Challenging Glas, Conference n Architectural an Stu Appliations of Gass Bos, outer, Ver (Organizing Commits) ‘wun BC det alsin Long span glass fin design John Keoymans Halerow Yolles, Toronto Canada, john.kooymans(@halerowyolles.com Dr. Jens Schneider, Dr. Holger Techen University of Applied Sciences, Frankfurt, jschneider(f,fh-frankfurt.de techem@fbl fi-frankfurtde ‘Suu gs fis have been wilzed in eeting aneparet wal sytem all ‘over the world. High i walls Have alays boas ellen dae olntations a ‘has sis for bth tmpoed glass an iat hss. This pope il ok a2 Sngue concept lo create tetra pas fis fora 18 oer ih ass wal whe limiting he wadional scl pie pa a the ls jas ‘The syste i snple in poeipa, Muses snple sae, catleveedcoonectons the bas atl op ofthe i, ad simple spa constuction in the ene pron of the fia The challenges ar cesta by the sear magatad of the elements aa turing the final despa secon to bald ‘Through computer mods an pial esting, his plas fn desig allows fr shallower fn depth ha fo conventional in eign enxiestanepreney fa alevits the ruement for uns stl connection plates a he splice ibeaiogs “The sniel aus to be addesed ate concabliy, safety” and Stability, wih he salty ue beg adresse bythe use of redundant srs of fas. This syst can ako be used for gs beam ot fis desans of varying Tents and cantlever proprio, 1. Introduction (Over the years, designers have continued to push the boundaries of structural glass design. The desire to ereate systems that are more ransparent, less dependant on other ‘materials, and more cost effective have provided the fuel Tor consulting firms to ‘become involved in these special structures and utilize their expertise and creativity to ‘offer new solutions. The structural glass fin is no exception. A simple concept that has ‘been utilized for decades, the desire to make them longer and more slender has ereated ‘anced for this Research & Development project. 2. The Concept ‘The glass fin design detailed in this paper is 18 meters tall and S00 mm deep. The design is based on a total wind load of 2.10 kNim pressure. The spun-to-depth ratio of the fin is 36. A very slender design with no lateral stabilizing elements other than the lass fagade perpendicular tothe fin assembly. ‘The system ulilized for this design allows for minimal exposed connection material, and assists in stabilizing the fin structure at the same time. As shown in Figure 1, the span is ereated by fixed end cantilever portions at each end of the span, and a simple span element in the center. The objective is to ercate a stable cantilevered section so that the intermediate span can be simply supported and independent of the overallcharacteristics of the span, thus allowing for minimal connection materials and a reduced fin depth, In this paper, the center span of the fin assembly is hung from the upper fin The lower fin assembly is gravity bearing at the base within the clamping device. A vertical sliding connection is introduced between the lower fin assembly and the middle fin assembly to allow the system to absorb vertical deflections imposed by the supporting structure, we Figure L Original Concept Skoteh3. Analysis ‘The overall assembly was frst sized based on standard hand calculations in accordance With the Australian Standard AS 1288-1994, Glass in Buildings. Stress levels were checked using simple statics prior to computer modeling in an effort to establish a realistic stating point. An overall computer model for an individual fin was modeled using the Finite Element program RFEM with platehell elements (see Figure 2) 10 review the buckling behaviour, stress levels, and deflections. nieces state te Figure 2; Overall View of FE Mod Fige 3 Fin Support Dati ‘The fin support detail at the base and top ofthe fin is detailed to be recessed within the floor and ceiling finish of structure. The amount of exposed steel at these locations Would be determined by the design team. Figure 3 shows the concept for the clamping device to be used, ‘The stability of the glass Fin was verified using a linear second order theory analysis (cluding P-Delta affects). To account for imperfections in the erection of the glass, a Tine load of 0.3 kNim was applied perpendicular to the inside edge of the glass fin ‘This value was chosen in order 10 simulate an out of plane imperfection due to ‘construction and manufacturing tolerances, In this model, the upper and lower glass fin assemblies are comprised of 4 ~ 15 mm. tempered lites with PVB interlayers. The intermediate glass fin assembly is comprised ‘of 3 — Smm tempered lites. The out-of-plane stiffness of the fin was reduced ‘conservatively by 50% to allow forthe composite effect of the PVB interlayer with the lass ‘The deflection resulting from the direct wind load was in the range of 23 mm, When the imperfection load was added tothe system, the deflection increased to 33 mm with some rotational characteristis, This is well within the deflection limitations over an 18‘meter span, Even the change in deflection at the inflection point is manageable. Figures 4a and 4b below show the resulting deflection diagrams, Figue da: Deflection ducto wind oad 23m) Figure 4: Deflction de o wind load plas inmpertection (33 ou) FigueS: Stes Distibuion of Middle Figure: Suess Distsbuion athe Upper Constant, (Gis Fin Assobly ‘11 was quite evident that the stress levels in the middle fin assembly would not be an issue in the overall design of the 18 meter fin. The simply supported bolted conection, siresses and bending stresses within this span are very low as shown in Figute 5. ‘The FE model was showing, as expected, much higher siress concentrations at the support constraints of the upper and lower glass fin assemblies. Figure 6 shows the siress distribution in our inital modeling. Note that in this model, two support points Were assumed atthe top ofthe fin, Stresses in this area were approaching higher than desired values due to the proximity of the holes to the bearing points on the gla fin, ‘As a result, it was devided that any further modeling and test specimens would have ‘one support point at the center