NNNV

SWrucWural Analysis of HisWorical ConsWrucWions – Jerzy Jasieńko (ed) © 2012 DPNH ProcławH PolandH ISSN 0860-2395H ISBN 978-83-7125-216-7

TeE NEt Yloh CoYSTAi mAiACEW IolN STorCTroE tITelrT ENdINEEoINd aonald criedmanNI dabriel mardo oedondoO ABSToACT qhe kew vork Crystal malace was constructed in NURPJ4 in imitation of the iondon Crystal malace of NURNI and held an exhibition imitating that housed in the iondon buildingK qhe kew vork building was an octagon in planI with two usable levels and a high roof surmounted by a large domeK ft was the first large building in the rnited ptates to have all of its major structure constructed of ironW the columns were cast ironI the floor structure was a combination of castJiron wroughtJiron builtJup girders supporting wood joists and plankingI and the building skin was plate glass and sheet ironK qhere was little formal engineering design of the iron structureK oatherI the architects created a framing layout and the various ironworks involved provided members to fitK auring the building’s brief life – it was destroyed by fire in NURU – there were reported problems with the structure that led to ad hoc repairs and reinforcingK qhis paper will analyze the major elements of the structure using EaF the techniques available in the rnited ptates in NURPI EbF more advanced ironJanalysis as was available in the late NUMMsI and EcF current analysis methodsK qhe analysis shows that much of the structure was not properly designed for full loading and that there is little correlation between the results of analysis and the more empirical performance of various building elementsK heywordsW Cast ironI trought ironI Column capacityI Beam capacity NK eISTloY lc TeE BrIiaINd NKNK Introduction qhe kew vork Crystal malace project was a literal imitation of the iondon building and exhibitionK qhe Board of airectors of kew vork’s “Association for the bxhibition of the fndustry of All kations” stated immediately after its formation in NUROW

ft is scarcely necessary to sayI that the idea was suggested by the brilliant success that the attended the iondon bxhibition of last yearKKK no nation gave more striking proofs of intellectual capacity and vigorI applied to the useful artsI than were manifested by our own peopleK ft wasI thereforeI a natural suggestionKKK that we should not only wish to see a like bxhibition in our own countryI but that we should desire to reJproduce in it the beneficial effects that had resulted from its great prototype xNzK

qhe kew vork exhibition was planned as a copy of the iondon exhibitionI and was to be housed in a building that would be a partial copy of the iondon buildingK qhe city government provided land in “oeservoir pquareI” waste land next to a large aboveJground distribution reservoirI contingent on the exhibition building being constructed of iron and glassK qhe exhibition itself was organized by a forJprofit private corporation xOzK kew vork became an innovative center in iron and steel construction during the era of the “fireproof building” starting in the NUTMsK fn NURPI the use of metal as building structure was still new and still partially experimentalK Cast iron was used first in fireJshutters and window framesI then storefrontsI and finally entire facades by the late NU4MsK Cast iron facade use grew steadily in the rnited ptates until the Chicago fire of NUTNI with kew vork’s games Bogardus and aaniel Badger as leading promoters of the ideaK trought iron developed laterI with the first use of fJbeams rolled specifically

N mrincipalI lld ptructures bngineeringI kew vorkI dfriedman]oldstructuresKcom O bngineerI lld ptructures bngineeringI kew vorkI gpardo]oldstructuresKcom

NNOM

for use in buildings Eas opposed to railroad railsF in NUR4 xPzK Before thenI industrial wrought iron production was limited to platesI anglesI railsI and small channelsK qhe structure of the kew vork Crystal malace therefore consisted of castJiron columns and beamsI and wroughtJiron girders built up of angles and platesI all supporting woodJjoist floors and roofsK NKOK Construction qhe building designI by architects deorge Carstensen and Charles dildemeisterI respectively aanish and derman immigrantsI had a twoJstory greekJcross core with oneJstory triangular infill spaces between the cross arms to create an octagonal planK qhe width of the building each way was NNN m and the central dome was 4R m high at a time when the tallest structure in the city was the US m tall masonry spire of qrinity Church x4zK

