor. The ground excitation is a simul. Hammer and H. An example structure is analyzed and its response computed for three different support conditions. A numerical methud of solution is followed that generates solutions at successive time intervals that are automatically adjusted to conform with preestablished accuracy criteria. and one that is not too well understood. possibly with additional fasteners to prevent sliding. INTRODUCTION A structure placed on the floor of the ocean must be adequate for the loading conditions expected during its useful lifetime. As envisioned. unmanned scientific or observation station. The structure Is assumed to consist spherical supported on legs. The problem is formulated by second order differential equations containing nonlinear terms. The footings would rest on or in the bottom material. 1  Simplified structore Preceding page blank 33 . which in turn rest on footings. The behavior of the foundation depends on the vertical forces and the load history. The results are shown in terms of displacement and accelerations of the structure and shearing force and vertical reactions of the supporting medium. It might be a selfcontained. it would consist of a negatively. Zwibel Naval Civil Engineering Laboratory Port Hueneme.ated earthquake motion generated from a random process of prescribed power spectral density.
gravity forces due to the submerged weight of the structure. Forces acting on the structure are the inertia of its mass and the added mass of the water. These range from 0.aken of the uppermost few feet of the ocean bottom sediment indicate very low shear strengths. The predominant sediments of the Atlantic and Indian Oceans are calcareous oozes. The quantity of data from Pacific Ocean sampling. These latter two compositions occur in the deep ocean and are pelagic sediments. However. plus the small amount of rock outcropping previously mentioned. The porosity was in the range 60% to 80%. velocities.5 psi to 2. The portion of the ocean floor that is solid rock is believed to be a very small percentage. It is believed that below this depth gradual hardening of the sediment has taken place so that the layers are intermediate in composition between the upper sediment and the basaltic bedrock underlying it. The loading situation is as follows. The water content varies from 80% to 150% in the majority of samples.The structure is assumed to be in moderate to deep water. and the vertical downward displacement of the center of mass is given by y. Submarine soils apparently vary as widely as do terrestrial soils. the drag forces caused by the motion of the water relative to the structure. described by the holding forces to prevent sliding relative to the ocean floor and by the vertical forces acting on the footings. The horizoraal position of the center of mass relative to the base of the structure is proportional to 0 for small rotations since the structure moves as a rigid body.[l] reported that only 300 or so cores have been obtained from the North Atlantic. / T / / . The footings are proportioned so as to support the weight of the submerged structure without exceeding the bearing strength of the bottom material. seismic surveys indicate a sediment thickness of only 0. All of the motion is assumed to occur in the single vertical plane that contains the horizontal axis of movement of the base excitation. The core samples that hav been .is even less representative. for example. Its total displacements are the sum of the relative movements and the movements of the supporting medium. 2. red clay deposits consisting of volcanic and terrigenous colloidal matter. terrigenic deposits that have come from rivers and currents cover the ocean floor on the continental shelves. The foundation requirements are 34 . About 82 of the area of the sea floor is terrigenic shelf sediment in an average water depth of 100 meters. [1]. with an being defined as a material containing than 30% organic material.greater   ' Fig. The ocean floor undergoes a violent horizontal shaking. horizontal translation.lem being considered is the nature of the material on which the structure is likely to be placed. The actual samples taken from the deep ocean floor are very few. The main drag effects of the water are assumed to be those acting on the spherical hull. PROPERTIES OF THE OCEAN FLOOR Probably the largest unknown factor in the prol. 2  Coordinate system The rotation clockwise from its initial position is given by 0. ___ __ I my °ooze .000 square miles. Ref. is that the floor of the Pacific Ocean is covered chiefly with elastic. [1] gives the following estimates as to the composition of the ocean floor. The mass of the structure is assumed to be concentrated at the center of the spherical hull. which averages only one sample for each 30. Based on this kind of rate. The remaining part of the ocean floor is believed to be hemipelagic muds. and the reactive forces of the ocean floor material on the footings. however. The spherical pressure hull is designed for the hydrostatic pressure at that depth. These ocean bottom sediments have accumulated over long periods of time. which considered about 250 reports and abstracted data from the most relevant of them. What is believed. which in turn shakes the footings and the rigid structure that they support. Ref. The position of the structure at any time is given by two coordinates as shown in Fig.5 psi. and displacements.5 kilometers. One of the most coinplete surveys of available data is Ref. The response of the structure is described by time records of accelerations. and vertical settlement of the footings. About 46% is an ooze and 28% is a red clay.1 to 0. the total depth of sediment would be expected to be a few kilometers. Closer to the land masses. The rates of deposit are of the order of a centimeter or less per thousand years. The system moves in a combination of rocking.
What can be done is to take a record of a strong typical earthquake and use it for study purposes as though it were applied to the footings of the structure. ...[3]. If a footing has settled into the sediment it will resist being pulled out because of suction formed beneath it. 2 . It is believed that the shearing resistance manifests itself in a way that offers passive resistance to slipping within the soil along some interior surface outward from the footing. When a negatively buoyant structure is 'placed on the ocean floor. . of course. ' >. = . The void ratio apparently does not decrease with depth.. In a specific location the capabilities of the bottom material could be compared with what is needed to prevent sliding. and the motion of the base itself. therefore. . It would be possible. however...1] had bearing strengths in the range 0. and the inertia force is F = 0N(hO + xb) 35 . given the properties of the soil material and the structural forces and geometry. its mass density. These breakcut forces are commonly experienced in anchor and salvage work. there will be a much slower. and it is therefore assumed that ignoring the moderating effect of the layers of sediment would not alter the structural response significantly if the bedrock input resembled a large land earthquake. to make a rough estimate of the horizontal restraint on a sunken footing. .. Lacking specific information about the sediment properties. and the velocity by a coefficient C. i The shear strength seems to increase almost linearly with depth below the watersediment interface. one can conelude that the material supporting the structure will be very soft with a low shearing and bearing. with the additional assumption that C is also proportional to the depth of the hole. . 3 shows the system of forces acting on the structure when it is displaced so that 0. . It is sometimes considered that the average foundation stress resisting pullout of a footing is comparable to the ultimate bearing capacity of the soil material. At the soilwater interface.. The center of mass is accelerated horizontally by the combination of two motions: a motion relative to the base as the rigid structure rotates. . Of interest is the fact that the peak accelerations of the movement do not change appreciably since they seem to be associated with the lower frequencies. This might indicate that the older. The analytical proedure established for this can be refined as information concerning sediment properties and depths and bedrock motions becomes available. is that one cannot predict what the disturbance will be. Poisson's ratio of the soil. This kind of passive resistance has been studied in a classical way. the size of the footing. It is believed that pullout finally results when a failure mechanism occurs in the soil material surrounding the footings.. Following this. elastic medium and derives expressions for a soil resistance function involving a linear combination of the deformation and the velocity of deformation. Sliding motion would be resisted by friction and by the shearing resistance of the sediment material. The total horizontal acceleration of the mass is thus h0 + xb. lower deposits have become stronger through age rather than by consolidation.5 psi.e. and to compute the horizontal restraint required during the response. isotropic. and the frequency of the periodic force that is loading the soil. and solutions usually used are those due to Rankine and to Coulomb (Ref.[8]). there will be some imnediate settlement due to an almost elastic distortion of the bottom material.. and was substantiated by actual experience reported in that study. and greater.. The deformation is multiplied by a coefficient K.[7J). The problem. Studies have been made of layered soil masses to show the effect on a seismic disturbance applied at the lowest layer and transmitted by shear from horizontal layer to layer until it reaches the uppermost layer (Ref. These studies seem to show that an input earthquake such as that recorded at El Centro in 1940 will emerge with similar low frequency components but with attenuated high frequency components.[51 assumes a homogeneous. K and C are functions of the shearing modulus of the soil. The thick layer of soft material overlaying the bedrock will tend to alter the nature of seismic disturbances occurring in the bedrock as they are transferred up to the sedimentwater interface. In fact 89% of the samples reported in Ref. The particular structural size ad type that is of interest here responds primarily to the lower frequencies. An equilibrium is eventually reached.[2].. Ref. y and xb are all positive and the motion is such that the three accelerations 0. There are several studies (Ref. even in the bedrock.Ti. it is convenient to assume that sliding is prevented. . and this condition is assumed at the time the seismic disturbance occurs. It is believed that the rate of loading has an effect on the static resistance of the soil mass. y and xb are also positive. Another possible motion of the footing (assumed to rest on or in a horizontal ocean floor) is horizontal sliding. [41) that have looked at the dynamic properties of soil under footings of structures on land. . settlement duc to consolidation of the material under the additional weight of the structure. Ref.[6] uses a similar soil resistance function to study dynamic field tests on a soilpile system.5 to 2. EQUATIONS Fig. This was assumed in Ref.
The total velocity uf the structure through twe stationary water is then h6 + . From the geometry of Fig. The quantity in the second parentheses is shown as an absolute value so that the velocity squared will have the same algebraic sign as the velocity. This resistive tcrque is The static resistance of the soil material is given by RS(A). RL and RR describe the static and dynamic resistance of the soil beneath the footings. There is an inertial resistance to rotation of the sphere about its own diameter. cross section presented to the flow. 3 b AL R  y 0  y + b0 2 These horizontal drag and inertia forces acting thrcugh the center of mass are opposed by a force FS acting through the base in a horizontal direction opposite to the algebraic sum of the inertia and drag forces.'_' *y Fig. where A.1/2 CD Tra2p ~h6 + )(h5 + 1)RR with H the mass.Force system (OyxbOyi positive) b where H is the mass of the structure and 8 is a multip)ier that incorporates the added mass of the water. 'Teyare the inertia force  10 yHF OMY 36 ? a ~ aaA~a' c ~ ~ a . The other vertical forces act through the center of mass. plus the base motion.b. and RD(Ali) is the additional effect caused by the rate of loading A and the depth of the hole A. The structure is considered to move through the water with a velocity resulting from the same two motions: A horizontal component of the rotation about the base. and the drag force will have a direction consistent with the direction of the flow. They are assumed to have the following form RL = RS(L) + t LAL) " RS (R i("R'4R) )+R 2 where CD is a drag coefficient. ra is the. and p is the mass density of the water. 3 . and the drag force acting on the sphere is FD . and A are the vertical movement of the fooings and AL and AR are the accompanying velocities. in which the resistive force depends on the vertical compression. and Y the proper coefficient so that YMa 2 is the mass moment of inertia of the spherical structure about its own diameter. a the radius.
(3) to find the time derivatives.i and . As far as possible the assumptions relating to the properties of the soil media are consistent with the reported data of Refs. the whole structure is assumed to weigh 418. ration whee x ar x an th acelraton nd . as previously explained. Haming's modified predictorcorrector method is used to advance through succeeding intervals of time. Summing the moments about the center of mass While Eqs. It is assumed that a steel structure Is to be placed on the level ocean floor at a depth of 6.y. If the spherical hull is designed for the hydrostatic pressure at 6. a number of assumptions were made as to values of the parameters.[91. After the RungeKutta pronedure has given points for four times and the corresponding first derivatives have been obtained.(i) and (2) can be transformed into four first order equations by making the following substitutions X X2 = Y/h AN EXA1PLE STRUCTURE In order to demonstrate the application of the procedure.[J1. The procedure used here is to begin a stepbystep solution using the RungeKutta method to predict values of each of the Xn coordinates at the end of successive time M2 M~b + _ ) R. it is X4 h 4g These equations can then be arranged in the form I f 1 (X3) 2 f2 (X 4 ) = f 3 kXI X2 'xy x 4 ) 4four assumed that the structure has four legs and 4 footings.and the submerged weight Eq. The particular subroutine used is given in Ref. nor was an optimum configuration sought.(3) and (4) cannot be solved directly. No attempt was made to make a detailed design of a structure. footings and interior loads. 1. with each footing about 12 feet square and having a bearing pressure of about 1. and angle 0 of 60*.b are the acneleration and valo city of the support medium. The initial conditions at time zero are Xl(to) =X2(to) "X 3 (t o) X 4 (t o ) " 0 (4) where Mg is the weight in air and a is a multiplier to Cive the weight in water.c~g  RL  RR where RL and RR are funecions of B. they can be handled readily by numerical methods. The values for any given time are 1/2 CDra20h(hi + hb)(h6 + :t) (1) (2) BH4y .000 feet and an estimate made for the weight of the legs. For symmetry.000 feet. 3 equilibrium equations can be written. It has the additional advantage that at each step the calculation procedure gives an estimate of the local truncation error. where b and . a base dimension b of 30 fect. intervala. (3) is a set of simultaneous nonlinear first order differential equations. Eqs. This method is a stable fourth order integration procedure that requires approximately half the calculations per step that other methods of comparable accuracy require.8 psi when resting in the static position.Il voe of of ater b onehalf e ate water of the the volume mass of the mass be displaced by the spherical hull. [2] and 17]. The structure resemb as that shown in Fig. From the force system shown in Fig. with a radius a of 10 feet. + h) O 2 hsubstituted into Eq. The drag The added mass of the water Is assumed to 37 . The procedure is therefore able to choose and change the size of the time interval based upon a preestablished accuracy criterion.000 pounds in air and 148. and a particular structure was analyzed.000 pounds In the water. and g is the accelof gravity.
CaacitY { limited bearing strength.n Z 6 in.000 lb/in. The coefficient c was asaumed to have a value rle i structuresoilwater system is then assumed to be excited by a horizontal motion of the base resembling a strong earthquake on land. The mass moment of inertia of the structure about a horizontal diameter of the spherical hull was assumed to be 2/3 Ma2 with no added mass effect when the sphere rotates. It is possible to model the effects of the sediment layers on the motion passing through thcm. stronger than that for which recorded data exist.000 lb/in.000 . It follows the pattern of Fig. Type (a) material was assumed to have un Types (b) and (c) were assumed to have an ultimate bearing strength three times the static bearing pres(b) Soft eiatoplastic bottom sure.Static resistance functions assumed for seafloor material soft material that deformed elasticially under the dead load of the structure but did not consolidate. This obviously will not be the same motion as occurs in the bedrock below the supporting material. The third type deformed initially under the dead load both by distortion and consolidation. This static slope was assumed to be 6. for both types (b) and (c). When the seismic loading occurs. Wi/n.[10]. When loaded further it followed the pattern shown in Fig. The second was a numerically equal to onetenth the slope k of the static resistance. and of course one is not sure anyway of the nature of the movement in the bedrock. In type (c) material an initial settlement of . The first was an almost rigid racklike surface that has the static resistance function shown in Fig. The pullout strength of type (c) mater ial then becomes the force necessary to lift the footin a distance equal to the consolidation settlement against a resistance of 6. (c) Soft llterir exhibiting Fig. In type (c) (a) bottom was assumed under each footing. 2 . that is believed to contain all the characteristics of a very strong earthquake. 4(a). Three types of support media were considered.kA + cLA. and the coefficient c was therefore taken as 600 lbsec/in. 38 . Rigid material (Ws4)(l/K)  b" \ """'oo 4 A 6. 4(c). the footings will tend to rock or slide. The earthquake Is of 120 seconds total duration and is designed to represent an upper bound for the ground motions to be expected in the vicinity of the causative fault during an earthquake of magnitude 8 or greater. An additional property of type (c) material is that it offers a tensile type resistance to having the footing lift out. plus the addition velocity dependent viscous force. The input used in this example is an artificial accelerogram. ultimate bearin capacity pullout for. The above assumptions imply that the structure placed on type (a) material will not settle. 4 .coefficient is assumed to be unity. r barltin uliabearinh material an initial settlemint of 6 inches was assumed due to elastic distortion. except perhaps for material type (a). but indications are that the layers of sediment will have a small effect on the frequency component that affect this structure the most. but this could be done if more specific information were available about a particular structural site. Using the base motion as though it were applied to the base of the structure is therefore not considered unconservative. taken from Ref. 4(b). but will not depress the supporting medium. Types (b) and (c) are assumed to be velocity dependent and to have a resistance function of the form R . The type (a) bottom material does not change with rate of loading so its resistance does not depend upon A. and an additional settlement of 18 inc'ies was assumed due to consolidation.
0. In Fig.6 Ig h"Iip 10 t . Fig. having the same support materlal. both as a function of time for the first 80 or 100 seconds of the earthquake. Undesirable frequency components were removed by filtering again. The same structure responding in air to the same excitation. The total horizontal 39 . b(a) As the response if tie supporting material is type (a). 7 shows that the structure is settling into a i" h1k& . and Fig. In Fig. This can be seen from Figs. An approximation to white noise was passed through a filter to give the process the desired frequency content as determined by the power spectral density. Finally the accelerograms were scaled to the appropriate intensivities of shaking. 6 shows the displacement xr of the center of mass of the structure relative to ith base and the angle of rotation in degrees. which are plots of the force exerted by the left and right footings as a function of time. The maximum lift at 0 .7* is about 31/2 feet.M 25 A 5M 4 L.j .0 means a footing has lifted off. It instead moved horizontally with the ground. none of the footings broke loose to lift off. 6(b) and Fig. The. 6(a) any departure from 0 . In this particular case. It is interesting to note that the reaction force can momentarily exceed the static bearing capacity becLuse of the dynamic effect. IN'' . for this case on both sides.  " . W . did not lift off at all. . and canted position even though the foundation reactions are fairly balanced.[i0] followed the following procedure.results in Figs. however. 5 . The record was then processed and corrected to filter the spurious long period componets. Response spectra were then calculated and compared to those of recorded motions. Comparing the final portions of the curves in Fig. The resulting stationary Gaussian process was then given the desired nonstationary properties by multiplying by a suitably chosen envelope. In Fig. tx. 6(b) is for type (b). 6(c) is for type (c).. 6 and 7 do not include any sliding relative to the support material.Assumed base excitation RESULTS Fig.Ref. which are used as the input for this structure. Fig. 5 shows the resulting simulated records. 7(a)and 7(b). Fig. In computing the response the assumption was made that the structure was restrained from horizontal sliding. 6(b) a departure from 6 . Liftoff did occur. V Fig.0 does not necessarily mean that liftoff has occurred because the rotation may be due to vertical movement of the footings. 7(c) tie response is less because of the resistance to uplift of the footings.
2.0 I II Iq I I I I I I I 0 0 20 to w0 2 W ~ I of02 6(b) tYr9b) 1.0 2. 6  Response of sample structure 40 .0 2.0L 0 201 0 20 40 50 40 0 70 0 typo Wp Fig.0 2.
^.0 0t Fig.000 lb 111. It would appear that types (b) and (c) might develop these forces under the footings assumed if all the footings were acting. The Input acceleration of the base had a peak value of about 0.600 friction between footings and support material. Whether or not horizontal1 resistances of the m. The horizontal force would probably be shared unequally by the four footings if none lifted off.189 g for type (c). in 7able I could be developed is.ignitude shown . I 0 .^.400 104. and by perhaps two of them if liftoff occurred.202 g for type (b). so it appears doubtful that the required frictional force could develop.44 g that did not coincide with the peak response. possibly not exceeding 1 psi. so the problem would require reanalysis.ot certain.AA 1O 20 O0 hmAAAh AA. The structure on the type (a) materiel would have to develop its horizontal restraint principally through Type of Bottom Material (a) (b) (c) Maximum FS if no Sliding 130. The peak acceleration was C. These horizontal forces must be resisted by friction between the footings and the bottom material. The force required would be nearly equal to the submerged weight of the structure. It is certain that materials of type (b) and Finally. 1. and by the dynamic shearing resistance of the material. These fasteners would of course affect the tendency toward lifting off.41OV Q0 50 GO 70 O s0 00 t I V a I I I (coo) ' Z 2.250 g for the structure on type (a) material. and 0. 7  Vertical reactions on footings restraining force required to prevent slid'ng was calculated. of the structure. however. . 0. and the values were as follows: Table I (c) would have a very low scatic shearing strength. 41 . the peak values of absolute acceleration of the center of mass of the sphere were computed during the response time in eaci of the three cases. Some kind of shear fasteners might be used to hold the footings against shear.0 o¢ 20 3 0 50 6 0 4 lo 02.
H. May 1962. IBM System/360 Scientific Subroutine Package (360ACM030 Version III.F I." Research Report No. "Analysis of Ocean Bottom Sediments. "Simulated Earthquake Motions. Jul. E. a more realistic analysis could'be made. 1969. K. Schmid.i about a particular structure and location. W. Korb and H.. Housner and N. Richart. p. H.. Even such a simplified analysis. OTC 1471. T." J. T." J. Texas A&M Univ. 93. 1968. Feb. Ratay. Fourth Ed. Ilsiech. should be useful. Peck. Contract N6239968C0044 for Naval Civil Engineering Laboratory." Final Report. Programmers Ma nual. "SlidingRocking Vibration of a Body on Elastic Medium. Liam Finn and P. Idriss and H. of Soil Mechanics and Foundation Div. 1252. The predictions of peak accelerations might show the need for shock mounting delicate equipment. and if more accurate information were kno. 94. Vol. "Seismic Response or Horizontal Soil Layers. California Inst. The maximum displacement and final displacement might be useful if precise positioning were a requirement for the structure.. G. REFERENCES [I] W. The possibility of a failure of a footing in bearing or shear might be predicted and avoided. however. Proceedings of the ASCE. "Foundation Vibrations. H. John Wiley & Sons. Jennings. Proceedings of the ASCE. The method itself is not limited to simplified abstractions.." Earthquake Engineering Research Laboratory. 1967." J. "Penetration of Objects Into the Ocean Bottom. CONCLUSIONS The kind of approach and solution described here could be extended and applied to the analysis of a great number of situations. [2] [3] [4] [5] 42 . Apr. Tsai. 1968. (7] [8] [9] [10] P. C. E. V. K. of Soil Mechanics and Foundation Div. In some cases. of Soil Mechanics and Foundation Div. Nov. 1971. Texas Transportation Institute. Terzaghi and R. Vol. Seed. Mar. 97. 1971. "Design Procedures for Dynamically Loaded Foundations. Continuing the approach described here to study structures of varying size and mass ought to provide considerable additional insight Into the behavior of ocean floor structures in general when they are subjected to earthquake forces. "Dynamic and Static Field Tests on a Small Instrumented Pile.. 22. B. failure by gross movement in overturning or sliding might be indicated if additional restraints were not provided. Vol. Jan. R. M. of Technology. Apr. R. Coyle. Vol. 337. C." Proceedings of the Institution of Civil Engineers. W. It can permit a rapid preliminary analysie of any structure being planned for the ocean floor. W. Proceedings of the ASCE." 1971 Offshore Technology Conference Paper No. I. Whitman and F. The purpose of this paper has been to discuss some of the problem elements and to demonstrate a method of solution. Soil Mechanics in Engineering Practice. 1968. Byrne. 1969. D. From the analysis rough design criteria can be formulated for the strengths required in the various structural components. 1948. [6] K.
AR a 0 p 43i I .NOTATION Symbol a b c g Meaning radius of spherical hull base dimension of structure velocity coefficient in soil resistance acceleration of gravity (32.f structure displacement coefficient In soil resistance ultimaate bearing stress of soil time motion of center of mass relative to base displacement. xb y.q mass of structure left and right vertical reaction forces on footings submerged weight of structure coordinates of transformcd equations CD FD F% FS t RI. velocity. xb. velocity and acceleration of center mass drag coefficient drag force inertia force composite force acting horizontally through base of structure' ..2 ft/sec 2 ) h k q t xr xb . A R vertical displacement of left and right facings vertical velocity of left and right footings 0 angular displacement. ' . and acceleration of rigid structure mass denqlty base angle of structural configuration AL. and acceleration of base excitation vertical displacement. height of center of mass above base . . RR Ws a 0 ratio of weight of structure in water to that in air ratio of effective mass of structure and water to rhe mass of structure 2 coefficient of MIa in expression for mass moment of inertia y Al. % . velocity.
The model looked to me like you had a weightless type of foundation at each support. Zudanz (Franklin Institute).DISCUSSION Voice: Did you include any kind of damping to account for the soil radiation? Mr.. 'You did not have any coupling between different support points. There were only soil springs to ground and not soil springs from support to support. but in this study we just assume 5% critical damping for each mode.' Hammer: We have run cases where we coupled the damping of the soil with cquivalent structural damping to get a new equivalent factor.Hammer: No. Is that correct? Mr. Mr. 44 .
limit to a very few the number of timehistories which can be considered in analyses or tests. First. however. pvoperly define system responses. INTRODUCTION Analyses and tests to determine the responses of systems exposed to transient disturbances often employ a response spectrum as a means for bounding the severity of all probable disturbances. in others it may be too broad in that other information concerning the disturbance may be In most analyses and in all tests. As the response spectrum describes only peak responses to the disturbance of linear. the spectrum does not define unique amplitudes and phase angles of the frequency components of the disturbance itself. hopefully. may not be sufficient tr. however. as transient disturbances are rarely unique. Saffell Robert YangM. each with different corrponent amplitudes and phase angles. undamned systems. as the spectrum bounds the peak motions only of li:tear. sometimes to only a single one. of course. If the strength of the disturbance is bounded by a spectrum. however. Au infinite number of disturbances. can generate peak responses in the linear undamped oscillators which are essentially identical. undamped oscillators expressed as a function of their natural frequencles. this generality of the response spectrum has been one of the factors which has prompted its extensive use as a means for defining shock severity.ndamped single degreeoffreedom linear system to a motion timehistory whose spectrum matches the design spectrum may differ significantly from those bounded by the spectrum. strict compliance with the spectrum would require that an infinite number of timehistories be ccnsidered in analyses and tests. Ralph Parsons Company Los Angeles. Indeed. the disturbance must be described by timehistories.DEVELOPMENT OF A WAVEFORM SYNTHESIS TECHNIQUE  A SUPPLEMENT TO RESPONSE SPECTRUM AS A DEFINITION OF SHOCK ENVIRONMENT Herbert R. Thus. and TheC. Third. while in some respects the generality of the spectrum may be desirable. each with frequency components whose amplitudes and phases were different. California A procedure is developed for synthesizing a timehistory to describe a transient disturbance such that the response spectrum of the disturbance matches a given spectrum and the amplitude ratios and phase relationships of its frequency components remain within assigned ranges. Peak responses of any system except an . spectra of the synthesized timehistories must closely approximate the design spectrum. responses of systems with other characteristics may be significantly different. Second. in selecting a few specific timehistoriles to represent the entire family bounded by the spectrum. chosen so that. critical responses of the system will be generated. the characteristics of the system which will be exposed to the motions must be examined and the parameters of the frequency components of the timehistories available which might be cmployed to reduce the size and perhaps the severity of the family ci possible disturbances. If all responses boundcd by the spectrum are the result of disturbances which must be considered in design. therefore. The magnitude of these differences will depend both on the characteristics of the system and on the amplification ratios and phase relationships of the frequency components of the disturbance. The response spectrum alone as a criterion of input motion in analyses or tests of many practical systems. The family of motions bounded by a spectrum can be reouued in size if an 45 . Practical considerations. shock test machines must generate motions which are describable in time rither than In frequency.
'rThe response spectrum of the excitation function W(t) Is a plot of the maximum responses of the system defined by Equation 1.ole in frequency regions lying between the selected frequencies wn so long as they do not exceed prescribeC bounds. consider first the case where all frequency EHquation 3 components are in phase. In constructing such a function. describes one forcing finction which satisfies all specified conditiors whenever N. s tht the fiets m ciosd response spectrum of the composite wavefcrm not only matches i. Or. an infinite number of terms would be required. for the ranges of possible amplification ratios and phase relationships. its initial conditions must be zero and Its terminal values of both acceleration and velocity must vanish. for pulses. is not possible since an infinite number of such functions can. Response spectra. such as a sine.~~~ 7 X 7. points on a given response spectrum but at inter it may be known that all the energy is introduced at essentially the same time. since the function will be used to represent a real transient event..77 77.. fn(t) = sin 2nb t sin PrNmb t = 0 0 < t<T t > 'I' m ( ) w(t) can be expressed as a linear functions combination of a. of a simple linear oscillator subjected to base excitation is: d2 Y 4t Forcing Function. THIEORETTCAL FORMULATION The equation of motion Princioles. the duration expressed In liz. . therefore.. the M). Synthesis of a timehistory of a ofM sequence select is to (m problemfro(t). This paper describes a technique for synthesizing oscillatory waveforms such that their response spectra closely match a given spectrum and the amplification ratio and phase relationship of each of its frequency components approximate specified values. and little problem is usually encountered in synthesizing a family of such pulses which will satisfy these three requirements. thus providing some insight into the phase ' mediate frequencies as well. 46 . a . even W(t) = H mA m (t) m (2) grossly. For some phenomdisplacements must also terminate ena. S For example. . where bm is quency of its Then Tm . w. ramp. the amplification ratios and phase relationships of the original disturbance is more difficult. however. Further. functions. all possible disturbances might be describable as where Am are constant coefficients deining the amplitude of the function r(t) In synthesizing a timehistory as amplification "oscillatory" ratios "pulselike the implying that or for the former will be low while for the latter they will be higher. such that The frequency of each component of the forcing function isN!mb and the freenvelope 1. versed sine. However. and W(t) is the base acceleration. If a unique forcing function is to be defined. of the forcing function. is any odd integer except unity. form. the problem of matching closely not only a response spectrum but also solution of a set of M simultaneous algebraic equations for M unknown value08 of A. Where the disturbance Is more oscillatory. with asso= l. . impact. as a continuous function 01 w. Is the natural frequency of the system. in nondimensional some measure of uncertainty.. 1 = I2b m ' This forcing function where N = 5 is shown in Figure 1. To match a spectrum exactly at all frequencies.2. perhaps. as the response spectrum to be matched usually involves realistic disturbance thus involves not only matching the design response spectrum but also incorporating specified amplification ratios and phase relationships for each frequency component. terminal peak. it is necessary to postulate a function fr(t) such that its amplification ratio and component frequencies may be varied to satisfy the equation. and others have been treated extensively in An analytical sclution of Equation 1 for the forcing function fo(t) which would reduce the problem to an explicit the literature. is T. at zero. t Is time. square. such as might occur relationships. y = 2 W(t) (1) where y is the absolute acceleration of the mass.celeratton r(t). the number of halfcycleb. small deviations may be accept. be found whose spectra will closely match the required spectrum.7T examination of the basic phenomenon which resulted in the spectrum indicates that limits can be established.
The first term of Equation 5 defines the principal responses of the oscillator to the forcing function while the second term defines its residual responses. substituting Equations 2 and 3.t!if c~s Ct 2Ift 2 (' .. As bm approaches either of these values.b) and 161<<i. however.(1m+1) + ~~121. the limit of Qm can be obtained by applying L'Ilospital's rule.5 +1 ["(i+i)t cos 27nTt + sin 2r.cos 2n(t [[+  [ao . (Equation N 1)2 Equation 5 is determinate for all . The solution of Equation I (Reference 1). is M A  where Figure 1.i) _N4 2 +6 M1n sin .t) + ? sn 2ntj (7) (N +1) n[1+26 (u 1 +1)j[6.+.)2(. Thus.. and t (11M+l)2 1cos 2 rnt . it] ( m where 6 = Wnl.cos 2 7.1).r.N5 .t)( + Nl(U. )b t ] Selected Acceleration Forcing Function. N . bm= .5 c. and bm except when values of' t..( I~~~ It tn . [1+' 6(1m1)] + M 14[2N + 6(Nm+l)2(Nm )] 1r f6t (N 1)2 cos 2nr t + 2 sin 2. co 2Z t cos 2nnt Con [ 2 n (t  (T) J+(iN 2b2  (TM t)2(5) = 24 . and H is the leaviside function..£ Solution of Equations. 47 =' .n/(l .Nr)"and b= n/(i + rm). f and +6 T where 6 n j6 <«1. + 16( m+a ) 1 n ] lJ211+4(11_1) 2 (Ifm+1)  .2 cos 21nt cos 2 n(l C(N )22 O. 5 may be rearranged os 2nt .
M (9).. Assume that values have been assigned to Yk. forcing function frequency Nmbm is selected to correspond to a proper ratio of the selected system frequency wn the amplification ratio at the frequency wn . Where Ii is equal to 3 or 5. the maximum response 9k at t " I . Usually. 7. The amplification ratio of each individual frequency based amplitude coefficients Al.. 4. Amplification Ratio.m =1. 7. that is. the value of the normalized frequency parameter bm/...19 respectively.0 SPCIU...0? . Normalized Maximax and Residual Response Spectra of the Selected Waveform Component for N = 3.M) Thus.3 ." 6. the choice of the number and relative 'values of the frequencies is not completely arbitrary nor is the selection of component frequencies bm. The iteration procedure employed to obtain coefficients Am is not necessarily convergent for any value of waveform component frequency bm. on the desired amplification ratio.8 1.. V  _TII]I 1 X Lkm] I'm Q = k. k is the maximum value of z. if the is even closer to 1/lj. 5..... I N. Zk. or 7 may be searched numerically for the maximum response occurring during the time period o Z t V (Tm + 1/ 2lTk) and for the time tk at which the maximum response occurred. 11 . Maximax responses are indicated by the solid lines while the residual responses are shown by the dotted and dashed lines. From the response spectrum to be matched.9. and that a trial set of [Qkm] .6 trial values of Am will not coincide with the maximum responses indicated on  given spectrum at the corresponding frequency..1. is equal to 0.. and Qkm is evaluated at t=tk. M numbers of Yk maximum responses corresponding to frequencies 1 (k . To generate large diagonal elements in can be selected...2. . at each iteration step.28 and 0. From Equation 8 it may be noted that if the diagonal Figure 2. or lYk.. A set of matrix equations can be written URk where component of the waveform can be determined from its normalized response spectrum (Figure 2)./ .1 . A2. A new set of Am values can then be calculated as jAm) = . .m = 1.01 . and b.2. 5..2. 48 value of 11for that component. the value of b. / frequency k calculated from the first . Procedures for establishing these parameters are discussed later. for a given Nm must be selected so as to produce a near resonant response. it each iteration step the new value of coefficient Am of each row is influenced principally by the required maximum response value Yk where In and k have identical index numbers and correspond to one system frecuency k on the given spectrum. For N = 7 andN = 9. IIi 4 1:0 j k' k.0 determined. a new set of values for 5k and t k can be . RSIDLAI.0 Using the modified coefficients. 6.[MINM SECIRA "  L..Z .6 . the diagonal the inverse matrix [Qkm] . the maximum response at a selected system frequency should be dominated by one frequency component of the forcing function and none of the other frequency components should contribute substantially to it.4 . and 9 It may be noted from Figure 2 that the maximum normalized response amplitude ratios for components where N is 3..0 . 1 (8 AQ. the variation of Am is a single valued function of Yk and the process converges.50 < 9 1 where S is an arbitrarily small number. or 9 are very nearly equal to the terms of the square matrix (Qkm] are large compared to tLe off diagonal trms terms of of the same rcw. have been selected.Amplitude Coefficients.' will also be relatively large.3 .. the normalized frequency at which maximum response occurs Thus. 00 ...h N1 * corresponding to Zk .6 .oIZED F0(%.0 3  : .C RATO Rb. This process may then be repeated until the calculated maximum responses are approximately equal to the on the response response values indicated spectrum. Either Equation 5. If the response spectrum is to be matched closely at intermediate frequencies.
Optimum separation ratios for different values of N arc shown In Table 1. the spectrum of the synthesized wave will lie close to the line which includes the three points. Frequently a design response spectrum. and the parameter 11. in the procedure. In the region between two match* points. has been found to give good results and is simple to apply.3 such that both the solution will converge and. if the components for the two match points do not have the same value for N. In the region of the spectrum between these frequencies. two adjacent match points are considered at a time. if the forcing function is to include components at other frequencies. However. the spectrum is influenced more by the component at the higher frequency. a semiempirical approach has been developed which. spectra of many occurrences of a phenomenon are frequently bounded in this manner. the value corresponding to the higher frequency component should be assumed to be applicable to both points. that the amplification ratio of a single component can only be an odd integer and that the minimum amplification ratio is 3 in order to satisfy the requirement that all motions vanish at t =T m. Selection of Frequencies. the spectrum of a waveform synthesized as described above will match M points on the given spectrum and the specified amplification ratios will be approximated at these points. a midfrequency range of constant pseudovelocity.1. if any. While this spectrum shape is not essential to the technique for selecting apnropriate frequencies. although perhaps not yielding optimum values. . Three Selected Match Points on a Trapezoidal Response Spectrum In determining the frequencies. However. 3 and the three collinear points which are to be matched. the stipulation can be made Yk k1 I) While if they lie in a line of constant pseudovelocity Yk Wk Y='1 If they lie in a line of constant acceleration. were varied. when plotted on four coordinate oaper (such as shown later in Figure II) is trapezoidal with a lowfrequency range of constant relative displacement.1. and . While spectra of few. where a n k]Zk. As an example. and a highfrequency region of constant absolute acceleration. The problem is to select the three frequencies. convergence of the iteration procedure can be assured only so long as the response of any single frequency component is not influenced substantially by those due to the remaining components. If the iteration procedure converges. If the match points lie on a line of constant relative displacement.in equal approximately to the selected number of halfcycle oscillations To demonstrate the method consider the trapezoidal spectrum shown in Figure N.2.43 Hz. The selected component should have N n 7 halfcycle oscillations at a frequency of bm a 10/7 .1. 0 0 0 C3 2 t 0 0 LOG FREQUENCY Figure 3. . of course Yk =Yki' Using these relationships. The complexity of the equations describing the response motions precludes the use of a closed form solution to identify optimum match points. assume that it is desired to match a given spectrum at a frequency of 10 Hz with a single frequency component with an amplification ratio of 7. 49 L . The problem which remains is that of selecting frequencies wk such that they are sufficiently separated to ensure convergence of the solution but not separated so far that at intermediate frequencies a close match between the two spectra cannot be obtained. a large number of cases were calculated in which the separation ratio a. Note also. it does simpilfy the selection. physical phenomena are of this exact shape. Thus.
amplification ratio. the effect of the time delay can be much more pronounced and match points selected to minimize error in the intermediate ranre are no longer valid for all time delays.leefnqd as the ratio or the time delay of two idjacent component: to the period or a hilecvcle or the comoonent at frequency NMb ''hujs m. Based on the haifcycle oscillations at the selected system frequency points. the time delay between the (M +l)th and the Mth components is td. in the example. For match point frequency ratios shown in Table 1. of course. Input data for the program consists of a definition of the required trapezoidal response spectrum. Using the optimum system ratios given in Table 1. Thus N m=1 fm= 0 = sin 2i. .6 liz and w3 = 1.3 '. bm(ttd) 0 < t < tdm sin 2. MWAVSYN Computer Program. Usln' the searatton rrenuencies ratios shown in Table ]. An Interpolation subroutine then establishes the required response amplitude.6 1. designated MWAVSYN. . ai indicated in Table . Phasing of Waveform Comnonents.. and the accuracy tolerance imposed on the spectrum of the synthesized waveform. These ranges were calculated for three frequency components whose frequency ratios wer.TABLE 1 (Reference 1) OPTIMUM SYSTEM FREQUENCY RATIOS FOR DIFFERENT VALUES OF N . Ranges of time delay ratios for values of N from 3 thru 13 which will minimize deviations from the design spectrum in the region between match point frequencies are shown in Table 2. 1. Mathematically.. = 1. the waveform with phased components can be written similar to Equation 3. Thus. '. Dr tdm i t <tdm + T m where td 5 Is tne time of inititation of the mth waveform component from time zero. TABLE 2 RANGE OF ACCEPTABLE TIME DELAY RATIOS FOR DIFFERENT VALUES OF N RANGES OF TIME DELAY RATIOS. if II = 5. ". and time delay for each frequency componert. and assuming . 7 9 11 13 3 5 N Dr (2Nmm) td. N. The time delay ratio Dr Is . 50 . a Wi S=1 k_1 1.35 1.16 Hz.35cose trum in the region between these frequencies cannot be achieved where the time delay ratio exceeds N.2 and N1 3 = 5.. the time delay term will have little effect on the responses at the spectrum match points. responses at the intermediate frequencies were calciilated for various time delays of any three adjacent frequency components and the ranves of time delays aere identified in which deviation was found to be a minimum.3 Thus. (M+1).6 x 1. M Note than when Dr . the timehistories of the two frequency components occur consecutively.1. the program selects the system frequency match points on the spectrum. 6.0 Hz. Equations b thru 7 remain valid even when the time delay is included in then. For other frequency ratios and combinations of N. other ranges of time delays might yield equally acceptable correlations with the design spectrum. however. . A computer program. 12 = 7. Tf tdm is the time delay of the mth frequency component measured from time zero.35 = 2. and whose values of N were equal. amplification ratios and time delays as functions of frequency. . Amplification ratios are reduced to Integers which are nearest to the oddnumbered halfcycle oscillations. and 7 for amplitudes Am and to perform the iterations necessary to ensure that the response spectrum of the composite timehistory match the design spectrum within an assigned tolerance 5. Phasing of the waveform components can be represented by specifying a delay in the starting time of each waveform component. should be 1. was developed to solve Equations 5. Hmb(tt) (10) correlation with the desigh specc1. (4+1). In the range between the match points.
Comparison of the Vertical Spectrum of a Typical Structure and the Spectrum Generated by the Synthesized Waveform resoonses.RIQU1NCV.1 to 20 Hz. r' lationshIrn were estimated f'rom Phase an examination of the nredicted timehistories a' the loading rhenc:. the shock environment within various zones of structures are  .the time delays are forced to stay within the rangps shown in Table 2. It was noted that 51 SNTHESITZo WAVCFOM . that as the airblast pressure wave approached closer to the structure higherfrequency modes were energized. and that the arrival airblast pressure wave at the structure was accompanied by the highestfrequency responses. As no single timehistory of the predicted responses had a snectrum which matched the envelope. In the design of shock isolation s'ystens supported by the third floor. such as the detail nature of the external loads. generated from the synthesized waveform is shown by the solid line in . EXAMPLE Dynamic analyses of structures subjected to earthquake or nuclear weapons effects involve the calculation of timemotion histories at many interior locations.o . To account for possible differences in the details of the responses due to possible variations in the parameters. responses calculated at any given location can be regarded as only one of a large number of possible  Figure 4. Based on a consideration of the frequencies of the modes of tht shock isolation system. achievable. The response spectrum. Ccmparison of the Required Amplification Ratio Spectrum and the Amplification Ratio Spectrum of the Synthesized Waveform the lowfrequency responses of the structure were the first to be energized by their exposure to an outrunning ground motion. it was desirable to synthesize a timehistory which would incorporate "worst" conditions. timehistories of the disturbance were required.  Output data consists of timemotion histories of the composite waveform expressed in terms of acceleration. the soil and site properties. Timehistories of the calculated structural resnonses were then filtered within this frequency bandwidth to determine their amnlilication ratios (Figure 5). and the dynamic characteristics of the structures themselves. so HZ Figure 5. it was concluded that It would be necessary to match the design snectrum only within the range of frequencies from 0.ena which nroduc Pd them. Owing to the many assumptions involved in these analyses. l eQUenD AIR SPECTRUM For example. Input requirements based on these estimates and of the calculated amplitude coefficients the synthesized waveforn are summarized in Table 3. by the dotted lines. Timemotion histories canobe punched on cards and/or plotted by a CALCOMP machine. and displacement. and as it was not feasible to investigate the behavior of the shock isolation system to all timehistories whose spectra were included in the envelope. A detailed description of the program is presented in Reference 1. velocity. A minimum accuracy tolerance of 15 percent has usually been found to be easily. an envelope of spectra I 01 05 AIR CT ( UM I I 100 I 200 of predicted vertical responses of a typical structure is shown in Figure 1.  frequently bounded by spectra enveloping the spectra of the calculated responses of all noints lying within that zone.
. Acceleration.. _ IrIT I a. velocity. As noted. has the important advantage of oermitting the systematic variation of amplification ratio and phase.. in Figures 6 thru 8. Figure 7.FT . The required amplification ratio spectrum and the amolification ratio spectrum of the synthesized waveform are compared in Figure 5.4 DisplacementTime Function of the Synthesized Waveform AccelerationTime Function of the Synthesized Waveform Figure 8.. i T ... Desnite these restrictions.. ..~ W_. ... the accuracy is within +15 percent at frequencies of .3 Hz and better than +10 percent at higher frequencies. _ ..~ . SUMMARY simplifies the synthesis of an oscil 47 d A technique has been developed which d latory waveform incorporating specified amplification ratios and time delays and  r . !.~A ~ .. in addition..i  _. _I j ~A. . . the technique is a significant improvement over the older cutandtry methods and.. . however... 52 . .WII Figure 6.. M4 .... F. VelocityTime Function of the Synthesized Waveform . can be included in the waveform and requires that their amplification ratios be equal to odd integers greater than unity. . _i ..  lower than 0. The nature of the basic synthesized waveform limits the frequency components which . two parameters of vital importance to the responses of most practical systems.1i il T matching a given response spectrum. . . I 14t i and displacement timehistories are shown.Figure 4. _JaJ.
Parsons Company.006 4.05 2116.42 31.555 0. 30 April 1971 I .1130 17. AND CALCULATED AMPLITUDE COEFFICIENTS OF THE SYNTHESIZED WAVEFORM NUMBER OR FREQUENCY COMPONENT SELECTED SYSTEM FREQUENCY Sm1z 0.310 13.111 219. sec 0 0 0 0 0 0 0 0 0.37 280.707 4.35 1 2 3 3 5 6 7 8 9 10 11 12 13 1l 15 16 1.02 1.37 0.70 153.13 312.706 1.120 11.078 0.90 76. TIME DELAYS.100 0.1111 0.152 3.239 0.006 4.110 7.51 16.165 0.889 3. Modification of the WAVSYN Computer Program.025 1.005 2.037 0.232 0. Document No. C.614 1.1188 0. COMPONENT FREQUENCIES.026 0. SAFU2.70 250. Yang.3115 TIME DELAY tdm./sec2 0.48 NUMBER OF HALFCYCLE OSCILLATIONS Nm 3 7 9 9 9 9 9 9 9 9 9 11 13 13 13 13 COMPONENT FREQUENCY bm Hz 0. The Ralph M.TABLE 3 SYSTEM FREQUENCIES.530 6.71 . R.033 1.612 0.115 23'.563 2.006 REFERENCES 1.793 1.160 0.023 0.054 0.925 AMPLITUDE COEFFICIENT Am in.336 0.1187 2.503 0.3147 0.950 10.028 0.1143 7.
were recorded.psi overpressure region from the detonation of the 500ton highexplosive shot of Event DIAL PACK. Background There is a requirement to store sensitive equipment if) aboveground hardened shelters. However. The objective of this test was to obtain information on the response of an isolated floor slab placed on asoil fill inside the concrete cylinder. the required strength of the equipment is controllkd by its response to the shock produced by the dynamic loads on the structure. Yet the unimpaired functioning of both per. 2. Shock Isolation An essential element in the design of protective structures is the provision of a reliable shock isolation system for personnel and equipment. transmitted through the soil fill. In the design of shock resistant equipment. It is important to limit the blast induced motions transmitted to the floor slab to reduce crete slab on asoil fill inside ahorizontal concrete cylinder is being considered as a possible means for diminishing the shock motion. Evaluate the survivability of the structure including the performance of the retaining wall. with ground accelerations over a hundred g's and ground displacements over two feet. inexpensive shelters which can be rapidly constructed and will eliminate the cost of providing expensive shock isolation platforms. An alternate approach is to isolate the floor slab by "floating" it iii a layer of soil! within the cylinder. Data from seventeen channels of active instrumentation. The equipment must be shock isolated from the shelter. The specific objectives were: 1. The cylinder is a closed structure capable of being constructed economically and supporting high overpressures with minimum disturbance. sonnel and equipment immediately following an attack must be preserved if the facility is to fulfill its intended mission.THE RESPONSE OF AN ISOLATED FLOOR SLABRESULTS OF AN EXPERIMENT IN EVENT DIAL PACK (U) J. This structure. A more serious limitation is the peak shock which an unsupported human being can tolerate.Ferritto Naval Civil Engineering Laboratory Port Hueneme. if satisfactory. Reduction of the data was made. velocity gages and accelerometers. Preceding page blank $ 55 . The motions of the floor slab are related to the motions of the surrounding concrete cylinde. Determine the absolute motions of the floor slab caused by the blast loading. Determine the motions of the slab relative to the cylinder. The levels of shock to which a system acting at its designed structural capacity may be exposed can be very high. will provide simple. Expensive conventional shock isolators could increase the cost of the structure to an extent that might be unacceptable. California This paper outlines atest of a horizontal cylinder covered with an earth berm subjected to the pressure and drag forces in the 300. as the cylinder is displaced downwardh theiefi owevisolated floir slab tdnas to reduce the peak accelerations felt on the floo. The floor slab must be strong enough to carry the equipment load and resist the longitudinal and torsional motions from the shock. INTRODUCTION Objective The prime objective of this project was to obtain information on the response of an isolated floor slab inside a horizontal cylinder covered with soil and subjected to pressure and drag forces in the 300psi overpressure region. 3. composed of a pressure cell. Very few devices arc capable of withstanding shocks of this magnitude without serious damage. M.
700 feet from ground zero.. a high explosive field test of 500 tons conducted at the Dcfence Research Establishment Suffield (DRES)._ . the second letter. The invert of the cylinder was 11/2 feet below the natural grade.1. ca is shown in Figures 6 and 7. amplified. := " 53' Instrumentation Seventeen channels of active instrumentation consisting of eirht structure velocity gages.. Canada.. An interrange instrumentation group timing system was recorded on one track of each tape recorder and later used in the data reduction. eight accelerometers. It consisted of a 6footinnerdiameter 7inchthick horizontal rightcircular cylinder aligned perpendicular to the direction of propagation of the blast wave. Ralston. . The data were conditioned. . indicates an accelerometer or a velocity gage. Jectionoam .Se renoreenomtte l ndaing.' o" . and Figure 5 shows the completed structure. . A springmass system..EXPERIMENTAL PROGRAM Test Structure The isolated floor slab test was planned as part of Event DIAL PACK [1].. pie cyinde .2'. 19'A 6i8' 1. One and onehalf feet of compacted soil was placed inside the cylinider. A detonation zero pulse provided by DRES was recorded as received directly onto the last track of each tape recorder to provide a reference for data located on different tape recorders. The test structure. Alberta. Figure 3.. The structure location and construction detcils are shown in Figures 1 and 2. and one pressure cell were used. Passive instrumen. Timing from an NCELdesigned timing generator was also recorded on one track of all recorders. tation consisted of tree orthogonally oriented reed gages mounted on the slab and a scratch gage (Figure 7) mounted between the retaining wall and the cylinder.. and a 6inch concrete slab was cast over the soil. The layout of the instrumentat.indicates vertical or horizontal orientation.B Nt. A or V. The pressure ceh was installed in a specially designed concrete mount which was cast in the wall with the heat shield flush with the face of the wall. A 2footthick retaining wall held the other end of the earth berm in place.' 6onstrucionnplan urY F3 I.. The cylinder was covered with 21/2 feet of compacted fill forming a tapered earth berm extending along the sides of the cylinder and one end at a 3:1 slope. _ _T_ u . B. was installed 4 feet from the rear end of the slab to simulate equipment on the slab. a bolton steel closure plate was incorporated in the retaining wall to provide access into the cylinder. Noe:Stelrenfrcmetomttendaing. The first letter. and recorded on 32track tape recorders located in a bunker 2. Figure 4 shows the structure under construction.2' 6" I_\& wallZ cylinderc berm "B'  420" closure Figur 1T 1''6 _ Section. V or H. located at an azimuth of 120 degrees 270 feet from ground zero. The gages were mounted on steel plates cast in or bolted to the structure. was at an anticipated sideon pressure range of 300 psi.
in..270ft ground zero I C L 270 ft 53 ft x 42 ft (berm dimensions) 120degree azimuth Figure 2. SecionAd Fgr .Srn~asla ytm Figure 4. 1/4.thlck end plates welded to beams A. Test structure under construction. Layout of Project LN322.OU0 lb. welded to beam flange ~2 Springmass system filled with bags to a total weight of 4. end plate" plywood bled to flanges x4 2 angle 0 2 x 2 x 1/4 angles. equally spaced.ln. 7"ItF ~~~ 3/4In. 1/4. 597I .
Lgaigefsei ags AV3 and 58 . test. Completed test structure prior to. groujnd zeroj AVi and AH V4ad V VH57 andVH AAV1 and AH6 Figur 6.!P7.  figure 5.
At D+lI hour project personnel returned to 6r.L LN322 site. Test gages. Approximately 18 inches of the berm on the groueMd Meo side were compressed and/or blown tway. 59 . slab to record rotational motions of the floor slab.~AV1 and AH2 AH 4 Figure 7. A vcrtici'l and two horizontal acclerMCM' etesre set at the center of the floor slab to record floor accelerations in three orthogonal direc tions. The two vertical velocity gages at eath end were installed near opposite edges of the floc. Twelve inches of crater ejccta were deposited in front of the ground zero end of the retaining wall. Horizontal and vertical accclcromcters wcrc set on thle cvI indcr ncar each cnd to measure thc horizontal and vertical components of tlic cylinder motion. The three orthogonal reed gages were used to provide records of horizontal and vertical shock spectra. This level was reduced to about 2 inches in)front of the closure plate.icld sidcon overpressure.. The lab instrumnentation was placed in tlac same crosssectional planes as thc cylinder instrumentation to evaluate the relative motions of the slab with respect to the cylindei. IThe maximnumn size of the ejrcta was estimated to be 8 inches. The Pressure gage Was Set in the outside (aec of the retaining wall to measure thc fre. The fireball Wablackened the retaining wall and the berm with an layer of carbon dust. The scratchi gage was used to record relative motion between the retaining wall and the cylinder. RESULTS AND DISCUSSION Observed Damagc Thc detonation of the 500ton highiexplosivc chiarge occurred on 23 July 1970.
