Carbon Fiber-Reinforced Elastomeric Pads for Building Isolation

Transcription

1 The 33 rd International Congress and Exposition on Noise Control Engineering Carbon Fiber-Reinorced Elastoeric Pads or Building Isolation S. D. Capbell Kinetics Noise Control, 6300 Irelan Place, Dublin, OH 43017, USA Abstract [535] A new and unique building vibration isolation pad has been developed utilizing alternating layers o elastoer and carbon iber esh. The carbon esh reinorceent provides several distinct advantages over traditional steel plate reinorceent. Lower natural requencies and/or saller operating heights can be obtained or a given load. In addition, the carbon esh ors both cheical and echanical bonds with the rubber, increasing the ultiate load or delaination. This paper briely discusses the analytical solution or stress and deoration in the new pads, and outlines the pad design ethodology. Tests were perored on both carbon esh and steel reinorced pads. Static load-delection curves, dynaic stiness values, and creep data are presented. Finally, the designs o a traditional pad as well as the new carbon esh-reinorced pad are shown or a realistic proble. 1 INTRODUCTION There is a growing interest in the use o vibration isolation systes or coplete buildings. Several actors have inluenced the increased interest in building isolation including governent regulation, increased consuer awareness o noise and vibration probles, econoics [1], and a lack o vibration-ree building sites in soe locations. Not only are the traditional candidates or isolation being considered (theaters, peroring arts centers, etc.), but coercial, industrial, edical, and residential structures are now potential users o building isolation systes. Soe o the interest is driven by the nature o the building; or exaple, the use o sensitive equipent over a large area, and soe is based on the project location (i.e. adjacent to rail lines). Regardless o the ultiate use o the structure, the need or isolation is based on the perceived beneit to the owner in either unction, econoics, or both. Several types o building isolation schees can be eployed, with steel springs or elastoeric bearings used in the ajority o projects. Each solution has advantages depending upon the source o the noise or vibration and the particular construction requireents or the building. Steel spring systes can be developed with lower natural requencies and can be easily pre-copressed to near their operating height, allowing or very little deoration during construction. Conversely, elastoeric pads are generally less expensive, easier to install, and have inherent daping that akes the ideal or any applications. 1/8

2 Elastoeric pads are typically constructed o alternating layers o natural rubber or neoprene and steel plates. Since the shear stress in the elastoer increases with thickness, it is not possible to use a single layer with a large thickness without overstressing the aterial. The steel plates provide lateral restraint to the elastoer layers, allowing thick isolators to be built up ro ultiple thinner layers, thus reducing the stress in any individual layer. However, this coes at a price since the thinner layers are stier and thereore less eective isolators. In order to achieve the desired isolation the total thickness ust be larger when lainated pads are used. In addition, the steel plates are heavy, with approxiately one-hal the total bearing weight coing ro the plates. The ajor reason or the excessive weight is that the plates are uch thicker (approxiately 3 ) than is necessary to provide adequate restraint due to the processing required to obtain a suitable bond between the steel and elastoer. This paper presents analytical and experiental background or a new pad design, using a carbon iber esh or reinorcing, that eliinates soe o the probles associated with traditional lainated pads. ANALYTICAL SOLUTI ON OF BEARING BEHAVIOR The design o lainated bearing pads is a quite coplex undertaking. The size o the pad, nuber and thickness o the elastoer layers, duroeter o the elastoer, and reinorcing thickness all need to be deterined. Several, oten conlicting, criteria ust be satisied including liitations on the pad size and total thickness, natural requency at the design load, appropriate saety actors when overloaded, and the changes in properties under long-ter or cyclic loads. It is thereore essential that an analytical odel o the bearings be developed in order to understand the behavior and correctly design the pads. Solutions are readily available or elastoeric pads restrained top and botto by rigid supports [] and are used or analysis o steel plate-reinorced bearings. The carbon iber esh used as reinorceent in the new pads is not rigid but can deor slightly along with the elastoer, requiring the reinorceent lexibility to be considered in the design. A brie overview o the analytical odel derivation and the resulting solution is presented below. In coon with the rigid support solutions, it is assued that the load is applied vertically, that horizontal planes reain horizontal, and that points lying on a vertical line beore loading lie along a parabola ater loading. The total horizontal displaceent in the x-direction at any point, u(x,y), is the su o the elastoer displaceent assuing rigid boundaries on both horizontal suraces, u 0 (x,y), and the displaceent o the reinorceent on the boundaries, u 1 (x,y). The solution begins with a copatibility equation, including the copressibility o the elastoer, relating the strains in the bearing to the applied pressure and bulk odulus o the elastoer as p ε xx + ε yy + ε zz = (1) K where ε xx, ε yy, and ε zz are the strains along the principle axes, p is the vertical pressure in the pad, and K is the bulk odulus o the elastoer. Cobining the copatibility equation with equilibriu equations or the shear stress, boundary conditions, and syetry requireents, the governing dierential equation can be derived as d p dx d p 4G 1G 1G ε + p dy + = E t t K t c () /8

