The moment-magnifier method applied to brick walls
Transcription
The moment-magnifier method applied to brick walls
4 . b. 3 THE MOMENT -MAGNIFIER METHOD APPLlED TO BRICK WALLS CARL TUR KS T RA JOSE OJINAG A Department of Ci,)il Engineering and Applied Mechanics , UcGill University , !1ontreal , Canada TllE MOMENT- MAGNIFIER METHOD APPLIED TO BRICK rvALLS DIE MOMENTEN-VERGROESSERUNGSMETllODE IN DER ANi.'ENDUNG AUF ZIEGEV1AUERWERK This paper summarises the results of a study of the ultimate capacity of vertically loaded brick walls . The study is based on the moment - magnifier method von Versuchen zusammen, die die Bruchlast von verti- which is used in steel and concrete column design in kal belastetem Mauerwerk bestüwnen sollte? Der Unter - North America . Extension to brick mas onry design would provide a con s istent approach in limit states grunde , wie sie auch für den Entwur f von Stahl - und design methods presently under development in Canada . Betonsáulen in den USA angewandt wird . In the method, bending moment- axial force inter action Eine Uebertragung dieser Methade auf den Entwur f von relation ships for shor t walls are used . However , the Diese Untersuchung fasst die Ergebnisse einer Reihe suchung liegt die Momenten- Vergrosserungsmethode zu- Ziegelmauer wer k WÜY'de zu einer konsequenten Annâne- bending moments found from conventional structural rung an die derzeitig in Kanada entwickelten Entwurf9- analysis are multiplied by a magnification facto r de - methoden fún r en . In dieser '4ethode wird das Zusammen - pending on wall height, stiffness and end constraints . wirken von Biegemoment und Axialkraft bei kurzen Mauerl<)erksteilen zugrunde gelegt . Jedoch we1'den die In the paper the basic assumptions of the method ar e aus üblichen Untersuchungen gefunderten und bekannten presented briefly and r elevant in f ormation on stress - Biegemomente mit einem Vergrosserungsfaktop multi- strain characteristics and br ick wall behaviour are pliziert , der von der 'Iandhohe , der Steifheit und r eviewed . der Einspanmmg abhá·n gt . The method is applied to a large collection of published test data from Europe and North America which has been set up in a computerised data bank . Test data includes a variety of brick- mor tar combi nations, wall slenderness and end conditions , and load eccentr1:ci ties . The limits of applicabil i t y In der vorliegenden Untersuchung werden die der Methode zugrunde gelegten Annahmen erortert. Es f olgt eine kurze Uebersicht úôer Charakterika der Spannungsbeansp~Achung und úôer das Verhalten von Mauerwerkskorpern . Die Methode bez ieht sich au f eine of the method ar e di scussed and conclusions drawn as to the advantages and di sadvantages of the approach in wall design o grosse Anzahl ver off entlichter Versuche au s EUr opa und den USA , die in e1:ner l<omputeY'ge steueY'tc<n Datenbank zusammengef asst wurden . Die Versuchsdaten schliessen verschiedenartige Mortel kombinationen , Schlankheiten, Einspannungen und ausmittige Lasten ein . Die Gr enze der Anwendbarkeit dieser Methode wird dis kutiert und schliesslich werden Folgerungen füp den Entwurf von Mauerwepkskó'rpern gezogen . 4 . b . 3-0 LA METHODE DU " MOMENT - MULTIPLICATEUR " DE METHODE VAN DE "MOMENT- VERMENI GVULDIGING " APPLIQUEE A DES MURS EN BRIQUES TOEGEPAST OP BAKSTEENMUREN Cette communication résume les résultats d 'une re- Deze mededeling vat de resultaten samen van een cherche de la capacité limite des murs en briques onderzoek naar de grenskapaciteit van vertikaal chargés verticalement . L ' étude est basée sur la belaste baksteenmuren . De studie is gebaseerd méthode du " moment - multiplicateur " largement op de 'moment- magnifici " methode welke in Noord- utilisée en Amérique du Nord pour le calcul des co- Amerika veel gebruikt wordt voor de berekening lonnes d 