. A summary of the peak values is presented iuTable I. Correction factors were subtracted in an attempt to remove the tilting effect. • .degree permanent rotation into the berm. Postshot measurements revealed that the slab rotated clockwise approximately 3 degrees. wave wai wall 317 above psi. The permancat horizontal translation of the cylinder with respect to the retaining wall was 35/16 inches away from ground zero.s Figure 8.. 6 v. give the shock frequency spectra of the slab. The reed gage consists ot'a number of masses on a rigid bs. The first three sections had been unintentiorally oriented so that the splice in the circular reinforcing steel was located in the region of the tension failure. several radial horizontal velocity plots. 1 . The wall had about 1. A region of major apalling and tensile cracking was noted at about 90 degrees from the compression zone (60 degrees from the top of the cylinder) on the windward side of all of the sections (Figure 8).A pattern of major diagonal cracks was noted on the wall.  ' 4 i AIn I. The permanent relative horizontal translation of the slab was 1/2 inch away from the ground zero side of the cylinder.000 msec. The The peak pressure had an arrival time of 32 msec and a total duration of the positive phase of 103 nsec. and measurements of the traces were taken and converted to displacements. and velocities were sampled at increments of I insec for 2. 60 . The actual duration of the positive phase was longer than the expected value of 60 nmsec. plotted in Figure 9. A horizontal line of compressive failure was located at about 30 dcgrees from the top of the cylinder on the lecward side of all the sections. The concrete cylinder sections suffered significant dam. Accelerations were sam. . These values. These cracks appeared to go completely through the wall. The floor slab was littered with spalled concrete and sand. The analog data tape was returned to NCEL where the data were converted to digital form using the NCEL AnalogtoDigital Converter. away from ground zero.. Instrumentation All of the active channels of instrumentation functioned satisfactorily. (: . The data were automatically plotted using a CDC 6600 computer.1 msec for 200 resec. Thus. Active I S. Th6 splice appeared to have failed. and to the right looking at ground zero. S. pled at increments of 0.. the velocity at late time did not retun to 7ero.vceal of these cracks were observed on the top surface of the v'all and extended completely through the wall. Positive values indicate motions down. indicating permanent tilting had occurred. the shock spectra diagram can be used to determine the shock isolation requirements of the equipment. Figure 10 is a plot of the sideon pressure data recorded at the midpoint of recorded the retaining theshock closure plate. At D+I day the closure plate was removed revealing additional cracking in the wall. The response of equipment mounted on the slab to a given shock can be determined fr3m the measured response of a reed having the same frequency and damping as the equipment to the shock. Postshot view of interior of structure. Visual measurement made after the shot indicated a permanent tilting of the slab of about 3 degrees clockwise when viewed from the retaining wall looking into the structure.. The accelerations and velocities were integrated. No filtering was used. and a section of the pipe was pushed inward approximately 4 inches along this line. The center of the slab remained at the same elevation relative to the cylinder. Some cracks in the concrete were noted around the closure plate. Reed Gages The plates from the recd gages were removed.
11000Shock Spectra HorzotalRaia 0~~~ 9W ElHorizontal Transverse SVertical 011 4 10 Frequency (cps) Figure 9. gage PSI. 61 . Shock spectra of floor slab. Pressuretirre curve. 100100 350 C 70' Time (msec) Figure 10.
4 .5 . it is believed to be accurate enough to relate direction of motion and approximate orders of magnitude of movement.8 173.3 10.0 11.4 2.2 13.0 0.4 3.4 5. Peak lnstrumer.8 6.0 0.0 0.6 2. Although there may be some error associated with this procedure.0 2.3 4.Table 1.5 0. Note: 1.8 3.7 10. and velocity 2.0 2.1 22.6 0.6 112.4 . Figures 12 and 13.3 4.0 1.0 0.8 3.6 9.4 23.5 19. were made for the slab movement using the horizontal and vertical displacements ubtained from the integration of the velocities.4 146.9 5.5 5.) Nive Positive Type of Instrumentation Gage Cylinder Vertical accelerometer Vertical accelerometer Horizontal accelerometer Horizontal acceleromctar AVI AV7 AH2 AH8 19. Displacement of cylinder.4 8.3 20.0 11.1 7.3 0.8 4.0 2.6 3.3 9.9 6.2 gage data taken for 2.0 1.0 0.3 5.7 13.) Figure 11.2 31 1 ground zero  3 down T lmsec 10 > =5 ! 7 '1 9 Horizontal Displacement (in.0 0. 62 .0 0.8 90. and to the right facing ground zero. away from ground zero.2 9.4 0.000 msec.1 5.tation Values Peak Acceleration (g) Negative I Posit Peak Velocity (fps) iti Peak Displacement (in.0 0.0 2. Cylinder Displacement The horizontal and vertical displacements obtained fromi double integrations of cylinder accelerations were used to produce a plot of cylinder movemcnt (Figure 1I).9 52.0 0.0 1.5 11.4 2.8 1. Horizontal radial velocity VH8 0.5 28.9 Floor Slab Vertical Vertical Vertical Vertical Vertical Vertical accelerometer accelerometer velocity velocity velocity velocity AV3 AV4 VV1 VV2 VV3 VV4 AH6 VH6 AH5 VH5 VH7 SpringMass 3.6 . Similar plots.0 0.7 0.5 5. Positive direction is downward.2 0.9 Horizontal transverse accelerometcr Horizontal transverse velocity Horizontal radial accelerometer Horizontal radial velocity Horizontal radial velocity 3.7 15.2 72.9 3.4 4.6 86. Accelerometer data taken for 200 msec.3 7.
Displacement of floor slab. reversed calibration. and 13 indicate tie motion of the structure was initially downward and away from ground zero. the movement of the floor slab was independent of the cylinder. The magnitude of the floor slab accelerations was substantially reduced compared to that of tile cylinder. i II ' I 1 1 I I I 1 1 1 I l l I 1 3 ground zeroundro down 3 5 0U > 5 Horizontal Displacement (in. The 22fps peak vertical vclocit) of the cylinder was reduced to about iI fps. Subsequent motion at about D+100 msec was upward and away from ground zero. for small rotations the influence is negligible. A possible explanation of the horizontal movement toward ground zero is the reaction to the magnitude shear wave which enveloped the berm producing planes of opposing shear forces. After this period secondary effects may adversely affect the data. The time to peak positive accelerations of the cylinder was about 12 msec after the arrival of the blast wave.ity gages experiencing rotation as slight as three degrees may erroneously indicate the cylinder was also dismissed as a possible explanation because the direction of ovaling near the gage location was inward rather than outward.e slab aid separated by 36 apparent motion of 100 percent of the anticipated value in magnitude. The motion toward ground zero is quite unusual and unexpected. These forces gave rise to motion as slippage occurred. confirming the ficld measurement. 12. Generally the data obtained in the first few hundred milliseconds are very reliable. Figur:. Accelcrometers are usually more reliable than velocity gages when rotation of the gage is suspected. Evidence that this has occurred is noted when velocities at late times (2 seconds) fail to return to zero. leeward side. Once set in motion.itinmg' on sand within a horiLontal cylinder reduced the peak a. The peak values and wave shapes in early time (several hundred milliseconds) are quite reliable and are relatively the most important. or an erroneous gage were considered and found not to have been present. the time to peak rositive acceleration of the floor slab was about 110 insec after the arrival of the blast wave. The initial downward movement was caused by the direct compressive wave. Displacement of floor slib. Table 2 compares tihe peak motions and shows the * 63 Ai . Thus. l)isplacenents from integration of accelerations should be capable of giving an order of magnitude of niovemnent and direction. but is believed to be valid. This effect was obse'ved on four independent sets of instrumentation. amid then moved downward until coming to rest on the soil. Accelerometers are relatively insensitive to rotation and are in. Pos.. the rotation of the slab began at about 125 msec and rotated the full amount at about 200 mscc. Ovaling of Horizontal Displacement (In. The amount of rotation shown is about 3 degrees. Integrations of acceleration data give reasonable indications of wave shape and velocity. Shock Isolation Usiiig all isolated concrete slab "flo.) Figure 12.uenced only by the prc luct of the sine of the angle of rotation and the component of acceleration iii the perpcndicular direction. inches. The integration of data represents a second level of confidence. sible sources of error such as gage rotation. Vcloc.) Figure 13. Figure 14 shows the relative vertical displacenient of two velocity gages located on tl.celcration of 174g's on the cylinder to about 2 8 g's on the floor slab. From Figure 14.11. windward side.however. The reverse movement of the structure was caused by a combination of rebound and direct induced ground shock arriving about 100 mscc after detonation. The slab appears to have remained stationary during the initial movements of the cylinder. chcn downward and toward ground zero.
6VV2 ~VVll 1.000 2 Time (msec) Figure 14. Comparison of Peak Motions Velocity (fps) Vertical Cylinder Floor slab Horizontal Cylinder Floor slab Acceleration WgS) ! 22 174 1I 28 in the shelter. The isolated floor slab significantly reduced the transmitted motion. May 1971." by J. . NCEL Technical Report R726. 2. CONCLUSIONS(p)(g) The objectives of this project were satisfied. Port Huenene. Shock spectra data for designing the required shock isolation system has been determined. M. Rotation of floor slab. and the survivability of the structure evaluated. All of the instrumentation functioned satisfactorily. Data were obtained to compute the shock isolation requirements for equipment to be stored Table 2. "Dynamic Response of an Isolated Floor SlabResults of an Experimental Test in Event DIAL PACK. The significant findings and conclusions arc: 1.000 2. Fcrritto. The motions of the floor slab were determined. California.shock isolation of the isolated floor slab. 11 4 90 19 REFERENCES 1. 64 ~I.
Ferritto: Yes. Ferritto: It is a very complex problem because the loading is quite complicated. as we are now looking at it. Voice: What is the highest frequency of the model you are putting together? Mr. Would this predict any of the high frequency phenomena which I assume to be present? Mr. it is a plane strain model. Voice: Is it correct that you were not trying to establish the effect of isolation using a layer of soil? Mr. finite element analysis with a quadrilateral element rather than another type of modeling procedure. The spring mass system shown in the slides is simply used to provide an equivalent equipment on the floor slab. The period of the cylinder by itself in compression is about 2 milliseconds. Keen (Bell Telephone Laboratories). This is one rescn we are using a plane strain.: The high accelerations ii your last slide uould indicate the presence of very high frequency data.800 degrees of freedom. and the time step is made as large as possible keeping the economics in mind in order to be able to run the problem. You also mentioned that you planned to use nonlinear finite element analysis techniques. I believe. The time step and other information we have used in sizing the finite element mesh has been scllected. does this refer to the isolated mass sitting on top of the springs. Vcice: Do you solve 4. The period of the cylinder in flexure is about 11 milliseconds. Zudans (Fraklin Institute): I am interested in your 4. We are running approximately 300 time increments to approximate about 100 milliseconds. The extent to which the berm participates in the problem still remains an unknown. Mr. The springs of the mass and the weight of the mass were selected to provide the frequency that we would expect from a typical piece of equipment placed on the slab. basically. so it is quite a large problem. or ig the slab sitting on top of the sol? Mr. Mr. It actually represented what we tried to approximate in the model. We have a very complex soilstructure interaction. We have a traveling wave.DISCUSSION Voice: Concerning the measurement on the floor slab. Ferritto: All the measurements were made on the actual concrete floor slab.000 dynamic degrees of freedom. The structure. Ferritto: Basically that is correct.000 dynamic degrees of freedom? Mr. Ferritto: Yes. to satisfy the structural characteristics. Ferritto: To this date we have been running a structural analysis primarily interested in the optimization of the structure itself without looking at the isolation characteristics. We are attempting first to look at it structurally. We have made several runs. 65 . The quadrilateral element that we are using has 12 degrees of freedom and approximately 400 elements. Are you modeling it as a twodimensional infinitely long type of strip? Mr. has about 4.
developed. Shockisolation costs themselves have also been reduced. 67 Preceding page blank .000 lb for skid buildings.A SHOCKISOLATION SYSTEM FOR 22 FEET OF VERTICAL GROUND MOTION* E. and overall reliability has been increased by this stat. and fielded several different shockmitigation systems. During the past several years we have designed.000 lbs for trailers and up to 140. resulting in considerable savings in diagnostic cable costs. C. Work performed under the auspices of the U. The optimum location. or about 22 ft displacement. geology. The vertical slapdown acceleration pulse is usually followed by a horizontal radial pulse. The new system is a modification of one of our standard designs. test results. REQUIREMENTS AND ENVIRONMENT For maximum utilization and mobility. Normally. Atomic Energy Commission. Horizontal tangential pulses are also occasionally significant. The ground motion induced by an underground nuclear detonation varies considerably. the experiment requires the equipment to be located relatively close to the explosive source. with respect to reliability and overall costs for diagnostic and other portable instrumentation bunkers located on the surface. These systems have allowed equipment to be located closer to the explosive source. with the usual crushable materials replaced by a columnar energy absorber. Jackson. and test results are discussed in this paper. These standard designs have allowed us to reduce signal attenuation and to reduce very large cable costs by locating equipment close to the source. but not more than half the depth of burial from surface ground zero. There is a large variety of electronic equipment and accessories. Figure 1 is a time history curve of surface ground motion for a large event. Two basic systems have been standardized and are discussed in this paper. depending upon yield of device. we have designed a shockmitigation system to withstand vertical ground motion up to 32 ft/sec.'L. Anew system has been designed for even more severe ground motion.000 lb to 72. most equipment falls within our medium fragility range . which can be either away from or toward surface ground zero. In this paper. B. INTRODUCTION Experimental data from underground nuclear detonations are conditioned and recorded by sensitive electronic equipment. ground motion induced by these nuclear explosions requires shockisolation systems capable of supporting heavy instrumentation vans and isolating them from a very severe threedimensional dynamic environment. This energy absorber allows a longer stroke without an increase in payload initial height.S. and location of interest. regardless of yield. The new design. but a few have been fabricated for higher shock requirements (7 g vertical). California Shock isolation of fragile equipment from severe ground motion induced by underground nuclear detonations requires special techniques for inexpensive. In many cases. reliable performance. Most trailers are conventional highway type. the instrumented testing program. When properly mounted. Shock fragility levels vary from 1/2 g to more than 50 g. Two shockmitigation systems that have been used successfully for the past several years are described. To meet requirements for even more severe ground motion.dardization.6 g vertical. Bernreuter Lawrence Livermore Laboratory. but the characteristics are typical of the spall region. is safely outside the subsidence crater. most electronic equipment is housed in trucktrailer vans or portable buildings on skids. we discuss the and new the design. Miller and D. In this area the surface usually spalls and follows a ballistic path. Loaded weights vary from less than 10. the instrumented testing program. A. University of California Livermore.
i.e.. and (3) design of a failsafe shock mount system to modify the shock environment when required. It includes permanent reusable wood cribbing with an angle iron base 68 . However. DESIGN METHODS The overall problem blends together: (1) definition of input. Decoupling is achieved by placing the system on surfaces with very low friction. The fundamental classification of shock and vibration systems is the manner in which it stores. a few methods that we have been using.. Dynamic effects of impact velocity on energy absorber forces are considered separately.. 1). A complete discussion of all .75. absorbs. Ideally.4 therefore. We have developed several shock mount systems varying in complexity and cost..1 V . FOAM AND REUSABLE CRIBBING SYSTEM For many events the estimated maximum ground motion is less than 10 ft/sec and we use the foam and reusable cribbing (F&RC) system for trailers and vans. The energyabsorber system does not necessarily have a factor of safety greater than unity with these maximum estimated input conditions. these absorbers should have negligible rebound characteristics.. I Typical time history of ground motion within the spall region. Vertical accelerations are isolated by constant force vs displacement energy absorbers. The respective nominal ground motion values are much less.. With this condition tha conventional analysis is relatively simple Ref. The maximum estimates are based on maximum credible yield and maximum scatter of applicable empirical groundotion data. excessive horizontal displacements are controlled by nylon tethers. This factor accounts for the elasticity of the shock mount structures.. accurate final position or alignment of the equipment with respect to the ground is important and must be considered in the overall design.. 0 h. The shockisolation design for a given event is based on nominal and maximum ground motion estimates in order to obtain a factor of safety based on energy. The selection of a system for a given event depends primarily on the estimated maximum vertical ground motion as shown in Table 1..The basic requirement of all shockmitigation systems is controlled relative displacement and force transmission between the shock input and the package or system to be isolated. Whenever possible. (2) fragility level or shock sensitivity of equipment to the input and 1. Accurate measurement of th6 payload weight and center of gravity and the appropriate sizing of the energy absorbers minimize the amount of differential vertical displacements (tilting) of the system. 1). equipment mounting brackets and the trailer structure. does it record or transmit through shock arrival time?). in most cases position and alignment are iot important. When required.e. ground shock parameters. the design is based only on peak input parameters since detailed time histories of ground motions are very difficult to predict. Energyabsorber deceleration set values are based on equipment fragility levels and a structural amplification factor of 1...(see "" Fig.6 o. The reliability of a system is strongly dependent on a foreknowledge of the input conditions. in this paper we will describe only . Prediction of surface ground motion for our purposes relies heavily on empirical data and methods (see Ref. A great many different materials and methods are used in shock and vibration isolation. This variance allows us to design around the threedimensional groundmotion environment by decoupling the horizontal shock components.l pareeters involved in shock mounting items for underground detonations would be quite long.ose (i. or dissipates energy.. except for the incorporation of failsafe features. %. 'dII " AV .. :. In some cases.. This paper concerns only these cases.
. We have found that the maximum horizontal displacements occur after slapdown and after the crushable material has compressed. . . Um . The crushable material sits between the metal disk and the beam as shown in Fig. 2 FARC shockmitigation system: (A) replaceable foam pad bolted to cribbing. The magnitude of this twisting moment is a function of the coefficient of friction the surface pad dnd crushable material height and crush load. All horizontal loads are transferred to the metal disk and therefore into the metal guide column. Measurements of the actual fullsize system have consistently indicated a 0. It is an energyabsorption system that can use almost any crushable material. ft 10 24 32 nu21/3 " 12 % 22 F Foam and reusable cribbing Universal guided column Fullstroke guided column tied to the trailer (Fig. the crush pads sit directly on plywood ground pads. 4. (E) reusable cribbing assenbly. Each end of this beam contains a guide in which a column is inserted. Fig. This forms a solid extension of the trailer structure to the normal suspension height and is capable of withstanding the horizontal loads. A typical installation is shown in Fig. grease.TABLE I ShockMitigation System Limits [Maximum System Vertical Ground Motion Velocity. A simple foam and plywood crush pad is bolted under the cribbing. /for  The bottom af the column is attached to a metal disk by means of a pivot joint.. .'. The crush strength of the polystyrene foam is constant the first of 140% deflection and increases for gradually to 35% about at 65%.. The vertically controlled crush load is transferred to the trailer via a cross beam. Horizontal loads into the disk are controlled by antifriction surface pads made of properly sized Teflon. (B) turnbuckle tie. the surface pad Is designed to have very low friction for a limited horizontal displacement while the structure is mounted high and then after slapdown the column moment arm is smaller and the system is allowed to slide onto higher friction c1 0 Fig. 69 . therefore.06 coefficient of friction. it can withstand the torsional loads. which imposes a twisting moment on the beam.'. 3. and acrylic. ft/sec Displacement. The horizontal force is transferred to the beam by means of the metal column. Commercially available polystyrene rigid foam has been the most inexpensive reliable material. . The pad design is based on many laboratory friction tests and field experience. This column extends through Jhe crushable material down to the surface pad. UNIVERSAL GUIDEDCOLUMN SYSTEM The universal guidedcolumn (UGC) system was designed for ground motions exceeding the F&RC system capabilities and up to a maximum of 24 ft/sec or about 12ft displacement. Because the beam is a boxtype structure. . (C) wood cribbing with plywood scabbing.: deflection.. it is strong enough to withstand the horizontal shear forces without additional guide structures. Styrofoam Is used because for the required heights and surface areas. Because verylowfriction surfaces are not required. '. (D) angle iron frame. 2). 3 Typical installation of Universal Guided Column (UGC) system.
B Momrent due to hrnl o The constantforce columnar energy absorber is called a TORSHOK and is manufacstage of this device consists of two concentric tubes with a coil of ductile wire forced between them.rgy absorber (see Fig. Nylon rope tie downs are also used to prevent excessive horizontal displacement and to absorb some of the energy. Vertical energyabsorption strokes have reached 40 in. with the crush materials replaced by a fullstroke. The fullstroke system allows tl! original height to be reduced by one third. surfaces. FULLSTROKE. constantforce proprietary energy absorber. 70 .) to the original height is designed to be about 35% for nominal design conditions and/or 65% for SECT AA Colun Rubber isolator Energy absorber Crossbeam maximumi credible conditions. (F) Teflon. Rotation of the wires is similar to rotating" the ring of a torus inside out.A primary difference in the system is the type and location of the ene. This design is a modification of the UGC system.'sec or 36 in. This system has been used reliably on more than twenty occasions with actual ground motions exceeding 12 ft. depending upon pua e Crushable material which case prevails. The crush materials bottom out at 65 to 75% deflection. Products. A single oC F IVeitical force Fig. (G)antifriction surface pad. 5). The resulting tensile and compressive strains are in the plastic range. (C) crushable material. The / / §//' Fig. Styrofoam. displacement. A. The structures for the fullstroke. 5 Characteristics of FSGC system. (B) crossbeam. guidedcolumn (FSGC) system are similar to the UGC system except that they are stronger. The interference fit between the wireand andto sliding the tubes is sufficient to prevent force the wires to rotate. and horizontal displacements up to 8 ft have occurred. and therefore the structural requirements for some of the basic components are also reduced by one third. All remaining components function the same as the UGC system. energyabsorption stroke relative etc. exploded for clarity. R. Forthe crush materials (honeycomb. (E)metal disc. 4 UGC characteristics: (A) column. GUIDEDCOLUMN SYSTEM A new shockmounting system has recently been designed to meet requirements for very severe vertical ground motion up to 32 ft/sec or about 22ft displacement. Contracting or extending the f tured by A. (D) pivot joint. Inc.
0 3.9 4 55 left side 77 right side t Gravity force not included Three energy absorbers failed and system bottoned oqit on right side.4 4 57 Actual Stroke (in.( tubes apart axially forces the wires to rotate many times. During the static tests loads were uniform and increased breakaway loads occurred only a few times. stroke. 5. TABLE 2 DropTest Results .) 7. Rubber isolators are used in mounting to insure that no bending moments are transferred into the energy absorber.0 Desi!n Conditions Stroke Force (in.7 t4. In all tests drop height. which was discovered before testing began. There were no changes in load during the tests. The axial component of the lateral resonant vibration was quite apparent during the compression drop tests.0 31.Average Values Deceleration Test No.5 38 74 12 180 Velocity at Impact (ft/sec) 10. 6 Prototype FSGC system before testing. Tilting occurred during the first three tests.) (g) 2. 6. on the FSGC beams and on the equipment inside the trailer. Thirteen accelerometers were placed on the top of the columns. L 71 ) .5 in.3 19. In one case.5 26.000 lb trailer was conducted. In order to use the fullstroke capability of the guidedcolumn system. and the average deceleration values agreed very well. 1 2 3 4 5 Drop Height (in. We are using threestage energy absorbers with a compressed length of 58.8 3.7 2. Test instrumentation consisted of acceleri ometers and highspeed movies.62 13. The FSGC system set up before drop testing is shown in Fig. Before incorporating TORSHOKs into a shockmounting system we purchased some for static and dynamic tests of individual units in both the extension and contraction directions. At some locations the clusters were arranged to impart very high eccentric loadings into the columns. Impact acceleration pulses (three times the average deceleration values) were measured on the payload in about half of the extension and compression drop tests. but averaged deceleration strokes agreed with test parameters. this was a function of two stages and not the overall length of the column. All channels were recorded on magnetic tape. f i cev aluatio n The drop tests are summarized in Table 2.000 lb. An erroneous center of gravity location. caused the tilting. and 8ft stroke capability with capacities up to 12. we attach the energy absorber between the top of the column and the beam. as shown in Fig. DROP TESTS A series of fullscale drop tests using a 30.17 7.000 lb.9 8.7 2.) 21. After the Fig. even though one unit experienced five complete strokes. The repeated cyclic plastic straining results in almost constant energy absorption per cycle of rotation (or inch of linear stroke) until eventual fatigue failure. with seven selected channels on m scietigraph . Each column can accommodate up to four energy absorbers with a total deceleration load capacity of 48. Various load ratings and cluster arrays of energy absorbers were tested.9 14 26.
Motion of the trailer during the third. . Two failed on the side that bottomed (see Fig. 72 Fig. Fifteen energy absorbers were used during the fifth test. Transducers 12 and 13 are at opposite ends of the trailer. showing two separated energy absorbers.6% of the total deceleration force. when the system bottomed..2 ced the end of their stroke and the spotwelded bands failed to retain the wires (see Fig. fourth. were 12 g (see Fig. 8). They represented 34% of the deceleration force on that side.third test the energy absorbers were reset with the correct center of gravity location. and one unit was (compressed) and redistributed in accordance _ 'q Fig. Impact acceleration at the top of the columns was 100 to 300 g. Fig.. Figure 7 shows the trailer after the fourth test during preparation forthe 15ft drop of the fifth test. They represented 26. . None of the equipment inside the trailer was damaged (including the fluorescent lights) and the only damage to the trailer was a wrinkled skin in one area requiring replacement of a few rivets. Several units were statically loaded to 25. Modified energy absorbers were then successfully tested. An initial peak pulse occurred only at one end during the second test and at both ends during the fifth test. Typical unfiltered acceleration response of equipment inside the trailer is shown in Fig. and three of them failed. 9. Peak accelerations inside the trailer. and fifth tests is shown in the movies. A AA A ff~T7~7T~N A. 9 Response of equipment inside trailer during tests 2 and 5. Failure of the three energy absorbers during the fifth test occurred when individual stages . 10). These initial peaks occurred randomly during the testing. 7 Prototype F5GC system ready for 15ft drop test.000 lb (twice the maximum absorption load). 8 Right side of FSGC system after 15ft drop test. 9).
" in Proc. I" Fig. Inc. the efforts of Mr. Normally we desire failsafe shockmitigation systems. 979993. we have installed rigidfoam backup pads. pp. Wer Nbof : 'o REFERENCE 1. R. This unit finally failed at a swaged tubing flange. 11 Field installation of FSGC system. H. Atomic Energy Commission Rept. Fuller and Mr.. 73 . (Nevada Test Site) in connection with the F&RC and UGC systems and EG&G. . P. Vol. Hulhall and Mr. These are under the protective polyethylene covers at the base of each column in Fig. ACKNOWLEDGMENT The authors would like to acknowledge the cooperation and assistance provided by Holmes & Narver. Jackson. Nakanishi of EG&G are gratefully acknowledged.impacttested to failure. Inc.. May 1970 II Fig. In actual field applications of this system. With Nuclear Explosives (Las Vegas. Montalvo H&N and Mr. Bernreuter. D. 2. C. CONF700101. B. L. U. on Eng. "Control of the Dvnamic Environment Produced by Underground Nuclear Explosives. Our comprehensive test program identified discrepancies in the asbuilt FSGC system. E. (Las Vegas) in the drop test portion of this program. 1970).S. Miller. and A. In particular. Symp. These were corrected before using the system under actual severe groundmotion conditions. 11. E... Nev. 10 Cross section of energy absorber at end of extension stroke.
Bernreuter: Yes we do quite a bit of calculation based on the weight of the trailer.DISCUSSION Mr. In this case they stroked 12 feet because they bottomed out. analysis. and we have performed a compute*. Mr. Peralta (Bell Telephone laboratories): In sizing the shock absorbers. Mr. Bernreuter: Yes. the actual extensionof the TOP SHOCKS themselves took up most of the shock. 74 . In fact. Fox (Barry Wright Corporation): How much deflection did you get out of the foam itself? Was most of the deflection taken up under the shock? Mr. Fox: You were only concerned about the inital shock"blast. We picked the trailer up 15 feet off the ground and dropped it. a couple of the units failed. Fox: Was that fifteen feet? Mr. do you do any kind of calculation as to how much energy they should be able to absorb? Mr. Bernreuter: Yes. I think their stroke was designed to be twelve feet capacity. the expected level. so it had a velocity of about 28 feet per second when It struck the ground. Mr. etc. You did not care what happened after theocital blast? Is that correct? h e Mr. Bernreuter: No. The formulas are worked out.
1 statictical qurntitJ. If it car.. bnlriv(! of the prond exactly the cam. uniform ground motion for tl. effects of seluinic excitation. Particular emphaiiis IN placel upon ire othe lya Thepro problem of the 4lynamic response of the utilizing standard atructural coessoitia nrd The recults ofr thi. seismic wave train must be connidered as it traverses the structure the assumption of uniform ground motion for the structurallarsembly can nn longer be Juatified.ure include the use of quanistatie load rar'tors. relatively slow.1 r) frxs'1 I'. A discuarion of ths fridpo mnI I r jv"%.iengt. n:hock spectra and.tructural nnalv' the ..~"':. six span. Tra..ultn eonidrit.noii.amo ieismic di.'.. A *1 A typical six span highway ibridge is presented. A finite element model of this bridge is develope4 including representation of soil/foundation stiffness and damping characteristics.1rration tlm. Mechanics *1 Los Angeles..rewhose ground motion varies between its supports and is dependent upon seismic wave propagation characteristics.e.ural 'voluqtion thin g:rond ttotion if. This paper Rd Considerable interest is currently being exprorced so to the behavior of civil structures uzidr thf. tri lo' e h.h time.r ur ar't.rn ie... '1 that the popoWf'tflrn t.. tr . that is.:. Research. to re. analytical techniques. uniform velocity..rint FI or an artifJ'Inlly i. tit each n. The method of:inalysin utilired for attacKintrcne or' aa jtruetural system to moving ground motion Ir eutlin. 1940 El Centro Earthquake * Moving Ground Acceleration 9 An Average Acceleration Shock Spectrum for Strong Ground Motion An analytical procedure is !outlined for predicting the response of a highway bridge structi. A 1trler ummary ir %r' Pte Is lndleate'.a .e.tsnumed ground motion is often given in the f'onri of an acceleration timn history recorded durini a neinmic .iior. The response of this structure is evaluated for the various ground motion conditions using normal mode techniques..irrlptleo of the bridge ansembly.pecial casve of time history analysis of a seismic 'aer.enerated time hi.e * Qro with ltuc l pre' 75 COPY .rirtit. The dynamic analysis of this st'ucture was predicted using the INTODUCTION tMHI/STARDYNIE Structural.Inri. This " motion nnalycsi technique are compared with more spectra and shock history time tra. 11.n.I.to the foundation and 'rfonol suppoktr with thr' diftlrJn n unif'orm manner.ho tte' imn .. . inelmior the geotrmry qn . ' . a1. rettined [. for wtrone.'n th" propalration velocity watvr 'movinfg !on(itdinally along the structure 1con::idered. rcura..iI urrtel was %Aste. liRults from this moving groun. ffort were then compare. For .tmutrel r . be anticipated that the ground motion may differ In phase and magnitude among several structural supports then the resulting structural response must be reviewed carefully.A 'Ir.ll1n . nt . in some eases. aseumetd aoil properti fvn'. m '.round mot. ped^ntal r:upport highway bridge structur..l analyris techniques applied to such :'tr et.'. California This paper compares the dynamic response of a highway bridge structure subjected to: * The Uniform Ground Acceleration Time History for the May 18. . a typiclu two lane.turbance and to reculto conil srldeng a typlenl average sheock speetrn. Analysic System.THE C014PARISON OF THE RESPONSE OF A HIGHWAY BRIDGE TO UNIFORM GROUND SHOCK AND MOVING GROUND EXCITATION (U) Galletly Robert D. This paper ad drersen n hypothetical case where a strong: elmic disturbnce propagates along the prounl supportino.es of truct.. For purpo. of a When tho .. 'y with certain nimrlni qn.. sinrh. porint. Inc.4 in detail. Johnson and Neil E.histories.itionil three techof all application the through niuen three ofa the iplicatinh tenique toh techniques to an idealizrd hirhway bridge.. generall:Y rpplre.
n Fg. in Table A brief descripton o f the norml r ode ethod an the ters in Tble T is given in Appendix A. toe supporte bridge sv paand in use by a large number of engineering firm tpe fodane a s. Note that a much more detailed model might be desired if an actual highway bridge were to be analyzed using the techniques described herein. vertical spxilaring r srutne s i eT dex The example bridge was modeled as a lumped atredatrtyia s n srcue tyia mass. 1. The generalized weight and modal particpaascmtrilwt h s onrt asm screi tion factors correspnding to each o these c mldes are also I given m. 1 " Blridge Physical Characteristics and Idealized )'thematica.2 23 Nodes with Lumped 2 Coordinate System Pig. ~ ~ ~ SpotclmsaBrid ~ Fg wa scaphrct ge risisan The is azdYN rgadvlpdb A mathematical model of a typical 5 ixspan. e 76 . 2 is n iealzaton in Table 3. Fg. These springs have been designed to give reasonable is a large (up based to 6000 itatic and dy oamic analysis system onDOF) the finite element. mode shapes and related properties arc identified.le 2. Normal mode techniques were utilized in the problem formulation. The bridge. The roadway superstructure and support columrns singe sppot pdesal hghwa brdgewaswere modeled as a series of beam mrembers. M4odel A finite element model of the bridge structural system was developed. The seldtehniued. if used. Foundation/soEf interaction was idealized as a series of linear support springs. Si. This modtl consists of 45 unconstrained nodes each having 6 degrtes of freedom. Only 2 n es were selct. prpit oilain o enocn te.Sch a meogeel Using the finite element model described Thew lnidealized bridgeconsiststh of six 120 foot above.cdeveloped to illustrate the analysis techniques tion properties evaluated for these beams are described herein.eTNE STA l Anas Semomputer ens ae assumed.. wnich is shown i~ Fi. Reinforced concrete construethegram. 9 m~ a:d isan dealzaton f f agiven aassume concrete as These the basic material with typical structure and is patterned after typical aprpit adfctosfrrifrigsel ~~~~twolane highway bridge fyeo designs. Such a model would probably include a more extensive threedimensional model considering both beam and plate finite elements and a more detailed description of soil/structure interaction including pile foundations. normal mode bethod of analysis. worldwide through CoNtrol Data Corporation Data typefoudatins. section properties Fi. ss poi locations dt in order bending or longitudinal are summarized in Table .ed as m.Centers. the natural frequencies and normal modes long toln spn wih heroadway superstrucof' the bridge were determined through the use of ture placed approximately 50 feet above the teSAON tutrlAayi ytmcmue ground urface. The STARDYNE Structural Analysis System and its auxiliary subroutines were used exclusively for the modal vibration analysis and the dynamic response computations. tion ~ asue. The natural frequencies. The assumed mass distribution summarized in Ta. finite element ap shown in Fig. The lowest sixty (60) normal modes were   translational and rotational stiffness properdermnd Ths oswihcnb gnral sri te aSsrued nye ob e Ta soi classed as either horizontal bending. Teodes her i. BRIDGE MODEL to minimize the number of dynamic degreos of freedom for the problem.: l utiled i the prolme raion.
60 h.. .35) . A modal damping factor of .5 SUMARY OF MASS PROPERTIES t Element Mass Property Fig.)47.AM PROPERTI ES Beam Type Area In J in 12 in I3 in Shear Shape Factors 2 Roadway 8111.8 !bs/ft 201. 77 .1450 x 106 lb/in Assume 2 sq. 1.282 b.2P'5 x 106 lb/in 1.240 0. )71. ..71 ). Each R3.2.1475 x 109 in lb S6 x 109 in ib 9.215 x 106 lb/In 2. TA= I SUMIARY OF SOIL SPRING CONSTANTS 8 4'6" Direction Soil Spring Rate pr[ 61J 121" X) X2 X3 1...44 1i3 ].05 was used for all response analyses. in.o5". Supp__rt Colun 23114 . steel per foot of length . 2  Typical CrossSection Roadway Superstructure Support Columns Pad Foundations 84.32 lbs/ft 23414.522.662 0.21l. it should be noted that only a portion of the normal modes listed in 17"0" Table4 were used in the various dynamic response analyses described below.76.4 10" 9.l'h. V2.Due to the symmetry properties of the idealized structural system.. .
0 0 oQ.l Be.4hil 14.434 7.ding Horizontal Bending Horizontal Bending Vertical Bending Vertical Bending bongitudi hal Vertical '.449 '.0183 Vertical Ber.709 3178 27 1.57)1 0 0 0 0 0 O. )' 54 1.41 9 2.1081 0 0.013 7.577 1..zontail iending orizontal orl zont i] 10.997 4. "173 *¢:2 on tai 1..13 "0I.0160 0 0 0 0 0 0.058 6.3 32 ]80 1 82 0 O.. 00.e1 5 15. 2') 0 or:or e I i.IA> ' "  ).718 1053 10.005 2.2141 7. "'.2 0 0 O.56 0 0.TABLE 4 RESULTS OF THE MODAL ANALYSIS Mode No.021 3.0588 0 0.247" 0 0 0.218P 0 0 lon1i tudinal 1 Vertical Bendin. * 0 0 dor zontal Ion.W r.774 1 .827 3207 58O 35P2 753 585 1227 620 5)2 701 ?)l 25 26 27 28 . i rwtl )' . ..782 1.512 6.Y17.413 0 0 0 0 0 0 0 0 0. 111 Horizontal Bending flor zontil Bending Longi tul nal llcrizontal Bending Vertical Banding Vertical Bending Vertical Bending Vertical Bending Vertical Bendingt Vertical Bending Horizontal Bendin nori zontal iforizonzal Rotation Hori zontal Pending Horizontal Bending Vertiva! Bendin. 0 0 0 0 iorl zo:It] 10 orn4(". 2 10"' 2 . V1' 18. 0'0.88 1188 14 45 40.'c . 15 . 0 0 0.0062 0 0 0.10.8))5 0. 170 0 0 0 0 0.0414 0 0.0. ) 00 0 I "orizontal 8ori zoistal Bendi n.492 7.731 Genralized Weight (kips) 3360 3h27 6557 3570 17211 1396 165) 2669 1958 '°84 2557 43 . ending !aoOh 42 4 13.1 "'4 '.981 7.717 3.73 12.037 3.4 0 8.176 13.7 i i.77 * 14 .: 51 '13 i .3 0 0 0 0 0 0 ). )! . f 5'0 ..50. .319 3. 0240 0 0 0 1) 0 0 1.6009 0 0.033.058 ).1 17 38 ]( 20 21 22 23 6.454 5.511 1 R.t 0 0 Q. 3618 0 0.itdinal 'iorni.P120 0 0 .20) 7.I 11. Ion. 315 I4.7? 38...t't ho .15) 1. ) 3204 15 A4 1811O) 2188 Modal Participation Factors x x3 o tbde Decription 0 0 i.' 1.. f 0 0 0 1. "i .(P3% 0 Vertical Bending 'ors i on 8. 'id.0444 0 0 .Horzontal Vertical Pending 0 .5317 0.orlzontal Longiltudinal Longitutdinal Horizontal Longitudinal 29 30 31 32 7.'dinal 1 .on'al BLndin 0 0 0 0 0 o. 47 48 I 1.259 3412 1773 . on1:t'l i'll ori.. 1 2 3 4 5 6 7 8 9 10 11 13 11h 15 i( Frequency (cps) 0.1)0 7. ) 3. 8or! zontal 0 0 0 0.185 3.ori.502 3.(l')5c 15 Y) 1.j' 7 )~f 0 0 0 ) .259 7.'43 17.0.).186 u.4O6 5.03b2 o 0 0 0 0.572 761 15)2 5 1250 3 34 35 i 18 3' j 7.'ontal Bendin.)5 0 0 0 0 0 0 0.181. (7)4 0. .278 7.676 7.) I121 P!I'.tP 78 .1431 3.256 7.
C .K f 1 f'C [c'1 (8) is a rectangular array of stiffness coefficients relating ground attach nodes to adjacent internal nodes of the system. Fh. where n nNq is the diagonal mass matrix of the system. Considering this effect. This is done for ease of the present analysis. the forces induced by ground motion on the constrained system (right hand side of Eq. nn n[C) is the damping matrix of the system. u. a rectangular array of dampini: coefficients relating ground attach nodes to adiacent internal noles of tree system. In the absence of ground motion the equations of motion are: Just as the terms on the left hand side of the equations can be considered as the forces acting upon the system in the absence of ground motion (lyl =jW =(Y' 0). This paper addresses the problem of the response of structures with large characteristic length and considers the case where unattenuated strong ground motion traverses the supports of a structure at a uniform velocity. Next. Introducing a new setof auxiliary coordinates Jz(. . ('.'rrr.. the equations of motion become scribed in many sources (5... describes the absolute motion of the mass elements of the system. (2). [K] XI=0 0) wcoefficients. For seismic analysis the ground motion is often described in terms of ground acceleration. (0S) are identical expressions since it was previously assumed that [C) =o{KI and 'c'] =a . is a transformation of ?onstant n1 nIz I is an auxiliary set of coordinates into Eq. I shows a typical system. the terms of the right hand side can be viewed as those forces Induced upon the system by the ground motions lyf when the mass points are constrained (lxi = .il = x] =O). (4). i *JK . the existenc. (3)) may be written K]IZI+ [CilC I =']lyl +['Jlfl () Substituting the transformation Izi = [T)JYj where r niT] (5) rmIYJI+CI1 .of iK1 is assured since !. The method is developed In a straightforward manner and is outlined below. (6) it may be written that [KI(TI =I]: and [C][TJ= !c') Eqs. assume the damping is small and proportional to the stiffness. it is seen that [K][Tflyl + [C][Tfl [ I. (1) be writken h. nn [K) n[x] is the stiffness matrix of the system. Consider a system with n degrees of freedom and r resilient supports to ground. This provides a straightforward way for constructing [c') and [').es 79 . (10. (7) may be rewritten to give (7) A For ground motion of the supports the forces induced upon the system are described in terms of' relative motion between the mass elements and the support elements.Wflyl + (5) computational tools are available to the structural analyst.eco. (3) may be rewritten in the following manner to accommodate these terms. the equations of motion can be written ~.Jjyl + . d'scribes thp absOlute ivround1 motions (displacements) nt each of the r supports.W 'KJlr7 + ( 1l ) hoarrn{ in. F'. Assume that the equations of motion are to be written in centerofmass coordinates and the mass matrix of the system Nt] is diagonal. Both transient response techniques (5] and shock spectra methods (6] have achieved wide acceptance in predicting the behavior of structural systems subjected to ground motion. other restricted types of damping could be considered.]iJi (61 Comparing like coefficients in Eq.'J (9) r n . 4]. modal damping will be utilized and modal coupling due to damping will be neglected.PRESENTATION OF THE METHOD OF ANALYSIS The earthquake response of structures is de Rearranging Eq.ay 1 r ly! j'zx . Modal techniques will be introduced.] repreents the stiffness %atrixof a constrained zyster. The two right hand portions of Eq.~( 'I[C]l where r n[kI] l c'IJ] PkI+'I'h +[K]ix! o (P) (T) .'in F1. . A variety of I cl' N 4I 1xl . Furthermore. Eq. Fig. For simplicity.
(13) may be written I! or +{c]I +[K]I f Nt [KM"I [k'1 J [D][k') l (15) 4 4.o Defining (11) Then IeI =xlf. Further.N4I'i+[C(iklfli)+[K(txlIzi). $s4 Plane Wave Front Time Fig. h shows the time histories 6ir a ground disturbance for the special case of several evenly spaced colinear ground points. "I =[T] I= Ill f[K (1i) Eq. is the angle between a line adjoining the first ground point and the ith ground point and the direction of wave propaga43 LAAA AA tion. it can be assumed that the basic ground disturbance has a plane wave front and a time history given by 5'(t). t where t di O tdcosoiI v die i (see Fig.2 = and subtracting 1JYiJ from both sides. (16) f the form 'f' where 'Ii (19) r di (j is an arraj of mode shapes o& the undamped constrained system. is the distance from the first ground point to tie i t h ground point. Fig.)rdinates. 4  '  tr . 3  Wave Train Propagation 80 .(t) (Y(t) t ) I Y (t. 3) (17) 04) 0 VVV is the time of arrival of the disturbance at the ith ground point. IIin a set of Penerallzerl (modil) co. (11) may be written ii= Y~) Ytt)(8 1 I'v~~~~~~~~~lY KII: I' cIIi Noting that Yi' . 0 0s. let the velocity of propagation for the disturbance be v and the time the disturbance encounters the first ground point be t D. For the case of moving ground motion.. 13.tr) Fig. C N14l +[C]II +IKJIB (16) 0 P4 H where (D] =N[K I1 is the dynamical matrix for the constrained system. 0 )]lY'(t){ y. v iz the velocity of propagation.. . the disturbance will arrive at successive ground points at a time 41 4 a . Fq. Then.(t .Orcuind t Time History for Moving Ground Motion Point ith Ground VPoint Consider a solution to Eq.
1 J*. * Transient response of the structural system subjected to the uniform ground acceleration time history for tL.JI I ( = The equations of motion become RESULTS Using the loading conditions described above. the following representative data are presented. in the usual manner. These results can be After solving Eq. The horizontal component of the ground ° acceleration in a direction 45 to the longitudinal direction of the bridgc with a magnitude equal to that for the May 18. = J I¢J=X ' ' .yrtn when both end supports have the identical ground excitation. Acceleration responses of a point on the roadway at the midspan of the bridge are presented Fig. This phenomena would increase the probability ofachleving n extreme peak response due toa fortuitous combinationof modal responses. Is show. 1940 El Centro earthquake.it We~ equal to tat for the May 18. To briefly illustrate some the results obtained and to provide a basis for comparing the three methods used.generally of longer duration for the Moving The loading conditions considered were: " The Lorizontal componcnt of the ground acceleration in the lateral direction of the bridge with a nagr. a series of nine independent analyses were performed. 12 show similar acceleration time histories using the Moving Ground Motion (MG4) method. 7. Eq. The vertical component of ground acceleration equal to 0.fTHY Iin INPUT DATA For this study three methods for solving the dynamic response of a structural system to seisinic inputs were used. (16) becomes SV 0) On defining. It should be noted that the acceleration values shown on these figures for various span locations do not necessarily occur at the same point in time. Some of the more significant of these are described below. The average acceleration shock spectrum for. time range of peak response of the bridge is for strong ground motion is shown in Fig.l Centro earthquake. Each of these analyses resulted in the computation of displacements. The vertical component of ground scceleratLion equal to 0. The methods wereFigs. For each method..ns at the various nodes on the I leal ~JI'i + + (21) ized bridge structure. CONCLUSION A number of conclusions may be drawn from a review of the analysis results presented above.6 times the horizontal component (Case3). * The horizont! component of the ground acceleration in the longtudinal direction of the bridge with a magnitude equal to Ground Motion cases than for the Uniform Ground Motion cases. Figs. 1940 El Centro earthquake.( times the horizontal component (Case I). 19O F. . that for the May 18. velocities and accelerat. 7 through 12. 191. three loading conditions were considered. 1. the generalized parameters . 9 and 11 show representative acceleration time histories using the Uniform Ground Motion (UGM) :ethod for each of the three load chses. An examination of Fig. I~x! 4. This is due to very largerol. For the Movinv Grounli case. Moving Ground Motion and Shock Spectra (SS) analysis methods. Shown on each figure are results from the Uniform Ground Motion. the longitudinal response is Aignificintly higher than for Moving Ground Motion. a reIur 81 . The vertical component of ground acceleration equal to 0. the absolute accelerations of the system become ' and 72) ] = li! . 1940 El Centro earthquake (NorthSouth component). by [O]T. 8. 19110 El Centro earthquake (NorthSouth component). 10 u. Transient response of the structural system subjected to moving ground motion with a constant propagation velocity of 400 ft/ see and an acceleration time history given by the May 18. strong grage mo eeratios n shon sp. [¢]T N([I = r'aj I JT(CJ( 4].IZ  (TJYJ further extended to include equivalent nodal forces and internal member loads and stresses. 7 through I that the Fig. Consequently. Substituting and premultiplyine.!. (21) for I1.alparticipation of" the first longitudinal mo'le of the . Figs. some care should be used when interpreting these analyis results. 1 for span locations).0 El Centro earthquake (NorthSouth component) is shown in It is seen from Figs. * 9 The time history of the May 18. Hespon~e of the structural system subjected to the average acceleration shock spectrum for strong ground motion 153.6 times the horizontal component (Case 2). 14 and 15 show the distribution of maximum values of acceleration for each load condition as a function of bridge span (see Fig. 5. that for Uniform Ground Motion. 'ay 18.
oo I 2a .8 ..4 analysis. Undamped Natural Period.he briIe.. The probblity of higher responses tends to increase as the number of supports inConsequently.. . 94 No. For a given ground motion time history respo~nse derived from shock spectra analysis will bound the response developed from a time history analy'sis for either the Uniform or Moving Ground Motion method. Hurtyand ieg. Amin tnd A. Robert L.h. 6response >Shock Spectrum Curves 01 . F. ThiS prnpnr'y n tte. Robert..i. For a shock ~utilized . This ii due S.M. heF.Vr*an . New eiersey.6 2.Jesrsy. n: Voals *wn m'd . April si in ceround it time h orides eects. o rn l oftib Applied Mechsnics Division. Procurement hpr a0y . Ang "Nonstationary Stoehn Efh. response spectrum selected for this con 0The parison was a composite average of acceleration : ..trtures. Harrnd. 'in. Wiegel. I1. .0 i  Percent of Critical Damping shock spectra for several strong ground motion events [3. mif dusturhane. * u[ .n in th2.c . 1970 P. the resiponses to 1. PrenticeHall.pen upon both the loafflnt.'. Since this spectrum does not bound that derived from the ground motion time history for this problem the resulting response does not necessarily encompass the results obtained from time history studies.round M'tion method.5  0 Fig. REFdRet . structural respnse than analysis the phase relationships are lost nd a pessimistic summation of modal response is be computed from a Uniform ground Motion _ _creases.In .. summary. reSponse rersult doot occur f'or 'i Mwvlni' g. s 82 . . :)ynai Hll. it can generally be concluded that" an evaluation of Moving Ground Motion effects Sresult in different. assumed. ASCE. enerally. For thi reason. 19140.50 0215 0. 1.ihout theei'pan of' . It can be further concluded that time history response analysis of either the Moving or Unifor Ground Motion types will give moretmehnondful predictions than those obtained from Spectra analysis when many modes contribute to the response of the uystem.. 1.. o 0 0.t+h.round Motion exhibit :. 1."I . realistic and economieal highway bridge design practice Ghould consider tine history effects. rl M..'rawle.8 1. Vol.iraitry prop..6 Ospectra = o4 '.would .. ptrticipation of the first longitucinal mode w tinn r 'ch nndupportwa distintly different. M. it can be recommended that Moving Ground Motion should be considered in the seismic design of multiple support brde . Sec to .oon . of the rosp travets's on gothe wiffer b n t hethev ' produein hcit . 5  10 Time. )~.* v'rltpian . Earthquake Fgineerin. PrentlceePt New 970 .lre tlon considred.rties . Uniform G. V'. shc 15 20 25 50 NorthSouth Coirponent of Horizontal Ground Acceleration of El Centro Earthquake May 18.. ue to 'he' bridpe symmetry.' . e =0. h~ 'n tiron ofc opnde l..r p observed when the ground excita Ic Models of Earthquake Motion".4 / '.tinst in.
e " 4 lpnertical 1 Mid~f tin ud ivll .25 Time. moo flt. see Kid SpunVertical.22 7v 8.2 . 23.3 . see Fig.oNc M kid Sjpn Longitudical Aceleraton History Moving GrOund W.Hid Span'Horitontal Acceleratioo History thiforn Oroua Mtion Fig.tio vvvv 3CM Came 3 . 0 .2 0 v. Ae AWAA Z ...1 ". 10 . . I 2 3 lraln. 8 Time.V __. see Kid Span Horizontol ActeleratlonHistory ov4'rg Grond lotlon Case 2  UGH Came 3M 0 0 2 3 4 50 1 3 Time see 4 3 Tie*. 2 0 1 Figure It 5 4 3 Tio'. 4 Ti. i d1 Span Longitudinal Acceleration History UniformGround Mtion Fig. Acceleration History Wlifors Crownd Flotion 2 0 1 ?Ig. 7 .
.oA. 1aximt1 . 0 01 1 I I I ..5 o  / _ . o It . ". a I .o\'.I.0*<. . 13 . ".5 1..0 1.O . . i .1/ / r%.94 ~' . . I I 0.0 .1'd ". 2 3 5 7 9 !I1 1 7 9 > Span iocation Fig. \. "0 .... .  G A .0 o .MGM [3.lMM aSS ... 1. . /1\a// l0.. ..0OO0 0 .. .. <)O <'..Maximum Vertical Acceleration Distribution (Case ") 84 .  0 ..0 A.UGM A.I ..4 ." . O ..o.UG' A0 V IJM 43/" . . I I 2.. 1 1. A. 1...lMhimun Horizontal Acceleration Distribution (Case I) 7 5 9 25 O. 9 11 15 1 1) 1 19 21 25 25 Span Location 1) .. Lonitudinal Acceleration Dintributon (Case ") 0. 1 3 5 Fi . .SS . I 3 195 11 13 15 17 19 21 P3 Span Location Fig.. < ot 1 I m I 10%" % ...
The rodal participation factorr :j~r) are fined to be . Mechanics Research. (1) The cigenvalues (natural frequencies) and eigenvectors (normal modes) of the system can be extrar'ted from the following reformulated version of Eq. Inc.. (r) is the r The generalized mass of the structure is given by The generalized weight is given by (4) where g s tho Gravitational constant. . Mechanics Research. 1968 APPENDIX A: DEFINITION OF ANALYSIS TERMS The STARDYNE System is formulated around the "Stiffness" or "Displacement" method of structural analysis. Richard Rosen. In using this method.1. 1971 Note that modal participation factors are computed for each of the threetranslational degrees of freedom. Inc.th th elgenvalue. Haelsig. eigenvector..mil'ir "a Mr where Pi(r) is the modal participation factr for 'he rth mode in jth translatIona direction. DYNRE IV User's Guide. m] is the oecntL of rars Ith d'oreo 'f freedon. r11r is 'C ~85  p.5. () K where o 2 1. Richard T. is the stiffness matrix. .( =j 0~ (2) is the 1. STARDYNE: User's Manual.soeiati with '1ir of rth 7o1al vector tho component ie a srocinte'l with i t !' dereop f frre'ioq the r'enerali7e'] mafrr for rth mo. 6. the structural system can be represented by the following equations of motion (for free vibration): r?4Ji 'j + [KF]Ix =1. is the system coordinate. et al.0 where M4 [Ki jxj is the diagonal mass matrix. and Control Data Corporation.