ofthe glass Fin (Figures § to 10). ‘As a comparison, Figure 7 below shows the stress distribution atthe support constraint With no holes inthis area, Both Figure 6 and Figure 7 show high stress concentration levels at the beating points against the glass Fin that ereate the resisting moment arm ‘The local service load reaction at these beating points is in the range of 15S KN to 175 KN. The stress levels shown in these images are conservative in this model since the support point is fixed t a single node, In the physical model, dhs bearing point will be 100 mm long and the local stresses will be reduced significantly as itis planned to use 4a resin with a Young's modulus between 1000 N/mas* and 3000 N/mm? between the lass and the steel bearing point, We will ely on the test results to calibrate our FE ‘model accordingly seeeeeeeeeel Figure 7: Ses Distibutio at Fin Consait oo bas)4. Test Specimen In order to verify the FE model analysis, our team decided to ereate a tes procedure to Validate the support conditions of the fin design. It was decided not to use a scaled down version of the cantilever, but to reduce the length of the cantilever only (3 m instead of 6 m). We therefore inereased the loading applied to achieve the same ‘bending moment at the support, This allowed for smaller test specimens and reduced ‘material costs, The intent ofthe test was to verily the earrying capacity of the support assembly and compare these resulls to our FE models and calibrate our models accordingly for use in future designs. It would also allow us to analyse the affect of imperfections within the glass manufacturing process and see how these allowable tolerances may oF may not reduce the capacity of the assembly. [A preliminary simplified finite element model of this test set-up is shown in Figures 8 10 10, This local model was analysed in MEPLA using multi-layered 9 node plate/shell clements capable of large deformations. Spring elements were intraduced to model the local slffiness of the bearing points for the resin between steel and glass. The hole diameter in the center was assumed to be 50 mm. Conservatively, it was assumed that the load would only be taken by three lites instead of four asthe local misalignment of laminated glass with tolerances of up to 2 mm usually does not guarantee a smooth load introduction, It ean be shown by this model that the local stresses in the glass are reduced significantly to values around 35 N/mm to 45 Nimm? (maximum principal tension stress) and 70 Nimne to 90 Nimm? (maximum principal compression stress) depending on the stiffness of the resin, As the behaviour ofthe resin was only assumed tw be linear elastic, the tess will have to show if this assumption is still valid inthe real situation afler assembling and if the related assumption of the stiffness is correct Within the model, the contact load transfer mechanism between the bolt and the glass was set with a tolerance of 0.01 mm, so that once there was any separation between this ‘contact surface, the load transfer would be eliminated since the resin is not capable of ‘wansfering tensile loads, Principally, local modeling of the load introduction for such high loads i still best correlated through testing procedures. Fig inte cement mesh for tet speciesFigae 10: Sis tajctro of png sess in the eal ae of lad iteducton 45. Testing and Results ‘The glass for the test specimens was supplied by Eckelt Glas GmbH. The steel shoe assembly was resinned in place with a Hilii HIT HY-70 resin. Two methods of resin installation were performed to establish the most suitable procedure, The first one injected the resin into the gap between the glass and the steel shoe with one closure plate removed so thatthe gap was exposed during the procedure. The second method ‘kept the closure plate in place so that the gap between the glass and the steel plate was captured by the assembly. Two holes were drilled into the stel plate so that we could inject the resin into one hole and allow the resin to exit through the second hole, thus ‘ensuring the void to be fully replaced with the resin material. This second approach was found to be the preferred procedure. Figures 11 and 12 show some of the elements and assembly procedure.‘Table 1 below summarizes the FE analysis results that will be used as a comparison to the physical test results. The MEPLA analysis was done using an E-value forthe Hilti Resin of 1780 MPa, Table 2 summarizes the specimen test results. The assembly ‘modeled and tested consisted of 3000 mm cantilever beyond the support condition (half the length of the cantilever portion for the 18 m fin analysed as a sealed down specimen). “able: MEPLA Rests LoverAmmat | Numberafismm | Lead An) | Tip Detection eam) | Max Principal Stes Supper un) ies Ps)Table2: Tet Ress Lover Amat | Numberat | Laan) | Tip Denson Comments Suppeet (nn) | 1S its (om) 400 ‘One ite purposely beaken 400 ‘One te purposely beaken 400 ‘One te purposely beaken 400 ‘Thre tes purpossily broken 400 ‘Thre ites purposely broken 400 ‘Thre ites purposely broken 400 ‘Thre ites purposely broken 400 Breakage ocared 400 400 400 Breakage of oe lite occured 400 400 400 400 “Toul beakage ocured 6. Conclusions ‘At the time of printing the Challenging Glass proceedings, all the testing was not ‘completed. A total of eight specimens were to be tested prior to analyzing the data and formulizing conclusions. Our preliminary findings based on the first two test specimens show that the cantilever fixing/structure is meeting the strength requirements and that the lamination is providing the redundancy required so that the fin ean still perform when one or two lites break. The preliminary results, however, show thatthe stiffness of the anchoring assembly is less than assumed and that larger than anticipated deflections are occurring with the test specimens. Further review of this anomaly is required, and perhaps the modeling of the resin within the FE analysis ‘may need to be adjusted. The final paper including the remaining test results and conclusions will be published at www. glass bk tudelft al for your reference