Bridges for Service Life Beyond 100 Years: Innovative Systems, Subsystems, and Components (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

173 A p p e n d i x d Bearings are an important bridge element that must be con- sidered when evaluating potential ways to extend overall bridge service life. Bridge superstructures experience trans- lational movements and rotations caused by traffic loading, thermal effects, creep and shrinkage, wind and seismic forces, initial construction tolerances, and other sources. Bridge bearings are designed and built to accommodate these move- ments and rotations while supporting required gravity loads, transmitting those loads to the substructure, and providing the necessary restraint to the structure. Background and problem Statement Newer types of bridge bearings using elastomeric materials have improved durability expectations. Steel-reinforced elasto- meric bearings have shown very good performance over the past 40 years due to their low cost and relatively long service life. However, with certain combinations of load and movement, the capacity of elastomeric pads to accommodate the required translation through shear deformation of the elastomer can be exceeded. In these cases, there is a need to provide additional movement capacity by means of sliding surfaces. Further, other types of bearings—including cotton duck pads and high-load multirotational pot, disc, and spherical bearings—use sliding surfaces to accommodate expansion requirements. Currently, polytetrafluorethylene (PTFE) is the material typically used for sliding surfaces. However, the service life of plain PTFE can be shortened due to eventual wear and deterioration of the sur- face, particularly when subjected to high pressures, fast sliding speeds, or cold temperatures. There are minimal data to deter- mine a life prediction model for sliding surfaces using PTFE. Research Objectives The main objective of this research was to determine the fea- sibility of achieving increased service life with bearings that use sliding surfaces for movement through the use of alterna- tive high-performing materials in lieu of conventional plain PTFE. The study compared performance and wear of alterna- tive high-performance sliding surfaces with plain PTFE over a range of bearing pressures, sliding speeds, cycles of move- ment, and total cumulative travel distances. The scope of this research was to conduct a proof of con- cept experimental program on small-scale sliding surface specimens. Performance characteristics such as coefficient of friction (COF), rate of wear, and total wear over a specified number of movement cycles were measured and compared. Two alternative materials were tested along with plain PTFE. The first was a relatively new German product, Maurer sliding material (MSM), which was shown by testing in Europe to have improved wear resistance over plain PTFE. The second was a glass-filled PTFE, Fluorogold, which was also described as pro- viding significantly greater wear resistance than plain PTFE. Analytical studies were performed to evaluate the magnitude of horizontal movement due to girder end rotation at expan- sion bearings under cyclic truck load. These studies assisted in determining realistic movement testing speeds with which to evaluate the various sliding surface materials. Bearing move- ment due to truck load is low-amplitude, high-cycle movement with fast movement speed, but movement due to temperature load is high-amplitude, low-cycle movement with low move- ment speed. Another objective was to evaluate the feasibility of devel- oping life prediction models for sliding materials that could be used for service life design. Design provisions were further developed for calculating the required thickness of the sliding surface based on the available literature, theoretical studies, and the results of the experimental program. Summary of Available Data Previous research reported in NCHRP Report 432 (Stanton et al. 1999) and earlier by Campbell and Kong (1987) Wear of Sliding Surfaces in Bridge Bearings

174 demonstrated the properties and wear behavior of PTFE. Key results of these studies are summarized in later sections, which are “Other PTFE Research” and “Phase Two Concep- tual Design of the Experiment.” This previously reported behavior was used as a starting point for the research reported here. Stanton et al. (1999) reported that low temperatures, fast sliding speeds, high contact pressures, rough mating surfaces, and contamination of the sliding interface all contribute to an increased wear rate of PTFE. However, fast sliding speed was shown to be a dominant parameter. Movement due to tem- perature change is low-cycle, high-amplitude movement, with a slow movement rate, and produces the least amount of wear. But movement due to truck load and associated dynamic effects is high-cycle, low-amplitude movement and has a much faster sliding speed. Wear rates associated with high sliding speeds have been shown to be significantly greater than wear rates at lower sliding speeds. Unfortunately, there are little actual field data to confirm the amount of movement and movement rates at expansion bearings due to truck load, and theoretical analysis is the only currently available way of estimating this behavior. Alternative high-performing sliding materials with poten- tially greater resistance to wear when subjected to high move- ment speeds were investigated as part of this research and were identified for experimental study. Current data relating to the behavior of these materials were available in some cases from the manufacturers, and