Hlaskell Vulnerability Laboratory U. i number of mcastirement.e.in reflected mine blast energy flux density. This method of analysis will facilitate vulnerability assessments and engineering design of armor subjected to mine blast attack. in base of the natural system of logarithms plate tensile no.ness i. normally reflected pressure from the blast wave. in nofracture thic. (1 * F YeF: 11 plate strain energy of deformation.i1' DE:OI1MATION AND FRACTURE OF TANK BOTTOM HULL PLATES SUBJECTED TO MINE BLAST Donald F. The method of analysis is based on large structural plastic deformation. ii r plate strain energy absorbed to fracture per ui. U • 1 a. accurate. amplitude of displacement w (sec Fig l). in mean thickness at hlich fracture occurs.it plate . a semiinverse energy method and a reasonable description of material behavior including a static stressstrain curve and strength failure criterion.835 .no fracture (951e probability that fracture %ill not occur) typial measurement mean value of tyi ical measurement.b e el: h h plate strain energy.e. and directly applicable tool for vulnerability assessments and engineering design. 0.S. plate thickness with a 951 probability that fracture witl h oc a . 11) width an'Jlength respectively of rectan gular plate [see Fig I).. thickness with 95: probability of fracture.315 . Plate deformation is calculated by equating blast energy imparted to the plate to the strain energy absorbed by the plate in reaching its final deformed shape.e. Md. plate thickness. ntc tI nb/ standard deviation of the average relative error.mean fracture. ip surefracture thickness i. see Fig 1).'mal fracture strain. 1. The results of the effort are presented in simple graph and nomogram format for rapid armor areal weight determinations and mine blasttank bottom plate evaluations. Preceding page blank 8 8"7 : _2 .. These formulations developed correlate within an average error of 7% with available aluminum and steel plate mine blast test results. 1'si M I Ih I standoff (platebottom to miletop distance i. inlb/in' ) H p S 1. upoa mit hat frtactr will not occur upon mine blast attack. An analytical study has been performed to develop a method for predicting the deformation and fracture characteristics of flat rectangular tank bottom hull plates subjected to blast attack from shallow buried mines located under the plate's center.rea.sure fracture (95% probability of fracture).e. The analytical method developed does not require electronic digital computers in its application.. LIST OF SYMBOIS A VE M Fy K maximum transverse plate deformatinn i. inlb/in 2 plate tensile yield strength. inlb mine explosive charge weight.. Army Ballistic Research Laboratories Aberdeen Proving Ground. Blast loading is characterized by the energy associated with the blast wave. and is an easy to use. psi plate fracture criterion constant with the following values: 1.791 .
The data consist of restuIts from armor plate . These equations should be particularly useful for quick vulnerability estimates and in the initial design phases of land mines and most armored vehicle design projects. To be efficient and effective. lit ot fracurt i. placed on top of two steel beams measuring apt ro~imate ly twelve inches square by sixty inches in length wi tir the beams resting on the ground. A rectangular Otccl frame heighing three and a half tong and an 8 ton plate for a total holddohn nation and fracture thickness: l I lam . edges that are subjected to blast from shallow buried mines centrally located tunder the plate (see Figure 1). an approximate method of aiialysis was selected to treat the subject problem. to 3 inches.23 0A4I e " 1 1.5 for the meedian fractur( 8. were recorded. The Amy BDARP mine category data collected described damage to the equipment including such factors as damage location.000 psi and 0.35 . deformation and the extent ef cracking. test plate a show in ijure 2 wa. and armored vehicle designer need a methodology for determining the effects of mine blast attack on the integrity of armored vehicle hull bottom plates. for or susure fracture.000 psi to 172. form so practical answers can be easily obtaied in a short time.7 ". probaih. I2 F 1/2 F A () h1 where  K I. the deformation and fracture claracteristics of armored vehicle hull bottom plates attacked by blast type mines.m m (l4) I'  (2) . The mine was then covered with approximatel:' 3 inches of loosely packed soil. while thickness of the aluminum plates ranged from 3/4 in.e. this methodology should be reasonably accurate and account for the various factors that influenL.. All plates here essentially the same size either .. The result of this treatment is a simple method for rapidly predicting. The particular equations developed are applicable to flat rectangular plates of uniform thickness with fully clamped. A mine %as then positioned centril) k ( I 1. tank bottom plate integrity in actual mine blast encounters. to 1 1/2 in. In the ast the information required to assess minetank bottom encounters had to be obtained by rather expensive and time consuming test programs .I) TF. by 65 in.10 in.' : that fracture hill not .. to 0.Ikor 88 2 55 . Tables I and II list the deformation and fracture test results of rolled homogeneous armor steel plates and 5083 aluminum plates.A'IION 01' ANALYSIS A. Charge weight ranged from 2.S holdmay (Tis downtotest plate weight ratio may be compared to the typical tanktobotton plate height ratio of 24. (3) Ii totest plate height ratio of 28 were then placed on top of the test plate. Yield strength and failure strain for the 5083 aluminum alloy plates here obtained from MIL 111)K 23 specifications. The test procedure was as follous. "ILST I)hSCRltIP'ION All the data presented in this report has obtained from available test results and battle damage records. 9. or builtin.SIULTS leformation and Fracture Relations. it should be simple i. measured or specified hardness values were converted to yield strength and failure strain by neans of well known relations (Ref.15 in/in. In the APG tests 'he mines included pancaketype service and experimental cased mines as well as experimental bare pancake A charges. The deformation and fracture analysis developed yields the foliloing relations for plate defor damage data collected by the BDAI6P (Battle Damaqe Assessment and Reporting Program) teams operating in South Vi.31. or 44 in.791 for no fractur(.ST RE. INTRODUCTION The vulnerabi!ity analyst. CORIU:I. estimates of the necessary parameters were obtained from the material specifications. if any. 95'. Such a methodology has been unavailable up to the present time. Correlation betheen these test results a:d predictions of the analysis i! discussed in the next section.. with reasonable accuracy and without recourse to a high speed electronic digital computer.analytical methods to develop this information were not available. All the aluminum plates vere apparently of the same strength and ductility whereas the teel plates ranged in strength and ductility from 90. respectively.5 to 24 lbs with a constant stan'off and burial depth of 17 inches and 3 inches. ght ie.tnam. that is. i). After detonation of the mine.23 in/in. In addition. by 75 in. This situation was caused by the very complicated nature of the problem. respectively. 0.mine blast tests performed at Aberdeen Proving Groutd as well as combat under the test plate in a predug hole leaving a distance of 17 inches between the top surface of the mine and bottom surface of the test plate. In regard to the steel armor plate. munitions designer. Since mechanical properties test results for cach of the plates tested were not available. The steel plates ranged in thickness from 3/4 in. To avoid these complicationa. systems affected and threat involved. 1.) 1is fixture provided eseit ally clamped (fixcd) boundary conditions all around Iprobahii the plate.
:est data is S . hlr. Figure 6 presents the conditions for plate thickness with a 95% (or + 2u) probability of no fracture.% probaIt is based on Eq. h" at %illch culate the a widthtolength a:pect ratio with R train energ) 89 . The test data is as follows: circles represent 5083 aliminum and the squares class 2 rolled hon. Figure 5 i.315 where this value of K has been determined by a least squares fit of the plate fracture data. these nomograms a ) tough ness II.6% and the average abiolute error is 13.2%. the theoretical curve is slightly below the least squares curve. plates of this Ia b 2 2 Fracture Correlation. namely. in Through five scales. Figure 7. limits as the dotted () lines. 1 as well as the deviation of these predictions from the test results.. the average positive error is 14% and the negative error is 18. A relatively simple and straightforward means for calculating the thickness of a given plate that will fracture when attacked by mine blast is presented by the nomograms of Figures S through 7. plate deformation predicted by the present theory is. 1. as indicated by Figure 4. For example. action. 2 with K r 1. Table I lists theoretical results for plate deformation calculated by use of Eq. the average error of this relation is 0.icture cri terion and neglect of bending alc tonl TLe procedure for using the fracture notnograms is as follows. was derived from first principles without making use of test data to develop the equation. To calthihtlthii% a plite of thioknes. inadequacy of the tr. until this analysis is improved to bring the theory into closer agreement with test. m That is. although the :xact value of the fracture criterion constant K was obtained from a least squares fit of test data.e method for predicting deformation is desire(* tile equation corresponding to either of the 2o limits could be employed. adequate fracturc predictions may be made by using the least squares data fit equation as the fracture criterion. These test and analytical results are plotted in Figure 3. 1. Plate thickness that will be breeched by 95% of the mine attacks may be obtained from Figure 7. i.4%. Besides. Lach data point represents a single test plate or the average of either two or three tests. posible reasons for this diserep. A least squares fit of the data is shown as the solid (. Eq.315). .sted in Table 11 is plotted in Figure 4. the equation for the 2o limit curve of Figure . 6.6%. Iowever. 2).2%.4% lower than would be obtained in practice. scaled standoff 1 '3 . This nomogram is based on Equation 2 with K 1.835 and 1. Along with the test data points this figure shows the least squares fit curve of the test data (equation 2 with k 1.1. or 95% probability of occurrence. In this figure the test data has beei. Eq. As indicated previously. The dashed line is the theoretical curve. The theoretical criterion for fracture is elsentially thickness have a 95% probability of retaining thines a e a m natty This their integrity after a mine attack. a nomogram for mean fracture thickness bosed on Eq. 2 with bility of fracture. In this figure plate strain energy absorbed to the point of fracture per unit plate area S is plotted versus mine blast energy flux cisity . Consequently. In adit ioi there are two unlabeled scales used in the * whereas the last squaiesdta fit equation for the "est iscalculation%. First principles were also used in developing the form of the fracture relation. use of the upper 2o curve to predict deformation would yield results that 95% of the time are higher than would be obtained in actual practice. As indicated. DISCUSSION General. it represents 9.e.) line with the 2o. plotted as reduced deformation amplitude A defined by: = 2A a 2 b 2 F hI11/2 Y versus mine blast energy flux density E m . As indicated.835. fracture thickness hi.. The surefracture nomogram. The deformation relation. the actual tank mine damage data points which were not used in arriving at the least squares equation and its 2a limits fall well within the 2a probability limits of the least squares fit equation.m according to Lq. If a conservati'. will be used as an example. As in Figure 3 the circles and squares represent 5083 aluminum and class 2 rolled homogeneous armor steel plates respectively.791) and the theoictial fracture Iine.Deformation Correlation. The aveiage absolute error between this least squares fit equation and the plate test data is 13.0. and mine charge weight h.791.on the average.1% with the overall average error 6. That is. These nlomograms have been prepared from the fracture thickness equatio)n (Eq. Plate fracture data li.iny between theory and test results. 2.ogeneous steel armor. This figure is a nomogram for sure fracture thickness. The average absolute error is 16. K = 0. 2o probability limits (equation 2 with K equal to 0. the lope of the least squares data f:t is s g~lht ly hl~'ler than half that of the theorctohi th teoi is~~~~clt titan hal tht here may be two iL i1 phte fracture critorio.533 I. Tile triangles correspond to actual tank damage from nine blast in South Vietnam. 2.
11 with 95% certainty under attack from a mine of charge weight W at a standoff R: 1. 95$ probability that the plate will not fracture. T. and draw a straight line through these points intersecting line a at point 3.tlon 90 . Ed. REFERENCES I. Draw a straight line through points 3 and 6 interesecting the sure fracture thickness scale hsF at point 7. the plate toughness on the U. Two equations have been developed. P y IV) Figuue t . Vol. 3. The average error between predictions of these equations and test results. point 1. Metals Handbook. one for deformation and one for fracture of flat. American Society for Metals. and the scaled standoff on the R/W scale. 1. From the fracture criterion a set of plate fracture nomograms have been prepared that yield plate thickness for 95% probability of fracture under mine blast attack. Similarly. is less than 7% with the average absolute error less than 17%. Draw a straight line through points 4 and 5 intersecting line b at point 6. point S. 2. Lyman. Locate the aspect ratio on the 9 scale. including both plate and tank data. scale. rectangular.. clamped plates subjected to blast attack from shallow buried mines. Fracture thickness and areal weight can be determined from these nomographs in little more than the time it takes to draw three straight lines on the nomograph. 1961. locate the charge weight on the W scale. point 2.MnmeR~ctongulor Plate confhqu¢. Properties and Selection of Metals. pp. 8th Ed. and mean fracture thickness. This va!ue is the thicknes which fracture occurs. 108110. point 4.
: Theory 91 . 0: 5083 Aluminum Test Data. 300 f 2o 200'11/ 0 0 01 'LEAST 0L6bI9. 0:Rolled Homogeneous Armor Steel Test Data.: Least Squares Fit of Data.oil SOS.PLAN VIEW OF TEST PLATE t HOLDDOWN PLATE FRM TEST PLATE MINE SIDE VIEW OF TEST ARRANGEMENT Figure 2 Typical Plate Test Arrangement. FIT  0 '4 100 00 010 E m~ los INLB/IN! 20 30 Figure 3 Plate Deformation.
4 1% + N'.0 03 ~ ~~ CL. N q r ~O ) r. r tfN N a) N 0 ID t..+1..1( +4 +I + 4 II 0 4J m 11 wo"l '.~.4C . 4..44 In m N'ae .   cl CC 9 W 00 4l CL o '4o 41 iin CSA4rNrtCD( rt. O 4 + (1 Nq 1 +I 4 . SNO) I.4 0 0) r. 0 *m ~ e N. 1"NL) too l It in Go N In TQO F.11 ~N m . 4) .a* *q 00O4 .* 4Jq t . .(1 I I 00 CS92 a0 .
0 2.6 .5 17.LeOst 00!0: 0.9 14.4 + 7.16 .6 .9 19.0 1.2.4.16 .0 2.23 .6 +55.16 . Ton Test Ooto6 93 .16 .0 1.0 +25.7e / rI~ .2 20.6 +11. Po'led homoi$ ecs Armor Steel T¢il T t :yt .20 .0 1.9 20.75 1.16 .7 0..6 24 19.4  Ri 0 ~ ~ ¢394 0 Fqgure 4 Plate Froture.2.u' s rI Ot "ti .8 Average Error * 0.16 .25 2.0 .16 .0 1.0 0.9 19.SO 1.0 1.0 1.S 14.175 .2 10.2 .8 +10.16 .1 21.20 .75 1.17 .16 . Steel 1.9 17 In.17 .0 16. 5083 Aluminum Test Dot 0.2 21.0 1.2 +15.0 6.75 1.1 .8 + 7.9 11.225 .13 .6.6 +14.21 Deviation From Least Sqs Material 5083 Al.0. o).0 15.8 17.6 14.9 19.9 19.4.1S .2 21.4 + 5.0 2.7 25.17 . Fracture Data StandOff W Test Case No 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Charge Weight lb 4. Mine Burial Depth) h Plate Thickness in 0.5 3.2% 30~ 2 /0 20 .0 +13.5 0.0 2.16 .6% Average Absolute = Error 13.5 2.0 1.5 12.S 20 20.4 20.7 19.5 .6 27.0 11. (This Includes 3 In.22 .LIN Ic Table II.0 a Plate Width in 44 44 44 44 44 44 44 44 44 44 44 40 40 40 40 40 40 40 40 44 44 44 44 44 a Plate Length in 65 65 65 65 65 65 65 65 65 60 60 65 65 65 65 65 65 65 65 65 65 65 65 65 Fy Yield Strength KSI 33 33 33 33 33 33 33 33 33 135 112 137 116 161 163 137 160 172 1FO 90 90 114 137 110 ef Failure Strain in/in .16 .25 1.155 .7 20.
. 1 co 00 1. . E 0 C0C 00 2 0 w 00 q 0 0 cr 0n < £0 _j3 W NC W >an W 00 0 W0smco ___ ___F__ x I*J<~ z < FW cI F w  U. 94 . 11 .. %on<..L . In T.S) . 0 0 In 0 y .. 0 A ~ 0 .
0 CL 11 K 0 0 E LL o cm00 Ir CL 0 fa i j 0 0 40(4M Zt w"ifu:E t 00  0 '2 ~ ~ N ~ ~ ~ CK) IDI ~ n 40 F) ( 0 j2 95~ .OD  No.
on 'I InJi ?. w0 0*Z I 0 962 .. 0.20 E 0 22 LL w0 00 U.
The conclusions made from these experiments were: I) The impulse from an explosive charge in the ground is a function of charge weight. The impulse is caused principally by the momentum in the soil particles surrounding the buried explosive. All of the analytical studies which have been conducted assumed that the loading is an air blast phenomena. was incorrect. the authors did recognize the limitations of using air blast data directly and attempted to account for the effects of shock wave reflections from the ground plane by doubling the explosive charge weight. Almost all past efforts have been purely empirical. standoff distance. 2) Buried charges demonstrate a greater efficiency in transmitting impulse to a target than does a blast wave from a spherical charge in air. Westine Southwest Research Institute San Antonio. In another study conducted at the Cleveland Army Tank Automotive Plant. INTRODUCTION Although the effects of land mines on combat vehicles have constituted a serious problem for a long time.. Numerous testing programs of this nature are listed in a recent bibliography [1].I THE IMPULSE IMPARTED TO TARGETS BY THE DETONATION OF LAND MINES Peter S. the same BRL data for impulsive loads from air blast were not properly used as in the Cornell 97 study. A technique is developed for predicting the impulse imparted to targets such as wheels and armor plates. however. These air blast data were applied without considering ground reflection factors or accounting for the effects of burial. In this paper. and depth of burial. The few analytical calculations which have been made assumed that the target was loaded by an air blast. A limited set of experiments measuring blast loads from buried 5lb TNT explosive charges was conducted by Aerojet General Corporation 14] and compared with air blast data. very few theoretical or careful experimental studies have been made of this problem in the past. chiefly that the loading wab caused by a shock wave propagated through air. and determined what thicknesses of various armor materials would be necessary to "defeat" given mines. In this paper. that is. one simply blew up vehicles or armor plates with land mines. however. Several Illustrative calculations are compared with experimental test results to establish the validity of this procedure for e. the author will demonstrate that the impulse imparted to a target by detonating a land mine is not an air blast phenomenon. very few analytical studies'have been conducted because the load imparted to a structure from a land mine explosion has never been determined. AllIson Division of General Motors 131. Aerojet General's conclusion that buried charges are more efficient was a correct observation. the writer shows that the impulse imparted to the target is not entirely an air blast phenomenon. however. they were .stimating the impulse imparted to a target from a land mine explosion. The basic premise behind these earlier studies. Personnel at Cornell Aeronautical Laboratories [2] used 13RL data on reflected impulses and pressures from explosive charges in air to estimate loads on tank hull bottoms from mine detcnation. Texas Numerous expensive testing programs have been conducted to determine how land mines damage vehicles ox' armor plates.
Theoretical studies such as those by Lutzky [7] indicate that explosions in an absolute vacuum would have these characteristics. I. The gradual rise time and rounded shape are caused by the mass of the explosive products impinging upon the plate and the pressure transducer in it. (1) is applicable to a land mine explosion because the mass of the engulfed *iat standoffs typical of mine attacks is very small relative to the mass of the explosive products. LOADING FROM A MINE BLAST The nature of the loading applied to P structure auch as the hull of a combat vehicle from a land mine explosion involves very complex phenomena. M. Baker [51 has shown that very close to the charge the normallyreflected specific impulse from a spherical charge detonated in air equals: I/ 2 (2 ME) 1 4iRS2 spheric density.0765 times 11. Thus. The density of air under ambient sealevel atmospheric conditions is 0. Observe that this normallyreflected pressure history differs markedly from conventional air blast waves. Peak pressures and transient loading histories are not ineluded. 0. (1). such as ratio of specific heats. The specific impulse imparted to the body being loaded is caused primarily by the momentum of the explosive products rather than being caused by a shoc. Actually this ratio of air mass relative to mass of explosive products is smaller than 0.ted over the surface nf several targets. can be found in Eq. I. A typical pressure trace as recorded by Jack and Armendt may be seen in Fig. a land mine represents a very similar circumstance. (1) is encased. Jack and Armendt [6] have measured transient pressures at the surface of a plate from spherical pentolite charges detonated in a vacuum. the loading is a very complex wave form that differs considerably from the classical. Rudimentary calculations indicate that the mass of earth and explosive products impinging upon the floor plate contribute considerably to the loading. 91 lb. Ifwe assume that a 20lb land mine is covered by 3 in. above the center of the mine. A shock wave propagated through air is only a minor cause of loading and not the principal one. The rest of the wave has a gradual rise time and a rounded shape. Two shock fronts may be seen at A and B in the pressure history in Fig. (1). and cannot be discussed until direct measurements are made of these effects. the area under the positive rressure history of the load imparted to a target from a land mine explosion. These limitations will not affect structural studies of hull plate response or vehicle overturning provided the response of the system being stiadied falls within the impulsive loading realm. which is very small relative to 20 lb of high explosive. the weight of an equivalent amount of engulfed air is 0. The mass term.4 paper will indicate how specific impulse is ditr* .91 to 20. Jack and Armendt feel that the initial. in diaimeter is 11. is thus the mass of the mine and an equivalent mass of soil. M. atmo . M. sharprising shock front at point A is an Initial reflected air blast wave that would disappear entirely were they to have had a cornplete rather than partial vacuum.0765 lb/ft3 and the volume of this air enclosed within a sphere 17 in. They offer 98 () where M = total mass of explosive and engulfed air = standoff distance from center of charge Sreflected specific impulse. Therefore. The philosophy represented by Eq. S This equation is applicable whenever the mass of the engulfed air is less than approximately onetenth the mass of the explosive products.9 or 0. because the effective weight of soil (approximately 100 Ib) is 5 times greater than the weight of the charge in this illustrative example. This loading is believed to be somewhat analogous to loads caused by land mines because the pressures are primarily caused by the explosive products. in Eq. i. The impulse imparted to a target from a land mine explosion is caused by the momenturn in the explosive products from the charge and primarily from the soil encasing the charge. or velocity of sound in air. Although this equation is Zor a spherical explosive charge in air.e. exponentiallydecaying wave that is usually associated with blasts. particularly wheels and flat plates. T41. of earth and that the hull of a structure is 17 in.incorrect in comparing the loads to air blast ones. the mass. represents the mass of the case and explosive. One can think of the ground which surrounds the mine as a charge surrounded by a weak case of soil. This paper considers only the ipulse. Later discussions will show howto estimate the effective maos term. Observe that no parameters defining ambient atmospheric conditions.k wave propagated through air. thin we can demonstrate the validity of this claim. If the explosive charge represented by Eq. 9 ft3 .
here. A land mine blast will probably give a loading somewhat similar to that seen in Fig. 1. these shocksare less intense than the first shock. the vast majority of the impulse is caused by momentum in soil products. The total impulse is then computed by integrating the differential areas around that portion of rim being loaded. Ordinarily. Eq. (2). 1. Probably in a land mine explosion. explosive products. under these circumnstances the wave has an atmosphere of gases from the explosive products through which it can expand. both of which are common targets. the Incident wave has almost no medium through which it can be conducted. consider encounter conditions as presented in Fig.. I that the impulse (area under the pressure history) is principally caused by the explosive products and that air shocks produce very little impulse. dA. imparted to a wheel or to a plate where the impulses imparted to differential areas on these targets are not normally reflected. SHAPE FACTOR FOR A WHEEL To calculate the shape factor for a wheel. Total impulse can be deterrilned by knowing the projected area of a target and the peak normally reflected impulse imparted to this area :)y using Eq. however. The second shock front may be a secondary shock (. in Eq. This equation does not tell one how to calculate the total impulse. 99 . LOADING IMPARTED TO COMPLEX TARGETS Before . we will consider how Eq. The wheel being considered is located directly over the center of the explosive charge. . the relative magnitude of first shock. Normally. Fig. can be computed. 2a. and second shock wave pressures will vary considerably.sometimes called a "pete" wave by other investigators). the initial peak would be larger because of the presence of an atmos.* Pressure3 Time Fig. (1) can be applied to real targets which possess complex configurations. (2) is a shape factor which is a function of target shape and standoff conditions. because of reflections being transmitted far into the earth and because the cylindrical shape of a land mine would disperse reflections rather than focus them as in a spherical charge. M. however. IR =iR Af (2) I The term I in Eq. 2b shows the 4 *caused Befoe dsct~sin wha nuericl vluespecific should be assigned to the mass. This shape factor will be calculated for wheels and rectangular plates. (1) allows one to compute the normally reflected impulse per unit area at various standoff distances. but the secondary shock would be greatly diminished. (1) impulse being applied to a differential area. We determine the total vertical impulse imparted to the wheel by considering the specific imnpulse imparted to a differential area located on the rim of the wheel.iscsslng wat num.Reflected Pressure History Taken In Vacuum Near a Spherical Pentolite Charge no opinion as to the cause of the second shock front at location B ini Fig. whereas. A strong possibility exists that a peak pressure will occur which is a shock rather than being by momentum from soil flung upward by a land mine explosion: however. 1. 1. One can observe in Fig. ipR.ecal valuethe so that the spLCific normally reflected pulse.
i. The impulse vector which is parallel to a line from the center of the charge to the differential area becomes parallel to the wheel when tan€ is a maximum. dA. This occurs at: d (tn 0 . it is valid that component of the perpendicular to theCo(9 momentum to the would be invalid were r (9) Let us define a nondimensional quantity. 2a indicate: r sin5 r(I . We are only interested in the vertical component as all horizontal components cancel because of symmetry. This assumption this to be an air shock.cos 0) d The cos ?/ can be calculated from the tan t and equals after substituting Eq. equals: dA = hrd0 (3) The specific impulse. of the impulse equals: dl = i cos (* + e) cos e dA (8) 2 (4) Geometric considerations which are seen in Fig. The impulse which is applied to the differential area equals: diA i cos I dA Substituting for d r in Eq. Because this loading is caused by soil particles impacting the wheel. i. dl. d (. as shown in Fig.r cote rflcosO)N rsinO~ ' L. suoh that 6 1 +A (10) because we are considering momentum from moving particles of soil. (8) over the loaded segment of the wheel.oer (6) =0 We will assume that only specific impulse which is differential area imparts target. 2b at the location of the differential area equals after substituting for S in Eq.rcose (d+ rr cos 0)+ r 2 Vin? or at cos 8 r . Wheel Traversing a Land Mine (b) If the wheel is a thin strip of thickness. 6. (9) and rearranging terms indicates that the limits of integration are: 100 . becomes tangent to the wheel. (1): (2 ME) 1/2 = 4 nr(lcos e) + d L cos This impulse may be divided into its vertical and horizontal components. all segments of the wheel are loaded until the impulse vector. (5) for tan t: Cos*r n r(lcosO) +d The total impulse imparted to the wheel is obtained if one integrates Eq. then the differential area. 2._ (a) Fig. h. The vertical component. however.
(3) and (4) into Eq. (11). and Eq. The impulse imparted to the differential area equals: 'A S2 = /X 2 7 $ ( 19 ) But Eq. (14). for a wheel equals: 0 Sird)2 f cos2# cosO cos(*+9) do (d+ rr cos 0) 2 Eq. i. (2) provided the shape factor. A. R. Eq. (18) into Eq. (13) is Eq. (21) by S2 XY and recognizes that 4 XY equals the projected area. SHAPE FACTOR FOR RECTANGULAR PLATES To calculate the ahape factor for a rectangulhr plate. This integration gives: e (12) e) cosO dA =fzi coso( I 0 Substituting Eqs.se. (22) is Eq. Because of the symmetry which exists when the charge is buried directly under the center of the plate. (20) yields: x Y f f (2ME) _/4Sdxdy 0 0 4r(x2+y2 5 2 )3 / 2 (21) Eqs. one obtains for the shape factor of a rectangular plate: ! a (X/S)(Y/S) a (Y/S) (XIS) iY/S)(IS) 2 1 2 4 1] i of the charge to the differential area equals: R F/(x24y2+S2) (16) 101 Ye.. X Y (20 4f f aAdA If 0 0 Substituting Eq. The slant range. from the center If one multiplies and divides Eq. of the wheel yields: 0 2 dO cos 2 $cos(*+0)cos0 _d J= (2ME) 1 2 A (d+ rr cose)2 0 4TS (18) 2 f Once again we assume that the component of this impulse which is tangential to the differential area does not load the differential area. dy dx Eq. equals: dA:. the total impulse may be obtained by multiplying Eq. (1) indiciaes that the specific impulse directed at the differential area is: i (2ME)I/ 41R (17) Because of symmetry. (8) by 2. The plate will have half spans of X and Y. expressed in Eq.e arc coo () () The differentl:n area. then: i (2ME) 1/ 2 A $ 2 r s Sdxdv 4n S2 00 (x 2 +y2 S2)3/2 But Eq. we will consider the vertical impulse imparted to a differential area located in only one quadrant of the plate. 0. (18). 0.0 and integrating the resulting expression from 6 equal to zero to the limit expressed by Eq. (5) and (6) are made into Eq. (14) for the shape factor may be simplified further if the appropriate substitutions from One calculates the total impulse imparted to the entire plate by substituting into Eq. consider a plate with a rectangular xy coordinate system having an origin at the center of the plate. . dA. (12) and recognizing that 2 rh equals the projected area. Making these substitutions and gathering terms yields: e (6 Cos 2 0= 62 1) cos 0 d9 (15) f6+126os3= Although this integral is a very difficult one to compute in closed form. (18). A. 3 where shape factor 0 for a wheel is presented as a function of 6. The results from a numerical investigation are presented in Fig. (19) and (17) into Eq. (2) provided the shape factor. (16) into Eq. (20). (19). a numerical solution is easily obtained on a computer. for a rectangular plate equals: rS3 E J X r f 0 x dy d3 y 2 + $2)3/2 After performing this integration. and that all of the impulse imparted to this differential area is caused by the normal component of the impt.
g. S.(1) and (2). F. to characterize soil properties. W. the density. indicates what soil properties should be selected. from our analysis. which was buried at some depth. we will use two of these three parameters. we will delete the acceleration of gravity. T. and the acceleration of gravity. d. In order to determine the impulse from the detonation of a buried explosive. The explosive cratering analysis must concern itself with both the momentum imparted to soil particles and the influence of gravity on the trajectory of these particles. c. however. length. one begins by listing the relevant parameters. Fortunately. Such an assumption is appropriate as the gravitational field of the earth is not strong enough to appreciably reduce the momentum in soil particles in the relatively small distance between most land mines and targets. and time. (1) and (2). Effective mass. and applying some experimental results to devise a numerical relationship between the various parameters. we are interested in the momentum imparted to soil particles. 1 8 10 20 40 60 80 100 3061 (I + dIR Shape Factor For a Wheel MODEL ANALYSIS TO DETERMINE EFFECTIVE MASS If one knows that value of effective mass is appropriate for a given size of explosive charge and depth of burial. L. c. this parameter is our response parameter. The most difficult aspect in conducting this similitude analysis is the selection of appropriate parametc rs for characterizing the soil.4 2 ~7/4. conducting a similitude analysis to interrelate charge weight and depth of burial. '/ 2 4 6 Fig. the seismic velocity of the soil. We wish to measure specific impulse. In this study. . The impulse is created by detonating an explosive charge of weight. will be determined by using the analytical results expressed by Eqs. p. M. The six paramete's which have just been presented as defining this problem of the impulse from the detonation of a land mine are summarized in Table A together with their fundamental dimensions in the engineering system of force. a recent analysis [8] of a similar problem. p. from the center of the charge. Impulse is to be measured at some standoff distance. g. and the seismic velocity. Deleting this term from the analysis is equivalent to assuming that gravitational effects are insignificant. from the surface of the soil to the center of the charge. To conduct a similitude analysis. the size of 102 crater produced by detonating a buried explosive. Reference [87 uses the bulk density of the soil. 3. he can compute either specific impulse or total impulse by selecting the appropriate shape factor and substituting in Eqs. i.
0 lb charges were buried at various depths and a rigid mass with an exposed surface area of 50 in.. The first pi term is normalized or scaled impulse. Eq. (26) may be rewritten as Eq. Before making this plot. because they are essentially constants. (27).(25).. A careful perusal of Reference [8] would show that relatively accurate predictions of crater size can be made when P and c are treated as constants. Seismic velocity for soil in a olid state does vary over a limited range. We will assume that the seismic velocity of all soil "fluids" is corstant. Eq. (27). (1) states that impulse varies inversely as the square of the standoff. W/d 3 .(2 (d 3 ) SCALED IMPULSE AS FUNCTION OF SCALED CHARGE WEIGIIT Reference 141 describes a limited series of tests in which 5.. (28) which is an expression for total impulse rather than for specific impulse..F 'r TABLE A Parameters For Determining Impulse From A Land Mine Explosion Fundamental Dimensions FT/L 2 FL L L FT 2 /L 4 L/T I d fI '6 73) . a statement of geometric similarity and scaled charge weight. Eq. 3 2 (i S )/(d ). (1) annot possibly be a function of S.Impulse From Land Mine Explesion i 'r r d Scaled impulse Geometric similarity charge weight . . Experiments must be conducted so the measured scaled impulse..1 Symbol i W d S p c Parameter specific impulse charge weight depth. 103 '~' g . however. (26). Table B lists three pi terms which can be obtained. (25). By measuring the displacement of the mass for different depths of burial and standoff distances. the soil which provides most of the impulse from a land mine explosion acts more like a fluid than a solid. Eq. Eq. iS2 = f"(. (27) is Eq. M. (1) without an explicit expression for the appropriate functional format. (f3 d W pc2 d 3 / (2) pcd A The soil parameters P and c can be deleted 'from Eq.Scaled pc d3 . (1) expressed in a slightly different form does not provide the desired functional form either. . (1) ir. M. This pi term io a unique function of the other two pi terms. (27) which is Eq. 2 3 f. in Eq... (25) because they are constants yields Eq. Eq. (1)does not tell us how effective mass. (27) will be modified by substituting Eq. thus. We write this functional relationship as Eq. to scaled impulse. IS Ad d 2 W Tn 3 . d.. must be obtained from experimental test data. relates to charge weight. d3 d3 (27) Many different textbooks present several different procedures for developing the three nondimensional ratios called pi terms from the six parameters listed in Table A..ity varies very little over a wide range oi soil conditions. .of burial standoff distance density of soil seismic velocity Eq. they could not rotate or translate horizontally. (26) defines a threedimensional space. 2 suspended at various heights directly over these charges. Soil d. In this paper. . Eq. and depth of burial. Inasmuch as the effective mass. we will omit the algebraic procedures and present the results.. (25) should be rewritten as the twodimnensional space. TABLE B Pi Terms .diated that the threedimensional space expi essed by Eq. (2) to form Eq. Deleting P and c from Eq.. W/d 3 .however. After performing this operation. it does not furnish the desired functional format for predicting impulse from a land mine explosion. (iS 2 )/(d 3 ) can be plotted as a function of scaled charge weight. Analyzing test data in terms of total impulse rather than in terms of specific impulse is more convenient. W. The functional format relating scaled charge weight.. The masses were not constrained or prevented from moving vertically. . Eq. . S. This space can be reducedi in a twodimensional space by squaring the second pi term and multiplying the result by the first pi term to form a new dependent parameter.
8 10.0 80. 106 lb to 0.125 1.6 10.51 51. I S2  The wheels were held by a yoke whose other end was attached by means of a torsio. A will be (in 2 ).79 1.3 44. Throughout the rest of this paper. depths of burial and standoff distances fromthe center of the charge were estimated. the following equation was applied and solved for the impulse imparted TABLE C W Versus .125 W d3 t I lb 101 102 105 106 107 108 109 111 113 114 115 117 3.07 3.5 75. The depth of burial was measured from the ground to the top of the charge rather than to the center of the chrge.46 1. The quantities (IS 2 )/(A~d 2 ) and W/d 3 are dimensional because the soil constants have been deleted from this analysis.2 4. The experimental .5 177. however.791 1. depth of burial.51 3. the shape factor § is essentially equal to 1. S will be (ft).51 3.1 51. and the calculated total impulse for these tests.125 1.51 3. *By assuming that the 5. 125 1. The final two columns in Table C present scaled impulse and scaled charge weight. In a 3wheel array.9 39.2 90. Bruce Morris at MERDC. V 2gh(9 Vwheel 2 Table C contains the test number.8 A second group of experiments which mea.5 202.125 1.95 35. the impulse was applied to all 3 wheels.3 9. This test configuration is effectively a springloa~ed pendulum with a as the ball of the pendulum. wheels were placed over buried charges.3 118.normally reflected total impulse could be easily calculated from Eq. these two quantities will be plotted to determine the functional format for Eq. By measurIng the maximum rotation of the yoke and wheel system as a result of the detonation of various size explosive charges.8 10.125 1. The charges were rectangular parallelepipedes with a constant surface area of 2" by 21" and a thickness that depended upon the size of the explosive charge. The middle wheel in any array was always over the center of the land mine. Soil Cover To Top Of Charge (in) 12 12 4 8 12 12 12 4 8 4 8 4 J Air Gap (in) 4 0 12 8 4 8 0 16 12 8 4 8 Total Impulse (lb.9 198.68 2. the units for I will be (lbsec).1 51. (28).1 3.125 1. Steel wheels which were 7" in diameter and 1" thick were in direct contact with the ground.791 0.9 170. In these experiments.125 0. (30) where tan to the center wheel.sec) 123.0 lb charges were cylinders with diameters of twice their thicknesses.a spring to a rigid wall.125 1.54 26.458 S (ft) 1.75) ic .79 j. On a few occasions the yoke held only one wheel which was centered directly over the mine.458 0. (29).46 1. sured impulse imparted to targets from land mine explosions are from a series of unreported model tests conducted by Mr.458 0.Using Rigid Mass Test Data 3 d A§d IS2 A§ d3 /Pss S.125 0. 1/2" of soil was placed over the charges.60 4. 468 lb of pentolite.791 0. 104 ic + 2i c sin 0 13 (L d1.3 154. the impulse imparted to a wheel could be calculated.79 t. To estimate the impulse imparted to the center wheel only.0 in these tests. air gap b~tween ground and bottom of the mass. The explosive charges used in this test varied from 0. d will be (ft).51 51.6 TestNo. Because the masses being loaded by the explosive charge and soil products are very small.7 d (ft) 1.5 10. In subsequent discussion. and i will be (psisec). Fort Belvoir.46 1.8 10. 75" centertocenter spacing between successive wheels.46 1.458 0. most experiments were conducted on a 3wheel array with a 1.5 166.68 46.\ it / 3.
4.275 1. 425.6 110.266 1.4017 .3 105.110 .9 d (ft) .5 91.0930 . and independent quantity.4 13. 274.0 3.5 W d3 Test No. 72 d0 . (30) proportions the impulso appropriately.387 4. 353.3847 . he obtains the effective mass term.25 4.6 18. one sees a plot of scaled impulse.5 53. 261. M(slugs) = 3.378 1. Center Wheels Wheel 3 3 3 3 3 3 3 3 3 3 1 1 1 12. These experimental data appear to collapse into a unique function as predicted by Eq. the charge weight. 4 gives Eq.25 3.9 23. The dependent quantity. W0 .319 1.3524 . 4 also shows that experimental resuits agree well with Eq.3847 . (1) and equates the left hand side of Eqs. the equivalent impulse imparted to the center wheel only.data indicate that Eq. 1. and that the upward momentum of outside wheels is caused by the vertical component of this impulse. In Fig.1 21.113 S (fit) . 353. one must look up the shape factor for the wheel in Fig. as a functlon of W/d 3 for the data contained in Tables C and D.5 155.6 27.0 '. 2 3 4 5 6 8 9 10 11 13 14 15 16 Total Impulse (lbsec) 23.4 x 10+6 W for E in Eq. If one substitutes 1. 311. Although data from 105 . The scatter is small considering the inherent scatter in experimental test results.5 92.3947 .0 5.117 . 90. we assune that the impulse imparted to the outaide wheels is applied parallel to. (28). M (lbs) = 112.6 14. The equation of this line as obtained by a leastsquares fit to the data in Fig. Fig. 261.0778 .5 130. In developing Eq. (30).2 13.3 4.49 W0 44 d1 .7 15. 113.4 3.208 1.0 3.1 24.4084 . and standoff distance. 353. (31).2 47. W/d 3 .0607 . 1 0. no one has measured the impulse imparted to the plates in such experiments.402 1.1 43.430 1.3749 . (32). (IS 2 )/(At d3 ).0832 .125 . 6 5 (32) Usually the reader prrers to think of mass in units of pounds. 84 I s2 #A = 1.402 1.7 15. (31) is the explicit expression for either Eq.0930 .1 71.354 1.1 10.55 4.9 20.4047 6 f ft3 400.9 Impulse on No.0930 . 725 (31) Eq.285 1.0930 . 240. I S2 A TABLE D W Versus  Using Wheel Data IS 2 AI d3 (psisec \ft/ 98.5 W 44 d I 68 (33) The author had substituted into Eq.6 3. In spite of spending many years and dollars on explosive plate bulge tests.45 4.319 1. (1) and (31) after making the expressions dimensionally consistent. (1).7 14.3695 . M when expressed in slugs is given by Eq.4 Tables C and D contain all of the easurements of impulse imparted to objects from land mine explosions that are known to this writer.4167 . Mass in pounds is given by Eq.117 . 286. 353. 131.4 31. 3. 90. A straight line fits the experimental data presented in Fig.1 16.7 13. Both 6 and # are listed in Table D for each test configuration. 4 very accurately.319 1.9 23. (31).1 4. (33) before making his earlier statement that the most significant mass in the explosive products from a land mine explosion was the mass of the soil products.2 102. Table D contains the total impulse measured by this test arrangement. the number of wheels in the test array.7 28. are computed from this information and listed in the last two columns of Table D.4087 . Before computing the total impulse imparted to a center wheel.103 .a line from the center of the charge to the bottom of the wheel. (33). (IS 2 )/(A§ d 3 ).3847 . 97.3847 . depth of burial.3 28. 475. (27) or Eq.5 32. M.
The charge weights in the wheel experiments were varied by a factor of 4. 0. Obviously. . W.600 ' ' ' ' '' 1 ' ' 400 200  x IMPULSE ON THREE WHEELS + IMPULSE ON ONE WHEEL 0 IMPULSE ON BLOCKS 100 60is' 40 r X _ 10 m7 f 4"4 44 \ . (I SZ)/(d 3). and the air gap varied from zero to 16". and S have all been varied in Eq.1l I1 2 4 6 10 20 W 40 60 Ilb"' 100 200 500 Fig. 4 and Eq. . Extending the leastsquares fit in Fig.8+ " 1. (31). a significant number of data points (25 points) from a variety of test conditions is used to develop this curve. if the charge is located at the surface with no cover over it. 4. W/d 3 equaled a finite value of 2570. if the depth of burial is too great. 0 predicts that scaled impulse. Three data points were not includedt in Table C. one can see that 4. the detonation of a buried explosive will not disturb the surface of 106 . (t) as applied to air blast with a ground reflection factor of 2. the impulse begins to be an air blast phenornenon. Because the center of the charge was below the surface of the ground. of 3227. WO7d o. 4 would predict that scaled inipulse. (31) do have limits of W/d 3 for which they apply. (I S2 )/(d 3 ). On the other hand. should equal 490. and no soil covered the charge. By combining these facts with the observation that theae data involve two different types of targets. Using Eq. 4. (I S)/(d3 ). d. Scaled Impulse Versus Scaled Charge Weight the ground. SUMMARY Fig.725 "s" + 2 I Ih II I I ii. The average of the three tests on the rigid masses was a scaled impulse. would equal 4745. Obviously this test was in a transition range where the impulse was changing from being only two types of experiments are used in developing Fig. Depths of burial on the rigid mass tests ranged from 4" to IZ" of soil cover. because in those rigid mass tests the top of the charge was flush with the stil'face of the ground.
Maryland. "Allison Division of General Motors Corp. Cleveland Army TankAutomotive Plant. 3 and the shape factor for rectangular objects such as plates may be calculated using Eq. The other major restrictions to this analysis procedure is that the standoff distance to the target must be sufficiently small so that the weight of an equivalent sphere of air must be less than 0. W. White Oaks. C. and B. 107 [I . "Summary of Armor Materials and Configuration Tests at Aberdeen Proving Ground. REFERENCES 1. (33) into Eq. Lutzky.K. Mech. Fort Belvoir. W. Armendt. B. The author is indebted to Mr. 3. J. Maryland. 2. et al. This final restriction does not appear to be very restrictive. 1970. Baker of SwRI is hereby thanked for reviewing this text and making several helpful suggestions. 7. (31) to be valid. for this approach to be strictly valid. Most targets are much closer to the ground than several feet. November 1962. 20 February 1968 (Confidential Report). 2. Jr. Virginia. the analysis procedure recommended in this report should be valid. (1967). W/d 3 . Cockrell. Dr. Provided scaled charge weight. Westine. Bruce Morris. 8. "Explosions in Vacuum. Under the terms of this contract. U. No. A. H. Aberdeen Proving Ground. (33). "Explosive Cratering. and Southwest Research Institute. 4551.. Probably this standoff distance can be doubled without causing serious error. F. (24). Baker. The specific impulse at any location on a target may be estimated by substituting Eq. R. "Prediction and Scaling of Reflected Impulse From Strong Blast Waves. Jr. Some finite soil cover is required for Eq. ". S. R. . May 1962. For a 20 lb charge buried 3 in. The shape factor for a wheel may be obtained from Fig. L. one must determine a shape factor which Isa function of the geometry associated with encounter conditions. pp. 6. W. Sci. for being given the opportunity of probing into this problem. Texas.. (31) to compute the total impulse imparted to any target. San Antonio. A rational design of a new roller system required that we determine the nature and magnitude of the loads from a land mine explosion. Army Weapons Conmand. and C. P.. this observation means that the standoff distance must be less than 38 in. 4. Anderson. J. pp. We have seen that the impulse is not primarily an air blast phenomenon. Wenzel. 919. June 1968 (Secret Report). M. Southwest Research Institute is to design. 1280. 7. E. The loading iscaused by momentum in the enplosive soil products.caused by momentum in soil particles to being caused by momentum in the explosive products from the charge. TR 3481. To calculate the impulse imparted to complea targets. Vol. Wilfred E. "Reduction of Blast Effects." BRL Report No. 1965. Jour.. Contract DA44009ENG4780. (1) and taking the appropriate component of the resulting specific impulse. 047701 (04)FP. Russell. In addition. " Int. falls between 1 and 1000. " Final Quarterly Report. Young. 5. R.Sova. "Structural Response and Human Protection From Land Mines (U). Kincheloe.S. "Phase III Parametric Design/Cost Effectiveness Study for a Mechanical Infantry Combat Vehicle (MICV). These shape factors are substituted into Eq.S. 1 times the weight of the explosive products given by Eq. ACKNOWLEDGMENTS This paper is a direct outgrowth of contract DAAK 0270C0579 between the U. "Cornell Aeronautical Labs. 28 June 1967 (Confi( itial Report). Alexander Wenzel. Report 6M2144H4." NOLTR 6219. Army Mobility Equipment Research and Development Center. "Measurements of Normally Reflected Shock Parameters From Explosive Charges Under Simulated High Altitude Conditions. "Combat Vehicle Mine Protection Conference (U). deep. In this paper a procedure has been developed for predicting either the specific impulse or the total impulse imparted to any target exposed to a land mine explosion. develop. SwRI project leader. the Army's technical monitor on this project and Mr. . and test a new mine clearing roller system. Journal of Terramechanics. 9. April. Jack.
the elastic response of the beam was separated into a quasistatic response and a sinusoidal response. poor agreement was obtained. However. Canada The response of a circular centilever beam subjected to a plane transverse air blast was obtained. The loading was separated into a diffraction phase and a drag phase. The drag coefficient will be a function of time due to a continuously changing Reynolds number and Mach number during the air blast loading.433) to 0. the drag coefficients were 5 obtained in the region of Reynolds number 7. However. each having pressure decreasing linearly with time to zero. Using this method. and 2 ft. and using the empirical equation for dynamic pressure. were tested at the 11 psi nominal overpressure location in a 500 ton TNT field explosion (Dial Pack) From the measured response. a mean drag coefficient was used for its duration.5 psi nominal pressure locations.S. The obtained drag coefficients were in the range of Reynolds number (Mach number) from 3. For a stiff beam.67 and 0. Drag loading is drag pressure times projected area and drag pressure is drag coefficient times dynamic pressure.7 x lO and Mach number 0. This duration was determined from the linearly decaying time function which maintained the same drag impulse as that obtained from the Brode [2] empirical relation for dynamic pressure.23. In this work. and 2 in. by expressing the drag loading (forcing function) Preceding page blank 109 . the drag coefficients of the circular cylinder under the unsteady flow conditions of an air blast wave were determined using the domain conversion method. It was mentioned that Reynolds number and Mach number appeared to vary jointly in such a way that the drag coefficient did not change with the time varying flow of the blast wave.75 x 106 (0. Ukrainetz Department of Mechanical Engineering University of Saskatchewan Saskatoon. To obtain the actual drag loading. However. respectively. the drag loading (drag coefficients) from an air blast wave was obtained using the elastic response record of a circular cantilever beam subjected to the air blast wave. respectively. Kim and P.25). a conversion of the response was made from the time domain into the frequency domain and then from the frequency domain back to the time domain.205).6) to I x 106 (0..48 for the 12 psi and 8. the predicted and the experimental results showed good agreement. INTRODUCTION Baker [1] obtained maximum responses of rectangular cantilever beams subjected to an air blast wave.5 psi locations.7 psi.41 to 0. The blast wave source was a 500 ton TNT field explosion. the drag force was obtained. Bishop [3] obtained drag coefficients of a circular cylinder and pressure coefficients around the cylinder when subjected to an air blast wave of nominal pressure 20. for slender beams. The drag coefficients oscillated about the mean values of 0. and diameters 3 in.R. The elastic response of a circular cantilever beam subjected to a transverse plane air blast wave was studied in this investigation. the drag phase analysis considered the actual drag loading which was expressed in polynomial form. The obtained drag coefficients were in the range of Reynolds number (Mach number) fiom 1 x 106 (0. The loading was separated into two phases as was done by Baker. The drag coefficients of a circular cylinder under steadystate conditions are already well known. In considering the drag phase. The results agreed roughly with steady state values. Two circular aluminum cantilever beams of lengths 5 ft.81 x 105 to 3. From this response.44 x 106 (0.CIRCULAR CANTILEVER BEAM ELASTIC RESPO1JF4 TO AN EXPLOSION Y. Mellsen [4] measured drag coefficients of circular cylinders by the free flight method at 12 psi and 8. the drag co efficients under unsteadystate conditions such as those from the air blast wave have only recently begun to be investigated (3][4]..
J Mellsen and Naylor [6] arrived at a nondimensionalized time for double vortices to fully form and start shodding. This is then the start of steadily decaying flow. This time is given 'by  D 2. is the time for double vortices to fully form and start shedding. the waves continue right back to the front of the . regular As the reflections shock front occur passes the cylinder. Mach stems arising from the incident and reflected waves begin to forn at each side of the cylinder. the dynamic pressure can be expressedwith reasonable accuracy using 8 a 2. the relation between the forcing 'fnction and the quasistatic r6sponse wasobtained. ~over Fig. the waves become very weak and are rapidly dissipated. In the case of stronger thocks (for example about 20 psi).  The dynamic pressure in the air blast wave is given by the emprical relation [2] t t/T d q(t) . In (2).q0 T1 d (2) d where q is peak dynamic pressure which can be obtaine3 from the RankineHugoniot relation q o X P.ylinder and croQ4s over once again at the leading edge.S  ~17. When the shock front hits the leading edge (stagnation point) of a circular cylinder. For small shock strengths tip to about S psi peak overpressure. At this stage.S.in a polynomial form. I RESPONSE IN TIM DOMAIN Small deflection. Thus the forcing function was determined from the "known" quasistatic response. the overpressure can be represented with reasonable accuracy by setting KR: 1. It was photographically obtained from the interaction of the air blast waves (oC peak ovcrpressures 10 to 17 iii)with circular cylinders. form their otn comfplex Mach stems. After this. until the angle between the shock front and the interacting surface reaches a critical value which depends on the shocX strength and the radius of curvature. due to the inertia of moving air. where P is peak overpressure. and T. the net transverse loading on a circular cylinder as a result of the air blast wave is separated into the diffraction phase and the drag phase. on passing back across the cylinder. T is the duration of the dynamic pressure an Is usually slightly larger than T. (3) where r is the ratio of the peak overpressure to the ambient pressure. T+ is the positive duration of the overpressure and K isa decaying constant. linear elastic theory (the Euler beam equation) will be used for prodicting the response of a circular cantilever beam subjected to a transverse plane air blast wave. in the wake flow. BLAST LOADING The overpressuretime record at some distance from an open air explosion source can berepresented by Friedlander type decay [5]'which is given by p(t) a P (I eKt/T.S (4) where U is shock front velocity.nl Xn(X) gn(t) where X (x)are normal mode functions and g (t) are time functions which can be determined ?rom Lagrange's equation. Mlach stems contInue to envelope the sur  2 . When the beam is subjected to a uniformly distributed forcing function ~These face until they cross over at the trailing edge and start to go back around the cylinder. The solution of the Euler beam equation with respect to deflection is represented by y(xt) . Fig. D is diameter of the cylinder. the waves. Thes2 never really reach the front of the cylinder because they break up and are dissipated 110 . (2. The time T1 is cbtained from (4). As already mentioned. The drag phase then follows the diffraction phase. 1 shows this type of loading. the pressure rises instantaneously to the noirmal reflected pressure. The quasistatic response was obtained by converting the time domain response into the freouency domain and then from the frequency domain back to the time domain. For low values of qo. In the approximate procedure for determining the net transverse loading during the diffraction phase. For low values of P (about 10 pAi). the net transverse overpressure is considered as a pressure decaying linearly from the normal reflected peak overpressure at time t = 0 to the drag pressure at time t =T I .