3 where G is the elastoer shear odulus, t is the thickness o the elastoer layer, E and t are the odulus and eective thickness o the reinorcing layer, and ε c is the average copressive strain in the elastoer. Given the value o average copressive strain the equation can be solved or the pressure. For a pad with diension a in the x-direction and b in the y-direction, the result is 1G 4 ε c π cosh α y πx p x, y = t 1 sin (3) = 1,3,5 α b a cosh α where 4G 1G π α = + +. (4) E t t K a The x-direction displaceent o the rubber and the reinorceent can then be derived as 6ε c cosh α y πx u0 x, y = 1 cos (5) = 1,3,5 α a b a cosh α 48Gε c cosh α y πx u1 x, y = 1 cos (6) = 1,3,5 te α ( π) t b a cosh α with siilar expressions or the y-direction displaceents. The noralized predicted pressure distribution or a typical pad is shown in Figure p/p ax x/a y/b -0.5 Figure 1: Noralized pressure distribution in a typical square bearing pad. 3/8

4 Deterination o the stress in the reinorcing is critical or selecting the required thickness o the esh and or provided adequate saety against debonding o the reinorceent and elastoer. The reinorcing orce per unit length is given by F ( x y) tp( x, y), =. (7) In this exaple, the axiu orce in the reinorceent (and pressure) is alost twice the average value. This is the case or ost bearings in typical sizes. Derivation o the solution or an actual proble is an iterative process since the total applied load (integral o the pressure over the pad area) is typically the known value and the aterial properties or the elastoer depend upon the strain. Thus, a value o copressive strain is assued, aterial properties are deterined, and the total applied load is calculated ro the pressure prediction as shown above. Based on the calculated load and the actual load, a new value o copressive strain is estiated and the solution proceeds until it has converged. For a lainated section this process is repeated or each layer thickness, and the results are cobined to deterine the overall bearing displaceent and hence stiness and natural requency. 3 EXPERIMENTAL TEST ING An experiental progra was designed to veriy the theoretical analysis and to investigate the iportant aspects o the behavior that cannot be atheatically predicted but ust be deterined ro physical testing. The testing progra deterined the static and dynaic stiness o the bearings, checked or debonding o the elastoer and carbon esh under extree loading, and investigated the long-ter delections through creep testing. All testing was independently perored at the Multidisciplinary Center or Earthquake Engineering Research (MCEER) at the University at Bualo, State University o New York. The test speciens easured 00x00 in plan and were coposed o ive elastoer layers (either natural rubber or neoprene) separated by either steel plates (3. thick) or a carbon iber esh (0.8 thick) and the bearings had a cover o 6.5 o elastoer on each side. The elastoer layers easured 6.5, 5.4, 9.5, 5.4, and 6.5 in thickness. Several speciens were obtained by cutting larger saples down to 00x00, thereby leaving soe edges without an elastoer cover over the reinorcing. A suary o the specien properties and diensions is shown in Figure and Table 1. Table 1: Bearing test saple properties and description. Spec. No. Elastoer Duroeter Reinorceent Description 1 Natural Rubber 69 Steel Plates Molded at tested size Natural Rubber 76 Carbon Mesh Molded at tested size 3 Neoprene 60 Carbon Mesh Molded at tested size 4 Natural Rubber 74 Carbon Mesh Cut ro large saple - 4 sides cut 5 Natural Rubber 73 Carbon Mesh Cut ro large saple - 3 sides cut 4/8