when relevant, were used as a basis for this study. Descriptions and data for the two high- performance sliding materials studied, MSM and Fluorogold, are summarized in the following subsections. Maurer Sliding Material MSM was primarily developed to accommodate fast bridge movements caused by high-speed trains. In contrast to conventional bridges, which experience slower movement due to live load, a bridge for high-speed trains will undergo fairly fast loading. The German Transrapid magnetic train guideway required the bearing material to move at sliding speeds up to 15 mm/s (35 in./min) and provide 80 years of service life. To meet these needs, Maurer developed a new sliding material that exceeded the performance requirements. Based on manu- facturer data, material such as MSM could be considered a high-performing alternative sliding surface to PTFE. The manufacturer has conducted tests in Europe compar- ing MSM with PTFE in accordance with provisions of EN1337-2 (Dansk Standard), which specifies the sliding ele- ments for structural bearings. Long-term sliding tests were carried out on dimpled and lubricated MSM and PTFE specimens. The test parameters for both MSM and PTFE are summarized in Table D.1. In these tests, MSM was subjected to higher pressures and move- ment velocities than PTFE. Figure D.1 shows test results comparing COF with total sliding path and various temperatures. The study showed that MSM could accommodate a much longer total sliding move- ment than PTFE and had considerably greater performance at lower temperatures. Figure D.2 and Figure D.3 show the surface conditions of sliding material and stainless steel mating surface for MSM Table D.1. Parameters in Long-Term Sliding Tests by Maurer Parameter Tested PTFE MSM Contact pressure 30 N/mm2 (4,350 lb/in.2) 60 N/mm2 (8,700 lb/in.2) Average sliding velocity 2 mm/s (5 in./min) 15 mm/s (35 in./min) Total accumulated sliding distance 20,000 m (12 mi) 50,000 m (31 mi) Figure D.1. Long-term test of MSM compared with PTFE.

175 and PTFE, respectively, after testing. The PTFE photo repre- sents a total accumulated sliding length of 10,000 m (6 mi), and that of MSM represents more than 50,000 m (31 mi). In these tests, the MSM surface was only slightly abraded, but the PTFE showed considerable wear and shredding of material. MSM also demonstrated high compression strength and could accommodate double the stress of PTFE. This ability could reduce the dimensions of the bearing, if feasible. Tests also showed that MSM experienced very limited abrasion, which extends its life to more than 50,000 m even at high stresses and high loading rates. From these tests on lubri- cated samples, it appeared that MSM could provide an excellent alternative for many bearing types that use sliding surfaces. Figure D.2. Shape of MSM after sliding path of more than 50,000 m (31 mi). Figure D.3. Shape of PTFE after sliding path of 10,000 m (6 mi).

176 Further testing on unlubricated samples, however, will indi- cate whether this product could be a true sliding surface alter- native for conventional-type bridges in U.S. practice, which typically use unlubricated, plain PTFE. There is concern whether such lubrication would remain throughout the service life of the structure under field conditions. Fluorogold Another high-performing sliding surface, Fluorogold, was identified for comparative testing with plain PTFE and MSM. Fluorogold is a proprietary glass-filled, reinforced PTFE with enhanced characteristics. Fluorogold sliding bearing systems are engineered products of Seismic Energy Products, Athens, Texas. Based on manufacturer’s information, “Fluorogold is com- prised of virgin PTFE and special reinforcing agents. This blend yields a structural material that offers significantly higher mechanical properties than plain PTFE. Compressive creep is virtually eliminated, wear is substantially reduced, and initial deformation is decreased; however, the low friction and chemi- cal inertness of PTFE are retained. This structural bearing sur- face is bonded to a back-up steel plate with a high-temperature homogenous epoxy system that is cured under precise heat and pressure in hydraulic presses. All Fluorogold bearings are factory bonded using strictly controlled, semiautomated pro- cedures developed to eliminate poor quality field-made bonds. The maximum design pressure of Fluorogold slide bearings, without elastomeric backing, is 2,000 psi. For neoprene-backed bearings, the maximum recommended pressure is 800 psi, and for the cotton duck–reinforced elastomer, it is 1,500 psi.” The decision was made to incorporate Fluorogold into the testing program as a high-performance PTFE-based material to compare with MSM and plain PTFE. Other PTFE Research NCHRP RePoRt 432, APPeNdix C: FRiCtioN ANd WeAR oF PtFe SuRFACeS In Appendix C of NCHRP Report 432, Stanton et al. (1999) describe research conducted on the friction and wear of five types of PTFE-based sliding surfaces: • Flat, unfilled, unlubricated (plain) PTFE; • Dimpled, lubricated PTFE; • Woven PTFE; • 25% Glass-filled PTFE; and • 15% Glass-filled PTFE. The testing program considered temperature, contact pres- sure, and sliding speed and compared COF and wear over the range of variables. The research program tested small-scale specimens (3-in. diameter). Wear was found to be primarily dependent on sliding speed, although the pressure was typi- cally constant in the wear portion of the study. Plain PTFE was shown to have the lowest wear resistance. Lubricated PTFE had the greatest wear resistance, but it was questionable as to how long the lubrication would last in