CM = H ..P J. The strain corresponding to g.4. (19)f = CO + c~t + C t2 n "Cn ..p(t) a P becomes f f(t).a0 =0 + (17) In (15). 2 dn(0) tn 4D 2 2 X n where En(X)n = 4 (V K) n X (L) 2 Tx1 ~ 2h (8)(14/ (9) 2 A_ ([g(O)J + [nO W g*(O) k. G (t)was separated into two parts which are g*(t) and a sinusoidal response. in(t)+ Co )2gnCt) (12) can also be expressed as a%6 modal where w are natural frequencies.2..m 6 if (Hm) is even if (Hm) is odd.. ) (13) Rewriting Gn(t) of (11) gives gn*(t) g (~)~ (t Po _ = ( An (o) + gn(O)cos W t + nGn(t) (0Gnt*ld (wn)'Mn(14) sin wnt (7) where n g*(O) cos wnt n n  n sin wnt si wn (IS) An g*(t) An sin({nt nnn' n) + where T is a dummy variable. The solution of (6)with nonvanishing initial conditions becomes gn n 113 i ml i0. t'M. participation factors. t.0.4. L is length of the beam. + n En(x).2. the nth mode response time function Gn(t) for 05t!T becomes n~.m = M.G (t) e(x. The strain (for zero initial conditions) at the stagnation point at distance x from the root of the beam becomes get) rO +M a f(t) + Ctt 2) 1 t n)nf ) M+ 6Mt 1f()t.(t) becomes to becme ait + a2 t2 + for 0tsT for t>T (10) e*(x. Pn are constants of the normal modes X (x). + CrV n el1) n n n sin wnt cos wnC .c1) "(n) m * 0.* a +i m m le n i=. n The normalized (inmagnitude at t = 0) forcing function f(t) is a polynomial of the Mth order in time. the time function equation PO nft n . (19) . * l m In (12) 0.2. gnCt) will be called the nth mode quasistatic response time function. C n where the coefficients are given by ('1)1/2 (mmi)l. Gn(t) is the nth mode response time function. Dots in the expression represent time differentiation.n On = tan n M2f(M)(t) H n In (8).1. in (9).t) .1. ) (t .2.. ) . f(t) . and Kn are related to the roots AfKO of the frequency equation.t) p g (t) En(X) P (18) g(t) can be expressed Inmatrix form as For this forcing function.M (1 e The constant colunn vector (Cn) can be expressed 111 . E is modulus of elasticity. and Mn are generalized masses.. I is moment of inertia. This quasistatic response is diffeent from the static response which can be obtained from the Euler beam equation by setting the inertia term equal to zero.CC n ..d t 0 n Ct) = onwheesint n (tTdsiwhere T (C g*(t) = (t) )si (Cn) n {tT = [. D is diameter.
With the consideration of damping.. (C where [B I is a M' x M' lower triangular square matrix which can be obtained frol (32) and (Cnd) is a constant column vector of C d (j = 0.t) and the damping ratios Cn are known. the nth mode quasistatic response time function..P (t)r nl n(x)[Sn]'(A) (21) 'NxH upper (.t) in polynomial form. the forcing function P(t) can be obtained from (21) by expressing e*(x. Thus if e*(x.)i / 2 &.__ i ( )u l (29) (.d 1(x.. the forcing function P(t) can be obtained from (36) by 112 . nd' Cd'nd. P (t)T(A) is the forcing function P(t).i)d where m (rnwn)( E j=o (Mj) +. in the actual case there is a slight damping in the response due mainly to internal damping and aerial damping. These are given by m~i * 0 (Cn'n)(m*i'j) In (21). Then substituting (19) and (20) into (18) gives e*(x.4.t) . A expanded irin infinite series. gnd(t) eCnwnt can be expressed in matrix form as n t gd(t) e = (t)T(Cne (33) (24) (34) ne =(Bn](C nd In (33). However.)e4nns(tr)in wnd(t. (Cnd) = ISnd]D](A) (35) (t)eCWt ecnwn t * t + *d Ac nd'oS(wndt Set .t) ) (l/tanend Cn(x t) • NOT} T = P't) F xn ' Dn)(A) (36) Coefficients CO Cd . • gd(t) e'nn e'%nJnt where nd n i " n ) * (25) (26) (27) where ca RSd an upper triangular square ] is which be obtained from (29).2.Z l4 by [C n) [Sn . So far dmping has not been considered. be + Ct + C2dt2+ nd nd 2 And '([gnd2(O)J + 2 (0) } 2 . (Cnd) can be expressed as '). The coefficients are thus n here (A) is a constant column vector of the l'ocin fuctin ft) nd S ]is triangular square matrix which can be obtained from (12). 4(t) can be expressed in an infinite as 9sIe *(t)en gn nt c0 1 2 C + t+ +C ne ne ~2 tc + 2~t + c' t ne (31) where End(x) ' En(X)/[(Cn ) 21 In (22)..2. g e'¢nunt. n Gnd(t) • 0nd 6t f(. (A) (20) substituting fjTxof (10) and the infinite serife expansion of en nI into (23). When damping is considered. cm 1 ne CndJ 32n 32) By taking terms up to the M'th order (!'Q M) in (23) *d( [g d(t)  0 ) g d(O) cos wdt sin wndt]oennt  (31). and ID ] matrix is a lower triangular square matrix which ca obtained from (30).t) is known.. the strain becomes ed(xt) P n l End(x) Gnd(t) (22) Jo ( j (30) where aj 0 for j > M. becomes cd ( I Wnd g9nd 0) C tan nd = tan ( n) .1. n are the constants of f(T)e nw In (29).. (28) are obtained by Thus if the quasistatic response es(x. G (t)is the daired nth mode response time funct gn which corresponds to Gn(t) in the case of no damping.non~hen the strain corresponding to g*d(t)..
i. 2 where F1(w) is FT of impulse response function sin w t of the beam. The forcing functon can then the be quasistatic obtained from response the derived between and relation the forcing computing function. if N' is taken sufficiently large. Knowing (hw). However. Forcing function f(t) is also zero for t < 0. likewise the forcing function. FFT Is simply an efficient method of the Discrete Fourier Transform As indicated by (43). Thus the respons. Actually. This Is oy taccomplished by taking the sampling duration as an integer number of periods corresponding to Gn(w). Thus Wn j " fn (w)= F{sin wnt u(t) a (0 ~ = ~ 2 ) [6(w + ton 6(w. In practice. R(w)cos h(t) . obtaining the forcing function directly from (45) is possible In theory only. respectively. RESPONSE IN FREQUENCY DONAIN Taking Fourier Transform (FT) of (7) with zero initial conditions gives an Pn W)Mwnfiw)Jf(w) n n (37) where • )n(w) Rf(w) n)2 ? 4 (45) (46) Taking IFT of a (w)using the real part only as given in (44) yfelds M gn  PO n (tan)2% [g~)# wnn 2 (t)] n~ (47) Substituting (37) into FT of (5) yields Px n i)( n((w) P(2) (38) a(t) and g2(t) are time functions obtained from n and d2CW). Let FT of f(t) be No= f + jf (40) Substituting (39) and (40) into (37) will give an(w) which has real and imaginary parts. if the response in the frequency domain is known. the error can be minimized. Impulses on the n imaginaiy axis correspond to a sine function and (c. 113 . FT of a real causal time function h(t)(71 Is () where R( R) NO cos w h(t) sin wt dt (42)' M(w) + jX( ) (41) Inverse Fourier Transform (IFT) of 11(w) is 2 o. lowever FIT does not have the characteristic property of FT of a causal time function. in the frequency where 6(w) is the delta function.)J n n (44) Impulses on the real axis correspond to a cosine function. the response in the time domain can be obtained from (48) and (49).Xf(w)6(w+wn)].is the frequency of the function.expressing e*(xt) in polynomial form.e. cos Wt dw  f(wn) cos nt (9 If the response in the frequency domain n(w) is known.n)] (39) g(t) J 0[Xf(w)6(wtn) . the forcing function f(t) can be obtaine. For this case. a sinusoidal the causal tine function function (sine and/or cosine) can be represented only the real or thefrequency domain. As shown in the Appendix. Then by taking IFT of Rf(w). Af(w) can be obtained from (45). This impulse response functdon is causal. an error will result due to truncation of the infinite series when taking terms up to the N'th order only. M(w) can be readily obtained because _2(w) has delta functions only. Also. The domain conversion is made possible using the Fast Fourier Transform (FFT). sin w t is zero for t cO. wt dw 2 RMX sin wt dw (43) domain obtained from the above analysis and that obtained using FFT will be different. If the imaginary part of %n(w) rather than the real part is used. the quasistatic response can be obtained using the characteristics of FFT. The real part of p R (w)) a ) } n {wn n ) (DFT) [8). a parallel analysis will give the same result. can be obtained using cainary t d u g imaginary part of FT.
5 rad. This "cutting impulse technique" can thus be used to obtain the quasistatic response of the beam.. A slightly damped sine function can still be represented by impulses at the corresponding frequency on the imaginary axis. . TS n_ 'nd nd or T. The result is shown in Fig. sampiIng was started at t = 0 (with sample intervzl being AT'.) Cutting the impulses at w 1 by linear interpolation using adjacent points implies removing the sinusoidal response corresponding to w! in the time domain and also implies that the quasistatic response in the frequency domain has a value at w1 which is the average of the two adjacent point values. 1. the fundamental mode is absent from this response. 2 the response (strain) of an aluminum cantilever beam of length S ft. the 114 . .... Thus. the response in the frcquency domain at w2 is not of impulse form. Thus. The result is shown in Fig. which is the result of dividing the strain by n W. / . L=5 ft. After cutting the impulse.. when slight damping exists in the response. using the imaginary part of FT only (refer to the Appendix).) " " .sec. 2.Cutting (or smoothing) thase impulses in the frequency domain corresponds to removing the sinusoidal function in the time domain.. (FT is shown for positive w only in all Figures). 3. From this modified response the quasistatic respnnse can be readily obtained and Is also shown in Fig. (which was also tested in a field explosion) subjected to a uniformly distributed exponential forcing function is shown. The units of the respore are psi.REAL MGNRPPRr function will have on the real axis cutting impulse remove this func . the error is quite small as shown In the Appendix.. 1500 .. 00a IMAGINARY PART A M.1024 21 & Nt tion. and diameter 3 in. If the sampling is taken fine enough such that an integer number of periods corresponding to w2 can be obtained for another sampling duration from this IFT... 5 ft. Since no damping was considered.D:3 in. 1) 3 in.. The sinusoidal response is removed by making use of this domain conversion method.3 in beam).. 3. As shown In Fig.l . However. k=20. 2 Responses of Circular Aluminum Cantilever Beam Subjected to Uniformly Distributed Exponential Forcing Function ekt (k * 20.. Since the sampling duration is not an integer multiple of the period corresponding to w2 . 2. The technique can also be used to only. except that there will now be a small side lobe across the impulses. the cutting impulse technique can still be used to remove the damped sine function. /NAT s0 2 z o a fI . 3 Response in Frequency Domain of Circular Aluminum Cantilever Beam Subjected to Uniformly Dn(x)" istributed Exponential Forcing Function ekt (W1 1 4S.CC¢ 499 4 Fig. 2 and Is called the modified response.. an impulse at wl on the imaginary axis corresponds to the sine function and an impulse at wI on the real axis corresponds to the cosine function. In Fig.. In this case an error due to the existing small side lobe will result. .. this time by using the real part of FT I. .. Pig.. The sampling duration must be the same as for the case of no damping.. the sinusoidal function must be made a sine or a cosine function only with a zero phase angle in order to use the cutting impulse technique. Here modes up to the 2nd were taken into account and damping was neglected. This is possible by starting the sampling at time C The FT of this response was obtained using 1024 sampling points (N) for the duration of eight times the fundmental period (the same sampling duration which was used for the compilation of experimental results for theS ft. A slightly damped cosine impulses and small side lobes in the frequency domain.% I FT was taken using both real and imaginary parts. nd : 'nd in (25) / o Soo0  / o0 4 0 FRoUECY Ia fa 600 In Fig.
0Uniformly . 1000 t. 0.0.+. The quasistatic response was smaller than the forcing function by approxi115 Fig. but higher on the beam) were Installed on each beam. The FT is shown in Fig. o o 1. D 3 in. a damping ratio of 0. this time taking the damping into account. REAL PART  AGINARY PART N . After cutting the impulses. 2. 5 Damped Response in Frequency Domain of Beam Subjected to Cantilever Circular Aluminum Distributed Exponential Forcing Function e kt (w. The result is shown in Fig.OU 1 +  o. To make the sinusoidal functioa a sine function (with zero phase angle) only.3% 3. Z All A ... ' 145.. Ralstog Alberta..0074..96 ms.. k = 20.S%. After cutting the impulses at w1 (of the imaginary part). ...<. 4. 4 are due to: 1. l Fig.) To avoid ambiguity.0. Two strain gauges were used for each strain output. The straip gauges were connected through bridges to r magnetic tape recorder which was located in . The forcing function e . 1024 OT. and dihmeter 2 in. Instead of using both real and imaginary parts for the IFF. This is duo. were tested at the expected 12 psi nominal overpressure location in a field explosion of 500 tons of TNT. The remaining sinusoidal difference corresponding to 1...23 sec... The beam response was obtained again..'++I o€* '¢u co..T Soo .. Thus at t .. Existing damping effect on the modified response. .. 4. . eiic of length 5 ft.. D = 3 in. An overpressure rec./sec. mately l... The maximum difference is about 2% (refer to the Appendix). The modified response is shown in Fig.. k = 20) For the S ft... 10 4 0.. I/N. W.. and the other t4o similarly located... and diameter 3 in. . Damping effect on the quasistatic response is very small and the difference due to damping was estimated to be only about 0. From this record it coa be seen that the actual peak overpressure is ipproximately 11 psi and the duration is ahoyo. Sampling starting time T .. . . L ... IFT was taken using the imagi nary part only.. Circular aluminum rods of smooth surface were fixed to steel bases using a shrink fit method. i 200 400 60  I"t011. EXPERIMENT AND EXPERIMrNTAL RESULTS Two aluminum cantilever beams.3 in.! 40*'.... 4 Responses of Circular Aluminum Cantilever Beam Subjected to Uniformly Distributed k Exponential Forcing Function e' t (I. 4.beam.0074 for the fundamental mode was observed from initial laboratory tests. = 0.6..... c. IFT was taken using only the imaginary part of FT. 2 due to the folding effect refer to the Appendix). bunker...001 at the end of the sampling duration. .1%..5 ft. sampling was started at time Ts . The similarity between the exponertial forcing function used and the quasistatic response was examined.0PO o .. the error in the quasistatic response due to the folding effect is around 0.t. 1970 at Defence Research Establishment Suffield. rbe steel bases were then fixed in the field by bolting then to concrete bases... and the other of length 2..Kt is around 0. D.5 rad..sinusoidal response corresponding to w 2 can also be removed if needed... The differences between the two modified responses of Fig. Four strain gauges (one facing charge one on the opposite side.. the real part or imaginary part only can be used..0 .d obtained from a piezoelectric gauge at lhe expected 12 psi nominal overpressure lo~ation Is shown in Fig. This modified response is different from the modified response shown in Fig. S. This test was conducted on JuJy 23. to the existing side lobe in the frequency domain when damping Is considered..S ft.5 ft. .. the exponential forcinp function and the quasistatic response obtained from the modified response considering damping were not shown in Fig. N f .
!. sampling was started for the 2..o.75 in.beam at the location 1. The actual quasistatlc response (up to 65 as) was obtained by eye from the quasistatic responses (which were obtained by taking two more and two less sample points at the in Fig.5 ft.5 388...1 144. sampling was started at Ts s 33 AT and continued for the duration of eight times its actual funJamental period.o TIE lSt€oftsi Fig.125 ms.. Then a conversion into the frequency domain was made using FFT. Strain outputs from the 5 ft. mode fundamental wave arrives the air blast shock front of the beim..3 in. The responses show that the ground shock arrives at the beam before the air blast wave. beam..5 ft. For the 5 ft.3 in.i The beam response from the air blast wave consisted of a slightly damped sinusoidal response (mainly of fundamental and 2nd mode type) superi mposed on the quasistatic response. After cutting the impulses (on the imaginary axis) corresponding to the fundamental frequency wl.beam at Ts = 16 AT and was carried on for the duration of twentytwo times its actual fundamental period.? obtained from 500 Ton TNT Overpressure 6 Explo)sion Fig. due to the difficulty of f'l=ing an integer number of 116 .. to an "averaging" method shown (refer response is Appendix).0 3955 2....23 sec.. and Diameter 3 in. beam is shown in a up of than is made response modified 8. IFT was taken using only the imaginary part to get the modified response. from the root and from the 2..125 ms.. Several attempts were made to remove the sinusoidal response corresponding to the 2nd mode (. type.0 in. The Fig.c response (dotted line). _____________I 1 Fundamental Frequencies To make the fundamental mode of the sinusoidal response a sine function only.3 in.3 in.. 9. . G ©4 . The response was sampled by digitizing using a sample interval (AT) of 0.5 ft. mainly of the Whenat is initiated. at II psi Location (AT = 0. 8 Responses of Circular Aluminum Cantilever BELan of Length 5 ft.. N = 2790) The modificd response and the actual quasistatic response for the 2... beam at the location 0. The actual fundamental frequencies obtained. I A 0 o t: &a . a 145.2 in.beam were obtained using the same 1. C Fig. The result for the 5 ft. are shown In Table 1.Predicted(rad/sec) Actual(rad/secl S ft. 7 Response (Strain Output) of Circular Aluminm Cantilever Beams Subjected to Air Blast Wave of Peak Overpressure 11 p.2 in.2 in . The time t I 0 was set to be the time when the shock front arrived at the beam.2) from the modified response using the technique described previously.. (slightly damped free vibrations now exist). from the root are shown In Fig.. However. The domain conversion method and the cutting impulse technique can be used to remove the sinusoidal response and so yield the quasistatic response.2 in. Fundamental frequencies (w ) of both beams were obtained by taking FT of the responses after 0. Field at the Location of Expected 12 psi Nominal Overpressurc. ithe using This by a dashed line. SMA Of LtI6TM * V nD SD4WTII *I ACTUAL OVASI. higher har *odes (which response sinusoidal Ibeginning) 2nd) superimposed on the quasistat. as well as predicted values. 7.5 ft. the initial conditions of the beam respOhseare no longer zero because of these small'free oscillations. The actual quasistatic response does not have any sinusoidal response remaining In it.ocedure and are shown in Fig.TABLE .STATIC IaCSPOS D 6 8IA WI~t~MACi1YIIMA Of LIOTP $FT AND 1AMIT 3 4%  .. Due to the excitasmall free oscillation of the beam..
beam.8t379..5 ft..81 x 105 and Mach number 0.6t 76. The regions of inaccuracy resulting (intc' section regions) are actually very short.267967. however.. Drag coefficients can then be obtained from Pd(t) Cd(t) = (t 1.3 in.41 (at time t 0) to Reynolds number 4.6t 2 .90 267967. Actual drag pressures were obtained from (50) and (51) using (21). the "equivalent" drag coefficient. The error will be small.82t7990.2 in.53t22080. 10 and II..7t 3 (50) =.18t22080. at 11 psi Location (AT = 0. The type of curve fitted was a polynomial obtained using polynomial regression. modes up to the 6th were taken.5 ms the flow the incident shock and hence there is no true is conditioned by reflection and diffraction of 2.25 ms). is useful for the purpose of constructing the loading configuration on the beam as shown in Fig. This difficulty arose as a result of the relatively large sample interval for U2 and the accumulation of error in obtaining FT and IFT using FFT. A smooth curve was drawn through the intersection point to connect the two parts for each case.beam 0 .15 ms: P (t) 299549. However.63042.t 3 +50734'8t 2. However.2 in. beam 0 .643tS87. With regard to (50) and (51)..35 ms Pd(t) a 11.87t7990.3 in. Mach numbers t 117 . + .beam 1. 1 <'   r' = . as it may be termed.23 (at time t 65 ms).1 Y 105 and Mach 0 tumber 0. I+* . beam nor after 35 ms for the 2. IsEco*os + Fig. differences will be small because the slope differences at the intersection points are small. than the because second damping were very hard of to DAAGo PRIsIIIR A S 10 1[ os i iobtain /UIIIJ'I1IM . and Diameter 2 in. cantilever beam for the duration of the initial 65 ms are for Reynolds number 7. DISCOSSION Drag Coefficients Drag coefficients obtained from the response of the S ft. Due to the "bumps" (a fairly rapid rise followed by a gradual fall) in the responses.55t(4774523t2+ 0 .369392. higherHowever.beam 2 + 1. 8 and 9 ind can be represented by the following: 5 ft.3 3 (51) 2 15 .7!"3  ocL0 00 003 0 I ~ 00 ..2 in.(21) was used instead of (36). it was not possible to construct an impulse at w2 in the frequency domain.39624. N a 2796) cycles of the 2nd mode from the modified response.. 9 Responses of Circular Aluminum Cantilever Beam of Length 2.571+563..01 and the damping effect was in fact very small.1Ot + 26127.coo0co 1 •IM. ..265+S72. In (21).9t 3 As mentioned when dealint with response.15 ms P 15 .5 ft. because the intersection region is very short.8t +50734.3 in.23t 0 .h ACtuAL ~ QASISTATIC os R[SIPONS[ NO1 IfV94~ stATIC RsPt either of the two parts of (50) and (51) in each case. 1.8t 3 2 2 For approximately the first 0.44631.125 ms. .35 ms Pd(t) = 1. the drag pressures (and also the actual quasistatic responses) were not obtained after 65 ms for the 5 ft. An empirical relationship was fitted to the actual quasistatic response time data (data were obtained using a time increment of 1. the existence of the intersection region introduces an error.006t4774. (36) should have been ratios modesused. the relationship for each beam was obtained in parts as follows (actual quasistatic response Ps in psi and time t in sec): 5 ft. there is a slight difference in the slope at the point where the magnitude of the responses Is made to coincide.35 ms: Ps(t) 3S 6S ms: Ps(t) Pd(t) 5 Pdt) 35ms 55ms 2 2( 5.986tS87. The intersection regions cannot be expressed by drag coefficient until a steadily decaying flow is established around the cylinder.68753.35 as: P (t) *3.10t2+ (52) 26127. Drag coefficients obtained from the responses of the two beams as a function of Reynolds number and Mach number are shown in Figs. .9t3 sExpressing the actual quasistatic response in parts using polynomials gives rise to the difficulty in matching the magnitude and the slope (rate of change) of the responses at the points where two parts meet.625404.08t 5625074. Bec'use this error will accumulate as more parts are taken into account.3 in. The drag pressures obtained are shown in Figs..5 ft. Since slight damping exists in the actual response.5 ft.
.2 in. Since the response was mainly of fundamental mode type. IAcH Fig. os 07 Elastic 0The 2 050oscillation IX04 0sponse the shear Centre and centroid coincide. boam 1. the maximum velocity at midsection was obtained using the maximum displacement at midsection and the fundamental frequency. As shown above. beam REYNOLDS (MACH No 0.41 (at time t = 0) to Reynolds 5 and Mach number 0..differences were observed: 1.. 11 Drag Coefficient vs Reynolds Number and Mach Number Obtained from the Elastic Response of Cantilever Beam of Length 2. Thus with regard to the steady state values of the drag coefficients.It"o. Thus.5 ft. The drag coefficients above the critical Reynolds number range are a little higher compared to the steady state values even considerIng the Mach number effect.5 ft. IIIIo1r 0 4838 REYNOLOS No.. oI 10101 103011 . .2 in. the obtained strain outputs from the strain gauges (placed in the direction of the blast wave) did not record any strain from the side oscillation.2 in. The levelling off of air blast drag coefficients at Reynolds number around S x 105 was not expected.41). .5 ft. Since ing vibrations about the two perpendicular axes are independent of each other. 3 in. t Reynolds number effect on drag coefficients is expected to be dominant in this Mach number r.. and Diameter 3 in.ocation are lower than the critical Mach number (which Is approximately 0.3 In.I 0J 01 06 I111 1614S. This side is very small compared to the rein the direction of the blast wave.o40 ) The maximum displacements at the midsectior.7 x lO 35 ms).01 ft.3 (at time t number 3. at 11 psi Location beam response for the duration of the initial 35 ms. the maximum velocities were: 5 ft. The upper critical Reynolds number is higher in the case of air blast drag coefficient values 2. the following 2. Ol4l 40*1.. It appears that the drag coefficient increases considerably before levelling off as Reynolds number decreases below 3. However. This may be the explanation for the "levelling off" of the drag coefficient at Reynolds number aroud 5 X 105. Thus. .2 x 105 and Mach number 0.004 ft.at 11 psi l.ngc Drag coefficients of a circular cylinder subjected to an air blast wave generally follow steady state values. are for Reynolds number 5. Thus.61 07 f# Wo 0.61 and a minimum value of around 0.5 ft.41 for the range of Reynolds number and Mach number observed. the displacements of the. and Diameter 2 in.55 ft/sec 2. 0conditions 01104 0.5 ft. Of course this region is in the critical region where flow conditions vary continuously.7 x 10 Response and Maximum Strain response of a cantilever beam subjected to an air blast wave will exhibit a side oscillation perpendicular to the direction of blast wave travel due to vortex shedding.2 in. beam 1. 106 . the bend 03 03 035 04 045 05 055 . 10 Drag Coefficient vs Reynolds Number and Mach Number obtained from Elastic Response of Cantilever Beam of Length 5 ft...45 ft/sec The velocities of the beams are very small com 118 .I 045 1045 J5 0o 4 45. any small shift in the flow can cause considerable changes. No (. a3. beam. Fig.5 08 Drag coefficients obtained from the 2. beams were very small and hence use of the Euler beam equation is justified. beam 0. of the two beams were: 5 ft. Variations in the drag coefficient (the same kind of phenomenon as noted above) were also observed for the 2. Drag coefficients as a function of Reynolds number and Mach number varied between a maximum value of around 0.
5 ft. is the Kronecker delta k*i k** Is k* and N . When X(i) and x(k) of (A.. mean drag [R{(i)cos WII{ (i))sin 2 i] i0 1 Ni kEO _N Ni 2k* 21 iN sk* 2k coefficients of 0. Therefore. I together with initial conditions.i=O l(X(1))sin i is an antisymmetric expression of WR(x(k)) folded 119 t! 2.5 ft.2) x(k) = i X(I) 3 N where N is number of sample points. it can be seen that T0 R((i))cos i is a symmetric The analogous Discrete FoL. The maximum strains so found were very close to the strains determined using mean drag coefficients. expression of R(x(k) ) folded about the folding sample duration tf and .ation tf(k z N)and the latter is antisymmetric about tf.3 in. Computed and measured maximum strains are shown in Table 2.1) is ~ Nj 2lI.48 and 0.3 in. For both beams."2 in. the good agreement suggests that the loading configuration used is justified. Instead of using mean drag coefficients with dynamic pressure of Friedlander decay type as was done above.beaml 490 TABLE 2 510 In (A.{Rqi)j the same 1s2 N 2 result will be obtained. * Thus R(x(k*)) becomes NI 2N k=O (k))}kk** 2o 1 N are analogous to the integral The Fourier Transform pair for continuous signals can be written in the form X(W) z . it can be shown that R(x(k)) is composed of two parts.1l) I R{x(k)N.cos 2xkydy respectively. maximum strains were computed using the loading configuration of Fig.. beam and the 2.2) are separatcrd into real and imaginary parts. were used.ft x(t) e'jwtdt 2 (A. 'beam (A..beam. Also..6 +N k=O k** kk** where 6kk.21Nk i sin 2k.. maximum strains aere also obtained using the actual drag pressures of (52) and (53).4).k* rk** ' I when k** is k* k x(t) z f' A(l) )Jwtdw = 1 when k** is N In (A.562 for the S ft. Computed(uin/in) 653 1EI I(x(k)csNi N k0 icO I NI N1 2k* 2k 'sin W. maximum strains were obtained near the quarter point of the first cycle The forner is symmetric about the folding sampling du. the effect of the velocity of the beams on Reynolds number of air flow around the beams can be neglected. The good agreement between computed maximum strains and experimentally obtained maximum strains implies that the obtained drag coefficients up to at least 4 of the period of each fundamental frequency are correct. .ier Transform pair to (A. the first term resulted from N1 R{X(i)}cos r 2nk* and the second term from iN12ak F . the air particle velocity was 504. 2rk* forms c s 0o 2"k*y'cos 2wky dy. Analytically. respectively. Ni 2Nk NI 2Nk 1 R(X(i))cos L i and iO l(X(i))sin L i.pared to the air particle velocity during the blast loading (at t a 0. Ni 2 1 k* lie cos Wicos N izO 2lk• Ni Experimental and Computed Maximum Strains 2r Nl 27Tk*. Icos 2wk*sin 2wky di s 2ak*ysin 2wky dy and 1sin 2aky. Mean drag coefficients are the mean values of the drag coefficient for the duration of a quarter of the first fundamental period of each beam.3) 2. For the drag loading.)0 X(i) NI x(k) 2A k kF 1 N~i) 1 x~k ) e N N=(A.2 in.3).S ft/sec). N sin ix*sn  Ni and t=0sin The computed maximum strains show close agreement with the experimental values. Therefore. .i N kE0 R(x(k)1ii0 sin N kWOk4p I NI NI 2vk* 2%k I S ft. Expanding R(x(k)) for a point k* (k I N gives R(x(k*)) * of each fundamental frequency.
there will be impulses at the corresponding plus and .t the frequency domain. the sinusoidal function Fig.. by taking IFT using I({(i)) only. The same is true for I(x(k)). functions very close to slightly damped plus and minus cosine waves (with zero phase angle) of initial amplitude 0.10) 0 .o 0 Fig. the error due to the existing side lobe was 2% at most. Whlen damping exists. The DFT of a sinusoidal function depends upon the sampling duration taken. the error dtic t. the harmonics between N/2 and last half of the time function can be interpreted as negative time.oo o .5 rad/sec and 41 = 0. A.02 was recovered.I. Since R(f) is inter   preted as being periodic from the sampling theorem. Thus.. "2" s # N . Likewise.I Slightly Damped Sine Function in Dmin FrAquecy UNI 40 j .ngle) of initial amplitude 0. A. . if the middle of the two recovered functions is taken (an "averaging" used). a function very close to a slightly damped sine wave (with zero phase . In Fig. taking DFT will not produce impulses. the existing damping can be minimized . NN 40 400 j REAL PART o0  Bt taking IFT using R{((i)) only (setting I((i)) equal to zero). The fact that the occurrence of impulses depends on the sampling duration of the DFT can be explained by the sampling theorem 171. respectively. . If the sampling duration is taken as an integer number of periods of the corresponding frequency. e "41W lt are shown for different sampling durations. After cutting the impulses. The sampling duration of I is eight times the period of w1 . for these cases. For II and III. Likewise.oo074 . R{R(i))is symmetric and IT{(i)) is antisymmetric about the fiding frequency ff (i N).4s 05 [ /ME. taking more or less samp points than the exact number makes the side lobe change sign across the corresponding frequency. the imaginary parts of sin wlt.02 were recovered. 200 S II8T . Sin ) 1t e 1 l with w1 = 145.so . the modified h x(k) which is antisymmetrically folded about t in the real part will be obtained.  pazt will be obtained.2. minus frequency points iii Impulses will be obtained on the imaginary axis for a sine function and on the real axis for a cosine function. IFT was taken using only the imaginary part for each case. Cos w t e' l1lt will have the same FT as sin Wlt C e "1lt except that the real part and imaginary part will be Interchanged. the Fourier coefficients between N/2 and Ni can be viewed as the "negative frequency" 1. Thus R(x(k)) :an be obtained using only the real part or the imaginary part of X(i) If Rfx(k)) is finite up to tf.1o24 .L2 * N N. NOA 2 . There are impulses at ± w I and small side lobes around ! wI in the imaginary part while the real part is far different from that when there . and the sampling duration of Ill is longer than that of I by TI As shown in Fig. Fer I. " 0 L (Si " .0074 is shown in Fig. WI W. For 120 0 W M" N . A..about tf.1455 N • 1024 T. A.f the sampling duration is not an integer imber of periods. when x(k) is real.2 Imaginary parts of Slightly Damped Sine Function Obtained with Different Sampling Duration cases I1 and III. Thus. A. is no damping (the real part is zero when there is no dimping). . the sampling duration of I1 Is shorter than that of I by AT. The magnitude of an impulse is .00o74 cannot be represented with impulses only in te 4 t frequency domain.1 . a modified h x(k) which is symmetrically folded about tf in the real 300 IMAGINARYPART .2.
W.. 6.7 psi.38/67. "The Effects of Nuclear Weapons". A. 1962. W. "Aerodynamic Drag Measurements and Flow Studies on a Circular Cylinder in a Shock Tube". "Drag Measurements on Cylinders by the Free Flight MethodOperation Prairie Flat".D. Incident Blast Wave 20. "A Guided Tour of the Fast Fourier Transform". S. Rowe. "The Interaction of a Long Duration Friedlander Shaped Blast Wave with an Infinitely Long Right Circular Cylinder. 1962. July 1969 ACKNOWLEDGMENTS The work reported herein was supported by the Defence Research Board (Grant No. 121 . The Fourier Integral and Its Application. S. Brode. Glasstone. 0 . Inc.S. 8. Mellsen and R. the sinusoidal a cofunction must be a sine function only or cutting the of use to make only function sine impulse technique. AWRE Report No. June 1955. Burnewitch. National Congress of Applied Mechanics. 249. 1969. Journal of Applied Physics. 853866.B. Naylor.. May 1969..B.From the above discussion. Bishop and R. Mellsen. 7169. Bergland. Suffield Technical Note No. Proceedings of the second U. S. 2. IEEE Spectrum. 1957. and a 16 cm Diameter Cylinder". REFERENCES 1. Positive Duration 50 ms. 3. "Numerical Solutions of Spherical Blast Wave". Ewing. the sinusoidal function does not necessarily have to be a sine or a cosine function only. McGrawHill Book Co. A3384). N.E.L. 167809) and the National Research Council (Grant No. Jr. Suffield Memotandum No. pp. Papoulis. "The Elastic and Plastic Response of Cantilevers to Air Blast Loading". The authors wish to thank Defence Research Establishment Suffield personnel for their helpful suggestions and for the opportunity to participate in event Dial Pack (500 ton TNT field explosion). New York.D. G. Revised in 1962. 4. Published by the United States Atomic Energy Commission. Jan. it is evident that when slight damping exists. J. V. Baker. while in the case of no dampA ing. April. 5. Hanna. and G.W. W. 1967.J. 7.O.
or Pi terms." Ie assume equality of blast pressure. Since accurate theoretical knowledge of this explosivetarget interaction is limited.* The utilization of mine neutralization hardware require that this equipment operate under the intense pressure of nearfield explosive detonations. Preceding page blank •. itwas necessary to determine the dimensionless products. These parameters. Hopkinson scaling was used to determine the proper charge size and location for oneiquarter scale blast tests.L aM 2 IL 1 p1/2 M112 Q Response scaling 1 2 I I P Lm=. Initial conditions and restraints . Westine [1] concludes that density and seismic velocity. Ten dimensionless products. Virginia 22060 Idimensional anaiysiswais peorfoRd to determine the physical scaItng parameters governing the response of wheels to blast loadihg. cM Soil conditions resort to fullscale explosive tests to evaluate 302P/ 7r4. Morris U. If the same material MM isused in the model and prototype wheels.S. Listed below isone set of dimensionless products or Pi terms.rt Irs 5 9 Zp fZ z7=E PO /f8 = Tig= = /i0 . ie. crater and impinging on the target.J INTRODUCTION IT tLl/2p 1/2 Time scaling . so this Center elected to use scale models to evaluate materials and configurations for mine clearing roller wheels capable of withstanding the blast effects of 30 lbs of explosive. are listed inTable I along with their dimensions ina forcelengthtime (FLT) system. The test and calculation procedures are descrbed.Pm Pp. along with others governing this phenomena. These constraints are then applied to the above Pi terms to establish the scaling law below. can be formed from these 13 parameters. parameters describing ambient air conditions were omitted from the analysis.. Since the target wheels are in contact with the ground.j MEASUREMENT OF IMPULSE FROM SCALED BURIED EXPLOSIVES Bruce L. and the scaled specific impulse was 1 clwculated. or Pi terms. T2" . It isbelieved that a major portion of the impulse imparted to the target iscaused by the soil being thrown out o? the. governing the interaction between explosive and target. E /7 where m and p denote model and prototype respectively. Amy Mobility Equipment Research and Development Center Fort Belvoir. best describe the soil conditions. rather than a stress parameter. the analysis isnot modified as only the algebra is deleted. 123 . I? = X 3. The total energy imported to the test wheels by the detonation was determined.L3. DERIVATION OF SCALING LAWS Inorder to correctly interpret the experimental results. There are many techniques for creating this list of terms. "0P 1/2 113. and no matter which method isused. This process is expensive and timeconsuming. For replica models. designers have had in the past to and prove their designs. .
. and the spring rotation wai measured by.6 blasting caps. ur= A. Figure 1.A7_ _ _ _ _ _ _ _  .4T LTI L FL']T LT? F FL LT2 FT FL"2 2 2 p  P TEST EQUIPMENT AND PROCEDURES The test wheels were placed in a 3. V2 p m/Op 773 cm= cp rr 4 (ri)m=(ri)p fl 5 gm I g 5 16 fm=Xfp 17 7 f8 Em.2 T FL.AE p m=l a p scribe1. The wheels were connected with a common axle and secured to the test box by rubber torsion spring with a total 'calibrated rotational stiffness of 1180 inchpounds per de. this term is distorted as an engineering judgement. The charges weredetonated by M. The explosive charges (2 inch square blocks of C4 explosive cut to provide desired weight) were placed under th.5by 2. and test yokes.JWp to produce equalblast pressures. Loads scale as square of length ratio to provide equal stresses.5 by 2.the scribes in Figure 1. Test box 5howing rubber torsion springs.center wheel with onehalf inch of soil cover over the Lharge. Thus. ykes orVeel 0 cag springs kmYl fr0 Tmp Hopkinson has shown that blast pressure is a function of standoff distance R and charge weight W as P=f(R/WI/ 3 ).  Description Units FL. Physical Parameters Governing ExplosiveTarget Interaction Symbol P t Blast pressure Time Mass density of soil c L ri M g f E a t (T1 <tm=t Seismic velocity of soil Characteristic length Shape of System Mass of target Acceleration of gravity Total load on wheel Energy absorbed in wheel system Acceleration of wheel under blast load Impulse applied to wheel Stress in target Time scales as length ratio Sam soil for model and prototype Geometric similarity Since model and prototype tests are conducted in same gravitational field. gree of rotation. 124 5l .. if Rm=ARp. scribes. i Table I.  "  .5'foot dirtfilled test rig as seen in Figure 1. '.
the solution of eq.K(Qp+ (1) (2) Substituting for A and B and solving for i yields i= 2K1 max I + 2 Mo (3) But Mo=K Op and the above equation reduces to eq. but the impulse on the center wheel alone is desired. (M o0+ 0 + ( .i moFigure 2. the wheel torsjons rtfig system is as shown in Figure 2. (6)and (7)yields A= Mo K and B= Ii 0 = 0.IMPULSE CALCULATIONS For computational purposes. d = depth to center of charge in feet 125 ~! . 2From conservation of energy. (5) is given by "(6) Q= A cos w t + B sin w t . i_ .E. x A 1 T torque Ignoring gravitational effects.1 K )Q= "(  ) (5) Boundary conditions on eq. (4). the total energy absorbed in the test system is given by E= ForceX Distance= [k(Op ) X Omax X ] +O V Y (B/A)2 K max). K. and the results are substituted into eq. I a mv VIEm. Thus. or when Mo A . (5) at tO are and 0=orig.= P. and V= 2E/H From the impulsemomentum relationship.Omax occurs at a time tma x when tan w tmax = B/A Sin w tmax and cos w tmax are calculated from tan w tmax . I borig t OO %% 1.ma x Equation (4)can also be derived using a conservation of momentum approach. Equation (4) gives the impulse imparted to the threewheel test rig. This is approxoy ic + 2 ic sin $ = i3 wheel where ic = impulse on center wheel VT 5ubstituting eq. (3)and realizing that Izintl yields 1= VYK1 meimated (4) 2W A1+ Qma. (6) to determine Qmx. T= 1 0 where 0 and 0 denote derivatives with respect to time.E. i5orig" For small angles (sin 0 0).Mo 0 9 prelcad angle maximum rotation spring constant total impulse total energy initial spring moment length of yoke mass moment of inertia o=A w sin w t + B w cos w t where 0max K I Substituting the boundary conditions into eqs. (1) into eq. Here. E.
.845 . (ft/lb" 3 ) Impulse on 1 Wheel (lbsec) 9 Specific Imp (psisec) Scaled Specific Impulse pstmsec lbl/3 656 875 594 782 766 1420 755 1165 1260 1880 744 900 1370 2 3 4 5 6 8 9 10 11 13 14 15 16 .00 3.631 1. 9 to 19.457 .7 13.460 . Vol 1 of 2.285 .00 5. These scaled specific impulses are presented in Table II along with the scaled distances and other data items and are compared to previous extrapolated data [3] for TNT in Figure 3.6 14.575 .054 . "Explosive Cratering. Table II.402 .420 .285 .068 . W.9 23. Fort Mc Nair.10 4. "Explosions and Explosives in Air. March 2425. 7.072 .4 31. NDRC.40 i .470 .S.7 15. Summary Technical Report.056 .420 . of "Impulse Imparted Targets on tn Detection and Neutralization.S.381 . Washington. The TNT data has been adjuated to C4 explosive.395 .Westine [2) has developed a method of calculating the impulse imparted to a target from a land mine detonation given by I= A0 REFERENCES [1].9 20. into Sympesium Land Mines.489 .1 10.496 .062 .056 .088 . These shape factors. pp.7 14.342 .specific impulse 2 A . This impulse is transformed into scaled specific impulse by dividing by the cube root of the charge weight." by Detonation P.55 4.25 4.060 . Data generated from these small charges is thus seen to fall within the limits of data generated using explosive charges of up to 550 lbs. Washington. Scaled Specific Impulse Shot Number Charge wt. Westine.60 3.2 13.285 .057 12.281 .062 .504 1. " in Effects of Impact and Explosions Volume I. D. D. (lbs) Scaled Dist.Westine. Terramichanics. (4). are used to calculate the specific impulse generated by the detonations. 2.25 3.1 21.projected area of target 7 in for these tests 0 = shape factor which is a function of target shape and standoff conditions [] [2). together with the total impulse as calculated from eq. 11o.9 4. impulse heeI where Ia specifi total impulse I .00 3.005 126 .1 24.058 .40 3. P.200 ." J. Vol.062 .285 .106 .1 16. Kennedy. 1946.45 4.062 . [3]. 1970. Shape factors are determined for the various configurations tested under this program. 1971.942 1.C.063 .30 4.00 3.229 .
5 LB rect blocks 10 LB bare charges Current tests  PUS UERL MERDC A P 2 1 . 127 .2 \ 11111II I .1 .6 1 2 3 lI I III 6 10 I 20 ! 40 4 SCALED DISTANCE (FT/LB1/ ) Figure 3.4 ."" o V) 1SYBL SOURCE TYPE OF CHARGE 42 4 8 A 6B8 D P U + BRL RRLARD DATA SHEET Bombs. all sizes 0.06 . all sizes 8550 LB bare charge Bombs. Scaled Specific Impulse vs Scaiea bistance.2000 1000 800 600 * 4+ +  iii 40 1  03 20B %"L .
they are easily solved for the harmonic steadystate response. From these equations. These system modes may be derived directly from a finiteelement model or through the modal coupling of a number of subsystems. The introduction of the momentum term produces a skewsymmetric. This method is based on the assumption that the modal damping matrix is diagonal. The introduction of the momentum wheel terms. The purpose of this study is to include the momentum term and develop a method to facilitate the calculation of the transfer functions so that their effects on the structure can be evaluated. There are times.DYNAMIC ANALYSIS THE EFFECTS OF MOMENTUM WHEELS ON THE FREQUENCY RESPONSE CHARACTERISTICS OF LARGE FLEXIBLE STRUCTURES F. Once the equations are uncoupled. The program was used to calculate transfer functions on the Skylab Apollo telescope mount both with and without the control moment gyros. rather than a diagonal.e. The structural characteristics are input In the form of normal modes of the system with nonrotating momentum wheels (i. Day III and S. analyses of this type have neglected the effects of the momentum wheels. This meLhod was used to calculate transfer functions for the Skylab Apollo telcscope mount (ATM) in order to determine the effects of the control moment gyros (CMGs) on the structure's behavior. The results are compared in this paper. Transfer functions with and without the C1Gs were computed. complex eigenvalue/eigenvector transformations are then generated and used to uncouple the equations of motion. These modes are sufficient to uncouple the equations of notion only when modal damping or damping proportional to either the mass or stiffness matrix is assumed. Colorado A computer program using existing mathematical techniques has been developed for use In performing linear frequency response analyses of large elastic systems that contain axisymmetric momentum wheels (gyros). damping matrix. R. recouples the modal damping matrix. The importance of this example and the reason for choosing it lie in the fact that these same transfer functions are used In designing the experiment poitting control system (EPCS) for the ATM. This is done using the state vector form of the equations of motion. A comparison of these transfer functions is presented. all translational Inertias and the rotacional inertias about the axes normal to the spin axis are included). Undamped normal modes are used to describe the characteristics of the flexible structure. Because of the proximity of the COGs (which are used for overall cluster control) to the ATM. however. when a structure has gyros or momentum wheels. In the past. their effect on the EPCS sonsors could be significant.NT The derivation of techniques used to generate acceleration responses or transfer functions for large flexible structures with momentum wheels will be developed in four parts: Preceding page blank 129 . and uses complex eigenvalue/eigenvector transformations to yield a set of uncoupled equations that are easily solved for the steadystate responses of the system. however. Tomer Martin Marietta Corporation Denver. INTRODUCTION The generally accepted method for deterraining transfer functions on large systems is based on calculating the normal modes of the system and using them to uncouple the equations of motion. The method makes use of existing mathematical techniques to develop the state vector form of the dynamic equations i motion. The program incorporates the state vector form of the dynamic equations of notion. D. for example. and a second eigenvalue s3lution must be determined. when the assumntion of a diagonal damping matrix is not valid. THEORETICAL DEVELO..
1 qol (4) consists where the modal coordinate vector of complex numbers yielding magnitude and phase information. o . These techniques rely on the orthogonality of the undamped normal modes of the structure or substructure to produce diagonal (0 1T ol0 [meqJ when the modes are orthoncrmalized on the wass matrix. Equations of motion for a flexible structure with spinning momentum wheels or gyros. T 2. The uncoupled set of equations formed by Eq. 6 into Eq. In terms of the generalized or modal coordinates.~ 1. when the normal modes are ortho. 2 yields a matrix equation relating responses to forcing functions in discrete coordinates: q where ml ()+ [rw1 P 0eq { . Equations of motion for a spinning rotor as a free body. generalized damping (F(t))  { (f(t)) generalized force. are related to (q) by the modal coordinate substitution (x)  [] {q). A brief review of the equations of siotion for the elastic structure without momentum wheels. Euler's moment equations can be expressed as follows: and the discrete coordinates.9~~  ~ 7~ 17777. Equations of Motion for Large Flexible . (3) Io x x + 0 I 0 y 0 O. (j~  4. The solution involves only 4W m o eq] t [k . the uncoupled equations are of the form Substituting Eq. 2a ](q) eq (1) {Xo}= [41  1 I + i 2. mentum of motion. As will be shown in the following including rotating moequations the modal fully couples wheels sections.iF(t)) 01 Tfol" {q} [(in]  (7) vector of modal coordinates. 3. Substituting Eq. Eq. (f(t)) fo e t. 2 Smeq 1 . ~] o i 2CUwoRJ) io} (5) (5)o 1. Equations of Motion for a Spinning Rotor as a Free Body Consider a rigid body spinning about one of its principal axes of inertia.T r1digital computers. 6 and 7 can be solved quite efficiently by m [4 1 T a . and for steadystate responses. 0 0Tc matrix multiplication and the inversion of a diagonal complex matrix.W. Procedure used In the digital computer program for uncoupling and solving the response equations. modal damping is usually assumed ao that the final equations are uncoupled and readily solved. 1 yields. {4(t)) = jqo} e ift.. (2) Now. 2. Solving for 4qoj equivalent mass and stiffness matrices. y 130 . 3 and 4 into Eq. (x). greatly increasing the computer cost involved in determining responses from this formulation of the problem. for a harmonic forcing function.generalized stiffness matrix. If the coordinate system is fixed In the body at the center of mass and aligned with the principal axes of inertia. since Structures Response analyses of large flexible structures generally employ component modal substitution or modal coupling [Ref 1] techniques to reduce the number of equationsrequired for the solution. 4.ormalized on the mass matrix. In ad lqol (I2r21 2 +I r2~cfl1) 10]T f T{o") (6) dition.generalized mass matrix.j Fcr iirgleunit input forces. 7 describes the steadystate transfer function.
8 becomes Ix M. A rvird. 6 and Oi' respectively. t @ the equations of motion can be expressed in terms of a state vector by making the olon subfollowing a: [ef. (9) (q~t)) ilol for steadyatate response to a harmonic input. If one pvocaeds as in section I ari lets int (f(t)) = If t e (F(t)) . 12 would be in the form  0.) I 131 . Equation 12. JR]  .Ox  Ao . The proc t b f e ul 4.0. 12 as {q) + [B) {) + [E] (q) where _ (F) ( (14) the coordinate system's notion. however. A. 00 L F 000 ( 10) matrix is inThe acceleration coafficient dependent of the spin rate and can be included in the system inertia matrix when the momentum wheels are coupled with the elastic structure. In matrix form. 9 becomes 1.'Eq.IFo1 e ift . 1 E __ (15) ! 30 1I (q)+ +2W ] [. y z 60 .mI z I + y V X In the previous section. 3. the determination of responses from Eq. X Ox + 6y ( Iz " 1s) X.(f(t)) B [E] [ + ]T[] [w. m] (x) + (c] + [r]) fx)+ (k] (x) . were un For an axisymetric rotor (I x y coupled and could bereadily solved for responses. The solution then involves invertiog a fully coupled matrix that is also dependent on the frequency of the forcing function.is dependent on the spin rate and will couple the system modal equations.undamped normal modes yields 10)€ (q). however. Eq.I T [. is fully coupled in the velocity coefficients. and [F] contains the skew'ymmetric rotationaldependent terms. the equations of X y 1X = Is) with (8) +6 1y  notion in modal coordinatei. 0n 0 0 0 z ~z Io A. Since the inversion must be performed for each forcing frequency.'s nn. Since the inertias about any axes in the xy plane are the same. 6.mal to the spin axis. these equations can be uncoupled by rewriting them in state vector form and generating complex eigenvalues and eigenvectors. . y. However. constant angular velocity X about the z axis. The angular velocities and accelerations.F]) . are then descriptive of [r10 02 T + i j onj + [] (13) Cr1 [ ] / fFo}. 2 stt q+ 1  where [m) now contains the rotor inertias about . (x) uz:6 "1v. The velocity coefficient matrix. letting y = I Eq. 13 is very timeconsuming and costly. while the spin rate. Procedure Used to of Uncouple System Equations Motion the Combined Rewriting Eq. relates the body motion to the coordinate system. . the solution to Eq. transformed to the system coordinates. nate transformation. Equations of Motion for a Flexible Structure with Spinnip Momentdm Wheels Wriig the equations of motion for irtng teeutosomoonorthe combined system in discrete coordinates. the coordinates can be fixed at the center of mass with the z axis coincident with the spin axis and the x and y axes nonrotatin&. 0 j(s .