'acier et de béton . Une extension de cette van staal en betonkolommen . Uitbreiding van deze méthode à la maçonnerie serait une étape importante methode tot metselwerk ware een belangrijke stap dans la direction du calcul des valeurs limites qui in de richting van de grenswaardenberekening die sont actuellement développées au Canada . thans ontwikkeld wordt in Ranada . Dans cette méthode on part du rapport moment de In deze methode wordt uitgegaan van de verhouding flexion - force axiale pour des murs courts . buigmoment - axiale kracht bij korte muren. Les moments de flexion trouvés par les méthodes de cal- De buigmomenten gevonden door konventionele reken- cul conventionnelles sont toutefois affectés d ' un methoden worden echter vermenigvuldigd met een multiplicateur, basé sur la hauteur de paroi , la ri - multiplikator, gebaseerd op wandhoogte , stijf- gidité et les tensions limites . heid en randspanningen . Dans cette communication les hypotheses de base de In de mededeling worden de basishypothesen van de la méthode sont brievement présentées ainsi que d ' im- methode kort voorgesteld, samen met de belangrijk- portantes constatations en liaison avec le rapport ste vaststellingen in verband met de verhouding tension - transformation et le comportement des murs spanning-vervorming en het gedrag van baksteenmu- en briques . ren . La méthode est appliquée à un grand ensemble de ré - De methode wordt toegepast op een grote verzameling sultats qui ont été traités par ordinateur et publiés van resultaten gepubliceerd in Amerika en Noord- en Amérique et dans le Nord de l ' Europe . Les données d ' essai comprennent une grande variété de combinaisons Europa die in een computer behandeld werden . De proefgegevens omvatten een grote variatie van bak- de mortiers - briques, d 'élancements de murs et de steenmortelkombinaties , muurslankheden en randvoor- conditions limites , ainsi que d ' excentricités . waarden alsmede excentriciteiten . Les limites du champ d 'application de la méthode sont De grenzen van het toepassingsgebied van de methode discutées , de même que les avantages et inconvénients worden besproken alsmede de voor- en nadelen voor pour le projet de maçonnerie . het ontwerpen van metselwerk . 4 . b . 3-1 have bean excluded . I [HRO OUCTIO~J To de velop m8thods of limit states design procedures for preoictlng average ultin~te load capacity. a know la dge of ths va riability of capacity about these aV3ratSS mUSL ~8 estaol ishe d . The objective Df thls paper is to sxamlne the LHe Df the momant magnifier a~pro a ch i n tne special ca se of singl o wythe unre illforc8d brid, \·la 11s subjected to 8ccentric vertical end loads . The design prob lem involved is the conventional one of relatin~ sactio n capac it y to loa ding conditions . end constraints, l'lull geomet r y. and the properties Df st a ndard reference specimens . However. the problem ia rela~iv9 1y difficult in orick wall design bar:ause of the ;,xis tence of two components with different ms chanical properties leadi ng to complex strGss distribution s and failure criteria [1,2) . For til e case considered . a numb5r Df simplified theories (3. 4 .5. 6 . 2) and emp iri ca l metilods (7) have been deveL::ped . PraGtical design methods should be relatively simple and consi ste nt ~ ith princip18s of strL~tural mechanics so that s ituations beyond test conditions can be tro~cod. Df cünsiderabls interest is the possibility Df using tne morr.ent -magnifier approach suggested by Yoke l. Ma they and Dikkers [8 . 