actual field conditions. Woven and glass-filled PTFE materials were shown to have significantly increased wear resistance compared with plain PTFE, but they exhibited a higher COF. Certain initial test parameters used in the NCHRP study, such as specimen size, contact pressure, and movement speed, were duplicated in this study to compare results, which are discussed in this appendix. PtFe SlidiNg SuRFACeS iN BRidge BeARiNgS Research performed on wear of PTFE sliding surfaces by Campbell and Kong (1987) indicated that the combination of pressure times velocity (PV) could be used as a base param- eter to predict the corresponding rate of wear. Their research indicated that there was a PV threshold below which there would be a low-wear regime and above which there would be a high-wear regime. This concept was evaluated as part of the study reported here. Research program Scope of Study This research topic included analytical studies followed by design, fabrication, and implementation of a limited experi- mental program to evaluate sliding surfaces used for bridge bearings. The experimental program included proof of con- cept testing aimed at comparing the life of sliding materials commonly used in bridge structures with certain alternative high-performance sliding materials. Specific material param- eters such as wear rate and COF were studied. The following three sliding materials were tested and compared in this study: 1. PTFE, the base material most commonly used in current bridge practice, was tested in plain and lubricated states. 2. MSM, developed by Maurer Söhne in Germany, is an alter- native material for bridge sliding bearings. MSM, which was developed for use in high-speed rail bridges, was tested in plain and lubricated states. 3. Fluorogold, developed by Seismic Energy Products (Athens, Texas), is an engineered product comprising virgin PTFE and special glass fiber–reinforcing agents. This material was tested only in a plain state. The research program was conducted in four major phases: 1. Phase 1 developed an analysis of a prototype bridge that computed and evaluated bridge movements and move- ment speed at expansion bearings due to truck loads. 2. Phase 2 developed the concept of an experimental pro- gram that was aimed at constructing a system capable of

177 performing wear tests on multiple sliding materials to sim- ulate various travel speeds and contact pressures. Test speci- mens were also designed as part of this phase. 3. Phase 3 constructed the test set up in the University of Nebraska–Lincoln structures laboratory. 4. Phase 4 conducted the experimental testing. Phase 1: Determination of Bridge Movements Finite element (FE) analysis and theoretical methods were used to calculate the movements that a sliding bearing might experience during its service life. Prototype Bridge Structure An FE model was developed to compute the simple-span bridge horizontal movement at an expansion bearing due to passage of a truck. Because the bridge movements depend on the bridge geometry and vary case by case, analyses were carried out on a 200-ft simple-span prototype steel bridge with four girders, as shown in Figure D.4. The FE package ABAQUS was used to develop a model of the prototype bridge. Reduced-integration shell elements were used to model the girders and the deck. Figure D.5 shows the schematic of the model developed. The passage of AASHTO HL-93 trucks was considered in evaluating the bridge horizontal movements at the expansion bearings. Figure D.5 shows the bridge with one truck located at the midspan of the bridge. Figure D.6, a, b, and c shows the verti- cal displacement contours due to the self-weight of the bridge and one, two, and three trucks, respectively, located side by side at midspan. One end of the girders was restrained with a pinned boundary condition (fixed end) and was free to move with a roller boundary condition (expansion end) at the other end. Table D.2 summarizes the movements and rotations at the roller bearings of the external girder (which experience the greatest deformations). A More General Case: Traffic-Related Movement Three major factors contribute to the traffic-related horizon- tal movements at bridge bearings: 1. Rotation of the girder ends. 2. Horizontal movement at the expansion bearing trans- ferred from the other fixed end of the girder. In other words, if one end of the girder is pinned and restrained against horizontal movement, its theoretical movement due to girder end rotation movement is transferred to the expan- sion end of the girder. 3. Dynamic effects, which include the following effects: a. A dynamic impact factor, which can be taken as 1.33 (AASHTO 2010b); and b. Vibration of the bridge after the truck is passed (dynamic decay). Figure D.7 shows a general simply supported beam with a concentrated load at midspan and the bridge girder cross sec- tion under bending moment in which qD1 is the horizontal movement of the bearing. Equations D.1 and D.2 can be written for the above simply supported beam as follows: Pl EI48 (D.1) 3 ∆ = Pl EI l16 3 (D.2) 2 θ = θ = ∆ Although the AASHTO LRFD Bridge Design Specifications (LRFD Specifications) (AASHTO 2010b) allow greater values, the maximum allowable live load deflection used by many states is span/800 for a bridge under vehicular loads. Therefore, for a worst-case scenario of a concentrated load at midspan of the bridge, the maximum probable rotation at the support can be determined as follows: l l 3 800 3 800 rad. 0.00375 rad.θ = × = = Figure D.4. Schematic of 200-ft simple-span prototype bridge. Figure D.5. ABAQUS FE model of prototype bridge.