Jl IT] . each forcing frequency.. and Substituting back to discrete degrees of freedom from Eq. 1.arlo.j. Now. For r harmonic input and steadystate responses (J) {(J}  J 4 ¢o e e i Odt. 17 then becomes (18) and q (Jul . ] This formulation of the equations of motion involves the inversion of a forcing frequencydependent matrix that Is diagonal.(o). (21) Since. for the elgenvector matrix [7. 24 yields B B and. 17 produces Ju]  [V] (0} (20) [ T lower half of the matrix the is 172] where I) w 1  .. If responses or transfer functions are to be generated for more than just a few forcing frequencies.(J).0 [ 0 .0 B "'. the homogeneous form of Eq. 21 by system of equations In the form 1 1mJ (F.i0.i.11. 17 is the standard formulation for an eigenvalue problem.] ( (z into Eq. xi  pressed as functions of the discrete forces by I Assembling the eigenvectors into a matrix [Y]. 22 becomes 132 . since (Z) = q} € [UB 0](H JR]  0 10 (f(t))$ The homogeneous form of Eq. each time than to invert the coupled matrix In Eq. (24) [H] {z} JR] (D) or JU] (Z) where [Rh {7)  (J). SAMPLE PROBLEX The following example shows how the method (22) Was applied to include moentum wheels when determining the frequency response of the Skylab ATH and associated substructures pictured in Fig. 12 and 17 into Eq.The modal equations of notion can then be ex pressed as hC or 3 ItoI 17 t ill JR) {b + [HI Preaultiplying by [R] (i) [R]i Z} = ID).) . 13 for which must exist. it becomes much more efficient to generate the complex eigenvalues and eigenvectors once and to t invert the diagonal matrix r'o .(. (17) L. letting {z) =( 0 eat. (19) {)[ q H{) the discrete coordinate accelerations can be exwhere the set of values nj are complex eisenvalues with corresponding eigenvectors .o. (23) ao that Eq. yields (16) = ! o a 170 iR 3 ] Jol.J.[] .(v) t . uncouples the 1[]premultiplying Eq.]) {(o . the coordinate substitution ia (j 1f 2 1 a  n. The ATM is the experlment pertion of Skylab that houses the photographic and telescopic equipment. [  .] .
eTZ TX+Z y FIGURE 2..I FIGURE 1... which are used to stabilize the entire Skylab cluster in orbit. various photoeraphic experimenLs. . the elastic motion of the CMG support structure will cause the wheel (gyro) to generate a torque that may be picked up by the EPCS sensors and cause the system to respond. which has motion sensors on the spar.... are the momentum wheels considered in this analysis... The goal of this task was to determine the effect of the spinning CHGs on the EPCS transfer functions. ... The GRA is controlled by the EPCS. dictated that the size of this model be reduced: first. in addition. . Figure 3 shows the ATM spar... These three MGs.. DEPLOYMENT Figure 1 shows the ATM rack./ . CM. it houses the CMGs. which have a spin rate of 9300 rpm. ASSEMBLY... . deployment assembly. and ATM solar arrays.. .TY. and The spar is the gimbal ring assembly (GRA)... structure on which the various experiment packThe GRA is used to aim the ages are mounted. Figure 2 shows the orientation of each of the three CMGs to the Skylab cluster..... of which the CHGs are a part. and the EPCS are independent control systems..SO0ARARRAYS ATh RACK. The design of the EPCS depends on transfer functions calculated for various points on the ATM system. . canister... SPIN AXES 133 . The rack is the main support. the computer cost to obtain an eigenvalue I.... and houses the structure containing the experiments... Although the cluster control system. Using existing models of the various ATM substructures for the vibration analysis led to a total structural model with 1321 degrees of Two factors associated with this study freedom..
rather than on a 1321degreeoffreedom system. 13 Rvv UiI/I 4 ULjJY FIG. or inertial modal coupling method. as can be seun in the figures. to determine the modal properties. In general. By using this method. Prcdictably. 4 thru 7 indicate that the CMGs on the ATh have a distinct effect on the ATM transfer functions. depending on the frequency range in question. The input points selected were the GRA flex actuator torque motors and the rotor imbalance moments on the CMGs./FIN. __= FREQUENCY (HZI FIG. FLEX ACTUATOR i FREQUENCY IHZ FIG.. we decided to use the constrained component mode substitution method [Ref.. 6. 6lR HlZ) RQUENCY (ttzI 5. CMG 134 .. the amplitude of the transfer function for a given input/output combination may remain unchanged. and the CMGs. 3]. The frequency of the modes obtained ranged from 0. the ATH system was broken into three substructures. CMG Z6 i  E 5USCMSO VS. 4.. For this analysis. as it is more commonly called.CMIG VS. . CG  4 . The output points selected were the fine sun sensor and the EPCS rate gyros (both of which are on the spar). the flex actuators. the ATM rack. the eigenvalue solution for the final modes and frequencies was performed on a 288degreeoffreedom system.solution to this large a model was excessive. and second. consisting of the spar and GRA.CMGVS. the deployment assembly and ATh solar arrays are considered part cf the rack.SPAR CENTER V1.. the effect is random: that is. FREQUENCY (1111 1 CMNTGVS. and the canister. . . . 7. the current transfer function program could only accommodate 50 modes due to computer storage limitations. increase.. II II I u  II Spunup I ii I I I CMS. the greatest effect occurs when either the input or output point is near the CMGs.FI. To use this method. RESULTS AND CONCLUSIONS The results shown in Fig. or decrease.  FREQULNCY IZ FLEX ACTUATOR ~~E6 E FIG. These 288 degrees of freedom represent component modes of the various substructures selected with a frequency cutoff criterion.009 to 68 Hz..FINE SUN SENSOR VS. * E5 Despun CMGs :I II . FLEX ACTIU 0 aR FIG. . To perform the transfer function analysis on the ATKA force input points and acceleration output points were selected on the basis of their effect on the EPCS. In light of these two factors.
upper half of [Y) s[2] .I 0 efficient matrix EI 0 statevector cotaefictoi mti paeent displacement 1.e coordinate ampli [k) . NOMENCLATUREegnaus  [1. either generated directly from a discrete coordinate model or indirectly through component mode aynthesis [m) n mass matrix in discrete coordinates (ki = stiffness matrix in discrete Vectors (q} .vector of rotational degrees of freedom 2m .Although the results of this study indicate that including momentum wheels or gyros in the [Ri  r [ = j1. i N m Jq .vector of discrete forces (x = vctrodireefceapius vector of discrete coordinates .undamped natural circular frequency of T matrix generalized stiffness th mode 135 .atrix  coordinates d aeqodiagonal generalized mass [1.complex eigenvector ] .vector of complex mode generalized coordinates rt  [T] = coordinate transformation matrix relating the rotor local coordinates (0) to the system coordinates xl [B] . Limitations on the scope of this study prevented investigation of some key factors.0 I [ = statevector velocity co calculation of transfer functions for structures containing them can influence the magnitude of the the transfer functions.vector of normal mode coordinates (F(t)) .vector of discrete force amplitudes vector of modal coordinate amplitude/ phase coefficients { .q _ j ness matrix .diagonal generalized damping matrix [ol  in rotor coordinates v2oW vector o of normal mode coordinate generalized force coefficients [r] .vector of discre.0 0 (D) {2}tJ)l [RJ ('*T .vector of moments in rotor coordinates (O .0] coefficient matrix [R] [H] . on magnitude of their the effect governing are not clearly factors understood.matrix of complex eigenvectors [a J = diagonal matrix of complex matrx of udanpednormaleigenvalues~f[~ 1711 .0J 0  ..matrix of assembled rotationdependent terms No.lower half of [I] [) =matrix of undamped normal modes.state vector J l [ 1[T (D) = generalized forces in statevector formulation of the equations of motion Y 0 0 .statevector characteristic matrix [I] . of Rotors Z} .diagonal generalized stiff 0. such as the proximity effect exhibited in the example and the effect of highermode truncation on the validity of the transfer functions near the truncation frequency.fol .vector of generalized forces in normal mode coordinates  Ii T MW (f(t)) .coupled velocity coefficient matrix for inclusion of momentum wheels _ [E] 0  1 ol coefficient matrix in (JI o e Scalars f = circular frequency of forcing function W .damping matrix in discrete.damings matrix in discrete coordinates [c) .0J if the normal modes are orthonormalized on the mass matrix tude/phase coefficients {O] .
I
mass
inertia about principal axis
e
j
 rotational degree of freedom about axis
 constant angular velocity about rotor spin axis y = angular momentum
a
complex eigenvalue
Symol
diagonal matrix transpose of matrix
 inverse of matrix
[
I
1
[]
 partitioned matrix  partitioned vector
REFERENCES 1. W. C. Hurty, "Dynamic Analysis of Structural Systems Using Component Modes," AIAA J., Vol. 3, No. 4, pp. 678685, April 1965. 2. P. W. Likens, "Dynamics and Control of Space Vehicles," NASA TR 321329, Rev. 1, Jet Propulsion Laboratory, Pasadena, California, January 15, 1970. 3. W. A. Benfield and R. F. liruda, "Vibration Analysis of Structures by Component Mode Substitution," AIAA/ASME lth Structures, Structural Dynamics, and Materials Conference, Denver, Colorado, April 2224, 1970.
ACKNOWLEDGEMENTS The authors wish to acknowledge the efforts of Mr. Jack Nichols and Mr. Wayne Ivcy, ShE ASTNADS, Marshall Space Flight Center, for their aid in obtaining permission to use the Skylab ATH modal data obtained under Contract NAS824000.
136
T
i'a
INTEGRATED DYNAMIC ANALYSIS OF A SPACE STATION WITH CONTROLLABLE SOLAR ARRAYS Joseph A. Heinrichs and Alan L. Weinberger Fairchild Industries, Inc. Germantown, Maryland Marvin D. Rhodes NASA Langley Research Center Hampton, Virginia An integrated dynamic analysis and corresponding digital computer simulation for application to a space station with controllable solar arrays are presented. The analysis and simulation have been developed for the primary purpose of evaluating dynamic load Interactions between the solar arrays and the space station which can result from orbital perturbations of the combined system. Integrated into the analytical formulation are the dynamics associated with the space station, the solar array flexibilities and their respective control systems. Application of the simulation is made utilizing present concepts of a space station with large area arrays and of typical control systems. A structural analysis of Ahe flexible solar arrays Is initially required to provide modal data for the simulation; and analytical results for an array concept are given. A verification of the structural dynamic methods used Inthe simulation is presented. This verification is accomplished by the application of the simulation to a problem of known solution, a uniform beam subjected to a unit step load applied at midspan.
INTRODUCTION The solar cell and battery system has been successfully used on many small spacecraft; however, space stations of the future will have power requirements which are much larger thin those within the present design experience of solar cell systems. Therefore, the solar cell arrays used on future spqce stations must be relatively large and be capable of tracking the sun in a manner that does not restrict the desired space station orientation. This Is usually accomplished through the use of an orientation control system (OCS) for the arrays. A potential problem exists duo to undesirable Interactions between the solar arrays and space station caused by required control and stabilization forces combined with external porturbations. Spacecraft Inmtabilities have been observed in the past when flexible appendages are par' of the satellite structure. This past experience is summarized by Likins and Bouvier (1) . Because of the requiremento imposed upon large area solar arrays, a weightefficient design rather than a stiffness 137
design results, and the primary array frequencies may fall within the control system bandwidth. A digital computer simulation for evaluating the dynamic interactions of large solar cell arrays and orbiting space stations has been formulated and considers the dynamic characteristics of the array structure and the required systems for attitude and orientation control. The objectives of this simulation were to (1) provide an automated methodology of Interaction loads analysis for use as a design tool, (2) analyze present array structural concepts which are to provide 100 KW of electrical power to future space stations and (3) obtain an indication of space station stability by its real time motions. The equations of motion for an orbiting space station with attached controllable arrays have been generated and were digitally programmed for solution by numerical integration techniques. In the development of the system's motion equations, a modal synthesization technique was employed whereby the elastic characteristics of the arrays were described by a finite set of orthogcnal cantilever ndes. Only rigid body motions of the space station were
'p
~~~~"' ~~ ~
~
I
~
4
~
4
C~
~
~
~
.k
~
~
~
Lt
aC2JSr~s<~
considered. The rigid body/flexible body Interface was described by space station acceleration in terms of Induced inertial loads forcing the flexible arrays; in like manner the flexible array root forces and moments acted as forcing functions on the space station. Structural mode descriptiona of the arrays were required as Input to the simulation; therefore, a structural analysis of the elastic system was required prior to the performance of this simulation, Provision was made In the simulation for closed loop attitude control system dynamics of the space station and OCS dynamics for the solar arrays. The latter control system provided the desired orientation of the arrays with the sun by controlling the rotation about the orbitadjust and seasonaladjust ax.. Outputs of the simulation include interaction forces and moments, magnitudes of all motion variables and control parameters as functions of time. The formulated simulation has been applied to an extendible solar array structural concept and space station which are presently undergoing separate engineering evaluations. Also, candidate array orientation control and space station attitude control systems have been mathematically described and digitally programmed for this application. Two attitude control systems were provided for the space station; they are the reaction jet and control moment gyro (CMG) systems. The necessary modal
synthesis analysis of the array structure was performed by a stiffness matrix method utilizing equivalent discrete element structural models representing 600 Inertial degrees of freedom. The modes utilized in the simulation were chosen on the basis of a significant percentage of load participation in Interface force and moment. Simulations were performed on the orbiting structural system perturbed by Initial attitude errors and external forces representing docking. To demonstrate the adequacy of the methodelegy which has been simulated, a problem of known solution was selected  the uniform freefree beam planar response to a unit step load applied at midspan. The flexible appendage solar arrays were represented as cantilever uniform beams, having the first five bending modes as flexible degrees of freedom. When coupled Inertially with the rigid body translatien mode, the cantilever mode solution yielded results for frequeny and loads which compared favorably with the exact freefree beam solution. CONSIDERED PARAMIETERS The presented analysis and corresponding simulation Is intended to be applicablo to future space stations with controllable solar arrays such as that shown In Figure 1. Structural concepts
t
I
II
Fig. 1. Space Station and Solar Array Structural Concepts 138
2,of
arrays and space stations as shown and associated control systems are presently undergoing separate engineering developments without regard to a total system dynamics criteria for minimizing interaction loads. One objective of the present analysis is to assist In the development of dynamics criteria for each of the cormponent structures from analytical results of total system characteristics. In order to account for all of the significant dynamic influences upon the space station and solar array load interactions, the following parameters were considered as basic and are accounted for in the simulation 0 Solar array flexible body dynamics in terms of generalized modal coordinates. Space station rigid body dynamics. Space station control system dynamics including guidance and command. Solar array OCS dynamics Including guidance and command.
liroviding sun aignment within a specified'time after leaving tie earth's shadow. In addition, the OCS mus, meet accuracy requirements despite experienced space station disturbances andf provide minimum dynamic excitation to the arrays. Two generic types of OCS drive systems have been considered in the simuiation and ares he continuous and nonlinear drive system,s.' The continuoustype drive system employs either a DC torque motor or a variablefrequeny synchronous motor as its drive element. A block diagram of the cortnuousdriveOCS model contained In the simulation is shown In Figure 3.
Conhns.aio "4colr
__
* *
nik M
*
It was assumed that the space station structural frequencies are significantly higher than the solar array frequencies and are decoupled from the control system. The analysis considerations can be best described by the block diagram shown In Figure 2. This illustration shows the breakdown of the basic parameters together with the considered counling paths.
a Auous
71  m Kn "
or lmecon.st
r co:fic.,.,
IAK
sr(,
,ttis
biourotation axis
for lopcorn;timtIon ti.s vm.,stanM &W g..?gear r4o
0,  sSe nfiu Fig. ., Contiruous Drive OCS Model The nonlinear drive OCS Is similar to the contindrive system with the exception that the control logic of the nonlinear OCS is operated in an onoff manner. When the array error exceeds some preselected threshold value, the motor Ia turned on until the array Is driven to a null position at which point the motor is rwitched off. This threshold value has been made a user option in the simulation so that its affect upon loads can be evaluated by parametric studies, Two generic types of control laws/orquers have been Incorporated into the simulation for attitude cotrol of the space station and are the CM, , and the reaction jet control system (ICS). The CMG control system is used for precision attitude stabilization against eyclic disturbance torques without the need for propellant exr'enditurc. It consists of three twodegreeoffreedom control moment gyros with parallel outer gimbals and with their momentum vectors Initially equally spaced In the orbit plane (Figure 4.). This partict,'ar CNIG configuration permits simple steering laws and a planer, rather than three dimensional, entihangup law. The CMG control dynamics included in the simulation (2] ha? a system frequency of 1.4 IN and a damping ratio 139
K  8E
or.ynarl
Control
,Sp¢ eIatlomi.
Cenerator
J:,,i
i
i
.Solar Array OrlevvIaln
Driver
"
Analysis ConsideratIons The constraint placed upon the representation of the array dynamics Is that they bp described in terms of their orthogonal cantilever modes. The orientation and space station control systems that have been Incorporated Into the simulation are those designed to complement the present space station and array configurations. The primary requirements of an OCS are maintaining a desired accuracy with the sun vector and
A
It is also an alternate to the CMG for controlling attitude of the space station. 707. The RCS is depicted in the sketch of Figure 5. 6. These provide the space station with a fixed orientation relative to orbit coordinates and point the solar arrays at the sun. 4. It is comprised of 4 sets of quad thrusters providing redundant control about the pitch. provision has also been made In the simulation to allow for the application of time dependent forcing functions.Jet Location/Function 140 . 3 PM Centrot Moment Gyro Array The RCS Is primarily used for Space station and solar array motion equations were formulated together with the Interactive dynamics provided by the respective attitude and orientation control systems. The flexibility of the solar array Is modeled by means of a truncated set of cantilever modes which Is excited by the acceleration of the array support. A difference equation technique [5J is utilized to obtain the modal response. RCS Model ANALYTICAL FORMULATION Fig. The allowed axes of rotation consist of those parallel to the space station roll axis and the array vane axis. The simulation model utilized for the mathematical system is described below. Kh Attitude Argle Galn. * The space station and the two arrays are each modeled as interconnected bodies with each of the arrays permitted controlled rotations about the spacecraft attachment points. * l'itch Fig. Both the space station attitude and array orientation guidance commands are computed external to the structural dynamics section. Rate Gala P v Control LoopDeadbad T a RCS Torque L l I w Spe Station Moment of tnertla MO Axisl IVASr Gimbl Axis Fig. RCS . The RCS model included in the simulation is depicted in Figure 6 and the corresponding control equations are given In Reference 3. * +yau + Z * %. reference attitude acquisition maneuvers and the momentum desaturation of the CMG system. 5. K*. M~aneuver and attitude control of the space station together with the solar array orientation control are modeled In terms of the transient forces and torques produced by closed loop guidance equations. of 0.Z Axis C Outer Gkim Spbas X Aded ot lomon Odter Gimbal KKsu) Axis _° 1+ Sc Attitude Anigle a Commended Attitude Artle: late 0. The method given by Likino [4] has been used as a basis for the methodology provided in the simulations.. roll and yaw body axes. In order to account for space station disturbarices such as crew motions and docking forces. The simulation developed from the employed mathematical models has been successfully run on the CDC 6600 computer. All maneuvers using the RCS are performed by firing the thrusters in pairs.
_ (jL) T rj xF FI r' F I 141  (fo. J is equal to 1 or 2 qare the transient forces and torques produced by the flexible array dynamics torque exerted on space station by hinged body along constrained axes TA J " T  mT Tr 0 0 B .total system mass . The vare driver axis is directly driven. axis parallel to the roll axis Is modeled as an ideal mechanical transformer. n.of solar array.t . ~r~~i&r . The generator is included In the simulation to provide a reference for guidarnco commands.~ Va ~ ~ z.. A block diagram representation of the simulation program is presented in Figure 7 where Important logical switches and function Interconnection have been clearly delineated. In Figure 8.vector from point of external force anplication to space station CG vector from space station reference to system CG Tl Interface torque on .Center of gravity for the entire system TH . CG CGj RO FR Li L f = + FR (1) h Vector from space station reference point to space station/ array interface from the space station/ array interface to the CG of the jTH array body (on space Space Station Moment jvector d (( )x ~( 11+ 0 'xI ) jstation) it (2)  ri + T Ai J rr FA rej ' I A  .rj x(30.J array body produced by OCS +T +T AC CNIG T Jth Array Body Moment d dt (J T .. System Force 2 2 mT d (Re+C CS. ~ 's4 . 9 The simulation orbit generator uses Lyddane's method (6) for near earth orbits. The following three vector equations are utilized.Eg ai! .Newtonian reference point .h ) force moment arm .. and either or both motions about these axes may be rigidly constrained. A complete fornulation c! all equations may be found in Reference 7.The array driver gear train for the.~ ra~n~z n~z ~ MudS~aas 3I~. The following notation applies J FAj  lndex.Spacecraft reference point Spacestationcenterofgravity .Interface force exerted by space  station on jTlI array body . The space station and array are modeled as a subsystem of Interconnected bodies whose motion Is described by the NewtonEuler equations of motion..Center of gravity for the J hinged body Radius vector In ECi R External force applied to space statioui sai Angular momentum of J array body abouts Its CG Angular momentum of space station about Its CG  RIGID BODY AND FLEXIBLE BODY DYNAMICS FORMULATIONS Salient equations and techniques used in the simulation are shown below. S. The connected space station and solar array bodies In the Earth Centered Inertial (ECI) coordinate frame are show.
Shown in Figure 10 is a sketch of the flexible array geometry utilized in the analysis. it is assumed that the particle masses of the array have negligible inertias and deflections are sufficiently small that linear structural analysis it valid. and T 1station Cj X I XA is the vector In solar array coordinates. The submatrfees (Aij represent the linear term coefficients. The rigid body scalar equations derived from those presented above reduce to the matrix form shown in Figure 9.0 o 09 RF space station attitudes are referenced to their respective coordinate systems and are periodically updated during the simulation. The flexible array dynamics formulations are adapted from the flexible appendage equations developed by Likins [4. The array and r "42. The principal coordinate frames and cosine identities utilized are as follows: s Lo X where iS the vector T CM G IX IXA 2 In the (ECI) coordinates. Rigid Body System Geometry Equations 13 were formulated Into a set of matrix equations for the facilitation of digital computations. direction reference frame.4 ' . torques (control torques Included) and all nonlinear terms. r. Xs Is the vector in space {X [ 5 J coordinates. The right hand side of the equations represent the applied forces. ~ t. an inertial reference frame control torque exerted by the control moment gyros (or reaction jets) Inittatisalion . This requires the updating of specific direction cosine matrices during the simulation. d2 F. The force on the ith mass Is given In Equation b. LPVZVL t>V. Vol and ol0represent the rigid 1ody translational and rotational accelerations respectively and LAI the rigid array rotational accelerations relatiA to the space station coordinates for the two unconstrained axes of rotation. W13 are the rotational rates about the ith coordinate frame axis. 8.777"N d t implies differentiation w.I'V~t4I  V . The system force and space station moment equations were formulated with respect to the space station reference frame and the array body equations were written In th array 142 F The terms used tare defined in Figure 10. o CNIG/teaction Jets o Appendage ynamics o Rigid Bodies Dynamics Blk 4 i Inisgration Packsage Fig. V V. 7. Controlling torque profiles are computed in the simulation at designated time increments by the appropriate control equations and are used to force the above matrix equations. w12. Simulation Flow Chart Space sttion  0S SeCO Space staion CG l. T dit Routine Ei Rupdated The direction cosine matrices are calculated in the simulation in terms of Euler angles and are periodically by the following equation: Miajor Cycle Furtiona o Appendage Equation Update o space Station Guidance o Solar Array Guidance IC i '113 2 12 i it 0mi Intermediate Step r 3 Vwherec Dynamica Equationa 411. = in1(1 +1  ON Fig. t.
..4 . X 1 matrices and M is the 7. I LA 1 . cor sion o T i t falp pnrice for urml. ri 1 2 N N] U2 U 1 3 2 1 2 2U 2 U1 N are solved sequentially employing a finite difference method. 10. Without this assumption the simulation would be required to he performed in discrete coupled coordinates with resulting manipulations of large order matrices. arc N1 Note that a modal damping term (2ecr) has been arbitrarily inserted in the classic manner of structural analysis. I'lexillet Arrav + 7 Pl. aippervIlsge t ior ileflhtlion of iast porticle Fig. A change to the Integration procedure in the simulation is presently being 1 143 Rloth the rigid body andi elastic equations 4.. the rigid body inertia tensor.A method of Reference 4. Rigid Body Matrix Equations hlef. Flexible Array Geometry Substitution of the appropriate direction cosine matrices and consilderailon of the appropriate properties resulting from elastic deformation gives the following: [MNl*q +[K~q where q " [G~q [11)4 + 1. Iling' It""% where Il.    M is the mass matrix [A] [ 5] [.~l If. [A 3 ] [ 7 It 11 [I. are small and have a negligible effect upon the resulting transformation procedure. The transformed equation then becomes 4] *A 21 Z522 Fig.) Do(3 K is the symmetric stiffness matrix L is the matrix of rigid body inertial loads (Reference 4) i 1 i 2 6 [ 1 i0 ] ri 14 ] ! (I 2 0o3 F2 B and G ace force coefficient matrices which are dependent upon rotation rates (Reference 4) The order of the above equations of motions are 3N where N is the number of discrete masses comprising the dynamical system. frequencies. 9. for a chosen nunlber of modes.hn Ir of IT t i The assumption has been made in going from Equation 6 to Equation 7 that the motion dependent . center of gravity and appendage attachment locations In the space station coordinate frame and modal properties of flexible appendages. The latter includes deflection coefficients. (6) The left hand sides of Equations 4 and 7 are constructed in the simulation from computed direction cosines. The discrete coordinate equations can be transformed to normal coordinates by orthogonal transformations prodticed by most automated matrix methods of structural dynamics analysis. h ass 1'oin I mat particle of flexlii asv. damping coefficients and masses. The system of equations in orthogonal coordinates cfhn then be truncated on the basis of some chosen i 1 A4 1 1 L1A 4A2 1 F4 F4 criteria or engineering experience. It Is of convenience to transform Equation 6 Into orthogonal coordinates representing cantilever modes of vibration. I 4r e'ctot i.r pirn. This procedure permits orders of magnitude reduction in the number of equations describing array flexibility. number of cantilevered modes utilized..] [A.3] .rmed  matrices which are functions of rigid body rotation rates. Following the  (4) . AqI.
An analysis was initially made for system elgenvalues and elgenvectors of the twocantilever arrangement in order that comparisons could be made with corresponding theoretical freefree modal properties. At that point. 11. control system torques and external forces. a basic assumption of rigid space station dynamics Is made in the mathematical dev'elopment. One solution was obtained by the method provided in 5 and the other by an independently derived method using a variable order Adams integrator. Figure 11 shows a sketch of the cantilever geometry and associated coordinates. These comparisons were necessary to determine if cantilever beam modes could he used to accurately represent freefree beam modes. As preylously mentioned.made and incorporates the simultaneous solution of the equations. Each functional subprogram Lyddane's orbit generator. two other solutions for the response of a freefree beam were obtained using numerical integration methods. The rigid body equations are in tirn forced by the application of the flexible body interaction loads. attached at their points of constraint. CMG space station control. OCS linear and nonlinear solar array drives was 144 A _ Z F) T(t) . In addition. The simulation computes interaction loads from the following definitions. It was necessary to approximate the freefree beam configuration in the simulation by two cantilever beams. the complete program was checked for continuity and a problem of known solution was then executed to verify the structural dynamics methodology contained in the simulation. Symmetric About x. F J A = 1. Cantilever Beam Simulation of a FreeFree Beam represented by cantileverfree beams. in progress for modifying the present simulation to include spqce station flexibility in terms of freefree modes of vibration. and Cj is the appropriate direction cosine matrix (internally coniputed). Work I. Excitation of Equation 7 is accomplished by the internal loading acting on each of the discrete masses resulting from the translational ani. SIMULATION VEJITICATION Verification of the simulation program was accomplished in two distinct parts. m 71 O [ " Transient force Transient moment T A r in defined by F F INTj TINT where Fw = II J H + C aReference exercised independently and the results correlated with known data.ao Fig. F A Aj T +J (CjJ JrC + j TA is the total force exerted by the th INTflexible array/hinged body combination on the space station. ' FIf . The "space station" mass has been set to zero and the "solar arrays" 0. The problem a freefree beam with zero damping subjected to a concentrated force at midspan was selected because it was considered to be a good test for solution convergence. X*(t)+ N i i (2 (i NA A O . These solutions provided a basis for verification of the simulations. T It are the hinge forces and moments *J Jon the jTiI rigid hinged body. The simulation computes all transient variables at specified tine increments with the option of automatically plotting these variables with the Calcomp plotter.rotational accelerations of the rigid bodies. TT total moment exerted by the J Ttflexible array/hinged body combination on the space station. It also provided information concerning the accuracy of the analytical approach and programming techniques. Motion equations for the system of two connected cantilevers were derived to facilitate the frequency analysis: these equations are presented below. Closed form solutions for the modal response of freefree beams subjected to concentrated forces are provided by Leonard (8J. The above completes the outline description of the formulations and the computations used in the simulation.
rification with known solutions atbandfo the onequarter 8] shear e histories ifr h s cmae h foi ha 13 use Figure were obtained. I 1 = 12. The rate of convergence in this frequency comparison is demonstrated by the successive number of modes used. approxlmates the iTil freefree frequency The solutions of Equations 8 and 9. in Figure 12 (t) = generalized modal coordinate for the ITI1 cantilever mode = coupled frequency of O. o p2 (MA I FQ 1 2 (1cos Qt)(12. The frequencies obtained by an orthogonal coordinate transformation analysis of Equations 8 and 9. of the system. Therefore. the degree of and convergence in the modal domain can be seer..Ft)( 2) 2 where MIA 0 2#1 and MA V1 ( i I. F The comparlson. are given by the following: F2 X (t) " + A A ( A _ Q1 cos g t (10) ~amplitude Vcorrelation I which is based upon use of three cantilever modes. Excitation at the base of the cantilever beams b. + X (t) = 0 1 2 (9) X M(14) *.0 slugs. where ft I A N5. ~(t)in the fTtt Modal cantilever mode at midspan of the simulated freefree beam 2F40 Fl t ) L 12 0 0 hEQ Ft fl2Fi (1coss t1) (16) F0 X(t) hEQ I applied step load at midspan = coordinate for rigid body motion = generalized mass of the ITil cantilever mode o = iT1 cantilever mode natural frequency (uncoupled) It should be noted that antisymmetric modes of the freefree beam were not excited due to the positioning of the disturbance force at midspan. the primary tijectlive of the slmulation was to obtain Interactio. for the cantilever beams. The evaluation of the numerical coefficients of Equations 816 was based upon a 25 point discretized mass representation of both the cantilever and freefree beamt. locatlotm beam 13. are given in Table 1.span (1) compires the shear hior "tainel Arom the of one cantilever mode It'he simulation with the shear calculated using the n. 566 rad/sec. Modal data for both the freefree and cantilever uniform beams were obtained from stzndArd reference tables (9] in order to compare the cantilever and freefree beam formulations. Fo = 0. loads between the space station and solar array fo desln evaluation. N it) F F Fi 2 o (15) EQF = Maass of the cantilevered beam L shear force coef. These frequencies. show that the cantilever beam (AMEQ VIF I 2 l ) formulation can be used to accurately descrbr freefree beam modal properties If a sufficient numiber of cantilever modes are usedl."dal acceleration method and one freefree mode. using only the fundamental cantilever mode. the corresponding analytical solutions for the perturbed freefree beanm were . Numerical data assumed for this comparison were: MA = 5. for purposes of :'. 2. along with corresponding freefree beam frequencies have been normalized with respect to the frequency of the fundamental freefree beam mode. The subscript "F" refers to freefree in these equations..bratior. 5 lb. 145 To A ) Similarly. Similarly. EQ 1 21 ca) I Although motion histories of the flexible appendages were of Interest.MEQ 1 1(1) () + M EQ (a 2 derived and are shown below.3. simulation. a zero nass space station posed no problem for ti.
401 5 Mode Cantilever Rigid en Body 1. em+ 1.548 13.812 Uncoupled Canti11~nm oe 0. 33r 3. Com~parison of Mode Shapes for a1 FrceFree Uniform Bean) and a 3 Cantilever Mode Approximation .t abyrnrI ______.000 5.958 11. 1..03 Us. 4 0 3.584 . Time for n Unit Step Force~ Applied at riMISpart.632 3. the' ditferevcet In loadt front the two tnds were not ovignttica.. shut C Figt.404 13.V 13Mise313t 1.6onquarter !>a Ptat .861 3 ModeCantilever Body Rigid._ _1_If__1 .31 loadr. 0003 5.__.6 . .965 41'. .410 25. the dert uilitv of the modal accelerr Pion metho'I ndcntly of the oyrautatlon.0000 5.ingMcdal Dipac Note Obse CalculaW. FilM. Com parison of Cantilever & FreeFree eam Shear tit 114 Sr~aai vs.6ree "s 4 3.652 36. With respect to computing total (quasidtcady plus Pointe out dynimir! ioads.q. 367 24.. The calculation of force hiatcrv co:reppondiig to the Modal diaplaecmrent Method was . 11 1.%placencnt methiods o)f computing totol .'p21. 0341 at (i fl I.0114 21.0. Symmet ric NorFe 1 2 3 4 5 FreeBeam 1. at The 146 .0000 5. 411 12. . Figure 13 ase contrastr thr.*. 0006 5.420 4 Mode Cant[lever Ba Body Rigid 1. 0 Timel. n I Jreifree . . Calcilated Coupled JM1ede Frequerncy Ratios 2 'Mode Cantilever Ba Body Rigid 1. 13.0101 .T'ABLE 1 F'requency Comparlisan of (Uniform Beam) Cantllor~r + Rigid Body Mode Representation of a FreeFree Uniform Beam Frequency Ratios: f Reference Feeqiency Ratios I Mode Cantilover Body Rigid em+ 1.814 3.344 24. 12. Bimpiinghoff (1io. 'L4o Sm. modal srcratIon Anx] mods~i di.010 1 7*t  13. Sm. mzari'?Indtl Fig.0 ZdSymomerc Mod.A.
. the maximum shear force given by the hitorles is in excellent agreement. 12 . The comparisons given by Figure 13 show an .1 excellent agreement between the simulation results and those obtained by an independent method utilizing the varlable order Adams numerical Integration technique. l __ 0 6f/o. . The above presented comparisons between cantilever an freefree beam response resuits demonstrated that the structural dynamics methodology contained in the simulation was sufficient for the accurate evaluation of Interaction loads.. I.1. each of the 40 .06 0 .3 . V (IA) Yl (0 ' ~ ~ ~ ~ . 22. '14.ode.3 . approximatltug the given number of freefree modes by the same number of cantilever modes. Uniform Bean Comparisons of Shear 1 icw00 90 Maximum her at Span x.2. Also. are attributable to the degree of convergence in l08 . good agreement exists Fig. teg firr ite Carfiler .. Variable Order Adams Numericul Integration Reults VaiI the Frat fo mntc Fr rn ree.2 .. I.' .. The low frequency waveform and magnitude in both histories is seen to be in good agreement. Modal truncation was not considered in this method. 14. I ) Fig. The small differences in the between the histories are attributable high frequency to the degree of convergence In approximating the given number of freefree modes by the same number of cantilever modes. the beam.e 16 . Again. Nlde Mro listories @ 1/4 Span for a Unit Step Force Applied at MidSpan Ilekrence 8 Figure 9 80 Sim latlon C70 ant lr Aroximaion Figure 15 presents a comparison of simulation reflults using five cantilever modes with results obtained using tho Nastran "Direct Transient Itesponse Method" [5].tlvr wed" S .. 15.10 . 1 (I/4. 2 I. .r. 1 .. . .lase . Scontent modes. the differences Indicated .08 ... Comparison of Modal Shear Force Participation for a Uniform Beam Subjected to a Unit Slop Load Appli.d at MidSpan between the frequency content of the two shear histories. ..o .4 r .r .0 . In general. I t IlM 11/4) 9(1) .ith  .6 . ts.. 14 . . Uniform Beam Comparisons of Shear istory @1/4 Span for a Unit Step Force Applied at MidSpan A modebymode comparison of the maxima shear force components derived from the simulation and those given in Reference 8 is presented In in Figure 16..l Tie . Time.11 *Fig.os.. .6e fIt. for use i this method. 16. 1 147 i' • 4 . 18 20 .o 0.Z V 41/4. . I ~ Figure 14 presents a comparison of simucantilever modes with independently calculated results using freefree ~~lation results using five 7 ? .uise frst rl.which Is the method used in the simulation. :". The freefree beam representation was discretized Into 40 masse._. . 4 " simlaltion Ietutt.. t)M (i/4) 9 (1S... 30 20 o 0 (ilHfdDlmly) l discr2te masses was allowed two Inertial degrees of freedom co. Ro.c. Beem Modes NI]2 h.. 8 .'responding to planar bending of 2 3 I F 4 5 symmetric hlode Number. .
The solar array is a large flexible structure composed of membrane strips stretched between the Inner and outer structural support members. Selected frequencies are listed in Table 2 in terms of inplane and outofplane modes.iX I 2. The solar cells and associated interconnects are cemented to the membrane substrate and generate the power required for space station operations. 7. The matrix method of structural analysis considered in Reference 12 was utilized. The stiffness of the array membrane was considered as a function of applied tension loading.the moment at the array attachment point due to a base rotational acceleration. Figure 191) shows the Initial array orientation errors with respect to the sun vector as a function of time. load participation . some preliminary results have been obtained. The array was modeled using a discrete mass technique with 200 mass points. Figure 19c shows the time variations of modal acceleration for two array modes .s x 1o sut. Therefore. I.4. 0i is the modal deflection coefficient and Rn is the modal mass.an outofplane mode forced primarilly by OCS torque and an inplane bending mode forced primarily by space station acceleration. This array was designed Iy the Lockheed Company under Contract NAS911039 and is reported In Reference 11. It Is seen from Figure 19a that I. seais Outter Strxtursl SRtdst) Support . This analysis will form the basis of a generalized dynamic design criterion for solar array structures.TYPICAL SIMULATION RESULTS An extensive analysis of the parameters that influence the structural dynamics of solar arrays is planned. The space station was perturbed by a force history the docking of an orbiting module. Space station control torques were not initiated during this simulation since the angular deadband contained within the reaction jet control system . Ithough this analysis has not been completed. • s. each mass having 148 the magnitude of feedback force from the flexible arrays upon the space station is small because of the nonoscillatory nature of the acceleration history. is given in Figure 19a. " 5. 17.all within the frequency bandwidth of the space station attitude control system. The participation is equal to (Z mi 0i) 2 /Rn Zmi where mI is the discrete mass. Fig. space Station Inrtias 1.890 Ib. The docking load profile this from and resulting roli axis acceleration. The solar array/space station conceot being evaluated is shown In Figure 17.090 Tb. 7%. The results of the analysis also Indicate that modes with a significant t .: UK_Fj. 5%of the total mass. The selections of modes to be used In the simulation was made on the basis of those contribiting a large percentage of load participation.a 2 Figure 19 is a graphical series of results simulation. The space station is a rather stiff structure (modeled as a rigid body) which contains 96. The disturbance force is directed along the X axis of the space station. 2! 3 degrees of freedom.4x 10 s lugPt. Load participation for symmetric modes is evaluated by calculating the shear at the array attachment point due to a base translational acceleration.  Array WetSs Space StationWeight 710. Space Station/Solar Array Concevt for Dynamic Interactions Study (Rollup Flexible Array) A structural analysis was performed on these solar arrays prior to simulation to obtain required modal data. A typical mode shape shown in Figure 18 for an outofplane antisymmetric mode. A detailed description of the structural modeling and resultant modal data is reported In Reference 3. This participation function is equal to (Zmo i ri)2 AZn mr12 where r[ is the distance from the vane axis of the array to the mass point.rt. Load participation for antisymmetric modes is evaluated by calculating . 'y 7 M ram @triph Inner Structural srepresenting 1104" s Selected results from one simulation are presented and correspond to the configuration of Figure 17 with small Initial attitude errors. coupling of the attitude control system with array modes can be expected. and the central boom was modeled as a beam column. An extendible boom in the center of each vane applies a tension load to the membrane strips..000 (Flexible Body)  PT control. The load participation factor for this mode was 67. The continuous OCS drive system was used In this simulation for array Solar Arrays Fully Deployed Sollup '4(10.
0557 Ha Particiption .5 6 31 57 62 InPlane Frequency .0.203 .0734 .340 Percent Participation 67.7 5.67.7 Symmetric Mode 2 Frequency (Hiz) 2. Timne (Second@) Simulation Rlesults for an Externally Applied Docking Force X2 Freq~vocy .287 Percent Priiton(Hiz) 50.0 Mode 1 2 7 8 Antisymmetric Frequency (Hlz) .12 Percent Participation 55.318 .291 1.4 7.71 4.0557 . A complete set of (data.1 6.7% oo X3 Fig. OutofPlane Antisymnietric Motion was not exceedled. and Interaction moment sbout the space station V axis.0 1.0 3.4 12.7 2500 2000 L 1 1h<' 500 1000Foc Applied Docking Force and Space Station Acceleration istories (Holt1Axls Direction) Force 1000 I~~  Accelerationx  500 0 0. This Is caused by a relatively high value (0.5 .26 Percent Participation 57. These limited results which arc presented show the type of design data that can he obltained from this dligital siniulation of the integrated dynamic analyis.3 7. 18.TABLE 2 Frequencies and Modal Participation Factors of Selected Roilup Array Modes Out of Plane _______ Symmetric Anti symmetric Frequency (Hz) 6 36 56 .94 2. 05) of the Inpuit modall damping parameter.5 2. 19 (a). corzrespondling to variations 149 .5 3. Figure l~ci presents the timc variation of interaction force along the space statim X axis. liiitia! high frequency transient loadhi are seen to decay rapidly.162 .5 4 6 0 10 "NC IFig.
in initial parameters and basic structural data. . 19. (b) Modal Coordifte Acceleration Histories CONCLUSIONS ) le Mio c e An Integrated dynamic analysis method has been developed and implemented in a digital computer program for eimulating the structural (Acceleration x 10) 0 lot Antlaymetric InPlan dynamic interactions between a space station and Bending Mode S2 controllable solar arrays. Simulation verification studies demonstrated that the analytical formulation and the modal synthesization technique employed provide an accurate method for evaluation of dynamic Interactions. 19. 3L 0 2 4 6 a 1O Time (Secors) Fig. In addition. the verification studies showed the programming of the almulation to be correct. (d) Time (Seconds) Fig. (c) 150 . interaction Force ad Moment Histories will provide the basis for ihe derivition of a otructural design critetion. 05  02 4 6 10 10 Fig. 19.1 Solar Array Anglar Error Hitorlos 200 Seasoal Adjust Axis .05 asOrbit Adjust Axle~ 0  .
" J. General Electric Company. 9. "Preliminary Synthesis and Simulation of the Selected CMG Attitude Control System. This means that even 0. California. Januay 15. Technical Report R104. Young and R. R. Wolf Research and Development Corporation. "Dynamics and Control of Flexible Space Vehicles" Jet Propulsion Laboratory Technical Report 321329. 10. Mains (Washington University): You said at one step that you were using a 600 degreeoffreedom system for analysis.. Astronautics and Aeronautics. NASA Technical Report R21. Is that correct? Mr. "Structural Interaction Simulation System". Germantown. John Wiley and Sons. "Tables of Characteristic Functions Representing Normal Modes ofVibrition of A Beam. Weinberger: In this particular case the docking force was applied at the aft end of the space station along the axis of the space station. Mr. Riverdale. But the docking force was applied along the axis of the space station. Ashley. Principles of Aeroelasticity. 6471.. There were some outofplane motions. Zudans (Franklin Institute): When you indicated on one of these diagrams the docking force introduced into the system. 1970. Los Angeles. Bisplinghoff and H. Inc.REFERENCES i. R. Washington. R. Maryland. Lockheed Missiles & Space Company. Bouvier. These motions were due primarily to a slight misalignment of the solar arrays that were active with a linear control system. W. Revision 1. 5. Jr." Astronomical Journal. 1970.1 of a degree attitude error would cause some motion of the solar arrays. NASA. Weinberger: That is correct. D. There was no eccentric force or loadIng. The Study of Dynamic Interactions of Solar Arrays with Space Stations and Development of Array Structural Requirements" Fairchild Industries Report 8581R1. Weinberger: Yes we did. "Attitude Control of Nonrigid Spacecraft. December 1970. Maryland. Fairchild Industries. C. Leonard.. pp. 5 Pg. New York. 8. 2. Likins. "Interim Report. of a solution of that size for this kind of problem? 151 . 1 July 19)49. Office of Technology Utilization." General Electric'Report EL506D. how was it introduced relative to the masscenter of the system? You seem to be getting the moments. May 1971/Vol. 8. 5 March 1970. P. 1962. Mains: Did you use a direct integration technique to get those response curves that you showed? Mr. W. Revision 1. DISCUSSION Mr. % 6. Mr.". K. 12. W. "On Solutions for the Transient Response of Beams". Likins and H. Section 11. 7. Volume 68. February 1971. L. Mechanics Research Inc. P. Felwar. 3. Document. but there were no indications on the slide as to how this was done. "The NASTIRAN Theoretical Manual". Lyddane. the reliability. "Stardyne User's Manual". Mr. and hence some torques Into the system. 344350. "Evaluation of Space Station Solar Array Technology and Recommended Advanced Development Program. P. October 1963. Mains: How do you have any handle on the meaning. "Small Eccentricities or Inclinations in the Brouwer Theory of the Artificial Satellite. p. 11. NASA SP221. Binghamton. 3." University of Texas Publication No. First Topical Report LMSCA981486. 1959. January 1971. No. D. 4. 9 No. If. Mr. 555. 4913.
other than for structural and vibration analysis. so the 12 elastic modes were used for the solar arrays plus the 6 rigid body degrees of freedom for the space station. very reliable.ucture that it could not be handled as a linear one. 152 . We did It to get the frequencies and the mode shapes and the generalized masses. Mr. Mr. Zudans: I would like to comment more on these questions. We have been able to modularize this program in such a way that it uses much less digital computer time in core than the NASTRAN program would use. in the NASTRAN program and users manual. Mains: You might be surprised if you would check the orthogonality of the vectors sometime. We felt it was more efficient in developing our own program. This is one of the areas in which we have made certain assumptions in linearizing. Clevensen: Why could not you have used NASTRAN exclusively and saved considerable work? Mr. Mains How do you have any handle on the reliability of an elgenvalue solution of that size? Mr. This was a Lockheed array geometry that we were studing. We did not have a 600 degrees of freedom model for direct intogration. Weinberger: We are examining this problem in connection with the tension loads that are transnritted through the boom and so on.000 and there is a perfcct orthogonality. However. Mr. I assume that you might be able to code subroutines which represent the orientation control system and the rigid body mechanics. Wr¢nberger: I think the reliability of structural models of that size is fairly well documented. because 600 degrees of freedom dynamically today Is nothing. Weinberger: That's right. Weinberger: Yes. From the structural analysis model. In this case we used 12 modes. Mr. Weinberger: The 600 degrees of freedom simply referred to the structural analysis model that was developed. the generalized mass and modal coordinates were used for the modes that were selected. Mr. The validity of these assumptions has not been established at this point. for example. Weinberger: Of course one of the problems that we looked at was the coupling of the control system. XrMjps Did you then do an elgenvalue solution on the 600 degrees of freedom? Mr. How did you account for obvious nonlinear atti'. Mr. I am not familiar with the NASTRAN program with regard to the demap instructions and the auxiliary useage of NASTRAN. The NASTRAN program and many other programs use the invrse iteration routine with spectral shapes and it is. This is the only way that we can have any confid ence in the linearization of the array. We get z correlation in that respect. you used NASTRAN for verifying some of your results? Mr. rather than to resort to the NASTRAN program. One thing that comes to mind immediately is the size of NASTRAN. From what we have been able to ascertain. we have chosen the model In such a way that the results agree fairly well with the test data. very. I wanted to ask a question. Clevensen (Langley liesearch Center): If I understood you correctly. You can handle 3.udes in your solar array? It is such a flexible st. although we do have some preliminary test data from Lockheed on the array of this type.DISCUSSION Mr.
that the material of the column 153 Pit) P 0 + P 1 (t) come(t) I u(x. Fig.. Of these few solids. M. (2) overall structure cost. Our discussion thus far has implied. 1. If an initially straight column. simulation and means of minimizing the effects of mechanical vibration on technical hardware by means of a theoretical investigation of the effects of hysteretic material properties on the response of parametrically excited systems. many are used in structural members. T. There are. however. however. it is desired to have machine or structural members perform multiple functions. Column response is detemined forthe stationary case withadditional assumptions on material properties and stress distributions. (3)overall structure weight. The response which occurs under these conditions is called parametric resonance. EvanIwanowski.PARAMETRICALLY EXCITED COLUMN WITH HYSTERETIC MATERIAL PROPERTIES . That is. P1/2(PEPo)) space. Is the first region characterized by v = 2o. It is well known that few solids actually obey Hooke's law very closely. For increasing values of P1 (t)the points of instability mentioned above cover instability regions.Column Configuration obeys Hooke's law. in (v/24. Mozer IBM Corporation East Fishkill. New York and R. For example it may be desirable to have a column perform both as a column and as an electrical or thermal conductor. Consequently this paper is concerned with the nature of vibratory motion of the column related to this region. is subjected to a periodic axial load. and thus the most important region. and if the maximum amplitude of the load is less than that of the static buckling (Euler) load. P(t) = Po + PI(t) cose(t). This phenomenon of parametric resonance may be expressed more generally as v(t)/2l = 1/. In this case the selection of the material will be based on some compromise between its load carrying capabilities and its electrical or . there exist continuous ranges of the values of v(t) for which the column will be unstable.3. PE. INTRODUCTION This paper is related to the description. forces and moments for aparametrically excited column. The largest of these instability regions. Expressions are developed for stress distributions.2n. certain relationships between the disturbing frequency 6(t) = v(t) and the natural frequency n of transverse vibrations of the column for which the straight column becomes dynamically unstable and lateral vibrations occur.n where m = 1. For a sufficiently small value of Pl(t) one such relationship between b(t) and n is 0 . Professor Syracuse University Syracuse. then the column experiences only longitudinal vibrations. In many instances.2. among other things.t) x u Fig. The principal reasons for their use are: (1) they display nearly Hookean behavior to a relatively high stress level. New York An investigation is performed to determine the effects of hysteretic material l.. 1 . properties onthe dynamic response of parametrically excited systems.
This paper deals with determination of regions of stability and instability as well as the lateral amplitude response of the axially excited column whose material properties are of the pointed hysteretic loop variety. V. (an exhaustive discussion is given in Ref. but these forces do not arise from the material properties of the column itself. The number and complexity of the material properties. Stevens and R.1). The Davxdenkov model is used to represent this material property.thermal conductivity. BASIC RELATIONSHIPS Davidenkov (5) developed relations to represent the behavior of metals in the form Snalready (1. K. Later K. It is important to note here that in this analysis energy dissipation occurs only due to the bending of the column and not due to its axial compression. and the deflection u is regarded as small.2) where M is the moment at any cross section and M = fA a y d A. In some instances the expression for the moment on any column 154 co co c Fig. 2  DaVidenkov's Model cross section at any time will be developed from the stress distributions. Detailed discussion is also given by Mozer (4]. The Davidenkov expressions have been used extensively by many authors. It is the objective of this section to develop stress or strain distributions for the parametrically excited column. In addition these two stresses will in general be out of phase. Several authors have considered material properties other than the simple Hookean case in the analysis of the arametrically excited column. The period of the total loop is the same as the period of lateral motion of the column. 1). Once we leave the domain of the Hookean solid elastic column. In Fig. for example. 1. If the bending is reduced the picture is altered to look somewhat like that shown in Fig.. Stevens [1). solves for the cases of the Maxwell Element and the ThreeParameter Model as material properties for the stationary case. . We have made the usual BernoulliEuler assumptions. [6] to solve several classes of problems in which the raterial is assumed to have hysteresis loops of the form represented by (l. Stress Distributions in Parametrically Excited Column. The strain dt any point is 22 U C = C + Y2 c + y K = E + y 2 (1. K. It is graphically represented in Fig. Evanlwanows'd [2] introduced the cornplex modulus material property representation to the stationary response of the parametrically excited column. Thus the period of the bending stresses will be twice the period of che periodic portion of the axial stresses. The case of a simple viscous damper in parallelwith a spring (Voigt Element) is shown in references treated for stationary and nonstationary modes. 3. If the bending is removed completely. 3 the small loop is due to the periodic axial load and the larger loop is due to the slower periodic bending. Such situations may present the necessity to use materials whose mechanical properties are nonHookean overall but the smallest stress levels.ZO a tx a xa (1. K. the two loops converge into a single two branch loop.3) X a So far the treatment is for any material property. V. 2. On other instances expressions for the temporal part of the motion of the column are derived.. 4. e. e know that in the principal region of instability the frequency of lateral response of the ishalf that of the excitation frequency. along with their relatively crude rheological models becomes overwhelming if one attempts to apply them with rigor to the problem of the parametrically excited column.. Bolotin (3] considers amplitude dependent damping forces due to viscous and dry friction at an end support of the column. a large field opens up even to the most casual observer. M. Analysis of the system configuration yields the partial differential equation a2 t 2 t a2 U x JR) + Pit)2u+m. A typical stress strain diagram for a fiber at some distance from the neutral axis is presented in Fig. The loop contains four distinct branches.g.1) cnol o a E(c t n (o + C)n Ccolumn These relations depend only on the amplitude of the strain. Consider now the stresses in the column whose material property is such that it is characterized by a pointed hysteresis loop. the period of which is j .
Each equation contains two or three terms corresponding to the different branches of the hysteresis loop present in a cross section. 5.Stress strain in a fiber of a column with hysteretic material. We make the additional assumption that the axial stresses are small and will not contribute appreciably to the column response. 4 . The stress strain distribution at a cross section of the column will be somewhat like that shown in Fig. Thus for a material having a pointed hysteresis loop. This assumption is a fairly good one for the slender colum. The use of the Davidenkov relations in the case of the parametrically excited column where axial stresses are taken into account is in general not possible since the constants in the relations will not always allow the loop to close after a complete cycle of the column.1) () M E I32u+ 2 3 X 155 . 3 . 0 AlI Fig. The stresses on a cross section of the column are such that at certain times three branches of the four branches of the loop are present.Stress strain in a fiber of a column with hysteretic material.Stress strain in a cross section of a column with hysteretic mterial. RESPONSE OF COLUMN WITH POINTED LOOP MATERIAL HYSTERESIS 4 Fig. we require several equations to represent the stress across the column at different times. However. Large moment. if one assumes that the strains due to bending and the strains due to the axial load are in phase. Fig. 2.the same as the period of P(t). Under the assumptions made. The evaluation of the constants in the resultant expressions is difficult since they depend on the distance from the neutral axis. then it is possible to write equations for a four branch loop based on the Davidenkov relations. Small moment. moment acting on a crosssection is a2u (2. y. 5 .
