Drilled Glass Bearing
Drilled Glass Bearing
Drilled Glass Bearing
Engineer
Session 11
2 = Engineering
Abstract
The process of engineering glass holes for bearing loads can be time consuming and difficult to visualize but is often critical in achieving design objectives. This presentation presents a collection of information from a variety of sources for determining tensile stress and probability of failure along the inner rim of a hole in bearing for some of the most common practical loading situations. This information can help facilitate fast, accurate and intuitive design, and allow the practicing structural glass engineer to focus on application and creativity. Discussion is also given to the effect of: 1) undersized/oversized hole/pin relationships, 2) bushing materials, 3) different schemes of bearing in laminated glass, and 4) application examples from completed projects.
When we move to a brittle material like glass, a more rigorous approach towards localized tensile stress concentration becomes essential as the average bearing stress alone no longer gives sufficient information for safe design. But even the most user-friendly finite element programs take time to generate answers and iterate refinements, time that can preclude variation and exploration early in the design phase. The focus of this presentation is an approach somewhere in the middle, which takes into consideration the most critical variables of the problem but also emphasizes an intuitive understanding and simple, fast solutions. An intuitive understanding can facilitate experimentation, which underlies the core objective of this presentation: to encourage new and better bearing connections and stimulate thinking about the next generation of possibilities for bearing in glass holes.
Introduction
It is rarely possible or desirable to construct large monolithic structures without joints. Bolting and riveting have been critical methods for joining together parts for thousands of years. Engineers traditionally design such connections in metal plates based on the average bearing stress (often referred to as the projected area method), a method common to most steel and aluminum specifications worldwide. This method has produced innumerable safe connections and equally important: has allowed engineers to move quickly beyond the complex stress distributions around a hole in bearing and focus on using the bearing joint creatively. However, the projected area method heavily relies upon two critical elements: 1) A dependence on localized plastic flow at areas of high tensile and shear stress. 2) Constrained values of edge and end distances.
Figure 1
136
the hole, =. Note that the intensity of tangential stress decreases extremely rapidly as we move away from the hole. The problem is highly localized and focused. Several actions can be taken to restrain ovalization. Increasing the plate thickness is often the simplest way to achieve this. In metal plates, a reinforcing ring can be welded or cast around the hole. At C and D, the lateral deformations can be restrained by increasing edge distance. Similarly the vertical deformation at A is restrained as the end distance is increased. It is important to note any reductions in the deformation at A, C or D reduces tangential tensile stress at the inner rim hole.
Figure 2 shows the deformation in a rectangular plate with a centrally loaded hole at B. Note that the plate is supported vertically (y-axis) along a line from E to F. Key deformations are: 1-Ovalization of the hole 2-Lateral Translation at C and D 3-Vertical Translation at A 4-Shear Deformation Anything done to restrain these deflections will lower stress along the inner rim of the hole. Stretching the inner radius into an oval creates tension stress along the sides of the inner rim. (Figure 3). The highest tensile stress point occurs where the curvature decreases the most. For the loading shown in Figure 2, the largest tensile stress occurs along the inner rim of the hole at =/2 and 3/2. The maximum compression occurs at the back of
Solution method
Today computers and the finite element method are used to determine the stress in plates and other complex problems; but early pioneers of mechanics had to be a bit more clever. The work
137
Session 11
r: Radial Stress : Tangential Stress r: Shear Stress Positive radial and tangential stresses are tensile, negative are compressive. All stresses shown in Figure 1 are drawn in the positive direction and all angles are measured from the yaxis counter clockwise. As we shall see, nearly all bearing connections failures (Figure 4) are due to tensile tangential stress along the inner rim of the hole (Figure 3).