178 (a) (b) (c) Figure D.6. Vertical displacement contours due to self-weight of bridge and (a) one, (b) two, and (c) three trucks located side by side. It should be noted that both ends of the beam will experi- ence the rotation q, and the girder bottom flanges will tend to move outward horizontally. With one end pinned and restrained from horizontal movement, the opposite end will be forced to move a total of 2qD1. For a range of spans and girder depths, the movement can be calculated as follows: • For D1 equal to 30 in., the maximum probable horizontal movement will be 0.00375 × 30 × 2 = 0.225 in. • For D1 equal to 70 in. (which is the case for the 200-ft span under consideration), the maximum probable horizontal movement will be 0.00375 × 70 × 2 = 0.525 in. For structures with the above range of variables designed in accordance with the LRFD Specifications, the maximum probable horizontal movement at the expansion bearing due to the passage of a truck (with one end fixed) would be between 0.225 and 0.525 in. for D1 ranging from 30 to 70 in. This range represents the maximum probable bearing

179 horizontal movement for a simple-span bridge, because most bridges are designed to have lower midspan deflections due to live loads (L/800 is the upper limit). Comparison of Approximate Results Against FE Study The horizontal movement due to one truck passage deter- mined from the FE study on the prototype bridge was 0.146 in., which is less than the above-mentioned range of horizontal movement determined from approximate methods. The lower amount determined from the FE study is con- sidered to be due to the presence of the other girders in the model, which increased the stiffness of the whole system. Effects of Vibration When a truck travels over a bridge, and after it has completely crossed over, the bridge continues to vibrate with progressively reducing amplitude because of damping. The principles of structural dynamics were used to determine the dynamic effects of a truck passage on the prototype bridge. Structures dissipate energy, mainly through friction, dur- ing vibration. The progressively reduced amplitude of vibra- tion is caused by the presence of damping forces that dissipate the input energy. Damping is a function of the structure, and its value has been determined from dynamic evaluation of various types of structures. In most actual physical systems it is very difficult to find the exact expression for the damping force; a viscous damp- ing model is widely chosen for its ease in mathematical treat- ment. In this model, the damping force is expressed as the viscous damping coefficient (C) multiplied by the velocity of the movement. In a critically damped system, the amplitude converges to zero as fast as possible without oscillating. The damping ratio is defined as the ratio of the damping coeffi- cient in the system’s differential equation to the critical damp- ing coefficient. The structure damping ratio is typically 2% to 10%, depend- ing on the type of construction (bolted steel ~6%, reinforced concrete ~5%, and welded steel ~2%). Generally, damping would be ignored for nontransient events (such as wind loading or crowd loading), but it would be important for transient events (e.g., impulse loads). After the passage of a truck over the bridge, the bridge will vibrate until the structure’s damping dissipates the amplitude of the motion, as described in Figure D.8. Assuming 7% damping, the dynamic decay can be deter- mined from Equation D.3: ln Peak@cycle1 Peak@cycle2 2 1 0.441 (D.3) 2 δ = = piξ − ξ = Having the dynamic decay determined according to Equa- tion D.3, the number of cycles (N) to transition from an Table D.2. Movements of Prototype Bridge Load Bearing Type Midspan Deflection (in.) Horizontal Deformation at Bearing (in.) Rotational Deformation at Bearing (Rad.) Weight of girder plus deck Roller 6.62 1.18 0.0148 1 HL-93 Truck Roller 1.28 0.15 0.0025 2 HL-93 Trucks Roller 2.04 0.26 0.0040 3 HL-93 Trucks Roller 2.31 0.33 0.0043 (a) l, EI P (b) D2 1D D1 Figure D.7. (a) A simply supported girder with a concentrated force at midspan and (b) bridge girder section under bending moment.