_ n 2 ax A L . This removes the necessity to keep track of the signs of y and c.where E iswhere a small parameter such that E > 0.3) E L~~~~~a 2 = f.6) f(t)nmn] ax =E I 2 We consider the special case where n u 2 2 :EX [(.10) + 3 u)n yn+l d A (2. we and get differentiating twice with respect We seek a solution to equation (2.t) .2) we get E 124Sn ZE(t) + Z 22 a x2 (3 2 3] "  [(f(t)m 4 f(t))n  .9) is always odd about y = 0 regardless of the value of n. Substituting (2.4) Inorder to make any headway toward solving (2. W" n' 2 L2 n [(n1) Sinn ax2 . 2 ) + ax 0 " 2nl ( ) (12u n (2. The arrows above t represent the branches corresponding to increasing and decreasing curvature. Thus the total integrand of (2. 2nl (.2). 2 EYi a X2 .9) is always even.nSin n ix L (I)n+2 [(f(t)m f(t)n and also noting that a2 u Y y . yn+l d dy 1 )n f(t)nm)) y ax in the case where the axial strains are neglected. X) aX is the dissipation functional depending only on the curvature and material constants.5) n4 where fm(t) is the maximum value of f(t).n Sin n nl in2I Substituting these values into (2. ydA 0 = .1) into (1.8) ax max a x2 t2 Substituting the value for u from (2. we obtain 2 a C 322 In ax 2 E W n+4 4 ( [(n1) Sinn IT Wn 2 2 n  1. Thus h/2 2 __2____ and the definition for the moment ".f)) (h)n n n  2 (2. we get 4.t 2nl(a 2 u)n yn+l d A (2.l)n f(t)n ]} yn .f(t) Sin (2.5) and substitute the resulting expression for the stress into M we get where W indicates the width of the column.1) we deduce that E f [( n _U a2 u n  A [( x max )y n+l d A ax u) .4) we need to know more about the functional . y d m( A (2.9) P(t)f(t)!gin x+ m i(t) Sin tx.2) of the form u(x. Due to this property we may integrate the above expression from zero to h/2 and multiply by two instead of integrating from h/2 to +h/2. We may now substitute the above value for c from (2.8) to X. d A + 2 a X max 2 max x +  n Sinn x [(f(t)m . We rewrite the equation for Davidenkov's model [(co . E<< 1 and (lnd her ) (a.  2 u n a x n a x max 2 ax =2 156 . E I " u + Z I" ax ax + P(t) + (.2 2U )M +.2U a x. 2nI (.0 (2.)n 2nI  on)) (2.7) i x2 max Comparing (2.. It can be shown that the expression + Ey/n{ )yn( in (2.3) into (2.7) with (2. 1? E 2 2y ~ U[. tegrating the above.
m+ (2.4) and apply Galerkin's method by multiplying the resultant by sin vx/L and integrating.constant which implies that a = 0 0.13) in the We now proceed to determine the stationary response of the column by investigating (2. that this term is actually a variable coefficient of f due to the alternating signs and we are thus justified in leaving it on the right hand side of the equation.14).12) and rearranging. y W h aCos2 . we get .19) and (2.20) become 157 . Stationary response is defined as the case where no system parameter changes with time.2 L 2El 24L f(t) .lar displacement term o/2 in (2.r) (2. We now substitute (2..18) (E . however.2 (2.F (f. The terms a(t) and t(t) are to be determined from the usual relations. We represent (2.11) represents the distributed lateral load due to dissipative material properties. u(t) .13) represents the temporal equation of the motion of the column.13) Equation (2.2 P(t) f(t) Z. ) After some calculation we find: E yW h 2) 6 (4a 2 _a 1E We also denote the loading parameter v as P1 (t) = 2 (2.+ L 1 + 2m)it! TEy Wh4 r5 48 Ls i2f(t)2 f(t) + f(t) Remembering that P(t) E ] + .17) 1 (r.15) 0 P. Itwould appear at first glance thit the coefficient + Z02fm could be combined with n2 on the left hand side of (2.14) 2 E(f(t)  fit) + f(t)m (2. a = EA (2.14) were expanded we would get the term + c B 2fmf. Note that if (2.12) and where f F + 2f12 lpfCos 0 m P0 L 2We seek a solution to (2. and is the subject of the following asymptotic solution.E + 2(12 [f 2 !2 ff Cos O) f 2 6 h4S 96W Ey L m . we get.15) in the form fit) =_ f = aft) Cos (0 + (t)) (2.19) and (2.1l) + Z2n 2 u Cos o f 96 L6 m 96 n 2 f(t)m]} where Es Equation (2.14) in the form f + n2 f = .16) is Eulers' buckling load and w2 El L'2V( is the transverse natural frequency of the column without axial load and the transverse natural frequency of the column loaded by Po is We have assumed analysis of the first instability region only by using the ang.20) in a special case.16). We rearrange (2. Explicitly we write v(t) = constant. cos . It must be noted. Under these conditions (2.P) E" 9 L6 m 2n a Substituting these relationships in (2.o. L .4form 3. ) 41v(t) + cBl(Tpa..P 96 L6 ma Stationary Response (2. Asymptotic Solution Only the first approximation will be sought in this analysis. [ f(t)2 2 and the term 2n2p has been replaced by i2n p.11) into (2.20) 0 (2.2 C 2 Ey W W6 h4' 6 {[l + 2 f(t) 2 1rX x 2 Sin + n2 f t 2) [f2 t 2ff .a.
saly sable stable or unstable branches.21) (2. 9 shows the change of the first instability region with amplitude of response. RESULTS AND CONCLUSIONS Discussion of Material Properties and Stress I Eliminating € from these equations while noting that am a a constant for the stationary case.frequencies. n B~nam am a ) V n. 10 shows the instability region in three dimensions.21) a using the value of a from (2. well Applying n oteratn heof. (2. characterizing soft systems. 8 that the point defined by da/dv = 0 separates the region for which nontrivial stable solution can exist to the right and the region for which no nontrivial stable solutions can exist to the left. for the parameStationary response curv trically excited column whose mwiterial may be represented by the Davidenkov Model with the special assumption that the axial stresses are negligible are plotted in Figs. Fig.Cos 2p. we obtain the following: For d ao/dv > 0 the solution is stable if 8. It is a.24). The stable and unstable branches of the phase angle ' are found to be (2. We may now determine the stationary phbse angle o from (2. 7 show the effect of varying the load parameter P on the column response. This means that no nontrivial solutions exist for y/2aless than that at which da/dv = 0.29) (nV)/d < 0 Hence the solution stable for v .8 shows the effect of varying the Davidenkov model parameter y.V)2   Pt] ] (2.22) region of stability is distinguished from the region of instability by the curve represented by (2. This result has not been frequently observed in the literature. 20 and d/dv > 0.26) tions which have also soft characteristics. It is further noted that as long as y is positive the backbone curve leans toward decreasing .24) 1 [ [2A .27) 158 . otherwise it is unstable. 6 and Fig. n. (Negative y implies that the material property is such that it generates energy over a cycle). Z 2___f?(2ov si(4n2+9)  = E wh 6 96L m Eywh4V 6 The boundaries of instability zone are: V [1 11 2p] determination of steps in we known bnownhes. Fig. .2 a. 6 and Fig. and the present work is concerned with systems near parametric resonance. ao 2 4 V3 <0 (2. 610.v) 2 2 + 9) 2) 04 U )p) 2 Distributions. P0 and y being constant. Fig.¢v a=n where 6 W EyWh4 L6 S 96 L m (2.v .v 2nP(4 f2 + = Sin9) 2 2 4 [(2 fV V 2+ 9 2 ) . comparison of results loses most of its meaning since Pisarenko analyzes systems near dynamic resonance. It is seen from Fig. and for d ao/dv < 0 the solution is stable if 2a .: T 2The 4 2 ITO( .0 2 S_0 Sin 2¢u.4)The significance of the harmonic axial load on column response is demonstrated.O (2 (2. although it can be shown that the response curve of a parametrically excited column having nonlinear damping of the form dff 2 where d is the nonlinear damping coefficient and structural nonlinear elasticity does close at the backbone curve in a similar manner to the present case. Stress distributions in the column are also illustrated for the pointed loop hysteretic material. while Fig. Pisarenko [6] arrives at dynamic response curves for various problems Using the Davidenkov rela 42 2 The t sign denotes the possibility of two solutions for a.).V ao  2 4  > 0 (2.28) . Swe get the following a 2r2 81(4 p {2 [(2n. u.25) The socalled "backbone curve" of the stationary response is that curve lying halfway between the two values of a obtained from (2.23) called "stability curve". but beyond this. 3.
D." J. 9 defining the region of instability shift to the right and become rounded with increasing amplitude. 7. see Mentel and Fu (8]. The Dynamic Stability of Elastic Systems. similar to the case where linear velocity dependent damping is included. due to the alternati:g sign. 1968.this model'the nonlinear elastic effects without difficulty. 6163. Damping of Materials and Members in Structural Mechanics. S. HoldenDay. Lo Ca Area of cross section of column Young's Modulus Depth of column in y direction Moment of inertia about Z axis Curvature I/ Length of column Mass per unit length Material constants Euler's Load Axial load Po + PI(t) Cos 0(t) Static compressive load Amplitude of dynamic load Width of column Axial coordinate end Distance from neutral axis. It may also be possible to incorporate into. Amterdam.. T. N. Maximum strain Normal strain due to Po 4 P 1 Cos 0 Small positive quantity This equation contains a nonlinear component proportional to co2 which is symmetric. San Francisco. V. 12. 9 that the curve defining the region of instability for zero amplitude of vibration is the same as that for the perfectly elastic case. If the addition of a 159 . G. 1969." M.13 (for the elastic case. B. 8. M. "Parametric Resonance of Viscoelastic Columns.S. Kiev. N. IzdVo Akad Nauk. From Fig. K. 1961.0013. "Energy Dissipation in Vibrations. Stevens. C. N. Inc. p. v/2n . 755765. 483. 6. Vol. R.It is seen from Fig.. J. No. Mentel. No. Under these conditions the system finds itself in a region of instability and a large amplitude results. (inRussian). Pergamon Press.I. Dissipation of Energy in Mechanical Vibrations." int. "Parametrically Excited Column with Dissipative Material Properties. USSR.odent damping found in most engineering materials and it results in a well defined mathematical problem at least for the single frequency case. for example. If one attempts to use a strictly linear mathematical relationship to represent a material whose characteristics are essentially nonlinear. 9 one can further note that in the present case regions of instability exist for large values of a which do not exist in the linear elastic damped case.. Dec. Journal. In the case consdered the energy dissipation per cycle is proportional to y a3 and so the curves in Fig. Thesis. Biot's linear hysteretic model is perhaps the best linear hysteretic model to extend for use with multifrequency excitations as encountered in the parametrically excited column or plate. 1962.1 and would have been rounded somewhat.A. in Fig. Mozer. our system is stable). Pisarenko. It possesses the characteristic of amplitude depe." A. REFERENCES 1. 5. One must approach the problem of selecting a mathematical model to represent the real material of the parametrically excited column (or in fact any vibratory system) with extreme caution. Vol. Fu. large errors may arise in the prediction of system vibratory response even though care is taken see as that dissipation per cycle is the to same the energy real material. Let the system now be perturbed such that ah/L > . p. K. Inc. Stevens. C. 8. T.9 and u = 0." ASD Technical Report. The coupling of the bending and the axial stress probably would have provided a more significant dissipation term for a 1 0 and thus the curve would have been shifted slightly to the right from u . K. 1964. 9 let our system be described by state v/2a . Lazan [7J indicates that the maximum point locus curve may be an appropriate one to use for the calculation of stored energy in material. In our case this "nonlinear elastic" term is probably the most appropriate one to use in explaining the nonlinear character of the column response. Lazan. 1966.Tech. 4.. For example. pp. 4. 1938. "On the Parametric Excitation of a Viscoelast'ic Column. Such nonlinear elasticity could be obtained from experimental data on the maximum point locus curve. EvanIwanowski. 6. Syracuse University. LIST OF SYMBOLS A E h I K L m r. 1969. The authors believe that this is due to the fact that the axial stresses were considered small.o R P t) Po PI(t) W x y Z c. Davidenkov. 2.. Syracuse. "Analytical Formulation of Damped StressStrain Relations Based on Experimental Data with Applications to Vibrating Structures. 3. London.Y.0. Solids Structures. K. Phys. J. •The downward shift in the curves defining the region of instability with increasing amplitude may be due to the nature of the maximum point locus curve for the Davidenkov model: tEco n 01 second frequency to the response of the model does not undully complic2tesubsequent use in the governing equations then such work would constitute a valuable 6ontribution. Bolotin. Coordinate Total normal strain. V. 5.A. J.0.
7 Effect of Varying uon Phase Shift p Fig.1.Efftaictt Fig.Cangeof Intabiity i icreasing Fig.0. nmltd Fig. 1 l0  Three Dimrensional Representa2tion of instability Region 160 .( FTig. .. Effect of Varying on Amplitude 8 ' oft Vayn 6 .Radius of curvature C1f~j O4 Anua dslcmetla All Frqunc Loa Oraet /(P P .
Stahle: The orthogonality referred to the abbreviated model test. Stable: This is the Earth Resources Technology Satellite which is a fairly small. Mr. Essentially it Is a modal model using modal coupling techniques to marry the spacecraft back to the launch vehicle. We were also concerned with some of the large amplitude modes like the first longitudinal which actually went as high as 55 Hertz. Was that mainly because of the use of experimental modes wihich hadnot been orthogonalized before their usage? Mr. V.DISCUSSION Mr. had gradually crept down into the pogo frequency range. The criterion that we had set up was that the measured experimental mode would check within 10 percent no' the analytical modes. Zudans (Franklin Inetitute): You surprised me with very poor orthogonality. through increased growth of the spacecraft. I think our main confidenco was gained from the fact that the frequencies matched up very well. The model that we have been using is the model that I presented here. Mr. and that the main structural modes agreed relative to the modal admittance through the base shear.000 pound. It was in this general frequency range that we were concerned. We had a number of modes which. space craft going up on the Thor Delta. It follows the basic inertial coupling procedures of component synthesis discussed in the literature to couple this analytical representation of the spacecraft back to the luanch vehicle. 2. The problem is the very limited amount of instrumentation used on the solar array panels. Prause (Battelle Institute): What are some of the important frequencies? We saw a Ioc of natural frequencies in the presentation but what rre the control system frequencles and what are the pogo frequencies for this type of space stations? Mr. Schrantz (Comsat Labs): Did you couple your model with the Thor Delta to check out the responses? Mr. *The paper was presented and discussed by C. The main pogo frequency varies somewhere between 17 and 23 Hertz. 161 . Stahle for the authors. Stable: This is done by Douglas personnel.
these conveyors could weigh from a few hundred pounds to several tons. absorbers. Dynamic forces and moments at the ends of each member are expressed in terms of the displacements and rotations at the ends. Civil Engineering Department University of Louisville Louisville. but. Then the structure is analyzed under the effect of the reactive forces at the points of support of the conveyor. Kentucky The dynamic analysis of the conveyorstructure system is presented using the stiffness method and an iterative scheme in which the structure and the vibrating conveyor are analyzed successively taking into account interacting effects. in some cases. and after a few cycles. the final results are reached. INTRODUCTION Vibrating conveyors are widely used in industry for conveying granulartype material (1]. Depending on specific applications. the dynamic analysis of the conveyor is performed. When vibrating conveyors are supported by a frame or truss. From this analysis the motion of the points of support is determined and used as boundary conditions in repeating the analysis of the conveyor. This iterative scheme cintinues. Louisville. The dynamic analysis of the conveyorstructure system is presented using an iterativr scheme where the three dimensional structure and the two dimensional conveyor are analyzed successively taking into account interacting effects. The main components are the pan or trough where the material is conveyed. giving the socalled eclment dynamic stiffness matrix [31. The nystem dynamic stiffness matrix is assembled from element matrices using the conditions of continuity between elements and equilibrium at the Preceding page blank 163 . Rex Chainbelt Inc. they are suspended from or supported by a structural system. depending on the type of conveyor (2). each having known elastic and inertial properties. the dynamic forces which are transmitted to the supporting structure result in vibration of and interaction between the structure and the conveyors. by assuming t' the supporting structure is rigid. Basically. An example of a structural truss supporting a vibrating conveyor is given. and the new reactive forces are applied in repeating the analysis of the structure. ANALYSIS The vibrating conveyor and the supporting structure are analyzed independently by the dynamic stiffness method where the system is divided into a number of members. or boosters. Kentucky and Oscar Mathis Design Engineer. Usually the conveyors are supported to a rigid foundation through isolation springs. First. this result can be achieved by solving the appropriate equation of motion. and a system of springs which connects the trough to a base. a driving mechanism which produces rectilinear motion at an angle with regard to the horizontal direction.DYNAMIC INTERACTION BETWEEN VIBRATING CONVEYORS AND SUPPORTING STRUCTURE Mario Paz Professor.
As a consequence of this absorbing action. the dynamic booster conveyor consists of a trough supported by isolation springs. Vibrating conveyors are manufactured in a variety of types. The dynamic boosters are tuned at a natural frequency in the neighborhood of the operating frequency of the motors. with double extended shafts to which eccentric masses are affixed. ranging from a single moving deck to three or more vibrating masses. The coordinates indicating joint displacements at the ends A A A A A C B B a B B Oe '/ D GLOBAL x G G G( G NG COORDINATES Fig. Diagrams for the basic elements of the booster conveyor are shown in Figs. and the corresponding dynamic stiffness matrices are given in the appendix. 1. the trough vibrates harmonically along a direction approximately normal to the orientation of the boosters. The n equations may be written as [S]{uJ = {FJ (1) [S] is an nxn symmetric matrix composed of terms derived from the dynamic stiffness matrices of the component elements of the system. the latter will include any applied external force and the forces from all the elements forming the joint. As shown in Fig. 2 through 4.l . The drive unit as well as the connecting leaf springs is considered to be a special beam element. The drive unit houses two motors. one above the other. The motors rotate in opposite directions producing a horizontdl hatmonic force. The trough is assumed to be composed of continuously connected beams with distributed and concentrated masses.(uJis a vector of the n independent joint displacement in the system.joints. a drive unit connected through leaf springs to the trough.Dynamic booster cenveyor 164 . and the isolation springs are treated as massless elastic members. Equilibrium conditions at the joints result in n equations relating the applied external forces to the independent joint displacements of the system. The boosters function as dynamic absorbers (4) for the component of the impressed force along the direction of the boosters. In this presentation the socalled "dynamic booster conveyor" is described in relation to the dynamic problem originated by the interaction between vibrating conveyors and aupporting structures. and a series of spri ngmass assemblies known as dynamic boosters. and{F is a vector of the external forces which are exciting the structure at frequency w. In the analysis of the conveyor the boosters are treated as rigid bodies connected elastically to the trough.
a nd e s or r fu inple di m a mple building floor systems. El A 5 2 A mgitudinal 4 1 L In general. thus. The dynamic stiffness matrix for the pinjointed bar element is given in the appendix. A dynamic booster conveyor supported by a truss type bridge between two buildings is presented to illustrate the interaction analysis.5 EXAMPLE  Pinjointed bar element 5 64 2 3veyor. the reactive forces are applied in performing the analysis of the truss to obtain the first approximation for the displacements at the points of support of the conveyor. Fig. It is assumed that for each individual element of the system these three deformations are uncoupled.4 Isolation spring element SUPPORTING STRUCTURE installations of conveyor supportcomplex three anr insionaltres dinbui fl yst omtr sim u sses to exs ms. the conveyor is analyzed initially under the action of the driving force and the condition of zero displacements at the supporting points.3  Dynamic booster element Fig. any member of the truss. under the action of inertial forces. The dynamic stiffness matrix for a uniform pinjointed bar element shown in Fig. although assumed to be ideally pin connected at its ends. 2 5 5 6 ELATOER 1~ 4 9 L Fig. Fig. may undergo. flexural deformation in each of the two principal planes in addition to the extensional deformations along the Ionaxis. A space truss is used in the interaction example presented. Then. The f irst four cycles of the interacting supporting structure and the conveyor are shown effects between inthe Table I. In the next cycle these dynamic displacements are imposed as external actions on the conveyor.of the elements are numbered consecutively.5 is obtained by solving the corresponding BernoulliEuler differential equation for flexural deformation and the wave equation for the axial and by introducing the appropriate boundary conditions. the dynamic stiffness matrix for the element of the truss is obtained independently for the two flexural deformations in each of the principal planes and for the axial deformations. 165 . 6 shows the schematic diagram of the space truss supporting the conAs explained above.
0370 0.0600 0.0077 0.1940 0. t 41 BWOTTO DIAGONALS Plan View BOOSTERSRO SPIG SPRIN "TOOL To.0390 0.0730 0.i590 0./ ' Ix 2 STAGGERED ' 7 /2' 8 Spaces at 30240" End View Elevation View Fig.1940 0.0998 0.0052 0.6 .0082 0.0800 0.0480 (a) Amplitude of truss vibration at conveyor suports fin.1890 0.0200 0.0022 0.) Cycle 2 3 4 5 0. *I  7: .0800 (b) Amplitude of conveyor reactive forces at supports (lbs.0620 0.1580 0.Example space truss supporting vibrating conveyor TABLE I Interaction Results suport4 support Support 2 Suppot up 3 t Supo Vertical Horizontal Vertical Horizontal Vertical Horizontal Vertical 1 y lorizontal 0 0 0 0 0 0 0 0 0.0365 0.Ph.0390 0.0198 0.0053 0.0620 0.0200 0.A j .0770 0.0045 0.0389 0.) Support 4 Support 3 Support 2 Support I Cycle iorizontal Vertical Horizontal Vertical Horizontal Vertical Rorizonta Vertical 1 339 172 203 T6 175 33 139 2 153 123 162 111 178 62 204 3 133 132 153 110 173 48 237 4 136 132 153 110 173 j 48 239 166 *'4 .1560 0.0830 0.0730 0.0710 0.
This study of the interaction of a vibrating conveyor and the supporting structure indicates that the method presented requires four to eight cycles converge to the final solution. the dynamic booster conveyor. at the same time the corresponding force term of this equation is replaced by the product of the nevly formed diagonal element and the prescribed displacement.NOTES ON COMPUTATIONAL METHOD A computer program in Fortran IV is developed for the analysis of the space truss. APPENDIX The dynamic stiffness matrices for the basic elements of the booster conand for the pinjointed bar element of the truss shown in Figs. CONCLUSIONS An iterative method for the analysis for vibrating conveyors mounted on supporting structure has been presented. The two systems are analyzed separately using as boundary conditions the deflections and forces developed at the points where the conveyor is supported by the structure. 2 through 5 may be written as follows: Isolation Spring 1 (6) IS THERE A SIGNIFICANT CHANGE IN DEFLECTIONS AT CONVEYOR SUPPORTSI NO YES vSET NEW DEFLECTION VALUES rCALL CONVEYOR PROGRAM TO IDETERMINE MOTION AND FORCES CALL TRUSS POGRAM TO DETERMINE MOTION AND FORCES IPRINT OUTPUT MOTION AND FORCES Kt 0 K K t t 0 K 0 Ka 0 0 K a 0 Kv 0 K Kv v 0 Kt 0 Kv Kt t Kv 0 Ka 0 0 Ka 0 Kv 0 KR/2 K v 0 KR R ~where: STOP The prescription of imposed displacements at the points of conveyor support may be programmed for a digital computer such that there results a reduction in the total number of equations to be solved. however. Several testing problems as well as actual cases of installation of conveyors supported by structural systems were analyzed by the iterative method. and the initerative procedure as described in this paper. The flow diagram of the computer program follows. Flow Diagram MachineStructure Interaction START 1(l) READ AND PRINT INPUT DATA 2) SET DEFLECTION AT CONVEYOR SUPPORT POINTS EQUAL TO ZERO 4to (3) LOOP I = 1 CALL CONVEYOR ANALYSIS PROGRAM (4) SET REACTION ON TRUSS EQUAL TO FORCES DETERMINED IN (3) iveyor (5) CALL TRUSS ANALYSIS PROGRAM] TO DETERMINE DEFLECTION AT CONVEYOR SUPPORTS Ia. In order to avoid rearrangement of computer storage. Ka = Axial spring constant Kv = 1/2 KtL KR = 1/3 KtL2 L = Length of the spring Kt = Transverse Spring Constant 167 . The diagonal coefficient of the dynamic stiffness matrix corresponding to the equation with imposed displacement is multiplied by a very large number (say 1021). it is often more convenient to proceed with the direct solution by implementing a computatiunal device due to Payne and Iron and referred to by Zienkiewicz(5).
Booster Element Jf mass moment of inertia at left end of beam Concentrated mass moment of inertia at right end of beam C2 K c +21 s M2 SC(KcKs) Cx c SSKe+CCKsmW2 i K.= .W 2 0 D (CaHca) D(HaSa) EABNbafW 2 0 0.C2Ks "ejc/12 c s 0 c (cK8)S 0 +I s e2 / 2 S = sinO Kc = Spring moduluscompression Spring moduluse..SIN(BL) b Cb Ub Q P = COS(BL) COSEC (EL) aHsa+Salca SalcaCaHsa L a Length of beam  Nb = COT(BL) M i . (1+e/l 2)j CK 1C 2 Cse+Sc K+S2X Symmetric SC(KK) s SC(KK)(SKcC 2 s)S1cl SC(K _(SI +C KCS 0 0 sc where: C = coso SC(KKs) S2 Kc. 0 (See Fig.aSaHsa Da2QmIW2 DPJf 2 o he: Da (HcaCa) where: E = Modulus of elasticity A = Crosssectional area B w m =(m2/EA) Angular velocity of forcing frequency Mass per unit length Ha = eosh(aL) Hs D sinh(aL) Ea/(lCa ca) S . 3) 3) 18 .A b. E moulushe~r s =Sprig m = mass of booster j= mass moment of inertia about Point 0 (iq.Concentrated mass at left end of beam a = (mw/EI) 1/4 Mf = Concentrated mass at right end of beam C s= Concentrated I Crosssectional moment of inertia a = COS(aL) Sa = SIN(aL) c. .Conveyor Beam Element E Nb iw2 RA0UDaSaHsa EABUb o o DaQMit2 Symmetric DS 0 "Da (S+Hsa) DPJ.
J. l.Pinjointed Bar A 1 0 0 A2 REFERENCES A2 0 0 Al 0 P(Z) 0 0 0 0 PY) 0 (1) Paz. 0. Z) DISCUSSION S= M.e. London 1967 (coeecharLCosecrL) acEI 2 r r  (r y. 0 Q(Z) 0 0 0 0 Q(Y) 0 0 0 where: P(Z) 0 0 P(Y) 0 0 Q(Z) 0 0 Q(Y) Al f EA~cot0L 1956 A2 EAcosec8L a1EIr 1 (5) Zienkiewiez. P. New York 1968 (4) DenHartog. and Cheung. although it is pinconnected at the ends. .Theory of MatrixStructural Analysis McGrawHill.. The Finite Element Method in Structural and Continuous Mechanics Q(r) P(r) Ur =  r Lcotha rL) (cOtr McGrawHill Company. M. J. The forces coming from the conveyer to the truss were dynamic forces and the truss was analyzed as a dynamic problem. AE understand your model.Y. C.Design of Machines PrenticeHall. 169 .Mechanical Vibratioas McGrawHill Book Company. S. But the equation to determine the dynamic stiffness for each element JiJf J'Jx k kaikc. . Paz: No. This is the distributed mass case It has a finite number of degrees of freedom because of the masses of the matrix method of structural analysis.yIy. I z = Crosssectional moment of inertia Mass moment of inertia moment of inertia Polar kfreedom members had distributed mass.d. Zudans: Did you have your truss representwith dynamic degrees of system ed as a lumped and notmass only static degrees of freedom? Mr. The paper shows the dynamic stiffness matrix for a truss which. Rolland . New Jersey 1957 (3) Przemieniecki.2Mr. T. Inc. still has bending due to the inertia effect.Conveying Speed of Vibrating Equipment Publication No. My impression is that dy Zudans (Franklin Institute): I do not think I NOMENCLATURE A = Area namically you only considered the conveyer and that the truss was taken as a simple static structurewithout any dynamics considered In the process of analysis. but this is not the case presented here. sir. . Paz: Actually this was done by the coauthor.k s Mimi'm j = = = Spring constant = Mass 0 = Angle w = Angular velocity takes Into account the distributed mass and elasticity. 64WA/MHI ASME 1965 (2) Hinkle. Mr. The truss shear I.h = dimension G = Modulus of elasticity in Mr. Inc.K. Is that correct? E = Modulus of elasticity L.
equations are derived for the response of a simply supported circular plate exposed to time dependent pressure and thermal loading. a sustained interest in the consequences of exposing a structure to timevarying thermal as well as mechanical loading conditions. GENERAL FORMULATION dynamic effect and the thermal loading become important may then be identified.J. The regimes in which the are time dependent but assumed not to vary in the inplane directions. Whereas the effect of a shock wave on various structures has recieved considerable study. In effect there appear to be two stressfree surfaces. The elastic properties have been assumed to be uniform and not temperature dependent. Results based on these equations are presented for several loading conditions and plate'geometries. a period of rapid and intense heating would soon be followed by a shock wave. the displacement w must satisfy the well known [I] thermoelasticdynamic plate equation together with the appropriate boundary and initial conditions. and M.N.thermal stresses may be compressive at both the front and rear surfaces.L. The analysis is based on the thermbelastic equation of motion for flexure of a circular plate subjected to time dependent lateral loads and temperature distribution. Results computed on the basis of these equations are presented for several loading conditions and plate geometries.N. Another result of the computations is the observation that the thermal stresses may be compressive on both the front and rear surfaces and tensile at the center.N. The applied pressure loading and the temperatures within the plate We wish to determine the displacements in a simply supported circular plate subjected to a uniform timedependent pressure p(t) and a timedependent temperature field T(z. Using classical plate theory.Cohen North Eastern Research Associates.RESPONSE OF A SIMPLY SUPPORTED CIRCULAR PLATE EXPOSED TO THERMAL AND PRESSURE LOADING J. One interesting result is that in some cases the rapid heating of such a plate by exposure to thermal radiation induces vibrations of substantial amplitude compared to the quasistatic displacements.t) which varies depthwise through the plate. One interesting effect is the presence of thermally induced vibrations having rather substantial amplitude in some cases. in the recent past. The problem shall be formulated along classical platetheory lines wherein. briefly. Upper Montclair.J. and tensile in the center.J. when exposed to such loading conditions. In addition it was observed that the. Classical modal series methods are used together with the MindlinGoodman procedure for treating time dependent boundary conditions.. the effect of rapid heating has been givin less attention. Hoboken. One may readily envision situations in which this type of loading might occur. the temperature distribution and history are computed using a finite difference technique. in the environment of a nuclear event.and Stevens Institute of Technology. Regarding the heat conduction problem as uncoupled. For example. both analytical as well as experimental. Upper Montclair.E. Preceding page blank 171 I . INTRODUCTION There has been. The object of this study is to determine the response induced in a circular plate simply supported on its boundary.Koch North Eastern Research Associates.
Thus we may say that T: VIMT=O in the above equation and omit derivatives with respect to) : we find that the normal function Rn(r) is composed of Bessel functions of the first and second kinds [3. Within this plate the displacement w must satisfy the governing differential equation a where e r ( () in which Rm(r) are the normal functions arising from the free vibration problem. 4 4. Since no generality is lost if the effects of arbitrary initial conditions are not included.l. on r=a for all t>0: /. 4 Furthermore. we impose the condition that the resulting displacements and stresses be finite at r0. Free Vibrations The displacements of a plate vibrating freely satisy the homogeneous form of the governing equation (3).() 6 Typical Circular Plate %4 M21 (7) Consider the thin circular plate of thickness h and radius a. ff% (9) The condition of finiteness of displacements at the center of the plate (at r=0) requires that the coefficients Cn and Dn vanish. (5) and the frequency ecuation (10) 172 . the center of the plate. alternatively: 0r where:...of or. gl(r) and g2 (r) are to be selected to satisfy the inhomogeneous boundary conditions. with its median plane in the rB plane and with z denoting the distance from this plane.kv 117P .)r.Dycr.41: .. we choose the following conditions to apply at tO for all values of r: " T  W=0 DYNAMIC DISPLACEMENTS We seek a solution to the governing equations together with the boundary and initial conditions in the following general form t= j. 1.rdr For the simply supported case we require that the moment Mr and the displacement w vanish on the boundary of the plate. this must be true of the thermal parameter MT as well. Applying the homogeneous form boundary conditions (4)and (5) we obtain the normal function d . Thus.e. i. NMT = ) A T* 1 Choosing a product solution of the form (2) we•?Rr) e Since the temperatures as well as tbe pressures are independent of r and l. fl(t) and f (t)represent the time dependency 8f the boundary conditions (2]. as shown in Fig..the homogeneous solution. .)T* Figure 1  a .t)a)0 (4) r& r )] Or .
.(4) ".. we find that Tm must satisfy the solution for Tm is readily from which found to be Ws.: Rs (As e it ~ o57  r Boundary Conditions The technique of Mindlin and Goodman [21 is used to treat the time dependent boundary conditions which appear in this case in connection with the thermal loading.7 ' IL V . equation (14) this to eliminate V becomes: I r . we placement obtain: w given by equation and the orthogonality conditions we may obtain: . if u. t 3z. stated below for the rth and sth distinct normal modes: (12).' lent to the following set of conditions on gl and g2 at r=a: atI / ( or r 0 From the homogeneous equation (8) we see that DV7R. circumferential. respectively.(+ P(0t m(t) 4 (t) where: Q() Equating the coefficients of Rm on the left and right sides of this equation.(. then we may write _____________________ ~t fjc . Orthogonality Of Normal Modes The orthogonality condition among normal modes may be deduced from Clebsch's thereom E5].) (16) R. . Substituting equation (70 into the boundary conditions (4)and (5) yields. Pa~  SR(14) fkl ...f&) Noting that the normal function Rn satisfieithe homogeneous boundary conditionh: these equations are equiva •_DEV"'. This is done by selecting the arbitrary functions gl(r) and g2(r) in such a way as to transform them into homogeneous conditions. Making use of .= (13) 4L S 3. 2. Substituting into the governing equation of motion (3) the modal series representation of the dis(7). Rj'rd 0. Response To Thermal And Mechanical Making use of the homogeneous solution now turn to the response of the plate to the actual time dependent thermal and mechanical loads. v and w are the radial. and lateral displacements. £ . '.0 j.'?m .) (11) on the right side of this equaThe terms tion are expressed in infinite series form: c( being the successive eigenvalues of thys equation. at r=a: . In 'e case of a circular plate.r= 0 these displacements into Substituting equation (12) we obtain the desired condition of orthogonality among normal modes: 6 R.C') Ra(r) q These series expressions may be substituted into equation (15) to obtain: G.t4) 2.pA'q9.)~ R Q. 173 .
./ ]rka') _ J(rL)r Jr 14 T(o) where K.) "14 WoL &17 10 "[drJ"i . to the the zero values ofcoefficients A and N: of Rm toobtainm we note from equation (17) that .) aAd carrying out the integratigns these become: . 9 f =0. for all values of r. For example. i . i1 .¢ . become 1 .. _n Evaluation Of Coefficients E' EM an and Qm.. The formal solution for w may now be written: / " z. etc..4). gl=l... 0A". '?' t1. from the first of equations (16) the expression for p is P 500 3.Jr.) (d. 1 (19) 0 Te R'+ Thus the coeficient may be computed according to the equation Yr )fk Similarly The second term of each of these equations may be expanded in infinite series form.L. .Initial Conditions Making use of the assumed displacement form (7) we find that the initial conditions (6) at t0. using the actual values p=p(t).. inte J( 2 • "1} J 174 Using these displacements w may now be . The coefficients contained within equation (21) may be computed in detail by applying the orthogonality condition (13) to the infinite series expansions (16) and (19).0= DT.) D. thereby simplifying these equations: " ~ Thus i TM(0)A m and ar.K grating with respect to r and making use of the orthogonality condition: [7121 Or i to both sides of this equation. 4 2 _.O) I './. . f = "Ir/(. . and substitute equatioiis (20) into (19) we may ejulre.) . (21) (23) .l'=. m) IL Finally.r T ] Am =Em Bm S* (22) EmIfII where L i' L ' r. and g =r0 A(. 4))o/'L' T Tr O& (Ihd. 61 R.)1 I.
Dynamic displacements were observed to range as high as 1.LM(C) Los&4 . another interesting may be observed.4)tu. The dynamic case is somewhat different.7 Av/"r J' 4s ) . but renders the plates somewhat d±simildr j_  Jiau.44O1zIC (24) ) r/ l) QUASISTialC DISPLACEMENTS The quasistatic counterpart of equation (24) in which inertia terms are neglected is readily shown to be: v.16 seconds) are presented in Figures 2 and 4./ (v).... These pulses rapidly rose to peak intensities. H .. 2.1 seconds and 0.rL flr 2"On 411. [the M.L 6755 . which occured at 0.. . these other plates may not be compared simply on a fundamental period basis because the variation in thickness does not merely change the period. (. /gradients and should be expected when rapid heating is involved. From the nearly symmetrical stress distribution in this case we also see that the static stresses induce no net moment in the plate. in which we may clearly see fp(r) I AnuW )dtJ Pk kL z(8*O i L) This may be further simplified by carrying out the integration of T and observing that 2) 2 . 1 RESULTS Making use of equations (24) for the dynamic case and (25) for the static case. 26 in thermal response as well. The displacementtime history of those plates having the greatest fundamental periods (ranging from 0.e)d . while they are tensile in the central region. the dynamic diminish as the period becomes effects substantially smaller than the heating time. . Note..a2 is the series representation of (r deduced in the first of equations (20): cc) thermally induced vibrations of a substantial magnitude.3L' _e Where NT .62 seconds to 0. or during the 175 . trJ a pulses. Clearly this unusual stress distribution arises because of the sharp thermal p" STRESSES the displacements (24) or (25) by substituting into the thermoelastic stressdisplacement relations (1): _ . displacements were computed for the center of an aluminummagnesium alloy 4K)I ob plate exposed to two radiant heating . however.l .1 47 1 3 (25) Q ._ . The stresses phenomenon within the plate may be compressive at both the front and rear surfaces. pulses having a rise time fairly close to the fundamental period.. Figures 3 and 5 summerize the results of these calculations as well as Strictly for one other thickness. The t~tal thermal exposures were 71 cal/cm and 145 cal/cm 2 respectively.AlL r . While still satisfying the condition of no net axial force in the plate.wr written as: " . that for a given thickness. and we see that there are now two "neutral" surfaces for stress.I_ f. .34 seconds. ") where{.. the basis of some preliminary stress calculations. are the quccessive roots of the equation ( .1. those speaking... 0 Pr). the magnification is greatest. which presents the static and dynamic stress distribution in a thin plate shortly after the thermal radiation peak has occured. a substantial moment is induced. PlAtes of a number of thicknesses and diameters were studied./ (26) Yr I. ._ Lo w. This is clearly seen in Figure 6.7 times the corresponding maximum static We see also that for displacements. The static distribution is almost precisely that described above.
Theory of Bessel Fucin. App. J.N. 3. p. Silver Spring. and the Naval Ordnance Laboratory. London.early stages of the heating period.Jahnke.. p. Lodn 966 5 Love.E.. 9 176 . A.D. Boundary Conditions".New York. G. Maryland. Tables of Higher Functions. 1950.g.Mindlin. McGrawHil. CambridgeUniv. R.A..72.Boley. and Weiner. 1960 4 Watson. New"York.H. B.. 377 Vol. Brooklyn. "Beam Vibrations With Time Dependent J. Theory of Thermal Stresses.Press. Press. ACKNOWLEDGEMENT This work has been supported by the United States Navy under Contracts N0014070C0019 and N6092171C0197 monitored by the Ntval Applied Science Laboratory. Mech. Cambridge Univ. New York. . REFERENCES a I . Emde. and Goodman.E. 2 .189 #4 '##*S'''t''' C. Wiley. and Losch. L. Theory of Elasticity. 1927..
A''' PH .i InMnit Having ACPeak cm Radan Pule 71 cal/cmiRadiant Puls 177I'.iti iI!~~~~llll 1 H 1IYI fiur 49VINUIX2 71 cal To ANC 2010 esponse ofT NYin 1/8i ! Thick lat At 0.
.1 ll di Il i vpIN p 1"! 1 111178l I .I'~~i tfI 1 il 11i HavWi A PekintniyAt03le 111 4 Ti! IJil ..1 14 ca/m Ra iant Pule il 1ni ll Ii fil 1i 1il 1!: !] . ll! I~ ......... fil it H)etr ......
.M wl Ii .2t0 Srs Ffue6 o ~T4C'~ C'' 'itM i lat !i Atl i t Cne Of /8 itiu n Pulse1 1!!o* AI71 cal/cm Rad Afe Expsur im 0 0 e .I CC( . 179.lfki iiI 1 i VI: 1iIff: iI'%VI11H.
not quite . Since the rise time is dependent upon the diameter. I wonder why the rise time for a diameter of 3 meters is the slowest. Mr. You showed the static stressesas being perfectly symmetrical. however I hace done some nonthermal work in the vibration of layered materials. Mr. Mr. The 1. I noticed that you bad fluxes of 71.5. I agree there is some lack of data on materials in the short time range. I suspect. Something confused me. Koch: One side of the plate only. Koch: I think they were 10 psi. Koch: No. Mercurio: There is also the problem of the properties changing over very short time durations where very little irlormation is available. Mercurio (Sperry Gyroscope Company): I would like to commend you on a very interesting paper. In fact in about a months time I am due to have a report on this subject. if you just turned the plate a bit.almost. through the thicknesis?" The frequency changes and the modulus Mr. Mr Koch: We are getting into it right now. meters is the fastest? Mr.5 meter one was quickest. Accidentally this looked symmetrical. Yang (University of Maryland): I have two questions. these happen to be particular weapon pulses. Is that correct? Mr. Mr.perhaps in the millisecond range. Mercurio: You have not shown the pressure stresses on the plate. so the whole dynamic picture might change. On one of your curves you show various diameters of 1. let's say. + 180 1 . lave you done any work in the thermal stresses of composite materials? Mr. and 145. Koch: No. Mr. This is what I do not understand. Mr. I have not looked at it. I had to get rather flexible plates in order to show the dynamics. Mercurio: Have you done any work in this area? I would like to know. We are also attempting to answer the question: "What actually happens if the properties vary with temperature. Mr. It was not actually. but the actual numerical value was such that it was not actually symmetrical. Koch: Yes.DISCUSSION Mr. Koch: I have access to some data which is in a range less than one second . Mr. 2 and 3 meters. Mr. I do not think that actually happened. it just looked like it. but I was promised the data. and I was constrained to use particular pulses by my contract. Zudans: How could the temperature distribution be anything near symmetrical If you bad heat flux from one side only. Zudans (Franklin Institute): I would like to refer to your last slide. Mr Koch: If you shifted the distribution curve you would have something that would appear to be nearly symmetrical by coincidence. and I want to clarify it. Zudans: How can that be explained? Was the heat flux coming to one side of the plate or both sides of the plate? Mr. Mr. Koch: I think I may have flashed the curve too fast. It happened. that the temperature distribution was such that. and I believe.5 meters Is in the middle and for 2 changes. Koch: Yes Mr. whereas the time for 1. Yang: The second question is. you get something that looked symmetrical. Mercurio: What were the associated pressure loadings? Mr. They would have wiped out these plates compietely. because some of te common materials that we are dealing with get into problems when you apply both the thermal and pressure loads simultaneously.
defines the most conservative design and the uppermost bound. optimum value in the sense of greatest freedom of choice of nacelle mounting stiffness for dynamic stability. It seems logical then. The lowest bound is found when the blade fundamental bending frequency is 4/10 This is considered to be an of the propeller rotating speed. at least partly. reported observations on the investigation of nacelle vibration on existing hovercrafto indicates that whirlflutter Further studies are could properly explain the results. In the past decade. the pea some V/STOL configurations. large hseli the controlling parameters of whirlflutter to assure an adequate design. The prdsent analysis was taken for under copters and vertical and short takeoff and landing aircrafts (V/STOL) came into existence.10]. propeller whirlflutter stability is one of the main considerations in design [310]. by the aircraft industry for propellerrotor dynamics is used to establish stability boundaries for variations of physical parameters of a propellernacellepylon system. This is attested to by the fact that on In the aircraft industry. The stability boundaries are presented in chart forms which can be used as design guideb. which considers ideally rigid blades. for a future large SEV equipped with a similar propulsion unit. due to the purpose of gaining a general knowledge of the propeller 181 . The uppermost and lowest bounds for stability Classical whirlflutter theory. sible occurrence of whirlingtype instability on propellerdriven aitcrafts was recognized as early as 1938 by Taylor and Browne [1].N4 [11]. are shown on these charts. This type of instabil On a large existing surface effect vehicle (SEV) such as SR.C.WHIRL FLUTTER ANALYSIS OF PROPELLERNACELLEPYLON SYSTEM ON LARGE SURFACE EFFECT VEHICLES YuanNing Liu Naval Ship Research and Development Center Washingtonr D. that studies be made to determine failures on the Lockheed "Electra" airSuboequent investigations revealed that propeller whirl flutter could cause total destruction of the propellernacelle system if the nacelle was not mounted with sufficient rigidity. 20034 A typical propellernacellepylon configuration installed on large surface effect vehicles is analyzed for dynamic staA whirlflutter theory deveAoped bility in normal operations. impetus of two early propellernacellepylon system is very similar to the propellerrotor system on a typical V/S'OL [3. recommended. INTRODUCTION the understanding of this phenomenon. I. Although direct comparisons of experimental data with these analytical results are not possible in the present analysis. the ity received a great deal of attentions in 1960 with the crafts [2].
respectively Dissipating t T U V Spanwise distance along the blade measured from hubcenter to a blade section Laplace transform variable Time Kinetic energy Resultant air velocity on a blade element Proveller advance speed y CdC c Ax .whirlflutter phenomenon as related to the design of the propellerrotor systern for a SEV.3. such as nacelle mounting stiffness. respectively Instantaneous blade section angle of attack A' Initial angle at a blade of attack section Effective flapping angle of the nth blade at its first bending mode. dL E h I Im(s) 1I11 A Blade sectional drag an d lift. k Effective k nacelle spring constants in pitch and yaw. . referred to the plane of rotati on Blade lock number.1. respectively ao an Position vectors for a blade sectin and a nacelle se'tion. a ACJ NOIENCLATJRE oa lift r sr Local curve slope of a blade section Equivalent damping matrix s Sectional blade airfoil drag and lift coefficients.2. . etc. The overall design for SEV must also consider the vibration characteristics of other mcchineries and the platform.. ~ A: AAA A~A ____________________4_ . respectively Blade chord Effective viscous damping constants on nacelle pitching and yawing motions. . nacelle inerti~s.~ ~. k8 Effective blade bending stiffness constant at hubcenter 1 mb Equivalent mass matrix Blade mass per unit spanwise length m N Nacelle (include shaft) mass per unit length Number of blades Blade designation number (n . s Normalized spanvise dis4 tance. respectively Potential energy Nacelle length ea Blade moment of inertia about hubcenter Imaginary part of s Nacelle moment of inertid including the mass of propeller in pitch and yaw. respectively. r/R D. Design criteria may be developed in order to define a stable propellernacellepylon system ae functions of design parameters. referred to the plane of rotation i [K] Equivalent stiffness Eualet . Kdol%. N) Propeller radius Real point of s n R Re(s) II. . respectively acR 4/b Exponential decay factor 182 IA ~AAA .c y fucinx 4 41 Xb.Kd 2 1 x y Drag factors Y Propeller disc pitch and yaw angles. and blade vibration characteristics.
A schematic representation configuration is for the Fig. It was this analysis rigidbladed propeller a flexibly mounted nacelle consists of thd pitching and yawing rotations of ti e nacelle combined with the corresponding motions of the propeller plane. and vibration are considered demonstrated by 183 I. I is a sketch which Illustrates the configuration of the (') ('b) (4) III. this system or any other propellernacellepylon are considered infinitely stiff.{ . THEORETICAL CONSIDERATIONS propellernacellepylon system on SR.e. i. 4X. V/RO Air density Nacelle instantaneous pitching and yawing angles. The pitching and yawing rotations of the propeller disc due to the nacelle flexibility destroy the central symmetry of the flow in a continuously varying manner and create additional undesirable moments which may lead to unstable nacelle circular motion in the direction opposite to that of the propeller rotation. aerodynamic loads used in be linear. X y respectively nB Blade frequency ratio (. in which the propeller blades is The on follows. respectively Azimuth position angle of the nth blade level flight.. dot between two vector quantities indicates dotproduct Superscript indicates differentiation with respect to the time variable Qt Tilt over a quantity indicates Laplace transformation Arrow over a quantity indicates a vector A. Fig. respectively Nonrotating blade fundamental bending frequency W . y $1 Propeller Nacelle W x y rotating speed natural frequency Nacelle natural frequencies in pitch and yaw.N4.) Dot over a quantity indicates differentiation with respect to time t. or. of The classical whirlflutter phenomenun. the resulting flow of air normal to the propeller plane is generally symmetrical and produces a symmetrically distributed propeller thrust.Nacelle damping ratio For a forward moving propellerrotor in A p Propeller inflow ratio. the backward whirling." which is to be distinguished from the purely mechanicallyinduced in whirling motion a rotating shaft s)stem. i r The aerodynamic source of this instability accounts for the usual designaa tion "whirl flutter. briefly explained as motion of the mathematical model analysis is shown in The mathematical model used tn this essentially that of a fourThe steady theory to degreeoffreedom system. The coupled motions of the blades and rotor form the theoretical background of this analysis. 2.q n y x Nacelle frequency ratios 0 4 /a) and (w y/). DYNAMIC STABILITY When considering the dynamic stability of a propellerrotor system a irore realistic representation of the system would be to take into account the flexible motion of the blades as well as that of the rotor.
It is felt that omit Mathematical Model for a PropellerRotor System ting higher order effects is justified and the stability boundaries resulting frod this analysis are valid. principle was used to derive the governing differential expressions for D 2 E + d dmb (la) s 0 (. + T (shaft) fR The nacelle ibility at the is is allowed to move in pitch and yaw directions assumed root of the and its flexE 2' JI nl 0 to be concentrated nacelle. c x kinetic energy.2 + Ic 2 4y (2c y 184 . 2  Schematic Representation of were small on the variance of stability boundaries.r blade is allowed to flap in its fundamental bending mode_1 and an equivalent system is f rmed such that the blade Is considered rigid but with an elastically hinge located peller restrained flapping at the center of pro2).Aerodynamic Forces on a Blade Element Lytwyn [7] that the nonlinear effects Fig. + The variational equations.FLAPPING HANGE . dssptopteiawrenedad p were formulated as T T follows. OF AXIS LADE C SHAFT SOTATING AWI 47 • L. The indi (bb bb n vidual propel. potential energy and .s( x'8 s ) dma k k 2 + 2 ky y E nk 82n n (lb) disk (see Fig. 2a  PropellerXotor Structural Mode! JL Fig. 2b . Hence. 1  PropellerNacellePylon System on the SR.N4 Kovercraft PARAtLTO I[ELANEOOfATOP Fig.6'96AD[ MEJINT PIT01 AXIS YAWARlS Fig.
Instability is considered to occur whenever any of the characteristic 185 . [Z . of all the modes as a function of different configurations may be expressed in rootlocus plots on a complex plane. corresponds to a mode of oscillation and can be expressed in the form (2 (4b) d 0 d1 2 a  Re(s) + jIm(s) (7) The variations of Re(s) and Im(s) The detailed analytical derivations for the mathematical model shown in Fig.Utilizing the Lagrangian equations. #. itxtrl Eof on Eq. (5). a a a C CK d Kd K + K d a + K d2a (4a) (6) vanish. 2 may be found in Ref. e the derivation of Theome Q5) th deiato aerodynamic forces Qi aerodynamic involved in the above derivations were based on the quasistatic aerodynamics. 2 are given by d ' E + (2) 8" C j c dt(~ q + aqj ilexternal if Silaerodynamic + oill y (5 is the generalized aerodynamic or the external applied force associated with The this A on used in the generalized coordinates qi" generalized coordinates analysis were 8c' 8. [7]. and [K] are formed from the coefficients in the governing differential equations of motions. The sectional lift and drag acting on the blade element are expressed in the form (see Fig. again. [C]. The governing equations from the derivations may be expressed in matrix forms. the equations of motions of the dynamic system shown ia Fig. The elements in the matrices [H].xf and constant rotating force is y *y. it becomes the shaft so as to produce a constant driving torque and this was the base in rnalRef. The characteristic roots a are those which where C£ and Cd are assumed in the form make the determinant of the coefficient Each elgen metrix in Eq. or the characteristic root. value s. 2b) foci 2 (s[H] + s[C] + [K  01 (6) dL iPcU C~dr (3a) y 1 dD  2 cU Cddr (3b) This represents a typical complex eigenvalue and eigenvector problem. they may be found in By taking Laplace transform [71.
occasionally the variatio~n of Im(s) is also needed in order to define whether the stability is a dynamic or astatic divergence case. However. 5 Nacelle Frequency as a Function (W 0 /In) 18j .. The roots which are located on the imaginary axis are corresponding to neutral stable convariation of Re(s) alone assa function of different configurations is enough //MMI 0.. In physical terms. 3 shows a typical rootlocus plot for blade bending frequency ratio 11 1.05 . 7T. 4 and 5 for different n1 values for illustrating * purpose. i. OARHRLN VNIMM/ MSCKVRD11IN 0. BACKWARD) WHIRLING 0 0046 ~ 02s1 Is 000 STA LE o UNSTABLE ___ 04 0.1 0.O j00 Fig.4 . 771 7. Fequency Bending Rati Plot for Blade 1.17 .001 . 3 RootLocus Fig.e._______________ ~ ~ ~ ~ 4 . 4 Exponential Decay Factor 6 as a Function of Nacelle Pitch Frequency for Various Blade Bending Frequencies Any particular configuration then may be checked for stability in the preliminary design stage by utilizing these results.0 WHIRLING0.0BldBednFrqncs PITCH FREONCY 0orjir0 u Fig. the damping value associated with a particular mode hsbecome negative.0) ~ N / / X"s to define the stability boundaries.10_________________ 0. roots enters the positive real axis *A_______ plane.00 ________0. Fig. when____ both Re(s) and Im(s) become zero.01 NCLLPIH ~ ""~c ~~/ FREQUENCYU#.4R 00 ___ STBL _O. sufficient variations in parameter for structural configurations are necessary in order to define accurate stability boundaries. 0.0FORWARD 0. These are shown in Figs.p r. Furthermore. YK Z ''~/ aK YON1'1/"""01/1 IACILL S 0.
One obtains B. i.. SPECIAL CASE  IDEALLY RIGID BLADES [K] S f)(8) 10) Ideally speaking. The ratio of natural blade bending fraquency to propeller rotational speed. (5). PARAMETRIC STUDIES istic roots of these equations. equations of motion represent the "classical" whirlflutter phenomenon of the twodegreeoffreedom propellernacelle combinations. parametric Static divergence of a propellerrotor system may be considered as a limiting case of whirlflutter phenomenon by conaidering only the static restoring ability from the structure.e. (6). Referring to the set of governing differential Eq. explicit expressions for stability boundaries may be obtained. as a design guide. an infinitely rigid blade is not possible in practice. (5) one may reduce these y A necessary and sufficient condition to have a nontrivial solution for the above equation is that the determinant of the coefficient matrix [K] equal to zero. This was done by assuming A practical design for propellerrotor systems would have physical properties such that their parameters were in the region bounded by the uppermost and lowest boundaries in the stability plot. 1 ~187 .. However. this condition provides us the uppermost bound in defining whirlflutter stability boundaries. By studying the character IV. This may be visualized by letting the "natural frequency" of the system become studies were made In order to define these whirlflutter stability boundaries.. n. C.A digital computer program 112] developed for the Naval Ship Research and Development Center (NSRDC) was used to solve Er.e. (9) was utilized in the process of mapping the static divergence boundsries. zero or by omitting timedependent terms in Eq.. These two Eq. for the case of ideally rigid This may be accomplished by I[K]I = 0 (9) assuming that the blade root constraint stiffness or the blade fundamental bending frequency is infinite. in both pitch and already been discussed as "Ideally Rigid ThereBlades" and "Static Divergence." fore. and its variations form a family of stability curves that are the results of such studies. is identically four equations to a special set of two equations blades. i. SIATIC DIVERGENCE The limiting boundaries have that the mass moment of inertia of nacelle was the same yaw directions.