5 6.5 (typ) Reinorcing Layer 3. Steel 0.8 Carbon Varies (73.3 Elastoer) 00 Figure : Diensions o bearing test saples. The test saples were based on pads used on a previous project and had a design dead load o 111. kn and live load o 89 kn. Testing began with a debonding test wherein the pad was loaded to 150% o the total design load (300.3 kn). Ater ive inutes o sustained load the specien was visually checked or debonding o the reinorceent and elastoer. A static load-delection test was perored next, up to a total load o 00. kn. Since the static and dynaic stiness o the pads can be signiicantly dierent, cyclic tests were perored at 5 Hz, 10 Hz, and 1 Hz. The upper loading requency o 1 Hz was based on the capacity o the testing achine. The speciens were subjected to the dead load, ollowed by cyclic displaceents o between 0.5 and 1.0, depending upon the requency. Finally, to allow creep estiations, a constant load (111. kn) was applied and the displaceents were easured over at least six hours. Results or all the tests, along with a plot o pad natural requency versus loading requency, are presented in Table and Figure 3. All specien results were adjusted to a duroeter o 70 to allow or coparison o stiness and natural requency between the speciens. The adjustents in stiness are based on the change in odulus with duroeter as reported in [3] and are approxiate. The creep, as a percentage o the initial delection, was estiated at 5 years. The ethod uses six hours o load-delection data to estiate the odulus, and hence creep, as a unction o tie [4]. Table : Specien testing results. Stiness values are in kn/, requencies in Hz, and 5-year creep in percent o initial delection. Static 5 Hz 10 Hz 1 Hz Spec Debond Sti Freq Sti Freq Sti Freq Sti Freq Creep 1 No % No No % 4 No % 5 No % No debonding o any specien was noted during testing, the restraint provided by the reinorcing reained eective or the ull duration o the overload. Figure 4 shows the deorations and 5/8

6 reinorceent restraint or Specien 5 during the debonding test. Typical values o allowable creep over 5 years range ro 5% to 45% o the initial delection, depending upon duroeter [5]. All the tested speciens easily eet the creep liit. 14 Spec 1 Spec Spec 3 Spec 4 Spec 5 Pad Natural Frequency (Hz Loading Frequency (Hz) Figure 3: Isolation pad natural requency versus loading requency or all speciens. The stiness and natural requency data, approxiately noralized to a 70-duroeter elastoer, shows the new bearings to behave coparably to steel-reinorced bearings. The static natural requency o all the bearings was within ten percent o the reerence bearing requency and all the bearings showed an increase in natural requency with increasing cyclic load requency. The dynaic natural requency o the pads ranged ro 3 to 31 percent above the static natural requency or the natural rubber bearings while the neoprene bearing natural requency increased by 50 percent at the 1 Hz loading. Most speciications call or a dynaic/static requency ratio o no ore than 1.4. O the tested saples, all the natural rubber isolators eet this criterion while the neoprene isolator stiened excessively under dynaic loading. Coparison with the theoretical values calculated as described above is useul or veriying the analysis ethod. The calculated static stiness o the bearings, or 70 duroeter rubber, is 9.6 kn/ with a resulting natural requency o 8.13 Hz. The calculation is sensitive to the elastoer duroeter, with an increase in duroeter to 71 leading to a stiness o 33.1 kn/ and a natural requency o 8.61 Hz. It is clear ro the results that a theoretical analysis can accurately estiate the vibration isolation characteristics o the bearings. However, care ust be taken in the use o the results, since the typical speciied tolerance on duroeter o ± 5 can have a signiicant eect on the isolation eiciency at certain requencies. 6/8