91 for concrete b lock rr. ~ 3'Jnry and b rick mesonry (10). Such an app roa ch would provi de cons istency with steel and reinforced concrete de sii~ matho ds . Mo reover, it would provide a r3tion al basis fo r consideration of general wall end ~nnditio ns. l~ading . and pcssibly the effects Df r einfCl r cing . This paper examines application Df the simplest posslble form Df tlle momen t-rrlag ni fi el' appT'oach tog2ther with re15ted wall p r operties. Attention is restr i cted to variations in wall capacity . The question Df safa ty factors to be used in design is not considered. SHORT SECTION CAPACITY Fundamen tal to any study Df wall beha viour is the capacity Df a short sRctl on subj ec tod to an axial f orce F' and a bending moment M about the centroidal axis . For brick wal ls. several cases must be considered dape nding on whether the section is in comprassion through its thickness or axp erien c es tensil e st r ain . In the latter case . bahaviour depends on ~hether or not the tensile capacity has been exceeded . For an assumed linear strain variation through the wall . resultant axial forces and bs nd ing moments can readily be calculated for any assumed stressstrajn charactgristics . Shovin in Fig . (2) are the axial fo rc e - banding moment relationships for the three ideal1zations Df Fig . (í) for cases corresponding to a ttainment Df the u ltimate rupture strain E. on the compression face . Results have been nondimensionalized using the reference ca~acity Po an d the conventional linear kern bending moment Po t/12 . Also sh:Jwn are wall test results fro rn Ref. (7) wi t :, a slanderness ratio kl/r less tnan 40 for walls tested axially . witn one end flat . or in double curvature, and lass than 30 f or single curvatura sonditions . Such walls can be expact~d to havB somewhat less than the short section capacity . Comparison Df the analytical results show that nonlinearity reduces the kern Bccentricity and inCI'eilSeS axial load capacity fo r a gil/en eccentri city . The ex periment a l resul t s in dicate t ha r cólpaclty decreases less rapidly with eccentricities up to t/3 tnan even a linear-recta ngul ar stress-strain diagra!l1 without te nsile stI'ength would predict . A simi l ar "straingradient effect " was noted by Yakel et aI (8) fo r concrete block masonry . Such effects may be oue to changes in transverse stress patterns under ncn-uniform strain and suggest thet the a xially 10i:l ded prism is an inadequate measure Df mechanical properties. REFERENCE PROPERTIES THE MOMENT - MAGNIFIER METHOD In the absence Df reliable models rela ting masonry prcperties to the properties nf bricKs and mDrtars . s mall piers or prisms hava be8n used as a standard measure Df masonry behaviour . The choice Df a standard rsfer8nce lnvo l v8s considp.ration Df the effects Df test conditions together with t he easo of fabrication and is somewhat arbitrary . I n this st udy . the single wythe prism with a haight to thickness ratio Df 5 te sted between flat rigid plates (11) was adopted as a practical reference . Al I wa ll capacities P per unit length are referred to the capacity P per unit length of s lJch prisms . o Numerous tests (12 .13 . 14 . 15) have shown that the stress-strain chBracteri stics Df brick masonry is generally non - linear . Some tests sugg es t that segments Df a parabola may be used out the behaviour in the region Df high compressive stresses does not saem to have heen extensi ve ly investigated . Tensile stress capacity has not been well defined . In exarnina tion Df short section behaviour the three idei:llize d stress - stra i n diagrams Df Fig . (1) were used. The ini ti al tangent modul us E has a significant random variat io n . Exarr.inat ion Df 142 experimental results (16) innira~ed a very good correlation with brick comprassive stre ngt h fb . The equation As wa ll slenderness increases . lat e ral deflect j.ons become s i gn-Lfican t as axial forces act throü"h the deflectians to modify the distribution Df bendirog rnoments along the \~all heigM . As a r"s ul t . the l ocation and magnitude Df tha maximum moment is va1' iabl e . For exarnple. an