of these early pioneers is instructive and their creativity is inspiring. Prior to the computer, the general state of stress in a plate loaded in bearing was determined analytically by solving the differential equations of elasticity for a general solution. That general solution would contain constants of integration, which could be determined from the boundary conditions. By substituting these constants back into the differential equations of equilibrium, the stress components at any point could then be determined. With a little intuition, the general solution could be used to creatively simplify and solve complex problems. Almost all stress concentration problems involving circular holes in plates have been solved with one equation: the general solution for a circular ring with a central hole (Appendix A), worked out initially by Lam (17951870). Before finite element analysis, this solution was the foundation of knowledge about stress concentrations around circular holes in plates. A stress concentration factor (K) is a scalar defined as:
max = K ave
where:
ave =
P 2 a t
in Figure 5. Reiners solution is accurate for an eyebar (as shown left in Figure 5) but it significantly overestimates tangential stress for the rectangular plate problem of Figure 2. Notably, this error does decrease as the ratio increases. The difference, however, is the shear contribution along the rings outer perimeter. Shear stress along r = b provides resistance to ovalization of the outer perimeter which contributes significantly to stress reduction at the inner rim. Earlier, Howland (1922) had solved the problem of an infinite strip with a hole subject to a uniform tensile stress with no applied loads at the hole (see CASE 5 in the following section). Theocaris (1956) cleverly coupled Howlands solution with something similar to Reiner and by superposition was able to accurately calculate a revised K factor for a rectangular plate (see CASE 1 in the following section) of 2.75 and 1.40 for. =2 and =4. To understand whats going on, imagine a ring (Figure 1) cut from the rectangular plate (Figure 2). The distribution of radial stress on the outer boundary would be very similar to Reiners assumptions (Figure 5). Additionally, the outer perimeter of the ring would have a shear stress similar to that shown in Figure 6 (center diagram). The resultant of radial stress (r) and shear stress (r) along the lower outer boundary of the ring is vertical (y). All horizontal stresses (x-direction) are neutralized.
Session 11
and: P=force applied by pin a=inner radius of hole t=thickness of plate The boundary conditions required to solve this equation are: r (r=a) r (r=b) r (r=a) r (r=b) The earliest closed form solutions (Reiner, 1933) of the pin/plate bearing problem were set up such that all applied boundary loads were radial:
Figure 6
Reiner calculated stress concentration factors K of 3.85 and 1.57 for =2 and =4 respectively for a circular ring with a central loaded as shown
Figure 7
138
This is can be more easily visualized looking at a plot of the shear stress, (Figure 7) which shows maximum and minimum values at 3/4 and 5/4. The shear contribution for =2 reduces tangential stress along the inner ring about 30%. For =4, the difference is only 10%.
by superposition some of the more complex problems encountered by the practicing engineer. The reader is encouraged to plot the tabular values for each case, which is very useful in visualizing the effect of variation in around the hole and understanding which areas of the hole are most sensitive.
Session 11
Case 1: Tension
As mentioned, this case seemed to generate a good deal of interest among early elasticians and is widely published. Petersons is an excellent reference with graphs and exact corrections for undersized pins. Very interestingly, for any fixed edge or end distance, an optimum value for load occurs near =2.4. It should be increasingly clear why the projected area method could be dangerous. When designing holes in glass, the engineer is usually aware of the minimum edge distance limitations for toughened glass. This value is the practical limit in being able to achieve full temper. It has nothing to do with proper edge distance for transferring engineering loads. If the hole is made larger and edge distance held constant, any increase in hole diameter meant to increase bearing capacity could be in fact reducing the capacity. The maximum stress concentration for Case 1 for any value of can be expressed as
K = 0.2880 + 8.82 1 1 1 23.196 2 + 29.168 3
Table 1 lists the maximum value of stress concentration K around the inner rim of the hole. Appendix B presents a table of tangential stress around the entire inner radius for each load case and listed in Table 1, which is useful in solving,
(Petersons)
Case 2: Compression
Figure 8: Tensile Load at Hole
139
Session 11
from Case 1 by changing sign on the terms of the inner rim boundary conditions and resolving for constants of integration. Even so, this case rarely appears in literature, probably because the concentration factor is less than one for all values of in Table 1. Caution should be used with this case as the design may be limited by large shearing stresses (see Figure 7) away from the hole. As the ratio of gets very large, Case 1 and Case 2 converge and the value of K approaches a lower limit of 0.30.
Application: An example
Consider the monolithic glass beam and loading shown in Figure 13:
Figure 13
A reasonably accurate estimate of the stress around the hole at section C can be determined from superposition of Cases 3 and 5. From statics, the reaction R1 at C is equal to 4.9 kN and the moment is equal to 3018 N-m. If the radius of the hole is 25mm, =100/25=4. With a thickness of 19mm and a section depth of 320mm, the maximum bending stress at section C (neglecting the hole) is 9.40 N/mm2. The tensile stress from bending at the top of the hole can be approximated
140
The results of a finite element solution were 11.3 N/mm2 and 1.22 N/mm2. From this example one can see that, this approximate method can be reasonably accurate and powerful. Its important to visualize that while the bending stress and bearing stress are both ovalizing the hole, they are acting perpendicular to each other not cumulative. This kind of visualization is useful in understanding the stresses around the hole, before going to the computer.