cigK N kew vork Crystal malace ECurrier C fvesI NURPF qhe frame totaled NPSM tonnes of ironI a small percentage of which was “moorish” ornament but most of which was base structureK qhe columnsI typically placed on a UKO m square gridI were octagonal in section and hollowK qhe exteriors were identicalI with the octagonal exteriors nominally OMP mm in diameter; the wall thicknesses varied with loadK qhe columns of a typical bay supported castJiron girders Eat the upper floor and roofF in the form of doubleJdiagonal trussesI which supported wood joists and woodJplank decksK qhe main northJsouth and eastJwest aisles were NOKS m wide and mostly open to arched roofs three levels aboveI but the upper floor extended across the aisles at the end baysK qhe upperJfloor long span across the aisles was supported on wroughtJiron girders in the form of doubleJdiagonal trusses composed of angles and platesK qhe curved roofs over the aislesI the central crossing domeI the vertical side wallsI and some portions of the flat roof were glazed in small panes coated with enamel to reduce transparency xRzK thile there were various municipal laws controlling constructionI there was no building code in force at the time of constructionI there were no engineering or construction nonJgovernmental regulations or regulatory agenciesI and there was no consensus in the country as to engineers’ role in building designK qhe architects specifically acknowledged the assistance of gulius hroehlI a dermanJborn kew vorker who worked as an inventor and engineerI in the iron designI but it should be noted that American engineering in NURP was quite primitiveK

NNON

cig OK lneJquarter framing plan of upper level at NWRMMK kote that EaF the building is symmetrical about both centerlines and EbF the hatched area is low roofI not floorK Beams marked AI BI and a are castJiron trussed

girdersI beams marked C are wroughtJiron trussed girdersI columns are indicated by circles

cig PK lneJquarter framing plan of main roof at NWRMMK kote that the building is symmetrical about both centerlinesK Beams marked B and a are castJiron trussedJgirdersI beams marked C and c are wroughtJiron trussed girdersI beams marked b are castJiron archesI beams marked d are wroughtJiron dome framing columns are indicated by circles NKPK aemolition qhe Crystal malace was destroyed by fire on lctober RI NURUK qhe heat damaged the building and its contents beyond repair or reuseK aespite the adjacent reservoirI firefighters were unable to stop the fireI in part because of a lack of water pressure in the nearby hydrants xSzK qhe fire burned only the flammable portion of the building contents and wood flooringI but the associated heat destroyed the