180 amplitude of xi to an amplitude of xi+k can be calculated from Equation D.4: N x x i i k 1 ln (D.4)= δ + The number of cycles required to reduce the free vibration amplitude to 10% of the first cycle’s amplitude can be deter- mined from Equation D.5: N 1 ln 10 1 1 2 ln10 0.366 1 0.366 (D.5)10% 2 2 = δ = − ξ piξ = − ξ ξ ≅ ξ For example, for x = 7% ⇒ N10% = 5 cycles. Figure D.9 shows the number of cycles required to reduce the free vibration amplitude by 90% and 95% (decay) for various damping ratios. As this figure shows, for a 7% damp- ing ratio, five cycles will be enough to reduce the first cycle’s amplitude to 10% for a free vibration. To calculate the total travel distance from the first cycle to 90% reduced amplitude, Equations D.6 and D.7 can be used: ln Peak @cycle1 Peak @cycle2 Peak @cycle1 Peak @cycle2 Peak @cycle2 Peak @cycle1 (D.6) e e δ = ⇒ = ⇒ = δ −δ ( ) = + + + + + = + + + + + − δ −δ − δ − δ − δ −δ − δ − δ Total travel 2 2 2 2 2 2 1 (D.7) 1 1 0.5 1 1 1.5 1 2 1 0.5 1.5 2 x x e x e x e x e x e e e e   Two main elements contribute to the cumulative total travel distance caused by each passage of a truck on a simply supported bridge (the discussions presented and equations developed are for simply supported bridges and could con- servatively be used for continuous bridges): 1. The deflection and rebound of the bridge (forced move- ment) during passage of a truck causes the bridge bearings to experience horizontal movements. The following items could be considered in calculating the total horizontal movement of the bearings during passage of a truck: a. For a simply supported girder, once a truck reaches the bridge midspan, the girder deflection is at a maximum, as is the horizontal movement at the bearings due to end rotation. As the truck moves toward the end of the span, the girder rebounds to its original position; con- sequently, the bearings rebound to their original posi- tion. Therefore, the total horizontal movement of the bearings during passage of one truck is equal to two times the horizontal movement resulting from girder maximum displacement. b. If one end is fixed against translation, then movement due to girder end rotation is transferred to the expan- sion end. If both ends of the span have expansion bearings, each end experiences horizontal movement due to girder rotation, and there is no movement transfer. In the case of the three-dimensional (3-D) FE model for the prototype bridge, the fixed and expansion ends of the bridge are represented in the bridge model, and the resulting movement at the expansion end reflects the movement transfer from the fixed end. c. Trucks are moving objects, and therefore a dynamic impact factor needs to be considered. In this study an impact factor of 1.33, as recommended by the LRFD Specifications, was used. Figure D.8. Decay of free vibration for a single degree of freedom system with damping. Figure D.9. Number of cycles required to reduce the free vibration amplitude by 90% and 95% (decay) for various damping ratios.