83 0.2 STATC DIVERGNC 8W~ RO * IC6N(atN~tC " CO. result...) 0. The stability boundary curve with n. a set of typical input CA.73bending 4 deg 0 The stability boundaries for the case of ideally rigid blade and static divergence were also evaluated and presented in Fig.4.in& 3tiffneas requirements.73 5.  0  Value Variable Variable 0 0.N4 represents a large highsoced SEV'system. rion shown in Eq.A 01? Ia ye N4 h/R 0.N4. 6 .. 04 0. 6. r and 188 '1 I l yrk  .N4 is used in this report to define whirlflutter stability boundaries and are summarized in 0. values resembling the propellernacellepylon configuration on SR. SIATIC VIIGCCOOUNARI$ / would not vary too much from those of SR. It is believed that design .3 l NACELLE PITCH FREQUEFNCY ftATIOq.ayvmum freedom to fulfill nacellesup. . equation as functions of n8 . value for blade frequency ratio n is found to be approximately 0.1 one can evaluate the variation of eigenvalues of different configurations and hence define the stability boundaries. By examining the Kd a a a an optimur. .Rotor Whirl Flutter Stability Boundaries for Various Blade Bending Frequencies K K d1 3 0._... . Note: Assume 1 1 X I Y Using the input values from Table 1.0087 0.Propelle. 6. values for the propellernacellepylon system on a larger and faster vehicle I . Solving this x . Therefore. and its structural configuration will probably resemble one of an anticipated design. In a similar propellerrotor configuration with a value of ng less than 0.: UNSTABLE N T1EI~i z 01 0.. a sudden reversal from the backward whirl instability to a forward whirling would be developed. The results are shown in Fig.2 " '1 0.s  STABLEi tTABLE i ley 1.4 encloses almost the largeat stible region and allows a designer thp .4 T ) Ta b l e 1 .4 5.~ Fig._.The propellerroto'rsystem on SR.0216 0.. 0. one may form the matrix equa(5). TABLE 1 Nacelle and Blade Parameters Used in Defining Stability Boundaries Parameter T1 0.4.
these figures * actually represent the locus of the motion of a Propeller hub.N2 or SR. the response due to any initial disturbance could be represented by some harmonic functions with constant amplitudes.N3 and The propeller cushion vehicles.N3. the results should be close e:. COMPARISON OF THEORY WITH AVAILABLE EXPERIMENTAL INFORMATION ON SURFACE EFFECT VEHICLES A physical interpretation of a propellerrotor configuration for dynamic stability is the transient response due to some initial disturbances. /. 6. This was based on the assumption that the nondimensional paramters used for stability apalysis those shown in Table for SR.no x y this line represents an These five test performed by NSRDC.N3 were obtained from Ref. 189 .* The stability boundaries of the isotropic mounted nacelle.e.2825. How ble system. it is very difficult to make any rational precise experimental evaluation of the applicability of the theory to propellerrotor systems.5 alfor SR. Nevertheless. with the limited structural information available SK.4. In these plots. the third one is on the neutral stable boundary. This may not necessarily be true. and the last two are in the unstable region. use n.3. shown in Fig. The results configurations on SR. SR. and 0. the Xaxis represents the experimental confirmation of the vklidity of the stability boundaries shown in Fig. Numerical evaluations were performed on a PACE Model2312 active analog computer urations to illustrate phenomena. For a sta Since test results and structural information on existing SEV are not readily available. A special case is the neutral stable condition. i.5. 6 is still lacking. 0.25. I line.N2.V. and SK.N2. The values of nacelle pitch and yaw natural frequencdes.ed from a vibration shake define the stability boundary and pick five points along the n In practice. 6. = 1 to values of the same parameters on SK.N2. for the purposes of preliminary evaluation. were obtained from the interpolation of those shown in Fig.N3. and SK. and the propeller blade fundamental beniing frequency onf SR. The only experimental pitch response and the Yaxis represents the yaw response. 0. however. Heuce. nacellepylon on SR. for five different propellerrotor configthe above ever.2. The first two cases are in the stable region. 7a7e. the amplitude of response would damp out in time. the following evaluations are made.N3. TRANSIENT RESPONSE ANALYSIS VI.5 were obtci!. 8 as well as those shown in Fig. These figures clearly show the backward whirling phenomena as well as the degree of stability for each configuration.N2 is essentially the same as the one on SR. SR.. were expressed through the usage of an XY plotter and are shown in Figs. 0.01 in the pitch direction was assumed throughout. 8.. but in an unstable system the dmplitude increases with time.ugh to give a designer a rough guidance. [I].5 were the same as 1. SR. A ]lagcn at NSRDC performed the SK5 vibration shake test in March 1971. The Referring to Fig. pointspropellerrotor y) 0. An initial angular velocity of 0.
        4 I $ . a 4 v 190 .4C  0.0 I ION SPaY Ma CC A4 0 __ _ 44 a 4 u 4w 4.1 ~ 4_ __ (44 ~0 1 I 14.
N2 or SR. 10 shows the responses in the same directions due to some random excitations.N2 and SR. In an evaluation of transient responses due to an initial disturbance and the performance ot a response analysis due to some random excitations were undertaken through the use of analog computer simulation.N3 has a ilightly stable one.5. SK. frequency range 0 to 50 Rz. Fig. the data point associated with SR. The forcing at tively small amount of damping in the system and Fig.N2.?L hise propellerrotor conilguration and SR. The above quoted statement clearly describes the propellerrotor flutter whirling phenomenon and the actual occurrence on some SEV. Experience so far has disclosed appreciable vibration only in the loweat modes. . however. (111.) uSTABLE . AR 1$7 How stable a system io can be judged by ~Ithe _ distance between the data point and its corresponding stability boundary S CA . 02 NAC 03 . . some of both the lateral and vertical modes the angular deflection at the propeller is considerable.T TDIVIRCCSCE __ .N3 were observed. Both SR.01. A ~e / toe ALLY RIMP KIAN . 111].. a . 9 shows the transient responses in both pitch and yaw directions due to an initial angular velocity 0. 10 also includes this particular noise output which has a rus value of 3.2 or SR. SR. and 3 JB cutoff.T STAIC DYEUN~CII UIS 1 and the condition is not serious . Fig. i.N3 Air Cushion Vehicles and nacelle vibratory motion on SR.16. Here the random exci tations were generated by attaching a Gaussian noise generator to the analog circuit at the point associated with the nacelle pitching angular velocity. 191i .N2 and SR. As the propeller is rotating at high npeed. icurve. or precessio.PropellerRotor Whirl Flutter Stability Boundaries for SK. 9 shows the rela The propeller is obviously the main source of forcing in these modes . 8 .N3 pylons can be seen to nod gently when running at idling conditions.5 has a very sta02 UNSTABLE* 0S.N3 is very close to the neutral stable line Fig.N3.. The pylon situation is further complicated to the extent that not only will it have modes in the lateral plane but also in the vertical plane. results the author could find were some observations on propellerrotor vibration in Ref. In order to have a better feeling of the actual physical behavior of the propellerrotor system on SR. p. PITCHREQUENYRAIO As one can see from Fig.N2 or SR. and SR. Fig. due to the gyroscope effect . 8 shows the same behavior. . 10 shows the actual response due to some random excitations these conditions is very low.e. . . 8. . (Ref'." II \. in one plane will induce a force in the plane at right angles. Fig. Fig. angular motion. 219.
9a Nacelle Pitch Response _.Analog Simulation of Propeller 4 iub Motion due to In ..N3 tend to agree with theoretical predictions. 10c  L L i Nacelle Yaw Response Figs.N3 Air Cushion Vehicles the observations in Ref..N3 Air Cushion . blade fundamental bending frequency plays an important role since it affects the stability mapping. Fig.. above.NACELLE PITCH FREO'JEOCaz q#0. A propellerrotor whirl flutter theory has been applied to pylonnecellepropeller configurations on surface effect vehicles.04 ili  0 t0 20 0 0 0 to go 10 to 0.Analog Simulation of Propeller Hub Motion due to Random Excitation for SR.N2 and SR. The classical whirlflutter Fig.i.1~tt~ N'~fl ~ l. some observations on SR.N2 and SR.mUI0. DISCUSSION SR.. These results agree with [11] quoted confirm the theoretical computations. Both Fig.N2 and VII.011 Fig. however. 9 . Wt. . .. 9 and 10 clearly illustrate the relatively large oscillatory motion in the nacelle pitch direction in comparison with yaw. based on a typical propellerrotor system on large SEV were evaluated for the purpose of preliminary design and guidance.i'a[4. No experimental data were available to theory in which blades are assumed to be infinitely rigid gives the most rigid and 192 NI. VehiclesIi1 which always exist in reality.. lOa  Random Exciting Forces 10 to 0 30 4 0 0 i I F 10 Fig. 10 . lOb .tial Disturbance 0 10 20 0 30 60 ?0 lb go 10 for SR..4..1 MACILL I YAW FREQUENCYs II AA^ IlFig.N..t'. 9b  Fig..1 tIj 0. Stability boundaries As indicated in the evaluation of stability boundaries.Nacelle Pitch Response Nacelle Yaw Response 0.
and unsteady aerodynamic loadings. Caution should be taken in the consideration of fatigue failure even Transient analysis by analog computer simulation gives a direct physical interpretation of the actual whirlwhen the propellerrotor system is operating i the stable region. a 0. Optimum fundamental blade bending ever.N2 and SR. increasing nacelle mounting stiffness to increase nacelle natural frequencies is another way to move a particular propellerrotor configuration from the unstable region into the stable zone. flutter phenomenon. How the theoretical analysis used in this report.overly conservative stability criteria. gives overconservative stability criteria which impose the most severe design requirements on the naceile supporting stiffness. underway test on a large SEV. value of 0. however. 6. Observations on SR. As implemented in Fig. 1. RECOMMENDATIONS In view of the importance of a sta due to initial disturbance and responses due to random excitations can be considered as the results of gusts and turbulence in the atmosphere or due to the irregularity in the ground/water surface. Study the effects on the change of stability boundaries as functions of 193 . and when the blade bending frequency is below this value.4 in order to impose the winimum requirements on nacelle supporting stiffness for a stable proptllerrotor configuration. Other dynamic con frequency with respect to propeller rotating speed should have e. An optimum design guidance. Observations of the nacelle motions on existing SEV tend to confirm value n. the optimum value for blade bending frequency is restricted in the sense that it would allow a designer the most freedom to choose nacelle mounting stiffness in order to avoid whirlflutter instability. 2. and compare the test results with the theoretical computations in this report. I. Perform experimental evaluations by using a model with the same nondimensional parameters used in this report. Perform a vibration shake test and aignificant for fatigue failure criteria even though the system is operating in the stable region. Analysis performed by applying the whirlflutter theory on a typical propellerrotor system of a large SEV resulted in a parametric chart for blade second bending mode. CONCLUSIONS 3. 4. These loads may be ble propellernacellepylon system on large highspeed SEV. leading to overdesign and possibly uneconomical construction. Classical whirlflutter theory siderations should be also included in overall design. They are in connection with dynamic loads and fatigue failures. forward whirling as well as backward whirling'could develop. 3. Transient responses IX. 2. Direct experimental confirmation is not possible at this time.N3 tend to strengthen this consideration. VIII. tht following statemenhts can be based on this analysis. and insufficient experimental data to substantiate the theoretical analysis of such a system. Extend the existing theory by including blade first torsion mode. 4.4 was found. the following efforts are recommended.
J. 2124 Apr 1969 Gaffey. 1963 4.Y.' paper presented at 34th AGARD Flight Mechanics Panel Meeting. Niblett." J. G. "A Graphical Representation of the Binary Flutter Equations in Normal Coordinates. T. H. The system also included an equivalent rotation spring at the hub. M. Aero. R.. "Vibration Isolation of Aircraft Power Plahts. propeller power setting. Lytwyn. J. Taylor. "Aeroelastic Problems of Flexible V/STOL Rotors. etc. Mr.. Buffalo. Early in your presentation you discussed the degrees of freedom you were admitting to the problem in reference to blade bending. Mr. Sep 2325. G. W. Hall. 2. Kvaternik. 11. No. E.H..... MarApr 1969 9. 4349.S. V/STOL and Ground Effect Machines. oubolt." Symposium on the Noise and Loading Actions on Helicopter. H. G.. the blade plane moves as a rigid body. III. E. N. J.. 6. are they not? 194 . I meant the flapping motion of the blade." CALTRECOM Symposium on Dynamic Loads Problems Associated with the Helicopters and V/STOL Aircraft. Vol.. H. L.. 333347. R. III and Bennett. 1970 England.. 1968 12. 3. pp. 6. A. T. and Devereus. Jan 1966 7. Southampton.... Inc." Royal Aircraft Establishment Tech. and Browne. E. I made an equlvelant system by considering the blade as rigid. Zudans: (rranklin Institute): Because of the yaw and pitch of the nacelle. K. T. 1969 REFERENCES 1. 1966 8. Rept.. No.. Apr 1966 DISCUSSION Mr. K. Jr. Aug 30 . Edenborough. Mr." Paper presented at the Air Force V/STOL Technology and Planning Conference. "PropellerNacelle Whirl Flutter. Elsley. Vertol Div. "Propeller Rotor Whirl Flutter: A StateoftheArt Review. Mar 1962 III." MacNealSchwendler Corporation Project Report. E. Did you not mean blade flapping as a ridgid motion? Mr. 66001.S." J. 5. "Propeller Whirl Considerations for V/STOL Aircraft. We do consider the blade ae an elastic blade but restrict it to the first bending mode. Sci. W. and the bending or flexing frequency was the same frequency as the first bending mode. Is it not? The blades themselves are treated as rigid. "Analysis and Model Teats of the Proprotor Dynamics of a TiltPropotor VTOL Aircraft. Vol. 6. Gayman: That is associated with the oscillation of the plane ofthe propeller disk. Hovercraft Design and Construction.A. Was M'at motion included in your hydrodynamic forces? Mr. but hinged at the hub. and Reed.C. Las Vegas. nacelle length. but only restricted to the first bending mode. blade inertia.." I. Cornell Maritime Press.. Jun 2627. Reed. Yen.H. Rept.. pp. Nevada." The Boeing Co.. Liu: Yes. Dec 1938 2.. "Investigation of TiltRotor VTOL Aircraft RotorPylon Stability. and 10. W. R. Tech.important nonClmensional parameters such as nacelle damping. Llu: Yes.A.Sep 3. Reed. Study the effects on the change of stability boundaries due to the overall structural dynamic characteristics of surface effect vehicles. "PropRotor Stability at High Advance Ratiou. H. Liu: Perhaps I did not make the point very clear.M. Brindt. S. A. 5. D80L95.. W." Journal of Aircraft. "PropellerRotor Dynamic Stability. 1965 5. Gayman (Jet Propulsion LaboratorX): I ask for a point of clarification. "SADSAM V User's Manual. Peterson.
The displacements of these nodal points are then the basic unknown parameters of the problem. T The theoretical background and &ssufnptions leading to the development of Matrix Equation (I) through the use of the finite element methodology is well known and not presented herein. Once nodal points and structural elements have been defined. Each row of Matrix Equation (1) repreaents a particular degree of freedom that. Evidence of this is the large number of general purpose finite element computer programs in use today.to solve dynamic response problems having timedependent displacement. These elements are assumed to be interconnected only at a discrete number of nodal points situated on their boundaries.ion problems is well documented. This matrix equation is then rearranged and partitioned to separate constrained and unconstrained displacement degrees of freedom. the solution of this discretized continuum elastic undamped dynamic problem becomes the solution of the hiatrix equation characterized as: [N) (A(t)) + [K.(F(t)) where IN) roC (F(t)) (A(t)) = mass matrix . then by standard matrix manipulations. Workman Battelle. highspeed digital computers and the increasing needs of the aerospace industry. INTRODUCTION Over the past decade and a half. the continuum is separated by imaginary lines or surfaces into a number of "finite or discrete elements".THE DYNAHIC RESPONSE OF STRUCTURES SUBJECTED TO TINEDEPENDENT BOUNDARY CONDITIONS USING THE FINITE ELEMENT METHOD George H. This example is used as a vehicle to demonstrate the versatility and effectiveness of this solution technique. The dynamic response of a bellows subjected to dynamic edge displacements and internal pressure. This includes both zero and '195 . complex structures. Two excellent books describing the finite element method have been authored by Zienkiewicz (1 ) and Przemieniecki( 2 f. The capability of this approach to obtain the dynamic response of compcC structures to known forcing functions and base mo. Once the general matrix equation has been properly partitioned. Ohio The dynamicmatrix equation of motion characteristic of structures modeled by the finite element method of analysis is vritten in general form. can be described as being associated with an unconstrained or constrained displacement. This paper presents a straightforward extension of the finite element approach. as determined by this approach.(A(t)] . for this study. the finite element method has emerged as a powerful tool for the structural analysis of large. with the advent of highcapacity.stiffness mtrix . The form that these discrete elements takes depends on the type of structural behavior assumed and the form of the approxication to that behavior.nodal timedependent forces . is presented. Displacements as used herein represent both translational and rotational motions.nodal displacements (1) At) nodal accelerations. Constrained displacements are those degrees of freedom for which the motion is known as a function of time. MErHOD OF ANALYSIS In the finite element method. Columbus Laboratories Columbus. the original mixed boundary value problem is transformed to a modified force motion problem.
This program uti 196 . respectively.  Fig. atrx Eqution (2) can then be expanded by rows and rearranged to yield [Mid ] (Ai) + [K id [L) . This system of equations can besolved by a variety of methods depending on therequirements of the problem. An earlier study of this bellows was conducted utilizing a continuum axisymmetric shell program as the basic analysis tool. Biggs(4 ) and Bisplinghoff..iC. Matrix Equation (4) yields the reactive forces at the constrained displacement. Once the t . nonzero motions. 'Therefore.history solution of been accomplished the tiatrix Equation (3) nodal displacements of the discretized problem are known. Ashley. t . The bellows is of uniform thickness. and 40 conical elements. (3) andIh Fj = I [Kj1) [Ai + [M 3 j + + CKj) IAj1 (4) k &W* N'ID Matrix Equation (3) is in the standard form associat. The top edge of the bellows is attached to the stationary vessel and the bottom edge is attached to the moving piston used to generate a pressure pulsation in the vessel. L. Three excellent books giving a number of these teihniques have been authored by Hurty and Rubinstein(3).[Kij3 (A J).0. Matrix Equation (1) can be rearranged such that those degrees of freedom associated with the unconstrained displaceQents and those associated with the constrained timedependent displacements are partitioned as shown below: BELLOWS ANALYSIS A dynamic anilysis of a bellows was conducted utilizing the method of analysis given above.0.3.062 inch. and is constructed of and sheet : E a 29 x 106 bin poisson's ratio . Figure I shows the bellows cross section.d with the elastic undamped multidegreeoffredom dynamic problem with a slightly modified righthand side to include the effect of the constrained timedependent displacements. L is  ' 4P [Fi) Mj [ . tp.]i J] J Fsteel (2) (2) K where subscript i is associated with the unconstrained displacements and subscript J is associated with the constrained displacements. I . accelerations and velocities of theme consuriined degrees of freedom are also known. Unconstrained displacements are those degrees of freedom for which the applied forces are known functions of time and whose resulting displacements are not prescribed. The bellows serves asa pressure seal between a vessel and a piston. so that the exact shape of the bellows in its neutral position was reproduced quite accurately. This shape was modeled by seven toroidal parts. and llalfman().Cross section of bellows used for vibration analysis (neutral position) Three finite element models of the bellows were created. 30. Each part was described by its 1 11 Lii ~J K IK J 1 + major and minor radii and the angular coordinates. The basic conical element used in this analysis was similar to thet developed by Grafton and 6 Strome( ) except numerical integration was employed to generate the element stiffness matrix. The resulting time history of the stresses and strains in the individual elements can be determined from the assumed behavior pattern of the individual elements. These models contained 20.
4 for .58 440. Also.I lized numerical integration of the governing continuum thinshell ecustions. eigure 2 shows the node locations and elements for the 30element model.44 30Element 44. deflection shape for first axisymmetric mode of bellows 197 . the modulus gf elasticity was taken to be 27. 4 # : 2197 004~ R4Cor. 2 30 Modal element model of bellows . respectively. 3 Fig.03 318.14 Utilizing these natural frequency data it was decided to use the 30element model for the dynamic analysis of the bellows.38 354.he bellws material and was assumed to 29 x 100 lb/in.20 376. \h \30 S.49 Continuum Model 43. The two lowest axisymmetric natural frequencies of the bellows were calculated for the four models assuming both boundaries fixed.96 318.38 40Elemen]t 44. Table I gives the results of the natural frequency prediction along with the frequency determined experimentally for the first mode. 21 204 4 4 24 ? 0 I 7 7 4 7 24 Fig.. the next four natural frequencies of the 30element model were determined and are given in Table 2. Frequency.96 2 3 4 5 6 TABLE I Natural Frequencies of Bellows Model Finite Element Muel 20Element Mode 1 Mode 2 45.4 for this study. The finite element models predicted natural frequencies slightly higher duc to the difference in the assumed modulus of elasticity.17 398. TABL9 2 Natural Frequencies of the 30Element Model Mode No.*. Figures 3 and 4 give mode shapes for the first and second natural frequency.2 311 Experiment 44. The meas of the bellows was luIsped at the nodal points for the finite element model and was evenly distributed along the length for the continuum model. cpa 44.95 318.96 318. Unfortunately.4 x 106 lb/in.
Biggs. and Rubinstein. and llalfman. M.5 to 1. 1968. Dynamics of Structures.80 inch. S. Mcrrawllill. 24Z Fig.C. Figure 8 shows the time history of the vertical displacement of Nodes 8. P.14 % Ithe 15 s element techniques. modal .ssure and edge displacement as a function of time. this stress ocurred at Node 5 ir the 30element model. J. REFERENCES 1. J. C. Figure 7 shows the tim. McGrawHill. 198 . McGrawHill. Figure 5 shows the internal pi. required to incorporate the dynamic mixed boundary value problem as formulated herein as compared with that for a typical force motion problem for a similar application is negligible. N. for engineering purposes. R. 2. history of the maximum stress occurring at th bottom side of the bellows. and 31. Prenticellall. Figure 6 shows the time history of the maximum stress occurring at the top aide ot the bellows. Zienkiewicz.n Figures 6 and 7 along with staLic stressesi. 116 I I7 / 12 ..__ 2Z" 2o 21/ 20 o_ /0 __ould 2 eration solution was not plotted because It er notediscrnibl nuecaelt not be discernible om from te the numerical integration solution.( 7 The other was the modal acceleration method. London.ed ". Ashley. The numerical integration solution isplo2. Aeroelasticity. H. Przemieniecki. Experience has shown that. New York.itical damping in the range of 0. L..0 percent gives realistic results. 0. New York. L. The matrix rearrangement and partitioning going from Matrix '"quation (1) to Matrix Equation (3) is very straightforward when accomplished on the digital computer.. W. The Finite Element Method in Structural and Continuum Mechanics. dropping to 10 psi as a versine.Two different techniques were emp!oy ed for the solution of Matrix Equation (3). This small amount of damping has a minimal effect for the first few cycles and therefore was neglected in this preliminary study.. By comparing the numerical integration solution and modal acceleration with only the first mode employed. F. Introduction to Structural Dynamics. Theory of Matrix Structural Analysis.. 24. The internal pressure )f the bellows. H. 4. only first vibration mode makes any appreciable contribution to the dynamic solution. "' One was the straight numerical integration of Matrix Equation (3) by a f urthorder singlestep RungeKutta method. 1965. finite :s 4 "0 % /solution \. it was clear that. and the vertical displacement of Node 31 is constrained to move downward as erin a veraine o ovedowwad a a 31 s cnsraied with a period of 33 milliseconds and a maximum amplitude of 0. The modal accel gel. One pulse of the piston was examined and no damping was included in this analysis. The constrained displacements are that the radial and vertical displacements and rotation of Node I are zero. the radial displacement and rotation of Node 31 are zero. to . 1964. for cnalysis of this type. 16. / 4.. that occurred at Node 77 of the model. Bisplinghoff. both engineering and computer. 25 . lurty. This paper has given and demonstrated a very straightforward and effective method of determining the dynamic response of structures subjected to timedependent boundary acodbounariy condisuete dtoimeependnty tions. The dynamic magnification of the statically calculated stresses is quite clear when examining the plots given in Figures 6 and 7.. This capability can be easily incorporated into an existing finite element coniie emn dnmi eady p pote proa puter program already progrotoed fo dynamic force motion problems. 1967.. 5. New Jersey.(5 ) Within the modal acceleration method only the first mode was employed. The additional time.S. Englewood Cliffs. A. The static displacement component ~~4 /3 4 wasn determined utilizing standard static. initially at 80 psi. is in phase %*ith the edge displacement. Addison 3. 4  Modal deflection shape for second axisymmetric mode of bellows The dynamic problem was formulated along standard finite element techniques for the forming of Matrix Equation (1)..
nilsecods "X_ ' 40 Dynomlc a ' . Time history of maximum at Node 5 ~d.¢.* **" " Fig. '' *' :Y'rv :/d. E.'!. D. Hassachusetts. . 1965... P.'. 1.' "". Reading. 7.Time history of piston displacement and vessel pressure 40 / Tkne. 5 .. I Fig. i .. '". pp 191202. New York.A. A First Course in Numerical Analysis. !'" l" L' ' ''' '' ' ' stress "'" '" . Vol. 1963. Ralston.t "" . 6 ''. and Strome. pp 23422347...'. R. McGrawHill. 1955 6. Grafton."'' . Journal A. A.t'/"d r 199 .A. leo 000 g0 10 / Tim milliseconds 39 40 so s i5 0. ' ..l. "Analysis of AxiSymmetric Shells by the Direct Stiffneas Method".Wesley.
miliseconds . R1 \% \.DiNe % Node 16 aNod 24 Fig. 7 .ICI D. 8 4of Time history of vertical displacement Nodes 8.€ Fig.16.Time h~stor fmaiu trs t oe2 Time.24 and 31 200 .
Schrantz (Comsat Laboratory): You said you used conical elements to define the bellows? Mr. That mode finds the dynamic component. It is a shell of revolution. actually it is a torold around a circular piston rather than a flat bellows. Workman: Rightl I used a series of conical elements. that boundary appears to be fixed. 201 . not the static displacement.. Workman: Yes.ed the equation after It is rearranged t. Mr. Workman: No. Mr.tA. If you use the modal method on that equation. Zudans (Franklin Institute): You mentioned comparison between modal solutions and you aid that the response was entirely in the first mode. so you must have done something in addition to that. Voice: Is this a toroidal bellows? What form is it? Mr. the third slids . Then the displacement is used in the static solution.DISCUSSION Mr. Yet the mode you showed was incompatible with boundrydisplacement. get it into the form that is solved.
I
VIBRATION ANALYSIS AND TEST OF THE
EARTH RESOURCES TECHNOLOGY SATELLITE T. 3. Cokonis and G. Sardella General Electric Company Space Division Philadelphia, Pennsylvania This paper presents a unique approach used for the launch vibraticn analysis of the Earth Resources Technology Satellite (ERTS) and compares the analytical results with experimental measurements. The ERTS is basically a modification of the Nimbus vehicle with solar arrA7 paddles unchanged. The complex paddle system could best be represented by measured data obtained from previous Nimbus modal testing. The successful extraction and subsequent recoupling on ERTS of the solar array paddle modes from the original Nimbus experimental mode shapes is given. The analytical model is described along with its verification by an abbreviated modal test. Good correlation between test and analysis was evidenced by frequency and mode shape comparisons. Some areas of discrepancy in the analytical model were uncovered which were subsequently modified to improve the analytical representation of the spacecraft. INTRODUCTION The Earth Resources Technology Satellite (ERTS) system (Figure 1) is being developed by the Goddard Space Flight Center (NASA) for obtaining data from near earth orbit immediately applicable to the determination of earth resources and their management. The spacecraft configuration will be launched on a Thor/Delta N9 booster. The structural subsystem is based on the flightproven Nimbus satellite using the same
philosophy of design. The greater weight of the
in mounted on an adapter which is bolted to the adapter Spacecraft ring. adapter launch vehicle separation occurs at the lower ring flange of the sensory ring. During the evolution of the Nimbus spacecraft from a 900pound configuration to the 2000pound ERTS configuration, the dynamic representation used for loads and clearance studies was based on an extrapolation of modal test data on Nimbus A. Analytical c" mparisons with
test results indicated tbat the
t
ERTS spacecraft and specific ERTS mission requirements necessitated structural modifications and some redesign. The ERTS spacecraft structural segments are shown in Figure 2. These segments are: 1. The Attitude Control Subsystem (ACS); 2. The Solar Array Paddles; 3. The ATruss Structure; 4. The Sensory Ring; and S. The Launch Vehicle Adapter. Structural differences with respect to the Nimbus were primarily in the center section of the sensory ring assembly. New structure was designed to accommodate the ERTS payload and other new equipment, Sensory ring and truss structure was modified for increased stiffness and strength. In the launch configuration, the solar array is folded along its longitudinal axis to fit within the sensory ring envelope and is secured to the sensory ring structure by a latch mechanism. The spacecraft
extrapolated modal model no longer predicted the dynamic behavior of the spacecraft with the required degree of accuracy. This was evidenced in the flight loads analysis for the POGO condition using the extrapolated model which predicted extremely high loads due to high coupling between the lateral modes and the axial excitation. To overcome these deficiencies, a new analytical model was developed. For an analytical model c the spacecraft the representation of the solar array becomes critical. Thir complex solar array was found to dominate all of the spacecraft modes. If finite element representation was used for the solar array, it would be difficult to derive and its accuracy would be questionable. This consideration was avoided, however, with a modal coupling approach using measured mode shapes. This required the extraction of 19 solar array
Preceding page blank203
L a
complete spacecraft (less solar paddles) matrix. To complete the model, the paddles were added by the modal coupling technique. Nineteen paddle modes and 11 spacecraft modes (less paddles) were coupled to produce the final set of complete spacecraft modes and frequencies. k' I. Spacecraft. The launch vehicle adapter structure was represented using over 160 internal joints in the MASS digital computer program (reference 1). These joints connected the S bam and panel elements used to represent the adapter structure. The resulting stiffness matrix was reduced to nine external joints which were selected to coincide with the sensory ring
external joints. The sensor, ring structure in
Figure 3 was modeled also using the MASS program. This representation consisted of over 200 joints connecting the beam and plate elements. The resulting stiffness matrix was reduced to 18 external joints common to the adapter, the truss, and other mass points of interest. In a similar manner, the stiffness matrices of the truss and the ACS structures were also developed. The complete spacecraft structure, less paddles, was then assembled by the stiffness coupling of each segment. The fixedfree natural frequencies and mode shapes were calculated using this stiffness matrix and the mass matrix given by the distribution shown in Figure 4. 2. Solar Array. A very accurate representation of the solar array paddles was possible since measured mode ohapes could be utilized. These measurements were taken from a previous modal survey of NIMBUS A (4) according to the pattern given in Figure 5. Since the ERTS spacecraft uses the NIMBUS solar array with minor modifications, these data were considered to be appropriate for the ERTS model, The extraction of the 19 paddle modes from the NIMBUS modal survey started with the formulation of the characteristic matrix vibration equation: n xn n x m [K) ] n x n n x m with [01 normalized to, [9] where [H) (9] = [tJ
Figure 1.
ERT8 Spacecraft Vibra, in Test
normal vibration modes from Nimbus spacecraft modal data. Such an approach is valid since ERTS uses a solar array which is structurally unchanged from Nimbus. Paddle frequencies (starting at 13 hz. ) and mode shapes having 154 coordinates were utilized for the new ERTS analytical model. Other major substructures were derived by a conventional finite element stiffness routine known as MASS (1). These substructures were then combined by modal coupling through a statically determinant interface in a manner similar to that used by Hurty and Hou (5, 8). The following sections of the paper describe the formulation of the ERTS spacecraft analytical modal and Its verification and evaluation by modal testing. Much of the modal dynamic analysis and testing activity conducted on ERTS!Nimbus programs parallels the welldocumented Mariner Spacecraft Programs (5, 6, 7) and have had a similar degree of success. DESCRIPTION OF ANALYTICAL MODEL
A,
m x
(
The analytical nodel of the spacecraft was developed by considering the major spacecraft segments separately. These structural segments are identified in Figure 2. Stiffness matrices were developed for the launch vehicle adapter, the sensory ring, truss, and ACS, and each segment was then combined to form the
(K) is a n x n square symmetric matrix of
[H) stiffness coefficients is a n x n square symmetric matrix of inertial mass items
204
so: :.::
.. L
AC ACS
ERTS/ACS
INTERFACE
ASSE 4BLT
L/V IhTERFACE Figure 2 E'ITS Structures Subsystem
A is a scalar of resonant frequency squared  2 [9] is a n x m modal matrix of eigenvectors (mode shapes) listed columnwise tiaere m < n [A3 is a m x m diagonal spectral matrix of eigenvalues (A, resonant frequencies squared) listed diagonally where T [ jIJ is a m x m identity matrix m refers to the number of modal degrees of freedom (modal courdinates) n refers to the number of generalized displacement coordinate degrees of freedom Equation I can be manipulated to derive the stiffness matrix if trices are known (2, 3). Post multiplying by: T the modal, spectral and inertial ma
C*]
[K] [9] (4 T [M)

[14]
[M4]
[
4
[Aj (M)
(2)
and utilizing the normalization (9
I"[t
[I]
(9JJ

[9]
11
 [4]
(3)
regrouping
205
2i
[]
!.]
1 T
[H]J
E
it)I
[I]
1#1
gives the identity []T [H]
which simplifies equation 2 to n x n n xn nxm
[K]
=
(J
[*]
mxm
mxn
nxn
[,HA(
[MJT
(H]
(4)
Utilizing equation 4 nineteen constrained solar array paddle modes were determined from the original measured modes of the Nimbus Spacecraft by applying constraints at all non solar array coordinates. The derived stiffness matrix was partitioned as folloit:
KI
K2
K22
XI
X2
[
0
M
O
22
XI
[21
Let
where:
X 1 ,
=
1
2
Y4
K1 1 is the desired cantilevered paddle stiffness matrix X is the generalized physical displacement vector ( )1 coordinates being retained; i.e., solar array )2 coordinates being restrained, i.e., spacecraft
12
I
K21
1
,
o
]
s]
I 1
,T
I [1 T
T
rl
I o
(I
(6)
2 T
K22]
=
[
N
M2
2'2
] A *
i *1 T H
T
[
0
H 1¢ 1 2
0
A
l
2J
F0
01
M2 j
1
[1
h .t2
1*T~.
2 1 h 0T~
01
T 2 M
;
o
A
2
2
MM2
(7)
(8)
2
"
2
A
Transforming Equation 5 in expanded form
[
1
Ml
A]
A1 A 1T
A
H 1
i
A
0T
*
02 T
2
i 2 [1 0
1
12 2
MIV2
42
A
*T #
N 1
* rhapes
A
~T
Y
,
m 1
*Jy I I#
ht Y1j'~
M
1
A
(9)~j
solving the above elgenvalue problem yields
and transforming
back to obtain the paddle
e'ode
I#ade
I
(10)
206
'Llak 4, 4L.1o Y O
. A total of ninety accelerometer channels were recorded during the resonant dwells. a simplified Hurty (5.. een modes nf the solar array were utilized in developing the complete system modes.8) attachment technique was used which entailed free mode coupling at statically determinate interfaces.... ..f . MODAL TEST DESCRIPTION Modal testing of the ERTS fullscale structural dynamic model was performed as a " FIGURE 4  SPACECRAFT ANALYTICAL MODEL part of the structural verification test... Basically. The primary objective of this test was to obtain fixedfree modal vibration data suitable for evaluating the analytical model of the spacecraft for use in the flight loads analysis..= . .. Having obtained the constrained modes..... .. . .z "t I 'NI I I I * I FIGURE 3  MASS COMPUTER MODEL SIMULATION OF SENSORY RING STRUCTURE (Crossbeam omitted for clarity) 3... This obtained the free modes. Modal Coupling.. The resulting eigenvalue equation form consisted of a coupled generall ... . These were located on the spacecraft structure in triaxial and biaxial groups at points corresponding to mass points I 4 207 .. . .. n I'. ..d mass matrix and diagonal generalized stiffness matrix with 30 degrees of freedom...... The free solar array and analytically derived ERTS spacecraft structures were then ready to be modally coupled to obtain the desired ERTS spacecraft vibration mode shapes and resonant frequencies. the statically determinant interface consisting of three translations and two angular reactions at the Solar Array Paddle Shaft and a latch line vertical tie to the sensory ring was released in the modal coordinate. Eleven modes of the structure and nin..... Sine sweeps for modal definition were performed with low level base excitation at a rate of 8 minutes per octave. This required the measurement of data sufficiently detailed to enable the response in the fundamental pitch.. roll and yaw modes and the lateral modes in the frequency range of the POGO excitation to be evaluated. .. The first twentyfour complete spacecraft modes ranging from 13 to 85 hz were subsequently correlated by vibration testing.. .
The measured responses for the nine modes obtained were considcred to behave as natural modes. The damping values obtained ranged from g =0. The paddles were instrumented at its extremities only since modal data for this solar array had previously been obtained. FIGUR 5  ACCELEROMETER LOCATIONS FOR PADDLE MDAL SURVEY NIMBUS Kof the analytical model. the ACS roll mode.81 hz. This required the measurement of the frequency of the peak inphase component of acceleration occurring just before and just after the resonance. Resonant frequencies for which dwells were not made were estimated from total response plots and are noted with an asterisk in the tabulation. This became significant in the higher paddle modes where the necessary coarse mass distribution was inadequate. These modes were then used for direct comparison with the analytical mode shape. These locations were selected based on analytical predictions of the paddle latch. This corresponded to the range of magnification factors obtained in the high level vibration test. Structural damping was determined by the method of Reference (9). p~il . From this group. Inphase and quadrature response plots referenced to the input acceleration were obtained according to the phase separation technique of Reference (9) for particular accelerometer locations. 4 208 E 4'V . This was bared on resonably good orthogonality shown by the evaluation of the generalized mass matrix which was calculated using the normalized experimental mode shapes. nine significant mode shapes were determined.. This is particularly true for the fundamental pitch and yaw modes. Since the testing was limited to the fundamental mode in each axis and ihle crossaxis modes in particular frequency tanea. The greatest deviation from the criterion was found in the higher paddle modes.jp dwell frequencies. 032 to 0. Frequencies of interest were identificd for resonant dwells. The average value of the offdiagonal terms for the nine spacecraft modes was 9.. These plots were of the type given in Figure 6 and they were made for all of .. Other frequencies were considered to be reasonably close to the corresponding calculated values. The test plan set a goal of + 1076 maxlmum coupling as evidenced by the offdiagonal terms for successful mode shape measurement. A comparison of test and analytical rcaonant frequencies is given in Table 1. 4 percent. the measured resonant frequencies are in good agreement with the corresponding calculated frequencies. Modal identification was established by •n 4 i2 1 i T 2" plots of the quadrature comppaent of acceleration.I I Illel 10161 TaftS pool$ W SMOM aIl 4 J l~~* I 4'4 I 4 I + A I 4 I I 1 I I 14ACJ4 V5W"A. 1 and the second paddle roll mode at 18. not all resonances had modal dwells. ACS and sensory ring responses. This result was anticipated since very few accelerometers were available for paddle response measurement. ''in '6r . (Previous Nimbus modal testing had recorded ZOO channels of data with over 100 data points on the solar paddles).4. In general. 12.
10 was assumed for all modes in the analysis. they could be traced to lack of p. better amplitude correlation'was obtained by varying modal damping coefficients. This was accomplished by the calculation of the modal shear for both test and analysis. CONCLUSIONS Analytically derived and test correlated mathematical models providc the most desirable data for use in ascertaining structural integrity of a spacecraft. The extraction technique was successfully demonstrated by the correlation with the modal test. L. especially in the lower frequency modes.A constant damping coefficient of g 2 C/Cc 0. There were larger deviations for those modes where the quantity of paddle measurements is critical.29 Hz made during the modal dwells restricted the scope of the analytical model evaluation. Refinement of the analytical model was obtained without significant changes in the frequency correspondence.lle measurements and not lack of validity of the mode. The limited quantity of measurements FIGURE 6  ERTS MODAL TEST PLOT FREQUENCY: 15. In addition to frequency comparisons and orthogonality checks. "MASS System . The ERTS/NIMBUS experience has shown that satisfactory vibration models of subsystems or complete spacecraft can be established by analysis. Also. Considering each area of comparison it was concluded that the analytical model was a good representation of the ERTS spacecraft. This resulted in subsequent modification and improvements in the model. however. Also. better response amplitudes were obtained by the selection of damping for individual modes. however. Beitch. Improvement was made in the mode shapes which was reflected in better correlation of the modal shear. as in the orthogonality check and modal shear coinparison. A need for variation of the damping coefficient was indicated. It was possible. This resulted in better modal shear correlation with test within + 16 percent. The results here showed the agreement between test and analysis predictions to be within + 29 percent for the primary response axis. Other possible areas of improvement considered as a result of the modal testing were in the mass distribution of the sensory ring and the representation of the launch vehicle adapter stiffness. I 209 . REFERENCES I. mode shape.as of discrepancy. Confidence in the model was supported by the good frequency agreement and the orthogonality checks. No criterion for this evaluation had been previously set. another evaluation of mode shape was made.The Computer Program for General Redundant Structures With Vibratory and General Static Loading. and amplitude correlation with the analysis. The basis of this conclusion is summarized in Table 2. test or a combination of both. Improvements in the amplitude response were suggested by the damping evaluation where a range of values were measured for individual modes. Limitations in schedule and cost. Where deviations from the criterion were large.. frequently force the dynamicist to resort to abbreviated test and analysis procedures. but the results were considered reasonable. The adjustnemts in the model were accomplished with insignificant changes in the frequency correspondence. to establish confidence in the model and to uncover some r. ANALYTICAL MODEL EVALUATION The analytical model was evaluated on the basis of frequency.
4%. 50% of all offdiagonal elements meet +10% criterion.76 62. c.71 47. & TEST Very good Good REMARKS Correlation within less than 12% in most modes. PADLdLES PADDLES. YAXIS X.27 39.00* 79. Frequency b. d.42 48. YAXIS X.RY.47 21.26 35.50* 90.38 63. PADDLE PADOLL. XAXIS X.54 66.50* 50.00' 24. ~ .92 53.00* 68. Amplitude Fair 210   .50* 30.53 40.81 17.00* 53. Measured daping ranges from 0.24 76.22 34. SENSORY PADDLE.48 19.77 38.10 for all modes. TABLE 2  ANALYTICAL MODEL EVALUATION BASIS a.57 55. YAXIS Y.29 18.81 24. overall average is 9. PADDLES SENSORY.72 15..00' 42. SENSORY ACS PNEUIATICS PADDLES XAXIS XAXIS X.00k 41.57 21.29 31.00* 35.IVII .T. XAXIS YAXIS ZAXIS Y. XAXIS ZAXIS ZAXIS YAXIS ZAXIS TEST NATURAL FREQUENCY CPS 13.03 to 0.12 according to mode. Those in excess lack paddle representation.13 18. Better correlation obtained by varying modal damping coefficients. 80% meet 415% criterion.. COMPARISON OF TEST AND CALCULATED NATURAL FREQUENCIES .50* 39. YAXIS X. XAXIS X.39 14.]BER 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1s 16 17 18 19 20 21 22 23 24 CALCULATED NATURAL FREQUENCY CPS13. XAXIS Y. SENSORY SENSORY.00' * APPROXIMATION FRO4 TOTAL RESPONSE PLOT.34 58. YAXIS YAXIS YAXIS ZAXIS ZAXIS ZAXIS Z.NU.TABLE 1.35 63.29 MAJOR RESPONSE REGION PADDLES PADDLES PADDLES ACS PADDLES ACS" ACS PADDLES SENS.IST SENSORY.09 33.00* 59. Orthogonality ANAL. PADDLES SENSORY SENSORY PADDLES.23 37. analysis used 0.24 43.03 29.24 58.00* 55.68 85. PADDLES PADDLES ACS SENSORY. Modal Shear Good Primary axis of response less than 16% deviation. YAXIS Y. ALL OTHERS DETERNIfiED PRECISELY FRO1M QUADRATURE PLOTS.25 28.00* 65.
T. J."Dynamic Analysis of Structural Systems by Component Mode Synthesis. Freelin." JPL Technical Report 32530. . 8. " Shock and Vibration Bulletin 39.. 8...J. "Review of Modal Synthesis Techniques and a New Approach. W. 9.J..tion Technique for Ground Vibration Testing. 7. 5. February. 1964. Hou. " 38th Shock and Vibration Bulletin." AIAA Journal. " Shock and Vibration Bulletihj No. T. Gaugh. and 6. "Theory of Incomplete Models of Dynamic Structures. G. Arthurs. R. C. T.General Electric Co." Aerospace Engineering. A. Dec. and Flannelly. No.. TIS R66FPD172. Holbeck."Structural Dynamic Analysis of the Mariner Mars '69 Spacecraft.D. 9. Jan. 1968. No. 1968. 21. S. F. 14811487. . W. 1968. TIS 68SD325. Vol. R..J. 4.. Smith. July. Jr..Nimbus Spacecraft Modal Survey Vibration Test Final Report. V.."Structural Dynamics Modal Matrix Methods for the Coupling of Spacecraft/Launch Vehicle Systems. TIS Report 68SD281. Cokonis. Vol. tember 13. and Gaugh... Stahle. 3. F. Berman. Hurty. W. 1971. August. 1969. Part 4.E. September.. H. Freeland. pp. Romano. .R. June. C. 1966. i 1962.J. " General Electric Company. "Modal Survey Results from the Mariner Mars 1969 Spacecraft. 211 .. 1969. 40. and Hutton. . Sep 2. General Electric Company. 7. Part 2. "Phase Sepax.
This structure of the disc suggests a possible appropriate mechanical model to be a sealed elastic cylinder containing a viscous fluid. The upward turning stressstrain curve implies that the discs will tend to shape pulses having submillisecond risetimes into shock waves and that shock inputs will propagate as shocks. Hartman The Johns Hopkins University Bal timore. Herniation of the intervertebral disc results in a loss of material fro the nucleus pulposus. which contains water in a mucopolysaccharide collagen framework[2]. Furthermore. and a fibrogelatinous core. . Most measurements of the dynamic mechanical properties of the disc have been restricted to smallamplitude deformations which are assumed appropriate for the perspective of linearized response. in severe cases. not enough is known about the geometry or the material properties to warrant such modelling. from about 1. the nucleus pulposus.4 the height of the vertebrae. The intervertebral disc is a nonuniform structure consisting of an elastic fibrocartilaginous envelope. The disc is capped above and below by bony endplates which are distinct from the vertebral bodies. The vertebrae are soft. allowing the nucleus pulposus to protrude into the spongiosa of Often this causes the the vertebral body[7. resulting mostly in injuries to the lumbar spine[6]. for the third cervical vertebra. The human spine is a nonuniformly curved column normally consisting of 7 cervical. 5 lumbar vertebrae and the sacrumcoccyx structure. which I shall discuss here. 60% of all disc protrusions are caused by injury. onethird of these by a fall onto one's feet or buttocks. The dynamic response of an intervertebral disc must be studied in order to predict the vertebral stresses that are caused by rapid motions or impact. Adjacent vertebral bodies are separated by an intervertebral disc which in the lumbar region is approximately 0.[3]. I here describe some of the large amplitude nonlinear characteristics of the intervertebral discs and discuss the implication of these on the dynamic response of the spine. I shall assume the disc to be uniform.2 in. However. the annulus fibrosus. Correlation with experimental results suggests that the application of finite amplitude wave theory to the shock loading of the spine should be further investigated. pertains to compressive loads and the resulting average strain.2W. Each vertebra has a cylindrical anterior part called the vertebral body and posterior archlike parts called pedicles and facets. decreasing in height upwards.for the fifth lumbar to about 0. FINITE AMPLITUDE SHOCK WAVES IN INTERVERTEBRAL PISCS William F. The disc seems mechanically symmetrical about its vertical axis. 12 thoracic.5). The latter guide and restrict relative displacements between the vertebrae for bending and torsion deformations but take up only 20% of axial loads[l]. These implications are explored for axial compressive impact of the spine. cancelluus bones. as can be inferred from the compression measurements of Brown et al. cancellous bone to collapse and continued loading produces a comminuted fracture of the vertebral body. having the average mass and size of the actual disccomplex. This can be caused by trauma or disease and is one of the frequent etiological factors associated with backpain and. During static compression of the lumbar spine the vertebral endplate frequently fractures. INTRODUCTION A knowledge of the mechanical properties of human vertebrae and intervertebral discs is fundamental to understanding body mobility and spinal injuries and to designing protectors and protheses for these spinal components. Since the only data. sciatica[4. Preceding page blarn I2.5 in. suchas is incurred during aircraftpilot ejection or duringa fall onto the buttocks. Maryland The nonlinear deformation of intervertebral discs is discussed.
30  E E E0 0'0 9 kg/mme e * II II A *0 00 0 0. 214 Y4.mI1callY Induced injuries could result from impact experiments or they may be formulated from an extension of the available static data. 1.disc. Although the viscous properties of vertebral bodies and intervertebral discs have not been documented for the large nonlinear deformations which precede failures. while Markolf and Steidel[9] report the damping of the entire intervertebral joint is small for the compressiontension mode of deformation. This suggests that the nonlinear response of the disc can be well defined. the dampIng properties have been measured for small deformations. 14 years ago and yet. this results in the stressstrain curveshown in Fig. so I calculate stress by dividing axial load values by the average area of the appropriate lumbar disc as given by Perey[7]. Fitzgerald and Freeland[8J have shown that the damping of fresh canine discs decreases with increasing frequency. Virgin's data is reproduced from his "typical" loaddeflection curve while Brown's is computed using the averaged values of four compression tests which are each described in [3]. The compression tests of both Brown et al. STRESSSTRAIN CURVES FOR INTERVERTEBRAL DISCS Loaddeflection data for the compression of lumbar intervertebral discs are given in [3.Hyptheses about Ehe type and mechanism of dyn.A.. to my best knowledge. Stressstrain curves for lumbar discs obtained from the data of Virgin.10. These results suggest neglecting viscous effects in a first analysis of the stresses caused by compressive impact.015 0.005 0. For the data of Virgin and Brown. It would be overly presumptuous to expect that accurate results could be obtained for a large. those of Brown et al. o. .05 Strain 0. 11]. Inwhat follows I shall estimate the impact response of intervertebral discs using static stressstrain data obtained from the literature.. Strain is calculated by dividing the deflection by the nominal height[12] of the corresponding disc. range of impact speeds. for these two sources of data were obtained bydifferent investigators in different countries using different testing machines and probably different loading rates. viscous effects are naturally anticipated for the deformation of a structure having the material composition of the. The tests of Virgin were conducted 20 years ago. 1. I present here the first correlation of any independently performed tests of this 0. Of course. The dimensions of specimenfs were not reported. and Brown et al. [31 and Virgin[lO] employei conventional testing mechines and used specimens which had the posterior facets and pedicles removes. but anyanalysis using onlythe data which is available is worthwhile if only to point out the necessity for specific additional experimental programs.15 (Strain) 2 Fig. Filledin symbols are stress plotted against squared strain. The two curves are more than similar for both are proportional to thesquared strain and the proportionality constant Is the same for each.
sex.E. However. 1.34 kg/mm 2 .04 I 0. since neither [3] nor [10] report the crosssectional areas of the specimens. . I2 0. based on Nachemson's measurements. disease.z:T. 4 4. assuming uniform incompressible deformation there. Nevertheless. The change in the slope of the stresssquared strain plot might correlate with the inapplicability of the incompressibility assumption.0.i kg/mm 2 0. For example.05 0.002 (StraIn) 2 0. Furthermore. and specimen preservation. which is the same value obtained from averaging compressive fracture strengths given in [3]. The longitudinal strain is calculated from the reported lateral bulge of the disc. the average vertebral fracture stress calculated from the data of Perey[7J is .01 I 0. 2. Variations are due to differences inage. Of course. would increase the strains for increasing stress. This p)codure results in the stressstrain curve of Fig.004 Fig. This isa poor assumption for large values of stress because Nachemson[13] has shown that the lateral bulging varies nonlinearly with axial strain and the disc's volume isknown to decrease[3].. The initial nonlinearity is precisely the same as that of Fig.from fourth lumbar disc data of Hirsch[ll].001 0. the degree of consistency noted above suggests that gross material characterization isboth sensible and ultimately useful. The compressive deformation of intervertebral discs is also reported by Hirsch[l1]. Normal deteriorations are not yet well documented and certain pathological disorders certainly go undetected. His 215 specimens contained half the upper and half the lower vertebral body and the corresponding intervertebral joints.'"i . 2. kind. thereby making the slope more akin to the initial behavior which agrees with the data of Fig. itis reasonable to assume that a general material description will aoply to vertebrae and discs only for the averaged data of several specimens from several bodies. The posterior pedicles and facets have been shown to take up approximately 20% of the axial load[l]. injury. Stress versus strain. 1 might be partially due to the use of inappropriate areas for one or both sets of data. it should hold approximately for small values of stress and give at least an estimate for moderate stress.04 1 . and squared strain s . o.01 00 O0 I . Therefore in calculating the stress on the disc from Hirsch's data.02 I Strain I 0. Reproducibility and consistency can be found inother mechanical properties of spinal components. the shift in the curves of Fig. I use 80% of the load values. The nearzero modulus for canine disc[8] isshown as .05 I+ ft ?[ ' i: 0.02 Imm 14 kig/mm 2 /9 kg/mm2 0. 1.i .06 0 0. A qualitative correction. size.
since the densities differ only by a factor of 2. 1 and 2. The onedimensional finite amplitude wave theory as developed by Karman and Taylor is given in (14]. then this pulse as it travels through the material will become continuously steeper since the larger strains propagate faster.WAVE PROPAGATION INTHE SPINE A distinguishing feature of the stressstrain curve for the compression of intervertebral discs is its concavity away from the strain axis.30 " " 4  E b E k 0. the strength of the disc isgenerally assumad to be greater than that of the vertebrae. are as follows. then the speed of propagation. Ifthe material is sufficiently long. the spinal curvature. The corresponding range from . the spinal components must be considered dynamically similar. (2)and (3). of the discs are such especially since the sizes that individually they are not rods. Also. 3.10 2 (1) 0 where a is the longitudinal stress.04 to . Taking the average curve of Fig. Since I have already indicated that I will ignore the not the discs. A perspective retaining the individual material properties of these components will result ina rather complicated picture for impact problems. there are important subtle 216 4 . Such material behavior is typical of many biological materials but here it has interesting implications regarding stresswave propagation along the spine. 3. I consider: (1)the pulseshaping performed by the discs. distinctions. twisting and viscosity are ignored. The gross weights and dimensions of fresh canine tntervertebral discs are given in [8].22 g/cm .'and #the response function. its axial dynamic response could be properly described as onedimensional only if its major components. (2)the prediction of the impact response of a preloaded comin and a specimen vertebraediscvertebrae given results parison with the experimental (11). behave linearly. if the material issubjected to a gradually risig pulse. were dynamically are sometimes The vertebrae similar.from the data of [7]. the vertebral bodies and the intervertebral discs. Using these. every pulse regardless of its shape will eventually develop into a shock. which isapplicable for nonlinear strainrate independent elastic or plastic deformations. especially for large stresses. I do wave of onedimensional nonuniform assume that hesitate tostructure propagation can be used to obtain reasonable estimates. their diameters being larger than their lengths. The average where p isthe density of the solid. discs for with theassumed in comparison as rigid bodies dynamic analysis[12]. (2)results in the curve of Fig. and Eqs.1 0. If the stressstrain relation iswritten as a= *() fracture stress 0. Nevertheless. of any strain level is C(C) (/p)l/ 2 00 0. a bulk density for the disc is calculated to be 2. bending.05 (2) Fig. The vertebrae.08 kg/mm 2 . and (3)the prediction of the impact speed which would produce fracture in the fifth lumbar vertebra. 1 and the above value for P. Such an assumption does not square with the stressstrain relation shown in Fig. The particle velocity. Nachemson and Morris[15] have reported invivo measurements of intradiscal pressures foFi'al loads such as those occurring during standing axial stresses and sitting. In all cases only compressive axial deformations are considered. Using the data described thus far. Both the deformation and the slope of this stressstrain curve are comparable with those of Figs. C(c). c the longitudinal strain. Some pertinent results of that theory. V(c). An interpretation of this result is that if a unitstep shock input isapplied to the onematerial dimensional maximum ofwill the remain speed it the then shocklike since stress is the greatest speed. This was constructed using the data on (7] for the compression of vertebral bodies. On the other hand. Therefore. Eqn. (2)increase with strain. associated with any e orp) can be written V(e) = c !dc' (3) If0(c) isan upward turning function then *'(c) increases with strain and so the wave speeds given by Eq. Stress versus strain for a lumbar vertebral body . Although the spine has the appearance of a curved rod. 4.