NNOO

building through thermal stress in the glass and cast ironI and warping of the wrought ironK By the end of the fireI the building frame was collapsed on the groundK bven before the fireI the exhibition was considered to have failed in two important regardsK cirstI the exhibition was a business failureK pecondI the building proven to be flawed both architecturally and structurallyK qhe complex roof leakedI a serviceability flaw that the architects blamed on changes made to their design for the roof drainage by the supervising engineerK Canvas was strapped over the outside in an attempt to stop the leaks but was both ineffectual and damaging to the building’s appearanceK qhe structural flaw found at the time was more serious and was addressed by the supervising engineerW the architects state that the wroughtJiron truss girders showed “defects in their lower flanges and the diagonals” under a test live load of 4KM kmaK ft is unclear what the word “defects” means in these circumstancesI although outJofJplane buckling of the compression diagonals was likely one piece of itK sisible excessive deflection or warping is a sign of potentially catastrophic overstressI even if no numerical analysis was performedK qhe general solutionI installed before the building opened to the public was to tie rods from the top of the aisle columns Eat the roof main springJpoint of the aisle archesF diagonally down to the first panel points of the trussed girder roughly OKR m from the girder endsK qhis reduced the clear span by 4MB and solved the problem to the satisfaction of the supervising engineerK cour of these girders were reinforced through the introduction of a tension rod installed below the lower chord of the truss in the manner of a tensionJrod girderK OK ANAiYSIS OKNK Analysis Context pince there was no governing code at the time of construction and there are no records of any engineering analysis performedI it is not possible to accurately reverse engineer the original designK Assuming there was an analysis performedI which is not entirely certainI we do not know what loadsI ultimate stressesI or allowable stresses may have been usedK pince neither cast nor wrought iron is recognized as a structural material by current codesI it is also not possible to perform an entirely modern analysisK diven these constraintsI we have attempted an analysis of three critical elements of the structure using the known dead loadsI a combination of currentJcode live loads and reduced live loads based on available information about useI values for ultimate and allowable stresses based on the literature of the era and the kew vork City Building Code for following decades that included iron useK Current code has a ground snow load of roughly NKN kma at kew vorkI as opposed to snow loads of NKV to OK4 kma used in earlier building codesK diven that the building had no insulationI was heated by iron stovesI and was at least partially occupied in the winterI the roof was likely warm enough to never collect much snowK Current snow load was used in analysisI as the absence of snow cannot be provenK An interior floor load of 4KM kma was used in the analysisK qhis is somewhat less than the currentJcode value for a place of assemblyI but exactly equals the test load used during construction and therefore is a load known to have been applied to the structureK qhe layout of the upper floor of the buildingI with relatively small aisle areas connecting exhibitsI suggests that the lower number may be a more accurate approximation of the loading than an assembly load based on uninterrupted crowdsK qhe three elements analyzed are a typicalJbay columnI a typicalJbay castJiron truss girderI and an aisleJbay wroughtJiron truss girderK iateral load analysis was not performedI but rather only gravity performance was checkedK aespite the height of the central domeI the building was broad relative to its height and had hundreds of potential moment frame connections Ethe trussedJgirder to column connections had topJ and bottomJchord connection pointsI but see the discussion below in section OKOFI meaning that the frame moments and shears from wind would be small in any individual memberK OKOK Columns Analysis of castJiron columns was addressed by the kew vork City Building Code from NUUO Ethe first comprehensive code in the cityF until the NVNS revision xTI UzK auring this periodI the codeJallowed stresses assumed concentric loading and used a straightJline formula to account for slendernessK ft must be noted that the NVNS code significantly reduced the allowable stress in the formulaI in reaction to several catastrophic failures and none of the known failures was caused by overstress from primary loadingI but rather were the result of foundation failures or eccentric overloadKxVz ptructural cast iron use is no longer allowed by American building codesI so there are no current design guidesK

NNOP

A typicalJbay column had an unbraced length of SK4m from the ground floor to the upper floor framingI and carried a total load of 4O4 kkK qhese columns were OMP mm octagons with O4 mm wallsI and the resulting section gives a compressive stress of PN jmaK

cig 4K lriginal drawing Eas reproduced in reference 4F showing column splice and lugs for a girder connectionK kote that distortion is the result of reproducing the fragile original and the twoJpiece bottom lugs

qhe only castJiron column formula generally available in NURP kew vork was that published by baton eodgkinson in iondon in NU4SI summarizing extensive research into the properties of iron xNMzK ft isI of courseI not known if this reference was used in the design of the columnsK eodgkinsonDs formula gives an allowable load of PTM kkK qhe fact that this value is inadequate is moot since the formula is restricted to columns with a maximum lengthJtoJdiameter ratio of NRI while the actual ratio is PNKRK ff we anachronistically use the NUUO kew vork Building CodeI which was written by a generation of kew vorkers experience in cast iron useI we are referred to qrautwine’s Civil bngineer’s mocketbook xNNzK qrautwine has a maximum stress of VN jma for a hollow round castJiron column with a length PO times its diameter; applying the safety factor of six required by the city code results in an allowable stress of NR jmaK qhe NVMN kew vork Building CodexNOz contains its own formulaI which leads to a resulting allowable stress of RT jmaI which is adequateI but again the formula is restricted to a lengthJtoJradiusJofJgyration ratio of TMI while the actual ratio is NMMK fn other wordsI the column stress is adequate by only one of the three formulas checked – that in the NVMN building code – but the columns used would be disallowed as too slender by that formulaK ft should be noted that the columns empirical performance was satisfactoryI but it is unlikely that they were loaded to their design loadsK OKPK Cast-Iron Truss dirders Analysis of the castJiron trussed girder depends on whether or not the girders are considered to be connected at the top and bottom to the columnsI providing end moment restraintK lur analysis was based on simple supports for two reasonsW slop and breakageK cirstI the distance between the top and bottom end connections was fixed at S mm more than the girder depthK pince the girders would sit on the bottom connectionsI the bottom connections had friction to help resist movement in addition to the single bolts that connected the column connection lugs to the girdersK qhe top connections were