7 Figure 4: Specien 5 during debonding test showing elastoer deoration and reinorcing restraint. 4 APPLICATION Exaples o the design o both steel- and carbon iber esh-reinorced pads are presented or a peroring arts center. The pads were located under the support coluns and rested on concrete piers. A typical colun load was 111. kn dead and 89 kn live. The pads were required to have steel plates lainated top and botto to allow or ounting, pre-copression o the pads, and the attachent o horizontal slider plates. The ounting plates required 5.4 o height out o a total allowable pad height o 114.3, leaving only 88.9 or the pad. The target static natural requency o the pad was 7 Hz ± 1 Hz under dead load plus 5 percent o the live load. The design utilized ive layers o natural rubber reinorced with our layers o either steel or carbon iber esh. The rubber layers were 6.5, 5.4, 9.5, 5.4, and 6.5 thick with 3. thick steel reinorceent or a total height o The sae laination was used or all the pads on the project with varying plan diensions based on the load. The pad described herein was 03 square. For 50 duroeter natural rubber this bearing had an estiated static natural requency o 6.0 Hz under the design load and 6.4 Hz under the dead load only. The steel reinorced pad et the design criteria and reained within the height liitation. A design using the carbon iber esh or reinorcing could take two ors. Identical rubber layers could be used, with identical perorance obtained, but with a total height o only Alternatively, the center rubber layer could be increased in thickness to 0 (total height 84.1 ) leading to a design load natural requency o 5.4 Hz and a dead load only natural requency o 5.7 Hz. This signiicant reduction in natural requency was achieved while siultaneously slightly reducing the operating height. Although the increased lexibility was not required or this project, it was desirable, and could be critical in soe applications. 7/8

8 5 CONCLUSION Vibration isolation o all or part o buildings is becoing increasingly coon. One option or the isolation design is lainated elastoeric pads. The developent o such a pad, using a carbon iber esh rather than steel plates as reinorceent, was described in this paper. The theoretical basis or deterining the deorations and pressure distribution in the pad, including accounting or the lexibility o the reinorceent, was presented. Also discussed are the results ro an extensive testing progra that investigated the static and dynaic stiness, creep, and debonding potential o the lainated bearings. Finally, saple designs o both the new pad and a standard pad were presented or a realistic proble. The new pad oers several advantages over the traditional steel-reinorced design. Since the carbon esh and elastoer or both cheical and echanical bonds, no processing is required prior to vulcanization o the lainate, and uch thinner (approxiately 0.8 or less) layers o reinorcing can be used. Pads designed or the sae natural requency are thinner and lighter than i steel-reinorced, due to both the reduced thickness o the reinorcing and the inherently lighter weight o the carbon. In addition, the pads are non-corrosive in ost environents suitable or natural rubber or neoprene and are non-agnetic. Finally, the pads are easily cut ater anuacture, unlike steel-reinorced bearings, allowing or last-inute adjustents o size and aster delivery ties. Testing has shown that the bearings provide perorance siilar to conventionally reinorced pads and that their operation eets current speciications. Additionally, an analytical solution or the pad behavior accurately predicts the easured response. In conclusion, the new carbon iber eshreinorced isolation pads do everything the standard pads do, but have any additional advantages. REFERENCES [1] L. Dunayevsky, Relations between dwelling price and noise actor, Technical Acoustics, pp , (00) (in Russian). [] J. M. Kelly, Earthquake-Resistant Design with Rubber, Springer-Verlag, London [3] Handbook o Molded and Extruded Rubber, Goodyear, Akron, Ohio, USA [4] J. Yura, A. Kuar, A. Yakut, C. Topkaya E. Becker and J. Collingwood, Elastoeric Bridge Bearings: Recoended Test Methods, NCHRP Report 449, National Cooperative Highway Research Progra, Transportation Research Board National Research Council, National Acadey Press, Washington, D.C., 001. [5] AASHTO, LRFD Bridge Design Speciications, Aerican Association o State Highway and Transportation Oicials, Washington, D.C., /8