eccentric load within the [,erro at the ends cm lead to eccontrici ties beyond tile I,ern at mid height . Details Df behaviour depend on the wall height 1. radius Df gyration r. end condHions. and the shape of the moment diagram found fro m elementary anal ysis. The moment-magn ifier method is a device for converting the bending moments alo"g the lsngth Df a beam-column to equi.val,mt short section bendir,g mO!l1ents . The section is then design e d using the shor~ 3ection interaction relationsh ips. the applied axial forces , and the larger Df the applied end moments or aqui valent "magnified " moment o Tha methoo was davelope d for steel sections and has been ad~pted to concret~ column design (171. For the load case under study . the equi valent moment M can be I-Iri tten in ter-ms Df the axial force and m~~~mum applied end eccentricity e in the form max M E 145 . 000 + 220 .6 fb (psi) (1 ) had a cor rel ation Goeffici ent Df 0.941 and a standard error Df e sti ma~ e of 360 .000 psi . Waak lima mortars mag LC Pe max (2) tl . b . 3-2 S t ress S tre 5S Stress +-. E . TE. E. I Str ain (a) ( b) Li near Strain Paraboli c FIG. 1 rDEALI ZED E. 2 St r ain I (c) Linear Rectangle STRESS STRAIN DIAGRA MS Axia I Force 1.0 0 .5 1.0 FIG~ E r ror (% po ) 15 10 ... -5 ... ... O X X O ... ... ... ... X ... "" 150 X X ... X Length Factor K = I K 0.7 K = 05 200 Slenderness Ratio = _k_1 r X O -2G FIG. 3 SHORT SECTION INTERACTION DIAGRAMS X ... 5 0 X - 10 - 15 ... ... 5 Effect ive 2 ERRORS FO R AX IALL Y LOADED WAL LS 2 .0 3 .0 4 .b . 3- 3 where 1 (31 ----- p- - L 1 - p- c, c ( 41 p ( 51 cr ll18 fectur L ll1r: l uooc; the effect. s of r'lateri éll stiffnass tllrough ths fIlodulus Df e lasticity t: , section goome try through Lhe moment Df i nertjoa I , the length x" emd en,J s'Jpoo,-t condi1: io n th l'ougn the effective lsngth filet!),- :'. . -;-he shape Df the e lfomentary í11Dme nt oiagralil l5 introouced b", the faetcr C. To dea l with nominally axiéllly loaded Cilses, a mllumum a r acci d",ltaI anel dccentri cit y G1l1s t be i ,-,troducB-J . lhe n~ thud was deve lope d fGr lineilr sl a stic sections ,;iLh c:onsta nt c r us s-s e C1:i OrI3 an d melst be approached ~ fth caution in mason r v as in concrete design o Non :Lj r,e ari ty Df mechan ica l pl'operties can increase l~ Leral deflections relativs cO an al ysis based on the j ~ itial tangent mod ulus. le nsi 18 cracking before failure Jea ds to var iatLuna in c:ross-sectional geo metry ':ln d can lead t,) stability failures n ot con ,, 1dered 1.;'1 the method . In con c rete design these pf +ects hciVO ~een i nclujed by the ~SB Df modified fl~x urAl riRid ities t I and J i rrdtations on the range 'o r app ] il' 6t LJiI l i: p recIedo ~;t clbi Iit:, f aj lures . 1\3 cn ir.:!.tiiJl stap i. n , ~\/Jluc t i'l n CJ f 'thc app l ication DF t~, e approach t u prEH.dc·;:. :i..l.lrl of \Jariat.ions j.n wall r:'lr,Bc it y , the sjompl e st possibln fo r ;]', wa s a pplied to plJblished wall data in ,,J hich uricK 3trength , pris:n stcength, a nd test condit1ons w a~8 elearly stated . oI ,;a ini tial modu l us frClm Eq o (1; ,483 uscd tog et her cá th a linear st r 2ss -s tr aJn ,j ~agr a rr" zero te nsi l e ~apacity and ~r:e un cracked s8c~io n r ad i us af gy ,'at i on . In p;enera l, íour caSE,S rêust toe sxam j.ned lef1dir,g to the fo llowing fo~r eq~etjO "5 Case 1 Fcr the give n values Df th e anti ecco n t~ i c 1ti 8s , the t heorstir:al capac i ty reduction fa e i,c r (P/P ) TH is t he least of tM8 fou r Bol utio ns t o these equat ~~n~ . To cC I~pa ct the re ~IJ I te for ô var'Lety uf brick dnd mor t ar strengtl's the deviatüm Df the theo r etical capa ci t y from the exp eri me ntal ca pacity hós oean calcu15ted as a p e rc e ntage cf tha pris m c~pa c it~ to obtai~ tl18 relativ8 8r rc r ( 101 A negati v8 valu8 cf t hi s Rrror in djcates