NNO4

fastened solely by those single boltsK Any misalignment or minor discrepancies in size EiKeKI ordinary connection slopF would affect the top connection far more than the bottom connectionK pecondI the top connections were a single integral lugI which was more vulnerable to fracture than the largerI twoJpiece bottom connectionsK cracture of integrallyJcast lugs has been commonly observed in a castJiron facade renovationsI likely from incidental moments created by connection eccentricityK qhe analysis was performed using the same 4KM kma live load used for the test load during construction and for our analysis of the columnsK qhe top chord of the girder directly supported the upper floor wood joists and was therefore subject to bending between the truss verticalsK qhe maximum local bending stresses are O4M jma in compression and NPR jma psi in tensionK eoweverI the overall truss form of the girder creates an axial compression of TS jma in the top chordI leading to maximum combined stresses of POM jma in compression and RV jma in tensionK qhe bottom chord has a maximum compressive stress of OO jmaI the verticals have a maximum compressive stress of OT jmaI and the diagonals have maximum stresses of 44 jma in tension and UT jma in compressionK

cigK R lriginal drawing Eas reproduced in reference 4F showing castJiron girder from end to center panel in elevationK kote that distortion is the result of reproducing the fragile original

eodgkinson gave ultimate stresses of VP to NRM jma in tension and SMM to NMMM jma in compressionK qrautwine provides similar valuesI and the NUUO building code proscribes a safety factor of S for compressive posts and tensile membersI or in other wordsI all of the components of a trussK qhe code also provides a safety factor of P for “beamsI girdersI and all other pieces subject to transverse strain” so it is not clear that the safety factor of S should be usedI but this safety factor was used in later codes for all cast iron analysisK qhe allowable stresses would therefore be approximately ON jma in tension and NMM jma in compression; the NVMN building code states explicitly that the maximum tension stress was ON jma and the maximum compressive stress NNM jmaK qhe stresses calculated for the castJiron trussedJgirder are less than the generally accepted ultimate stressesI but considerably more than the allowable values using any of the sourcesK qhis analysis does not take into account any possible live load reduction Esuch as would be caused by fixed exhibition displays that weighed less than UR psf and blocked human access to portions of the floorFK OK4K trought-Iron Truss dirders Analysis of the wroughtJiron trussed girders is similar in most respects to that of the castJiron girdersK eoweverI the individual chordsI verticalsI and diagonals of the castJiron girders all have tee or cruciform crossJsectionsI while the wroughtJiron girders were built of angles and platesK qhe top chords were pairs of angles reinforced with pairs of platesI while the bottom chords of the wroughtJiron girders were pairs of plates and the diagonals were single platesK qhis section implies that the girders were originally analyzed as simplyJsupported beamsI since the bottom chord plates would perform badly under compressionK qhe verticals were castJiron teesI obviously intended for compressionK fn our analysisI those diagonals subject to compression were removed from the modelI since they would buckle in serviceI shedding loadK te reviewed the wroughtJiron girders as originally builtI and as modified after load testing by the introduction of diagonal rods to the columns topsK qhe topJchord bending stress from the local joistJsupport moments was NPM jmaK Axial loads provide a maximum compressive stress in the top chord of 4MM jma and tension in the bottom chord of OMMjmaK qhese axialJforce stress values are not realisticI as they exceed the ultimate stress in wrought iron using any of the sourcesK then the reinforcing rods are addedI changing a singleJspan girder to a threeJspan girderI the local moments in the top chord are unchangedK qhe axialJforce stresses are reduced to approximately

NNOR

UP jma compression or tension if we assume that wroughtJiron yielding allowed moment redistributionK qhe earliest general reference is qrautwineI who provides ultimate stresses of OMM jma in compression and at least PNM jma in tensionK mer the safety factor discussion in section OKOI it is not clear whether the NUUO kew vork code required the use of a safety factor of P or SI although P is closer to the values used later for wrought ironK rsing either safety factorI the stresses are too high even with the reinforcing rods in placeK qhe location of these girders provides one possible explanation for their adequate performance in serviceW they were used exclusively in two bays in each quadrant of the upper floorI connecting the stairs to the main portions of the upper floor where exhibits were locatedK qhese girder were loaded exclusively by peopleI not exhibitsI and would only have reached full design loads if the floor were packed solidK ft is possible that in the limited life of this buildingI such loading never occurredK

cig SK lriginal drawing Eas reproduced in reference 4F showing wroughtJiron girder from end to center panel in elevation and planK kote that distortion is the result of reproducing the fragile original