Pf = 1 e
Because the stress is uniform throughout the thickness, the surface integral reduces to a line integral and the bi-axial correction factor c is constant. The B term of the risk equation could then be rewritten:
Bushing material
A material soft enough to conform to the micro surface of the glass hole without causing micro stress concentrations, is usually not stiff or strong enough to efficiently transfer bending forces in the pin. For this reason, a pin and bushing are generally used. Ideally, the elastic modulus of the bushing should be close to that of the glass. For this reason, annealed pure aluminum can be good choice. When using aluminum as a bushing material, extreme care should be taken to specify a grade that wont strain or time harden. The bushing must be dimensionally stable over time. Anything that swells or shrinks will raise the stress concentration. For this reason, nylon, which absorbs water, can be problematic in humid environments. Delrin, a Teflon impregnated nylon,
B = k c m P m h t dA
m A
where k and m are Weibull parameters, P is the applied load, and is the hole diameter.
I=
m t
dA
Values for I are given in Table 1 and could also be calculated directly for any value of m from the table in Appendix B. Reference values for m and k for which the US
141
Session 11
as 9.40 N/mm2 multiplied by the ratio (60mm/ 160mm) equals 3.52 N/mm2. From Case 5 the stress concentration factor at the top of the hole (= /2) is 3.0 respectively. The average bearing stress at the hole is equal to 4.9kN divided by the projected area (19mm times 50mm) equals 5.16 N/mm2. From Case 3, the stress concentration factors at the top of the hole (=0) and side of the hole (=/2) are equal to 0.0 and 0.84 respectively. The stress at the top of the hole and sides of the hole are then:
Session 11
is a better choice as it does not absorb undue amounts of water and wont bind in friction. Delrin has a modulus of elasticity about 1/10th that of glass and for this reason a delrin or other synthetic bush should be made as thin as possible to ensure the best possible distribution of radial bearing stress along the inner rim of the glass hole. Petersons also has good information on modifying the stress concentration factor for different ratios of plate elasticity to pin elasticity.
Yurakucho Canopy, the Cam Bezel technique for bearing in laminated glass utilizes two nested bushings, each with an offset hole. This method is well documented (Wilson, 1999). The nested bushings are each free to pivot (ideally during installation only) and the superposition of rotations for each offset bushing allows a two dimensional variation of pin position. Increasing the variation capacity requires larger cam hole offsets and a larger glass hole. The cam system could be designed to meet any tolerance, that is, to fill the hole but as the cams get larger in relation to the pin, load sharing drops off and bending stress in the pin can become the limiting design factor. Of course ductile failure of a pin is certainly preferable to brittle failure of the hole and replacing pins is typically easier and cheaper than replacing laminated glass. Even when offsets are small, the cam bezel requires a larger hole in relation to the diameter of the pin than any other method presented here and because of that and the number of components involved is typically more expensive. Additionally, installation of the bezel can initially be difficult but generally, once understood, the mechanism generally requires less fieldwork than many of the others methods discussed. When properly designed, manufactured and installed, the bezel has the potential for uniform load distribution among glass plies.
Epoxy/Urethane filler/bush
Figure 14: Yurakucho Canopy
Epoxy has been used for years in critical structural anchorages worldwide and is emerging
142
Figure 17
design element began with the Pilkington Planar Fox Plaza Canopies (1995 Century City, California). These glass canopies were designed to shield occupants from glass falling from the adjacent 100m high rise office tower that had a particular problem with nickel sulfide breakage. (The manufacturer of the tower glass was not Pilkington) The impact criterion was certainly the most interesting feature of the project, but the canopies also had to withstand unusually large in-plane seismic racking due to the flexibility of the supporting steel columns. The fittings were
Canopy was directly inspired by Dewhurst Macfarlanes Yurakucho canopy in Tokyo. The canopy is constructed similarly of multiple laminated beams pinned together. The 86th Street Canopy uses the method of separate drilled disks to transfer load into the laminate beams. In the shop, each beam was fabricated, assembled and drilled. The beams were then shipped and erected as an assembly. The cantilever span of the canopy (wind uplift condition) is 6 meters.
Figure 16
Figure 18
143
Session 11
as a new option in structural glass connections. Cast epoxy or urethane filler bushings could be useful for bearing in triple-lam constructions, as unless oversized, offset in the outer ply holes would prevent installation of any other type of bushing (see above). However, this method can be very messy and it is often difficult to properly align, set, and hold a pin through multiple holes. The designer must also understand (and verify) the long- and short-term interaction of both the glass and the laminate interlayer with the particular epoxy to be used.
designed to force displacement along a single axis, which generated unusually large in-plane shear forces in the 18mm diameter countersunk hole. Cyclic racking tests were successfully performed for this project, which demonstrated the strength of the hole for elastic (+/ 15mm per silicone joint) and inelastic (+/ 30mm per silicone joint) seismic events.