5 illustrates the pulse shaping which can occur as a stress pulse propagates along the spine. Of course any pulse whose initial rise time is less than 0. Fig.1 msec tail at the first thoracic vertebra.. Fig. The experimental points shown in the figure are plotted taking the equivalent impact speed to be twice that of the drop speed.+(R [a ll + [ RI (] . A lumbar disc and half of each adjacent vertebra including the posterior processes were preloaded in compression and then subjected to impact by allowing a weight to fall through a known height onto the loading lever of the testing machine. The resulting expression is: 2[Vii = [aD] + $C(C')dc +z( c where Z. The remaining stress is easily accomplished through the multiple reflections dt the infinite mpedance.1 T] T(4) . 5. 4. Example of shock formation by discs.35 kg/mW . is the impedance of the . In this case a linearly rising pulse increasing from the normal sitting stress to a value which is less than the fracture stress steepens from a 0. let [a n * 0. However. while pulses whose rise times are longer than several msec would not be noticeably shaped. This orr sornds to a dynamic jump from an tnilial stress of .30 kg/mmz. this gives 217 j .5 m sec 1Fig. Using the curve of Fig. Therefore.1 msec seem plausible for dynamic stressing such as those occurring during !iaircraftpilot ejection.08 . Using Eqns. which is 90% of the fracture stress. Thus the shock wave formation as described here could account for the fact that vertebral fractures occir inthe thoracic vertebrae during pilot ejection even when [12]. Assuming continuity of stress and particle velocity at the vertebradisc interface the following Jump equations apply for 100 E 50 . (3) permit a check on the consistency of the present mechanical perspective. 4 shows that the wave speeds double over the range of probable dynamic stresses.26 and c. 2 and Eqn. 4. Using a constant vertebrae impedance. Since Hirsch did not report the dimensions or the raterials of the testing apparatus. Neverthehis recorded values of the dynamic bulging of the disc together with the stress strain curves of Figs. then the effective crosssectional area is reduced and failure will occur at a lower nominal stress. Rise times less than. no damage is sustained by the lumbar spine Some impact experiments on spinal segments were performed by Hirsch[1l]. 6 is obtained. (3).30 (kg/mm 2 ) mless. The porous structure of the soft cancellous bon2 of the vertebrae make them inherently weaker in tension than in compression.rtebral body.05. c = 0. Variation of wave speed with stress. In the above. the linear shock equations.17.5 irsec rise time et the fifth lumbar vertebra to a partial shock an 0. 0. the response curve of Fig. During sufficiently gradual application of compressive forces the deforming vertebrae uniformly distribute the load effectively over the entire croEs section due to the collapse of voids and the reduction of porosity. the speed sufficient to cause vertebral fracture from impact onto an infinite impedance may be calculated. the effective impact speed at the vertebra cannot be calculated. respectiely. if compressive i 1 + [VR i VTJ where 1. a . XL5  X=Li XaT! ~~a shock input.04 kg/mm2 to a valie. (4). The agreement with the theory further suggests that it might be rewarding to pursue experimental studies of wave propagation in the spine. which cin be obtained from the data of [7]. the stressstrain curve of Fig. vertebral fracture stress.stress isapplied sufficiently rapidly. such as a stress pulse whose length is less than the average pore size. is .8"10. 50 .10 Fig.5 msec would become a more fully developed shock. 2 and 3 and Eqn. R label the incident and reflected shock in the vertebra and T labels the transmitted shock in the disc.
05 Strain increment 0. Herniation. Scand. Suppl. "The Dynamic Characteristics of the Human Intervertebral Joint. Harkolf and Robert F. J. 1932. 14. J. an impact speed in the vertebra of 12.6 6A Calculated Measured (Hirsch 1955) E4 0. 6. Fribert. Oct. REFERENCES [1) Af. Ortho. which corresponds to a free terrestrial fall from approximately 7." Acta Ortho. 5." to appear in J. Keyes and E. 67. Compere.. "Anatomy of the Intervertebral Disk. 25. Vol. Since buttock and pelvic elasticity have been ignored. Fitzgerald and Alan E. Ortho. Nachemson. 830. and Rel. A. Edwin R. (5] (6] [7] 1960. Scand. Bone and Joint [9] 218 .06 Fig. 338. Coventry. Suppl. 915..5 meters. 1957. A. Perey. "Viscoelastic Response of intervertebral Disks at Audiofrequencies. On the other hand. 1957..04 0. (2] M. (8] Surg. 0." Acta. pp. 897938. Res. 1951. Eng. No. this value is surprisingly large and certainly does not seem to agree with ordinary experience. The case shown is for a lumbar discvertebrae segment prestresed 0. Bone and Joint Surgery. Hansen. Ned. 43. L. "Low Back and Sciatic Pain Caused by Intervertebral Disc Acta Ortho. pp." J. Scand. C. Keith L. Bone and Joint Surg. Steidel.. O'Connell. "Fracture of the Vertebral End Plate In the Lumbar Spine. J. Vol. Suppl.. S. Freeland. "Some Mechanical Tests on the Lumbosacral Spine with Particular Reference to the Intervertebral Discs. [3] Thornton Brown. "The Normal and Pathological Physiology of the Nucleus Pulposus of the Intervertebral Disc.2 m/sec. 25. the existence of finite amplitude shock waves in nonlinear intervertebral discs has been shown to be plausible and its study is potentially important in understanding dynamically induced spinal injuries. (4] D." ASME publication 70WA/BHF6. ol. and Biol. R." Clin. impacts which are truly axial are seldom achieved and the consideration of bending and shear would reduce the injury speed.. pp.005 kgria2 . "Protrusions of the Lumbar Intervertebral Discs. Vol. 39A." 1957. Yorra. The calculated additional strain increment due to impact is nonlinear with the speed. "Lumbar Intradiscal Pressure.. B. E. 1969." J. c ile mch of what I have discussed will be considered speculative because it is based upon minimal appropriate data. pp. 1970." J. 11351164.
" Oxford University Press.. Vol.130143." J. Uone and Joint Surg. pp. 37A. pp. "A Mathematical Model of Spinal Response to Impact. "Some Mechanical Properties of the Lumbar Intervertebral Discs.[10) W. 1953. 46A. 338. pp. 1962. Vol." J. 1955. H. 4. Alf Nachemson and James M. Virgin.11881196. Joint Disease.607611. David Orne and Y. Kolsky." J. Alf Nachemson." J. Vol. pp. pp. J." Bull.. 23. Vol. Bone and Joint Surg. Bone and Joint Surg. [12) (13) (14) (15) 219 219 a't a'4~I . 1951. 1964. 10771092. Biomechanics.. "The Reaction of Intervertebral Discs to Compression Forces. "Experimental Investigations into the Physical Properties of the Intervertebral Disc. "Stress Waves in Solids.4971. Hosp. Vol. 1971. "In Vivo Measurements of Intradiscal Pressure. [11] Carl Hirsch. Morris. King Liu.
Classical bending equations for plates. the peak values are attenuated and much of the veryhighfrequency response appears to dissipate rapidly. a prevalent practice for analysts is to construct a mathematical model of the structure assuming that an entire panel can be represented by a single degree of freedom. nine modes should adequately represent the plate's true response.6 times the first mode response i. Whippany. are assumed to be applicable and these lead to a timedependent. 15 March 1957. However. In this regard. When damping is incorporated in the solution. simply supported plate acted upon by a uniform stepfunction pressure is analyzed here to determine the validity of this approximation. what would be calculated using the onedegreeoffreedom analog. Inc.t A direct consequence of this approach is that the calculated peak acceleration response at the center of a square plate element exposed to a step overpressure is Peak Acceleration Peak Overpressure 5Mass per Unit Area of Panel (1) . hence the onedegreeoffreedom approximation is appropriate for these quantities. and it is established that the peak acceleration response can be as much as 2. Acceleration responses are plotted as functions of time to determine the contributions associated with frequencies higher than the fundamental. In prac*. that is.ACCELERATION RESPONSE OF A BLASTLOADED PLATE Lawrence W. Fagel Bell Telephone Laboratories. it is a usual inherent requirement to consider these plate elements to be externally loaded by blastinduced overpressures* which may in some cases be approximated by a step function for determining earlytime responses. Plate dimensions are assumed to be such that classical bending equations apply. the maximum amplitude of response is not obvious from the solution expression. indicatingsolution inshould not be ignored. planestress equations. New Jersey A solution for a simply supported plate loaded by a stepfunction pressure is closely examined to determine contributions to acceleration the plate's higher modes of vibration. double infinitetrigonometric series solution in which each term of the series represents the transverse vibrational response of a different mode. computed accelerations may be nonconservative b up to 40 percent if only four modes are considered and by more than 100 percent if only one INTRODUCTION A commonly used technique for calculating the approximate response of a plate subjected to blast loading is to consider the plate to be springequivalenit to a onedegreeoffreedom mass system. are sometimes regarded as composites of plate elements.ical situations where the damping ratio will be at least 1 percent. Peak acceleration response for an undamped plate appears to be about 2. the amplitudes of some of the higher modes of acceleration response are significant compared that these to those of the first mode. Moreover. for structuralmotionresponse studies.Pressures in excess of ambient.e. Because the volves an infinite series in time. Preceding page blank 221 . for shelllike tective and analytical purposes.. Corps of Engineers Manual EMl110345110. Comparison of the relative amplitudes of each modal response demonstrates that the first mode predominates for the dynamic responses of displacement and stress. tThis practice is recommended in "Design of Structures to Resist the Effects of Atomic Weapons." U.6 times the peak resnse of a onedegreeoffreedom analog. however.S. the frequency of which corresponds to the fundamental frequencyof the plate. MOTIVATION Structures which are designed to withstand nuclearweapon effects are often either shelllike structures orstructurei are encapsulated within prodesign whibh.
t) = fore there is perhaps a need for a more quantitative definition of the number of modes which. The intentions of this study are to determine an upper limit of acceleration response values resulting from the consideration of higherSignificantly modes. although also significant. the value of this coefficient depends on edgesupport conditions. In situations where structural responses are computed for the sole purpose of specifying shock environments for accelerationsensitive components within the structure. i...t). plane stress equations apply. i. cates that Eq. (t) 4q(t) I sin sin ab J f a 0 0 ocs b a { dxdy b ii Pi(t) = q(t) Eh 12(1v ) 2 V4 w =q(x.) An examination of a platevibrationproblem solution that includes modes indiwhich are higher than the fundamental. then Pij(t) = 16 PoU(t)/ Iij f for odd values of I and J.y.y.. w =E i**1 E Jla 1= J=l sin'a'snJT.! .t) then becomes The derivatives of w are Sin Ly b w J .j t)Sil)in a (xv y t) 16P Ut 16P 0 '0 i=1 j=l sin sin).e.y. a 0.e. (1) characterizes the response in the first mode. The expression for q(x. (2) ( ab Therefore.x sin 15 dd dxdy.t) in a double Fourier series involving x and y. i1* b The Fourier coefficient Pr(t) is evaluated as Pi(t) !b fa q(x. for the stated practical purpose.y. J T 0 A deflection which conforms with the simplesupport boundary conditions at x = 0 and a. y =0 and b is Pi (t) = 0 for even values of I or J.y. and Pij(t) = 6q(t) for odd values of i and J.t) varies with time but is constant with respect to x and y.t) = q(t). b (3 If q(t) is a Heaviside step function of amplitude Po.(1. and to establish a quantitative cornparison between onemode and manymode acceleration response solutions by considering a simply supported plate loaded by a stepfunction pressure. it is of course necessary that all modes which contribute significantly to acceleration response be accounted for.y. the amplitudes of acceleration responses in modes other than the first.y. have been ignored. however. 0 40 unfortunately somewhat ambiguous. Assuming that the classical bendings. the governing differential equation [11 is Vh w + g L1 J=I Pij(t) sin jr a sin LT.t) 0 0 sin i. to ascertain theto number of modes which contribute accelerationi response. should be considered. q(x.t y (5) (4a) 222 .5 is approximate to about *10%. q(t) = PoU(t).e. and theee q(x. ANALYSIS OF A SIMPLY SUPPORTED PLATE The response solutions for an undamped simply supported plate acted upon by a stepfunction overpressure are directly derived as follows. i. This apparently reasonable requirement Is w = go  4 ^' w +Y NxAcxoy ij(t) i= j=I sin I= sin a b (4b) Expanding q(x. For uniformloading cases where q(x.
L o sin k inI 1. for odd value ofi 0 for even values of i and j. and (5) substitute into Eq.co s Wijt) (Lb ( are rivallyequl 6a) teoForodd2 tozer byEq. (9 susiue intE. (2) to form The initial conditions are 0i(0) Solving Eq. (4).Mdlt dispacemnts ~j fr evn vaues f i223 ) ijd(S ) i(+ for o Salu(0s 7). ._3)yild n 1=1 in a j1of I nsin for oddb values I and j. (6a). ( 2w 2w\ w.+ L valus frm i ad jthea quaionsarein Ez  W 2 . bi sin irx sin ! y I . 16PoU(t) and e (6a) ii 2 W/g 2. all initial conditions with and j of I Modal values for even displacements 'o Ez Ez 2w 2w are trivially equal to zero by Eq. 0 i_ (t) sin " a sin i g 2 b 2N +W S (9) Eh 121 (ifr)2 + (ig)2]2 t1i(t) =2(I  Cos coj t) sinsinsi 0 ~t a b Eq.  .Eqs. For odd values of i and J the equations are in the form 2 16 o U(t)= AijU(t) ij2 2(a (7) b t (l The Laplace transform of Eq. (8). (3). 2 l (10) and each model differential equation becomes Ii 2 2.2 Using the6 stressstr'an relations for plates: Ez.(b 2 1Y~ow a b  2 Mryy +.U IL6I Ehthe a ++ b (11) +V ij16 being zero. (7) is wihaliiilcniiosbigzr.
compared todisplacements are negligible mental stresses associated with the fundamental I=1 jl ({I. pared MY ( 1G Po Ez ". Note that 00 o in 2 2 izl.  i where i and j are odd.Cos 2 a j it) . (10) through (14) 1 and J are odd nd #h ff a2 3ij\ v 2) i + \2 (. and acceleration.tion: _i n~p IF 2W/ig k r (2k (2r + 1) w(first mode) iJ' 224  16 P'O Wg IT2 P0 =W' "  ~ £ . Information from these graphs concurs with the philosophy that displacements and stresses resulting from mode shapes other than the fundamode. we have: Displacement: S wi. Wmidpoint (14) 6/go16 2 W/g 0 0 sin sin L4 i I=1 1= Cos Wit (15) In Eqs. the contributions to acceleration from some higher modes appear to be significant.(first mode) i 2' (I)k 1 (2k + 1) Stresses: _ . ros wit I1 for all i and J. On the other hand. .W/"g t=1 j=1 2o J 2 Comparing highermode amplitudes to fundamentalmode amplitudes for displacement.r2 a b COS Wijt ._ . COS cos jXy a b (13) The acceleration response is sin Go 16 o W/g~ J4I Interest is focused at the center of the plate where the singledegreeoffreedom analog had been presumed to be applicable. and Eq. At t 2 0.3.. I nor from Eq. therefore this phenomenon will now be more closely examined. The relative amplitudes of highermodal values comto the respective amplitudes associated with the fundamental mode are plotted on Fig. Just how the highermode acceleration amplitudes supplement the acceleration response of the first mode is not obvious from Fig. where l and j are oddJ. At the midpoint the acceleration is .etc. (15) degenerates to " wmidpoint 16 E E . . (14).. (first mode) 1 k Accelerg.xy = 2 w Ez.5. k0O tan1I / j (+j) : 2 (I+ ) 2 ) .2W/g1+v +and I 00 O0 2 where I and j can only be odd values. 1. xxtlj ax (first mode) 4 j j4 so thata t . 16 PO (g (_)r +I' . stresses.
.. . ... ... . 1. . . . . . . . . . . .. ...... . . . .. . .. . . . ... .. . . .. . S. .4 . .. .. . . . .. . I . . . .. ... . ... .. ... .. . . . .. . . . ... . .. .. .. . . RELATIlVE AMiPLITUDErSOF MODAL. .. . ... . . .. . . .. 3..... . . . .. . . . IT . . .. . . . . .. .. . I . .. . . . . . " ... .. . .. . . . . .. .. . .... . . . . . ... . ... . . . . . . . . . . . . .... . . . . .. .. . . . . . ..... .. . . RELATIVE AM/PLITUDES OF MO&. . .... . . .. . . . ... . .."lESS RELATIVE AMdPLITUDES[ OF MW AL ACCELEPAITIONS Fig. .. . . .. . .. . . . . . . . . .. . . . . . .. . Ai . . . . II. 15 . . . al . . . . . . . . . . .. . . . . . 2P . . .. . .. . ... .. ..7//777/7T.. .?. . ...... ". . . 9 .. . . . .. . .. ".. . . .. . .. . 113. . . . . . . ..'. . . .. .. . .. . .J . 21 . . ... J . .. . . .S. . . . . . . ... . .. . . . . . ... . .. _j i2. .. . . . . . . . . . .' . .. .. . . . . .. . . . . . .. . . .. . . . .. . .7 . . . . . .. . .. . . .. . 7. . . . .. . .... . . . .. . . . 27 . . " . . .. ... .. . ... .. .. . 1 .AL$HlEAM S.. ... .. . 1 . 25S.. . . . ... . T. .. . . . . . . . . . . . . . . ... ... .. •~~~~~~~~~~~~ .. . . . . . ... . . . . . . I .. . . . . . .j' . ... . . . .. . . . . . it . . . J ' . . . .. .. . . . . .. .. I . . . . . . . .. .. ... 19 .. . . . . . .. . . 7... . .. I17 . . . . .. . . . . .. . .. .. . . . . . . . . ... II. . ... . . . . . . . .... . . . .. . ... . .. . . 1. . . .. . . . . . . ... .. . . . . . .. . . 27 . . ... . . . . . . . .. . . . . . . . . . .. 2$ ... . . * ... . . .. . . . . . . . . . . . .. . . . . . .. .. . 21 ... . .. . . .. .. . ... I. .. 23. . " .. . .' . '... . . .J . o 1 • • • • • •I ~O ... . . . . . . . .. . . .. .... € . . .I. .. ..... . . . . . . . . . . .. . . . ... ISl. ... . . . . . . .. .. .... 19 I .. . I Comparisons of the Relative Amplitudes of Modal Responses 225 . . .. . . . DlISPLACEMENTS RErLATIVE AMPiLITUDESl Of MOAL NORMAL STRE~SrS IT5. 9 . . . ... .. . ...
therefore. and in fact. That is. 3 as a function of the number of modes. certainly all structural vibrations. however. It Is generally conceded to be a combination of air damping and ctructural damping and although approximate solutions exist which incorporate these effects. DAMPED RESPONSE The previous discussion pertained only to undamped vibration. plots of normalized acceleration versus normalized time show a contrary trend. to and including 225 modes are considered in computing these values even though it is recogtary inertia is (2). (17) results the total response w and its timederivatives are summations of the modal values oij and timederivatives of OiJ respectively. that Is about 2. it appears that the onemodeonly representation of response Is reasonable and conservative for the initial value of acceleration. some discussion regarding this topic is appropriate.and iJ is the p~rcentofcritical viscous damping ratio. definedso that Eq. An effort to consider structural damping is presented below. Up and the modal acceleration is . (17) 2 P( ) J=l * j 2 where 2i w0 is the modal viscous damping force.12 Po/(W/g) as m and n become large. (16) for increasingly larger values of m and n demonstrate the effect of considering higher frequencies in the acceleration response. 2.ofthe contributlonsafromuall middeb the should remain bounded by the amplitude response of the first mode. The solution to Eq. These timevarying plots of Eq. indicates that the acceleration response increases substantially when higher modes are included in the calculation. At t> 0 however.V  1 g c 1 1 ij1 CO t .  in iJ t +0 / ' 3 1 kj Wicos t+0 For small values of PiJ 0 v/2 and the acceler ation can be approximated by knowledge. The peak values from each of these graphs are plotted in Fig. it is not'obvious that the absolute value ofthesum. no unified mathematical treatment of the problem exists which exactly defines the physics of such damping. (1)]. (17) in cos wijt (I and j are odd) (16) plotted versus wot/2v on Fig. which not neglects shear for and the roveryhighfrequency mode shapes.Bycomparing this with the firstmode contriPo/(W/g) 1. 226 tural vibrations is complex and to the author's The phenomena Involved in damping of struc Ajt elijwijCt =A(t) ij e coB 16 P . and as will anthmolacertoni 4'iJ = A elij=oijt iji + 4 ( 2 i be shown. . 0 tan "I 1 " 9I li onem~odeonly representation of response.with Eq.6 Po/(W/g) button at t =0 [(16/) which approximately agrees. Nonetheless. Damping is thereby defined so that Eq. this simplifying assumption permits the calculation of an approximate upper value. When the governing partial differential equation i.. computed accelerations must be conservative since all mechanical vibrations. probably appropriate t nized i that Eq. (7) is repFaced by 4'j + 2(3 wI 4. the equation m Wmidpoin Po/(W/g 6 r2 n sin Lr sin Jr I they are based on damping forces defined'In a manner which preserves thelinear nature of the modal differential equations of motioi by ntrodicing a viscous damping term.Iof Ii + W4'=1 I Ii i Mi 2 16' Po 0 i t (t) U~ . and this curve appears to be asymptotic to about 4. are impeded by damping.6 times larger than the ij'(t) 1 = A 1 1 e I  sin 2 2o i (ji 1ol I3 t + . th consideration of damnping can greatly reduce the contributional effect of these veryhighfrequency modes.
0 0.000  a. 1 MODE CONSIDERED MI n 1 0 PERCENT DAMPING S 6.sYU.600 0.200 1.220 '.60 .000 0.000 .000 PLATE VIBRATION.5 S w. . 8I MODES CONSIDERED n 17 0 PERCENT DAMPING inn17 600 40000 .8010 1. s~ '. 2 4 oprsnotceeainRsos Cosierin One and Many Mode  2.0o 1 00 0. 0.4 ~s.000.0 .0 .9 MODES CONSIDERED nlz 5 0 PERCENT DAMPING Mz5 S6.i s~s' .000 1.400 0.000 1.600 0. . .PLATE VIBRATION.0 .200 1.200 0.400 (TI FUNDAMENTAL PEPIOD) 1.800 2.600 LOW0 2. s~x.. .800 1.600 1.0 .0 .0.4~. ~ '. '~4(5.400 0.400 1..0 (TFUDAETA0PRID Fig.200 0. .00 ~V0.PERIOD) PLATE VIBRATION.000 CT/FUNDAMENTAL..
however proportional damping was experienced while testing a different structure.TN[E [XACT 9OUyION:. however. . tisepszdthths reslt eca ared aJ't isemofhadsie thtohi result is arrived at because of a desire for mathematical simplicity rather than from an exact definition of the physical phenomena for energy dissipation. however. While testing building structures Alford and Housner [21 found that the damping ratios for the higher modes were the same as for the fundamental mode.a %j. . Kimball [61 found that internal damping is neither proportional to frequency nor constant. in fact. ACcL[ERATION . tis i e n e l x ep bn tI' it INDE[X OF m AND It is w i g 1 4 9 16 29 3 49 44 IN 10 111 144 149 196 22 NUtM[M OF MODES$ €ONS1IIO RAr tFig.'.. Regarding values for modal viscous damping ratio iJ. and Sesan.   ~ ~ 'ir.~~~~xstkvt~ c A c. but rather that for most materials it increases to a maximum value at some frequency and then decreases for all subsequent frequencies. and there is apparent worldwide disagreement in the interpretations of results from such experiments asevidenced by a sample of some of the Thus a damped modal vibration can be considered the same as an undamped modal vibration except that it Isenveloped by an exponenUaldeay e i w i t..I Jul i S 14 IN F SW I €O* 4 lIt' i THE SUMMATION LIMITS ARE REPLACED BY VALUES. Crongradi Diaconu and Strat [31 concurred with this finding in a separate study On the other hand. The damping forces in structural vibrations are not. and it is a standard method for evaluating damped multimodal vibration responses. Kawasumi and Kmi[41 claim to have experienced proportion:l damping by doing similar experiments. VALUiS ARE CALCULATED FROM. and modal Eq. Adamson's [8) observations are similar to Kimball's. this simplifying assumption provides a means to an end in that it yields approximately correct dynamic responses. the literature on structural damping contains different opinions regarding this subject.PAKACCEM[RATION AT THE MIDPOINT OF AN UI09D SIMPLYSUPPORTED PLATE SUBECTED TO A STEPFUNCTION PRES*UK. OF m AND a FOR I AND j REAiCVILY. a Kelvin (alIso known as a Stokes) model could be assumed for the material (stress is proportional to strain rate as well as strain). Po/(WI) WHERE FINITE isf11 711 . If the viscousdamped assumption were completely true.t  1 Adt3.[7[ recommend using a constant damping ratio. Mindlin.! . (17) would evolve where P J is proportional to wij. the governing biharmonic equation would be modified by adding a (/ t) V4 w term. Adamson has a compromise recommendation to use proportional damping if only a few modes are inherent but a 'I * . known to be viscous. Stubner and Cooper. 3 Peak Acceleration at the Midpoint of an Undamped Simply Supported Plate Subjected to a StepFunction Pressure damping in real structural vibrations has been the subject of extensive experimentation. Nielson's [51 experiments showed that the same damping coefficient was evidenced at different modes during tests on one structure. Whether or not the modal damping ratio is indeed proportional to 228 literature on the subject.
the acceleration response at the midpoint becomes sidered. the peak acceleration response calculated for a 2percentofcritically damped plate that is approximated by using only one mode and considered to be subjected to an impulsive load. They of course reduce the peak acceleration responses. W16 midpoint LL P W i=1 j=1 e m n sin ir sin! I J  iJ t CosWfactor where I and j are odd. If the excitation is oscillatory the above recommendation is not intended to apply. when both air and structural damping are con 229 f1 . If dominant frequencies in the forcing function coincide with frequencies of the plate's higher modes. for practical purposes usually be adequate for dynamic response calculations involving plates subjected to impulsive loads.without damping. 3 that the response considering 225 modes is equivalent to the response considering an infinite number of modes. For example. For materials which have three or more percent damping. therefore. 5 suggest that larger values of critical damping ratio produce two desirable effects from a structural analysis viewpoint. If fewer than nine modes are used to represent plateresponses in a dynamicanalysis study. however. The advice of Mindlin and others who recommend the use of a constant damping ratio. There exists. the curves on Fig. these frequencies will contribute more significantly to acceleration responses (displacements and stresses also) than if the structure were impulsively loaded. For these situations it most certainly would be prudent to have the mathematical model of the plate contain modes which at least include the dominant frequencies of the forcing function. the difference between the peak responses using 9 and 225 modes is indistinguishable. the error introduced by using only a few rather than many modes to represent dynamic response is also reduced. CONCLUSIONS The contribution to peak acceleration response of a blastloaded plate from modes other than the fundamental can be significant. 5 where it is inferred from the asymptotic nature of the curve on Fig. From a practical standpoint. therefore modal values associated with frequencies above the fundamental should be incorporated in blastresponse computations.constant ratio if many modes of vibration exist in the response solution. a variety of opinions regarding how equivalent modal viscous damping ratios vary as a function of modal frequency. independent of frequency is used in the investigation below todetermine the effect that damping has on the acceleration response of a vibrating plate subjected to a stepfuncti'n overpressure. The veryhighfrequenicy characteristics quickly disappear and the peak responses are attenuated. If Pij = 3 for all values of I and j and j << 1. This is shown on Fig. 5 are indicative of the possible amount of nonconservatism which should be expected. it is recommended that the mathematical representation of the plate contain at least nine modes if blastloading conditions exist and acceleration responses are desired. 5 indicates only about a 15 peicentd isagreement with the 225mode response at 1 percent damping. The curves on Fig. If the excitation is impulsive (blast loading). The ninemode curve on Fig. Computed accelerations may be nonconservative by up to 40 percent if only four modes are used and by more than 100 percent if only one mode is used. the calculated peak acceleration responses will possibly be nonconservative by an amount dependent on the percent of critical damping and the number of modes considered. 4 which shows the midspan response considering 225 modes with and . The effect that P has on response is demonstrated in Fig. Of particular interest is the comparison of the amplitude of the dampedacceleration peak response considering few and many modes of vibration. therefore ninemodes should. most materials (even steel plates) probably have minimum equivalent damping ratios of about ['percent. could be nonconservative by a of two. For engineering purposes where at least 1 percent of critical damping may be assumed.
4. .. 1. .000 (T/FUNDAMENTAL Fig..000 4.000.200 0. .000 S2.° 3R° 4..200 0.000 ..  I1.200 1...00 2. 6000 0000 0. .600 1.200 1. 225 MODES CONSIDERED 0 PERCENT DAMPING n:29 mx29 S6.000 0.000 .000 .. . ..200 0. o . V* ...400 0.000 1.600 1. .. ....600 0. . . 2 000 ..000  .600 ... 1.. .000 1.000 0. 225 MODES CONSIDERED 3 PERCENT DAMPING m s 29 n 29 6. 00 0 0.400 0.800 1. .000 K.. . .600 0. 4  Comparison of Normalized Acceleration Response Withot jid With Damping 230 . .0001 H.000 0..000 .000 Z 4. i 1200 1..800 1..900 2.00 1..PLATE VIBRATION. ...PLATE VIBRATION. 1 inA AA / : 0.400 (T/FUNDAMENTAL PERIOD) .600 0..225 MODES CONSIDERED I PERCENT DAMPING nz29 mz29 6. . . .400 (T/FUNDAMENTAL PERIOD) PLATE VIBRATION..000 0.400 0. . . .600 2..800 2... . .. .400 PERIOD) 1...
U ) to NOmsE CONSINNRED 9 AIOSS CONSINERI 0 I i I ! I i 7 6 1 3 4 2 I PIERCENT OF CRITICAL DAMPIIiN i I S Fig. 5 .Ratio of Peak Acceleration Response Considering 225 Modes to Peak Acceleration Response Considering Fewer Modes 231 .
" ASCE Structural Engineering Ccnference. 2. n ienl ~ Junlo 3.nering. Alford and G. Fasc. "A Method for Measuring Damping and Frequencies of High Modes of Vibration of Beams. 8. 8." Proceedings World Conference on Earthquake Engineering." Proceedings of the Society for Experimental Stress Analysis. Vibration Problems in Engineering. Cooper. Nielson. 7. "Experimental Determinations of Natural Periods and Dami in Buildings.APPENDIX I . J. Kana. Part V  Fito Mechanics. H. 1955. "A Dynamic Cf a FourStoryReinforcedCo lTest ncrete Building. Stubner. Diaconu. Van Nostrand. 1548. L. 2." Buletinul Institutului Polite cniciTomul X (XIV). J 1Applied 5. R. W. W. A. F. L. January 31. 6. Kimball. 1964. No. B. 34. Housner. 4. Crongradi. D. "Small Amplitude Vibrations of Actual Buildings. 1966. "Response of Damped Elastic Systems to Transient Disturbances. Vol. Kawasumi and K." Bulletin of Seismological Soiciety." Publication of International Association for Bridges and Structural Eng. B. "Vibration Problems. I. 5. N. Timoshenko. D. 1941. Adamson. Sesan. N.REFERENCES 1. January 1955. 1956. Strat. Mindlin. Vol. "Damping in Multistory Buildlngs Determined from SteadyState Vibration Tests. and H. 2 . and L. A.
z = plate coordinates y = density V = del operator =iJ jjth frequency 0o i.y.j = modal indices Pi (t) = Fourier coefficient Po = overpresmure q = pressure function S = Laplace transform variable fundamental frequency = Poisson's ratio = normal stresses y xy = shear t = timeq U(t) = Heaviside unit step function stress 233 .NOTATION ab = plate dimensions in x.y directions Alj = modal coefficient of forcing function E = Young's modulus g = gravitational constant h = plate thickmess w = transverse displacement w = transverse acceleration W = weight per unit area =yh x.APPENDIX II .
more than any other. So is the highintensity noise usel frequently now to simulate them. if the responses were not similar in the simple resonator. shLck and vibration testing can be a useful engineering tool when it is planned intelligently on the basis of suitable criteria for realism. Morrow Advanced Technology Center. It is not possible to obtain vaid criteria for realism by a test case that involves only a singlepoint input. To accomplish this instead by theory. even when the damping of the strip is large enough to prevent significant retu:n reflections from the ends. for vibration and shock. The timehonored test case. (U) No environmental simulation is ever completely realistic. U .age potential is less critically dependent n damping and the resonator is frequently assumeL to be undamped. i"inadequate. of whether a test item would survive and function as intended. the amount. the formulas remain applicable when. Statistical Energy Analysis. for an item of equipment subject to shock or vibration. In spite of the limitations of the simple test case cnu the compromises that are necessary to achieve a practical environmental specification. j Charles T. The simple mechanical resonator. It is shown that in an effort to design a realistic test. The same would be true for a complete space vehicle excited by turbulence or by highintensity noise. it is generally possible to prescribe a useful test. narrow by comparison with the lateral correlation distonces of the exciting field.EFFECT OF CORRELATION IN HIGHINTENSITY NOISE TESTING AS INDICATED BY THE RESPONSE OF AN INFINITE STRIP (U) V . and independent excitations at the different antinodal regions. Inc. with a minimum of computation. (U) Turbulence and rocket noise are distributed excitationc. but the precise relationship and the fundamental theory behind it. It is not true that two shocks or two vibrations that produce practically identical . is conceptually suited primarily to the reverberant field. Yet. of dazmp'ng is an important parameter. and some modes of failure might be overlooked. since normal modes are assumed in the exciting field as well as in the structure.'espon3es in a simple resont. possibly according to the square root. one must consider tile pointtopoint correlation of the applied field as well as the sound pressure level. in addition. an exponential decay of correlation with distance in either direction along the strip is introduced. These criteria are obtained by calculation of the responses of a simple structure as a test case. is the simple mechanIca] resonator. The prediction method that has received the most intensive development. For vibration. they would have almost no rhance of being similar in the more complicated item of equipment. INTRODUCTION (U) Environmental testing is intended to provide an evaluation. The strip could Le general Preceding page blank . (U) The test case . Dallas. Texas (U) A narrow strip (bar or ribbon) is taken as a theoretical test case for the realism of highintensity no4 se testing in much the same way that a simple mechanical resonator is commonly taken as a theoretical test case for more conventional shock and vibration testing.ilized in this paper is a simple strip of structure. Three types of correlation are investigated in particularcomplete correlation at the coincidence angle for the given frequency. The velocity of transverse vibration along the strip is assumed to vary with frequency. (U) This type of analysis is also potentially useful in the prediction of responses. With minor reinterpretation. or will necessarily do the same in any item of more complicated internal dynamics. in practical use. are not criticaQ lo the discussion. even if the excitation iq conceived as a force acting on the mass. For shuck. It may eventually be possible to calculate response corrections for reverberant versus flight fields. would be time consuming and inaccurate. Nevertheless.
ized to two dimesinns. the transverse waves propagating in either direction from alternate antinodes should be of opposite phase and tend to cancel each other. or. and subject to attenuation of the transverse waves on their way to the observation point. cospectrum or real part of the crosspower spectrum. wA+B = wA + wB :2wAB w WA + B 1!2 AB WB) (1) where wA and wB are the individual power spectral densities. As with shock and vibration. normalized to unity variance by division by the respective standard deviations oA and OB. (U) Although the analysis based on this aimplified model does not permit accurate prediction of the response of the strip to continuously distributed excitation. At the other extreme. and cAB 03 the nar rowband correlation coefficient.t fields. For completely uncorrelated excitations exhibiting no dominant phase angle. an expression to equation (1) must be derived for analogous the sum of an infinite number of random signals. the more antinodes contribute significantly to the summation. (U) Let a sinusoidal pressure p ut the i'th antinode result in a transverse 4elocity 8 * vi = i P" V. producing an interthe sum of mediate behaviour. the more complete the cancellation. and averaging over time * * 2  (A/ 2(1 I B/OB 2 A 2 = A2/O 2AB/OA B 2 + B 2 = > 0. The lower the damping. The important field characteristics will turn out to be not only the power spectral density of the sound pressure as a function of position and frequency but the pointtopoint correlation as well. but the ultimate purpose Js inference about practical situations. For a completely correlated wave at normal incidence. correlation density. This is a convenient simplified model to visualize and to use for inferences in advance of any calculation. as in Figure 1. for a completely correlated wave incident at the coincidence angle so that its trace velocity equals the velocity of propagation of transverse waves in the strip. will be an important parameter. For the present study the strip will be assumed infinite in both directions. I carried out a simplified analysis of the infinite strip by assuming that excitation took place only at the antinodes. attenuation with propagation distance. more properly. It is therefore worth summarizing here as a preliminary exercise. is of magnitude not greater than (U) That c unity msy be proved by squaring the sum and difference of two random signals A and B. The lower the damping. As we are concerned only with relative responses to differez. AB (2) Hence B IcAB i (3) (U) In the analysis to follow. pi i=2 i=1 0 i=1 i=2 k=2 FIGURE 1 k=1 0 k=l k 2 EXCITATION ONLY AT THE ANTINODES OF AN INFINITE STRIP. it does provide insights into the nature and effect of correlation. Damping. subject to a phase reversal according to whether the integer numbering a particular antinode is even or odd. the mechanical impedance of the strip on aisabsolute scale is unimportant. (U) At one time. the waves propagating from the antinodes should combine as the square root of the squares. the transverse waves propagating from the antinodes should all be directly additive. which will be chosen as the origin. the simplest test case (U) It is well knovn that the power spectral density for the sum or difference of two randon signals is given by yields mucb of the information one needs to know forthis. 236 .
and all possible cospectra. indicating a phase reversal before combination at the origin. We obtain pik w(f) = K2wp(f) tanh (15) = 0k = 1. for the k in antinode. this would be . generalization of Equation (1). we will assume a constant power spectral density wp(f). The latter are shown also as geometric means of all possible pairs of power spectral density. If. all cp(f) = 1. the quad spectrum or imaginary part of the crosspower spectrum of the sound pressure has an effect. ( Vk = Ok Pk' K5 1 (10) (U) Now assume random pressures p if(t) and Af. 6i a decay constant. cpik(f) = (I) ik.at. and the choice of sign for the exponents i and k has no numerical effect and will merely produce a form more similar to that of a derivation to come. and a transverse wave propagating in either lirection from each antinode and decaying exponentially. Equation (6) becomes (U) The square is V f2(t) = IP~ (t ) " Pf p8 (t ) wf) =K 2 w (f k i e" IiI+IkI) (12) pci(f). if the excitations are uncorrelabu out.. the origin.and we obtain at the origin. As phase shifts in propagation are not limited to 0 and 1800. EXCITATION AT ALL POSITIONS (U) We will now turn our attention from this preliminary exercise to the analysis of a more realistic model excited continuously as a function of position. Except in special cases. By expressing Equation I k=w pik (U) Average over time and let Af approach zero to obtain the power spectral density of the total response (10) as the product of two summations in i and k and applying the expression for the sum of a binomial series and the definitions for the hyperbolic functions. the positive sign for the power spectral density of a sum. 237 . (U) If all i Wvcf = 2p(f)/tan h 2(=12) (U) Finally. for any pair of I and k. multiplied by the corresponding narrow band correlation coefficients. taken witi. (U) For the problem at hand.k=YQ I k wpik(f) wvn(f) tanh 2 (/2) (13) 1/2 (f) Wpk (f) (f)Ic Fik [wpi pik M i=k= (8) (U) For incidence at the coincidence angle. a phase reversal when the integer i or k is odd. with a real quantity. indicating an odd number of half wavelengths spacing from the c. I Sl k PiAf (t) PkAf(t) (T) (U) For complete correlation at normal incidence.ign. In short. 8 8 k = 1. f hicions r ted Fillt Cp (f) =fI when I = k. SimiwI = larly. The Pkf(t) within a narrow bandwidth corresponding total velocity at the origin is and K (l)ke kl $ J (11) v Af(t) SpiAf(t) i=M k=w 0 k PkAf(t) (6) where K is a constant of proportionality. it would be a generalization of the power spectral density of a difference. All possible power spectral densities (ik) appear In the summation. we obtain = 2w p(f) wv(f) L i=. the summations become integrals. direct (U) These results are consistent with the trends predicted in advance but are not accurate for a strip excited all along its length.
the final result is i X(f) Hwp (f) 22 d ) X2H2W(r) a2+T cf. ~v a(Igl+lhla) (16) p 2 2 (0 +h. to weigh the realism of exiLting a lti'1i1ture by a field of one charuIcterintie at. (U) On the other hand. and A is the wavelength of transverse vibration in the strip. have been completely correlated except for phase effects. If we take the antinodes to be the centers of half wavelength segments of the strip. not confined to the integral values i and k.g. the final result is wvu M dlI2wp(f) P a2 + 2 2 U 1u . 238 i. the final result is X2 H2w(f) _ GU) vn = a 2 (17) low.~is (U) In Appendix 1. as in Figure 2. A2 2 (U) Up to this point.2) (8 (f)cos(IgjIhj 0)dh. I1 field (l' it 111'" imulation of tile effect of ferent characteristic. and Cpxy(f) w 0 for all other pairs of x and y. and excite each sagment by an independent normally incident wavo no that the excitations of different segmantgP aro uncorrelated.2yfl FIGURE 2 ANTINODES AS CENTERS OF HALFWAVELENGTH SEGMENTS. # to Equation (8) (U) The equation corresponding (U) For a completely correlated field incident at the coincidence angle such that the trace velocity equals the velocity of propagation of transverse vibration in the strip. . we aru partiulaPrly In 2 1 0 1 2 0 2 0 1 0 o 0 0 1 1 ° 0 2 I 11 . IgI . H is a constant. and g and h are continuous variables along the strip. We would like an indication of what Lppens to the response as the correlation distance of the pressure field approaches zero. The expression (")akc i (f) P dg +1 g w (f) v qIhd g+Ag (19) c( Igl+lhl )dh. the derivatior of a gcneral equation for the response of the continuous strip and expressions for three specific cases is carried out in detail.i for g Ag Let Cpg(f) = +1 and 0 I h < g + Ag. is replaced by the continuous expression C which approaches zero as Ag approaches zero. correlated field at normal completely (U) For a (e. the fields considered (f) describe the correlation where 0 and C charactelistlg of the incident field.3 ' pg (f)cos(igjhj0) h again dependent on both the transmission characteristic and the correlation of the incident fiela. incidence incident normally from a single distant source).
so that mn=ft. . . for verifying the derivation of Equation (20).. it would also show a marked difference in response to the different fields according to whether the number of antinodes is odd or even. . The ratio of coincidence to uncorrelated response is W (ff vc • (a 2 2 a2 (a +4V ) le. Meyer. Analyses such as the one given here may be useful in suggesting the amount of correction to apply. . however.2' .. .. The return wave from the m'th reflection at the far end. Introducing an exponential decay of correlation along the strip has the same effect on response as increasing the internal attenuation. an integer. . Warren A. Rn =Wvu(f) 2+3e'ae 3 .. (U) A simple resonant system decays according to (U) Consequently.. after stoppage of the excitation. . realism must be a compromise limited by practical constaints. (U) It follows that the correlation of the pressure field as well as its sound pressure level must be considered in establishing a simulation. . . immediately that vn W(f) 2 le "2a It follows CONCLUSION (U) The effect of the type of correlation of the correlation of the pressure field on ctructural response is by no means negligible of typical Q's or even for infinite structures structures long enough so that return reflections are negligible... ' . bracketing the possible responses of an infinite strip or a strip with enough internal attenuation tc make returns from ... As usual. (21) is the ratio of normal incidence to uncorrelated response. If the analysis givenhere were extended to cover a strip of finite length.. ..2a (22) 2+3e'oe'3a wVU(f) (U) We can find an equivalent Q corresponding to the decay constant a in the following way.... I am indebted to my colleague. is proportional to ACKNOWLEDGMENT is The time of arrival of the m'th return wave t = mn/f. Quite large effects of the type of correlation occur for Q's that are typical of airframe structure.. o ' .terested in ratios of response. the equivalent Q is Q = /2u (23) (U) The two ratios are plotted against both a and Q in Figure 3. .nd reflections negligible. When it is not possible to coritr0l the correlation closely. .. Consider a finite strip starting at g=O and ending at gun.. it may be desirable to introduce corrections in the SPL to compensate. 239  ."" ..
50
0.2 11.
0.4

0.6
0.8
1.0
a
10
S1
"

1
Rc FOR COMPLETELY CORRELATED . EXCITAlION AT THE COINCIDENCE ANGLE
20
10
5
0
2
o.1 7
€ ".
EXCITATION AT NORMAL INCIDENCE
0.1 0
0.2
0.4
0.6
0.8
.
FIGURE 3
RATIO OF RESPONSE P.S.O. TO RESPONSE P.S.D. FOR EXCITATION CORRELATED ONLY WITHIN HALF WAVELENOTH SEGMENTS.
240

V
APPenIDIX I INFINITE STRIP ANALYSIS OF All EXCITED AT ALL POINTS
(U) The equation + e
2 ft
very simply. (U) The analysis to follow will be expendited by using the complex exponential.
n
=
cos 21ft + J sin 2nft,
(24)
(U) Imagine a loop of magnetic tape, with a
where j is a unit vector along the axis of imaginaries, expresses a rotating unit vector sample of random pressure signal, played back continuously so as to generate an arTificially
as a complex sum of projections along two
orthogonal axes. The symbol e is the base of natural .logarithms. It follows that
periodic function, with energy at the fundamental frequency and at each harmonic. Let us examine the n th harmonic, which is among those within a small bandwidth Vf. Let the pressure at position x j (27ff)
ft eJ 2 11 f+ eJ2sft
2
(25)
p
=
P MP
(30)
or more simply, the cosine function can be extracted from Equation (24) by taking the real
at the k th antinode result in a transverse velocity response J (2rfmlxm)x
part.
(U) A phase shift 0 can be expressed as complex multiplier: e. o
J
&vm a
P
e
AX
(31)
e
j2
ft
e(2tft0) e
at the origin. The yXM is taken to be complex, for all possible phase shifts can occur in propogation Prom various x to the origin, but it is assumed constant over Af, equal to
cos(2,nft0) + J sin(21ft0)
Yxfo
(U) The velocity produced by all pxn within Af is
or
o e
0
J2ift . ej(2,7ft)
fI"
AvJ(2nf
Af (27) •f x J2nf
0) mAV mxm
cos(2ft0)  J sln(2dft0)
2JI x~ e x Pxm e Ax
( 2 (32)
so that
e
t~
cos(2nft0)
+Let J(2nf0) ej(2nt ) + e J (21ft 2
0)
(28)
"Jexm
(33)
Note also that d dt j2nft J2nfr j2 n ft '
(U) The total complex velocity at the origin
is (29)
241
/V
Fi1
Vxf
Af
Yxm P xm
m
J (2i~f ) dx 0
the transmission characteristic of 'the strip, are alwy and simple functions of frequency and position way be assumed constant over Af as follows:
I f
xn Af(34)
*
J(2f"Q0=)d j (20
dx
2021
X
xn e
G
G
(38)
x=e X (U) The total real velocity at the origin Is eyn
0
(39) (40)
y
VA f
G
P
cos(2 1vfe 0
)d x
eI "
x
y
eJ(" '
+)
"J"
Gxmui
xn Pyn e
0
dy
(4
j (211fro" Ox0+xC) ]dx,+eXW +x •M
E (3)A xn Pyn
•
dy (l
xm~d
(35)
which can be written in terms of y as the independent variable and n as the subscript. Obtain the square by multiplying the integrals in x and y together, and average over time
2
f f "=
J
yn)
P
P +
OS
0 0
sin (
V
f 2 dx f
Gx G P PAf xf m yn xm yn
+ 2f
+
rj
___
n
Iy
2
fe xy mon+ e
• dx
dx
xw yn xn yn Idy
exm
P
PP
P
2yn
0y) As( f
dy h3)
yn sInn(mxmy
G
Pxn
Pxn Pyn (U) 11ow let the length of the sample increase beyond limit and Af approach zero, more slowly so that the number of spectral lines within Af also increases beyond limit. The power spectral density of transverse vibration at the origin is
coS(xm.Oxn+0n
)dy" y(36)
(36)
Pn are random quantities, Pxn a and nnaerdo thle Now, te (U) (U)No, The 0 anid 0 are also random, although their xn ynoradj difference may may not be a simple function of frequency and ponition, depending on the degree of correlation of the incident wave. flowever, Gxm Gyn n 0 and 0 , characterizing the
Wv(f)
dx
JJ.L
GxGy
e0O
W
P Y
(f) M
+ jqp j
PXY
(f)j (hi)
y xn
yn
242
+ eJ
e
[Wpxy(f)  Jqpxy ()]1
dy
1
where w
is. the cospectrum or real part of
x is the quadIt follows that
J(0) dy
Wp(f)I2 1w/() c
PC PY u(f wL
f
"/
d
o"d
x
GXY
the crosspower spectrum and spectrum or imaginary part.
)k(f where tow
xy
) 
f
 '€
0
J
.PY
COSW(0)dy (5)
,
(45)
w z( r) )+ [ xx( ]l/
+q XY(46) and
let
g= 2 x/ A
 2y/A
sa
(53)
PXY PX adh
= Wx
qpxy(f ) (f 2wf.2
where A is the wavelength of transverse vibration in the strip.=
f)]/2)52
G(j 6f, (fkn
f
d9)Wvf
(U) Further, the magnitude W y(f) is obtained by mutltiplying out the complex conjugates and taking the square'root. Wpx(f) " PxM P o~lmn Let
)
9
hW9ph
]1/h2g~ Cxyfcs(~id CpYf)o(OOd (5")
WpXyf

Cos(00
)
Lt
= HeIgl
(55)
___
COfl
Px Xyf
nG
h0)+E Ieah (56)
l2o:(19llhl) and
xm
yo(9f
P
sln((xm. yn)0 yn(D"' 2 Sin(0 m0
=
(57)
fW
((f)
9f) = wplf), (58)
(U) It is important to note that within etch product the m's and n's of one summation are independent of those of the other. Without actually assigning new symbols to one pair of subscripts and multiplying out, we can see that the result would reduce to a single summation of terms each involving four pressure amplitudes multiplied by the squared cosine of a conpond difference angle, and, if all these difference angles should be zero, reaches a maximum value Max W (f) Wpx (f) w (f) W f) 1/2 , (19)
a constant with distance.
2 A2 11 w (f)
w (
v
4
d
.
I
ea(Igl+lhl) .f
Cpgh~fes(
gl.l
)h
which is the same as Euation (16).
w when Af approaches zero, which can also be inferred from Equation (3). (U) If we define a new coefficient Cpx(f) p
*1(f)/
(U) Consider first a field completely correlated so that Cpxy(f) =1 for all x and y, and normally X incident so that 0 o.
Wpxy
Wpx(fW (OW y M
f) Mv/2  A2CWP
,px
f
ea(Igl+lhl)
,
"
(50)
243

,
p~ which~~5 12n~ .g9h)J Finally cosie a4on il uhta X2H2W (f) is the Eqaton(1). g+2rgdg )2 2w M (f aO) 0 Jc(g+h)4 4 Iffo f&5hJguhdh] 2 2 A HW M 2 U0 xc~i 5120 ~zJu 1 2 A ~gh)+ (f) .~ poiie2 in A H~w I 2 2 (f ''gjIhI)A24 e~ whic 2 22 f i we )ajrhd3  'Jdg Boh)r istesaea qato 1) 4o A2HHVw (f) d e1d U ialcnie 1_gj on il uhta k4 k+2 PX excep haeefetHs fo ht 1 (f) w 2 24 X2H2W cos~gjjhjh~d +2 (f gf k2 8 k _c(j~jjhj+J.r P 2 e0 d9H~ +f (.j~jjhj ed ~j  [ r~ii .. jh~j)Jv 'j h . ma~ja aS .
.1 (.j (k+i) e + A2[J /2 + j k4)] ka + C0 jJ  AH~wV)~ + e  +e/ e +~I/ ~2+~. dik .kyp +C 2 4. 2 l~e2a edg kfki 2uw 2  dg 2 kl ejIhIJ eI dh which is the same as Equation (20). .3 2a 3 f 2 ) 2H w (fM 2 2 ~~ Hw (f 2) +e~ a I . 2 )22H W  * 2  1a 2a)2+2e2t e 2 3c 2 2 2m +e:c 2 +j~1 +0 +0 2 e [1. . 2 2 2 HW ef~ 5 cg+Jirg Mdg f 2ahjuh e dl' + I.+f 2 9.k fT IF Cag+J1Tg d k4 +2 ahith dh 2 2 _g e +2~ I a I 2 +~~~~[.2 12 245 .chl+jwilhlh e .
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.