PK ClNCirSIlNS kew vork grew rapidly in the second half of the nineteenth centuryI not just in populationI but in the complexity of the built environmentK fn the decades after the Crystal malace was built and burnedI the use of first cast ironI then wrought ironI and finally steel would make possible buildings of exceptional height and with exceptionally long spansK qhe experience and expertise that led to the the skyscraper boom that began in the NUTMs did not yet exist at the time the Crystal malace was constructedK qhe design of this building was instead based on experience with castJ and wroughtJiron bridge designI the only significant iron structures in the country at that timeK qhe Crystal malace did not serve as a model for other buildingsK qhe exhibition was a commercial failure and the known problems with the building created an impression that it was a constructive failure as wellK Animosity between the architects and the boardI resulting in part from the poorlyJcoordinated building processI showed that the design and construction community in kew vork was not yet ready for a building of this scale and innovative typeK ft is difficult to say how well the building would have performed if it had not burnedK qhe relative extremes of temperature in kew vork – average low temperatures in the winter of JP°C and average high temperatures in the summer of OV°C – would have caused the exterior portions of the building to “work” in repeated expansion and contraction cyclesK qhis movement has damaged the connections in castJiron facades in the city and would likely have caused more extreme damage in a building without other structureK jore importantlyI continued use over an extended period of time would almost certainly have included changes in load patterns that could have exacerbated the problems with the trussed girdersK Analysis shows that these members wereI under full loadI overstressed by both the standards of the midJ to lateJnineteenth century and the standards in use todayK qhese members appear to have performed adequately during the building’s limited life only because they were not loaded to their full capacityK qhe introduction of reinforcing for the longJspan wroughtJiron trussed girders during the course of construction shows the true nature of American engineering at that timeW it was a mostly empirical processK bven though the trusses were designed by an engineerI the lack of standards for allowable material stressesI the lack of standards for loadingI and the lack of generallyJaccepted analysis methods meant that these girders could be completely inadequate as designedK qhe experience that later led to the construction of impressive ironJ and steelJframed buildings wasI specificallyI experience of failure of inadequate structuresK fn this contextI the Crystal malace is one of the early steps towards modern iron design in the rnited ptatesK

NNOS

oEcEoENCES xNz dreeley eKI edK ENURPF Art and industry as represented in the exhibition at the crystal palace kew

vork NURPJ4K kew vorkI oedfieldK xOz eirschfeld CK ENVRTF America on exhibitionW the kew vork crystal palaceK American nuarterlyI V

EOFW part NI NMNJNNSK xPz ConditI CK ENVUOF American buildingI 2nd editionK ChicagoI qhe rniversity of Chicago mressI UPK x4z Carstensen dKI dildemeister CK ENUR4F fllustrated description of the kew-vork crystal palaceK

kew vorkI oikerI qhrone C CoK xRz nKaK ENUR4F crancis’s new guide to the cities of kew vork and BrooklynKKK kew vorkI CKpK crancis

C CoK xSz nKaK ENURUF qhe crystal palace fireK The kew vork TimesI NO lctober NURUK xTz nKaK ENUUTF iaws relating to buildings in the city of kew vorkKKK kew vorkI qhe oecord and

duideK xUz nKaK ENVNSF The code of ordinances of the city of kew vorkK BrooklynI qhe Brooklyn aaily

bagle CoK xVz criedman aK EOMMSF qhe darlington building collapseW modern engineering and obsolete systemsK

fnW mroceedings of the Qth Congress on corensic bngineeringK oeston sAI American pociety of Civil bngineeringI PPVJPROK

xNMz eodgkinson bK ENU4SF bxperimental researches on the strength and other properties of cast ironKKK iondonI gohn tealeK

xNNz qrautwineI gK ENUUPF The civil engineer’s pocket-bookK mhiladelphiaI bK Claxton C CoK xNOz nKaK ENVMNF The Building Code of the City of kew vorkK kew vorkI kew vork aepartment of

BuildingsK