Visually it was very important to lift the canopy glass off the beams. It was felt that for this structure, disconnecting the structural elements reinforced its lightness. It was important that each beam layer had its own identity, therefore each the shapes are distinctly different and the distance between fins exaggerated. Rather than the two-pin connection used on Yurakucho to connect the composite beam to the canopy, the 86th Street Canopy used a single point, rotating connection. These stainless steel connections are free to rotate at every point except the rod tieback, allowing for expansion and contraction as well as installation tolerance control. Each fitting was able to accommodate installation
Session 11
were used to develop seismic forces. As a result, the new structural glass system in the atrium had to be designed to absorb large differential seismic motion between the two existing structures effectively functioning as the seismic joint. Specialty bearing connections, developed exclusively for this project, were designed to transform unpredictable, random ground shaking into a predictable, directed and controlled glass displacement pattern regardless of direction or magnitude the earthquake. The system successfully endured simulated earthquake cyclical racking at the Pilkington test labs.These tests demonstrated the capacity of the fittings to drive
tolerances by adjusting up to five degrees and 20mm. One of the difficulties in design was the separation of the fins in the transverse direction, which increased bending load in the pin. The diameter of the through pin had to increase to that it was stiff enough to transfer loads uniformly (enough) through individual plates. This increase in diameter forced the pin center farther down into the section of the beam.
Figure 21: IBT Test assembly
144
that intimacy often gave insight into creative methods and simplification. Today, an intuitive feel of the problem is no less necessary and fast methods of approximate analysis are essential in facilitating variation, creativity and also for the checking and understanding of computer analysis.
Session 11
Acknowledgments
Dr. Michael Botwin, California Polytechnic State University, San Luis Obispo Tim Macfarlane, Dewhurst Macfarlane and Partners David Shea, Dewhurst Macfarlane and Partners Jonathan Sakula, Dewhurst Macfarlane and Partners Philip Wilson, Dewhurst Macfarlane and Partners Michael Mulhearn, TriPyramid Structures TJ Dehghanyar, Advance Structures Incorporated
References
Beason, W.L., and Morgan, J.R. (1984). Glass failure prediction model. J. Struct. Engrg., ASCE, 110(2), p. 197-212 Beason, W.L., Kohutek, T.L., Bracci, J.M.,(1998). Basis for ASTM 1300 Annealed Glass Thickness Selection Charts J. Struct. Engrg., ASCE, Feburary, p. 215-220 Pilkey, W., (1997) Petersons Stress Concentration Factors, 2nd ed., John Wiley & Sons, Inc., New York, p. 343, 391-394, 422424 Reiner, H. und Strauch, Fr., (1933) Ringplatte und Augenstab Ingenieur-Archiv, vol. 4, p. 481 Theocaris, P.S., (1956), The Stress Distribution in a Strip Loaded in Tension by Means of a Central Pin, Trans. ASME, Appl. Mech. Section, Vol. 78, p 482 Timoshenko, S., (1951) Theory of Elasticity , 2nd ed.,McGraw Hill, New York, p. 55-123 Wakui, Satoru (1999) Architectural glass as a structural element in Japan, ASHAHI Glass, Glass Processing Days, Tampere, Finland ISBN 952-91-08850 Wilson, P., (1999) Bearing technology in glass, Centre for Window & Cladding Technology, Bath, UK p. 53
Conclusions
The essence of the problem of bearing in glass holes is minimizing ovalization. The most sensitive variables are: the relationship between hole diameter and edge distance, the fit of the pin in the hole, and the quality of the surface of the hole. In the past, intimacy with these problems was a stronger requirement for accurate solutions and
Appendix A:
Equations used for generating stress concentration graphs Differential Equations of Equilibrium in Polar Coordinates
1 1 2 + r r r 2 2 2 = 2 r 1 r = r r
r =
+ a n r n + bn r n + 2 + a n r n + bn r n + 2 cos + c n r n + d n r n + 2 + c n r n + d n r n + 2 sin
' ' ' ' n=2 n=2
145
Constants of Integration:
an , bn , cn , d n , an , bn , cn , d n
'
'
'
'
( r ) r = a = A0 + An cos(n ) + Bn sin( n )
n =1 n =1
Session 11
( r ) r =b = A0 + An cos(n ) + Bn sin( n )
' ' ' n =1 n =1
( r ) r = a = C 0 + C n cos(n ) + Dn sin( n )
n =1 n =1
( r ) r =b = C 0 + C n cos(n ) + Dn sin( n )
' ' ' n =1 n =1
1 a0 = 2 an =
f ( x)dx
1 f ( x) cos(nx)dx
1 bn = f ( x) sin(nx)dx
Appendix B:
Unitized stress concentrations: (Appendix B.xls)
146