vative tn eor8 tic~1 predi utio n . ~ conser - As mentioned pre viously , ml. nl mUm eccentrtcities must be assumed to predict the heha viour uf nomi nal]y axially loaded walls . Since the form J& such eccen tricities a long t he height of ô wa ll mu st aIs o be assumed , a va riety Df ch o ices are possiole . In analysis it was conservat.ivel y assume d that the rúni mum eccentricity was constant A1Dng th e height 18arl ing to single curvature . Show n in Fig . (3) are the errors ob cair led for t h e da ta cf Ref . l71 wi t h a min i mum eccentricity 8f ~ per c8n t Df the wal l thickness . lt can be seen that lhe methcd generally predicts the v ariati on Df capacit y wi th heig h t end end conditi o ns with an abso lut e error thdt mi ght docrease with slenderness . tiy su i ;:able cho ice of mi n imum e ccentri c iti es , t hB grror i n p~ediction c an be adjusted tow ar ds more cons8 rv~t t lie r2su l t s ~nd the variat i cn with slerderne~ s ca n be 0'':lr:ged . ECCENTR IC.A,LL'( L.O,\,Jéll !t/.A, ~ LS In the case Df eccentrically 10 30",rj 'NC" ls, t!,fe. bd~ Le variables are the effective Rlend e~ne s s r dti n kt/r , t h e r e lati ve end eccentricities eJe~ ,md the maxin:um 8nc1 eccentricity e . Shown ü , F i ~ . (dl ""re U-,e absolu te predic ti~n ~~~ors obt~inB ~ fo r the Ja ta oi' Ref . ( 7 , 181 . These data indicat e t:hilt 'lhR rnethod is gensrally conse rv ê:ti V2 wi t I"! A i") q rl' ~ r 7 t:bt Ch::c:'Ra:::y,; with sle ndernBss , ~n d Uncracked S8ct i on . Fai lure Rt DISCUSSIClN [ ~ 01= c Ca'-ip 2 0.5 < 1 + 6 e I!lax ~- /t < 1 ( 61 o Uncrilcked ::J8cti.Cln , Failure Al c n g I-ieight <' ~ n. 5 __ p _< (7) o Case 3 Crilcke d Saction , Fai l ure a t En d ~ L Célse 4 4 [1 - Crack ~rl 2 8 /t l max <[J . 5 0 -< ~ p - ( 81 u SRc t i on , Fai l ure Along Length [J<~ <0 . 5 o (9) In evaluation Df the ability of th'l morr.enr. - m,~gni fi8 1 ' me th od to pre~ ic t ~aria tions in wall capaclty, th9 assumptio lls adopted rnust be considered . {l,s shovm iq Fig. (2) , use Df a linea r stress -st rai n dia gram sV's te mati cally und erestimates section capacity for 8ccentric loads . USB Df the i ni tial modulus Df p Jes ticity to c:a l c ulate criti caI l o a ds le a ds to un ds; estimation Df t h e effe c ts Df lateral deflection s wlth decreasing erro r as t he slenderness increases, As ~ res ult , the method can be e xpecte d te oe c on se rl:et~ 'ie for relatively s hort ec c sntrically 10a deG wa l ls , As s l enderne s s inc r easE:s I.htJ noncor,5el'võtisr.1 Df tnG effects Df the laterõ o" [J8Fl 8ctiom; comUones w~t,-, the conservatis!l! of sh ort e e C'tin ll ~ r:)p",; · ti es l Cilding to a reou ce d total el"rr:('o The r e sult s Df 'lhis i> t l!dy SLl g6est l.r,at USE c,f Ju,oa r elastic mode l for h ri.cK masonry 1s no t war('anteo . ,4 more realistic defi ni tion ~f t he st,e s s-strai n charac teristics i s required . S"C'.cessfL!l app l i cõtion Df' t.he moment-magnifi e r app roacr wi ll ; 'equire coth lmprol/C'c shol ,t section prc;p8rtios and ~ cetal l ed ~tudy (1f l ateral defl ections TO ystabllsh ~hp val~e5 cf f rlnd r to De use d in designo 4.b.él-4 Fl nallv ~t shQ~lrl be noted that some variations obtsined resu lt fr nm the fact that prism capa ci ty is normally found from amall samples . As a result the ratio (P/r lEXP may indicate a greater vari atio n in wall capac~ty P than ac t ual l y exists. Related studies (19 ) and repeoted wall testa suggest t ha t wall capaci ty may not be as vari able as uni t a nd priam props rt ies . REFERE I~CES 1. Hilsdorf, H.F., " Investigation i nto the Failure Mechanism of Brick Masonry Loaded in Axial Compression°, Oesigning, Engineering and Co nstructing with Masonry Products. F . B. Johnson. ed., Gu lf Publishing Company. Hou5ton , Texas , 1969 , pp . 34 -41. 2. Sahlin , Se ven, Structural f'lasonry , Englewood Cliffs , N.J.: Prentic Hall, 197 1 . 3. Chapman, J . C. and Slatford. J .• "The Elastic Buckli ng of Brittle Co lumns", Pr'oceedings of t he Institution of Civil Engi neers. January. 1957 , Vol. 6. Paper No. 6147, 107-125. 4. Poulsen, E. and Risager. S " "Tl;e Bearing Capacity Df Linear Elastic Brittle Colurnns". Bygningsstatiske Medde leser . Vol. 36, No . 3 , Copenhagen . Oenmark , 1365. 15. SCR . "Compressive and Transverse StrenEth Tests of Eight-Inc h Brick Walls" . Structural Clay Products Research Foundation. Research Repurt No. 10, Geneva , Illinois , october 1966 . 16 . Eskenazi , A., and Ojinaga, J . and Turkstra , C. J .• "Some Mechanica l Properties of Brick and Block Masonry ; Interim Report" , St ructura l Masonry Series 75-2, McGi11 University , 1975. 17 . MacGregor , Jarnes G.• Breen, Joh n E .• and Pfrang. Edward O.• "Oesign o f Slender Concrete Columns". American Conc reta Instituto Journal, VaI. 67, No. 1. January 1970, pp 6-28 . 18 . Watsein, O., and Al l en, M.H., "Compressiv8 Strength of Brick Walls with Large Eccentricities", A. S.C.E . Nation al Structural Engineering Meeting , Meeting Preprin t 1400, Baltirnore, f'la ryland. April 19 -23. 1971. 19 . Fisher , K. • "The Effect of Low Strength Bricks i n High Strength Brickwor~," . Proceedings Df the British Ceramic Society. Load-Bearing Brickwork (4), No. 21 . April 1973. E rror (% Po ) 5. Hallar, P .• " Load Capacity of Brick Masonry" , Oesigning Enginee ring and Constructing wit h Masonry Products . F.B . Jo hnson, ed ., Gu lf Publis hing Company, Houston. Te xas, 1969. 10 ~--------~-----------r---------'~---~---~ 7. 8. 9: Monk, Clarence B., "Column AGtion of Clay Masonry Wal ls", Oesigning, Engineeri ng and Construct i ng with Masonry Products , F. S . Johnsoll, ed. , Gulf Publistüng Cmnpany, Houston , Texas, 1969 . -3 O Struct ural [1 ,,'1 Produots Insti tute . Recommended Practice for En gineered Elri~ f', f1asonry . McLean, Virgi nia : SCPI , 1969 . -50 Vokel, Fe l ix V., and Mathey , Rebert G., and Oikkers, Robert O.• " Compressi'Je Strength of Slender Concrete Mas onry Walls ", Nati onal Bureau of Standards , Building Sc~encB Series 33. Oec. 1970. Vokel. F.V. and Oikkers , R. O.• "Strength of Mason1''1 Walls under Compressive and Transverse Loads ", National Bureau of Standards (U.S.). Bldg . Sci . Ser. 34, March 1971. 10. Vokel. F . V. and Oikkers , R. O. , "Strength of Load Beari rlg Mas on ry Walls ", Structul'al Di vision, Proceedi ngs Df the A. S . C. E .• May 197 1 . Vol.97, 1593-16Ll9 . 13 . SeR , " Compressiv8 and Transve rse Tests of Five Inch Brick Wa lls" Structural Clay Products Rosearch Fou ndatio n. Research Repurt No . B. Gene va , 11lino1s, May 196 5 . 14 . SCR , "Comp ressive,lrans vérse, and Racking Strength Tests of Fou r-Inc l; Bri ck Walls " , Structural Clay Prod ucts Research Feundat ion . Research Re port No. g, Gbn'l 'J a. Ill1n015 , Aug . 196 5 . Curvature X,e Max =t/6 ... e Mox = t/3 X -40 E rror (% pc) 10 50 O O -10 -20 -30 11. Canadian Stan dard s Association , "CSA A23.2.13 , '1973. Te s t for Compressive Strength cf Moulded Co ncreto Cyllnders ", Rexd~le, Ont ari o , 1973 . 12. Glõ,',vi lle. W. H. • an d Ba rnett. P , W., "Mechanical PrOpeI'ti85 of Bricks and Brick\<lOr k Masonry" , Oepartme nt of Scientifi c and I ndust r ial Resea rch. Building Research. Specia l Report No. 22 . Building Research St atiorl. Ga rston. Watford. Herts, His Majesty's Stôtiorlery IJffice. London, 1934 . 200 X e 1 /e2 = LO Single -20 6. i50 100 50 - 10 kl O X ... X "- ... X 8 xlOoA ... X~ X XX X 150'" 200 kl ,-- X e 1 /e 2 = O \ " Max- t /6 ... e Max = t/3 O e Max =tl2,4 X ... ... e Error (% pol 10 ... -10 ... ... ... -20 X -30 50 X ... ... X ... kl ,& --r--x- 200 e I le2 = - 1.0 Double Curvatut'e X e Max = t /6 ... e = t /3 Max 100 150 X X FrG, 4 ERRORS FOR ECCENTRICALLY LOADED ~IAU . S