181 2. After the truck clears the span, the bridge bounces back to its original position, and then continues upward to a deflection slightly less than the initial downward deflec- tion caused by the truck during its passage. The bridge will continue to vibrate up and down until the structure’s damping dissipates the amplitude of the motion, as shown in Figure D.8. The term 2x1 in Equation D.7 is the amplitude of move- ment (x1 is the maximum initial horizontal movement at the expansion bearing). Note that the amplitude of the movement is two times the initial maximum horizontal movement. Equation D.7 includes all the above-mentioned effects except the 1.33 impact factor. This equation can be plotted as shown in Figure D.10. As this figure shows, for a 7% damping ratio (on the horizontal axis), the total travel dis- tance at the expansion bearing will be 10.1 times the maxi- mum initial horizontal movement. This factor is 7.4 for a 10% damping ratio. Therefore, every time a truck passes the bridge, the total movement, including the movement due to vibration, will be approximately 10.1 times the first deformation due to the truck passage for a 7% damping ratio and about 7.4 times for a 10% damping ratio. For the prototype bridge, the maximum initial horizon- tal movement (one direction) due to truck passage was cal- culated from the FE analysis as 0.146 in. A dynamic impact factor of 1.33 should also be applied to this movement to account for the fact that the truck is a moving object. Note that the prototype bridge FE model was a 3-D model with one end of the girders pinned (restrained against horizon- tal movements) and the other end of the girders free to expand. The horizontal movement of 0.146 in. was directly extracted from the 3-D model at the actual location of the expansion bearings under the bottom flange of the girder, and it already accounts for the horizontal movement transfer from the fixed end. Thus, the multiplier 2, for the transferred horizontal movement from the fixed end of the girder, does not apply to this case; therefore, the total cumulative horizontal movement due to one truck passage is as follows: Cumulative movement 0.146 1.33 7.4 1.44 in.= × × = The 7.4 factor shown in the above equation accounts for free vibration of the bridge and is obtained from Figure D.10, assuming a 10% damping ratio. According to National Bridge Inventory (NBI 2013) data, the average daily truck traffic (ADTT) of most bridges ranges from 50 to over 5,000. For an ADTT of 1,000, the total hori- zontal movement of the expansion bearing, in miles, over a 100-year service life could be estimated as follows: 1,000 365 100 years 1.44 12 5,280 800 mi( )× × × ≈ Table D.3 summarizes the results for similar calculations considering ADTTs ranging from 1,000 to 5,000, with vibra- tion included. For comparison, the table also summarizes the total cumulative movement without vibration, considering only one full cycle of movement (i.e., initial deflection and rebound to the original position after truck passage). It is concluded that total cumulative horizontal movement at an expansion bearing due to truck load can be very signifi- cant, especially when movement due to free vibration is also considered. Figure D.10. Graph to determine total travel distance (to determine total travel distance of the sliding bearing, the maximum initial horizontal movement is multiplied by the number read from this plot). Table D.3. Approximate Cumulative Movement over 100 Years Due to Truck Load ADTT Approximate Total Cumulative Travel (mi) With Vibration (10% damping ratio) Without Vibration (one cycle) 1,000 800 200 2,000 1,700 450 3,000 2,500 700 4,000 3,300 900 5,000 4,100 1,100

182 Bearing Movements Due to Temperature Variations Temperature fluctuations, including daily and seasonal tem- perature changes, also contribute to total horizontal movement of expansion bearings. The accumulated movement of the pro- totype bridge bearings with 100-plus years of service life due to daily and annual temperature variations was evaluated. dAily temPeRAtuRe VARiAtioN Assuming 40°F average daily temperature change, the accu- mulated bearing movement can be determined as follows: L L T2 . . 2 7 10 200 12 40 1.34 in.Daily 6 ( )∆ = × α ∆ = × × × × × =− For 100 years of service life, the total daily accumulated bear- ing movement is given by the following equation: 365 100 1.34 48,910 in. 4,076 ftTotal, dailyL∆ = × × = = ANNuAl temPeRAtuRe VARiAtioN If an average annual temperature variation of 110°F is assumed, the annual end movement due to this thermal load can be determined as follows: L L T2 . . 2 7 10 200 12 110 3.70 in.Annual 6 ( )∆ = × α ∆ = × × × × × =− For 100 years of service life, the total annual movement is given by ∆ = × = =100 3.70 370 in. 30.8 ft.Total, annualL totAl ACCumulAtiVe BRidge moVemeNt due to tHeRmAl loAdS Finally, the total horizontal movement at the expansion bear- ing for the prototype bridge over 100 years of service life due to both daily and annual temperature variation is given as follows: L L L 4,076 308 4,107 ftTotal Total, annual Total, daily∆ = ∆ + ∆ = + = Speed of Bearing Movement for Truck Load The speed of travel for the sliding bearing can be determined from the general formula given by Equation D.8: Average travel speed Distance traveled Travel time (D.8)= As previously discussed, high-speed movement causes the majority of sliding-material damage. Bridge deflection dur- ing the passage of a truck, and subsequent free bridge vibra- tion, results in different sliding speeds. SlidiNg SPeed duRiNg tRuCk PASSAge The time for a truck to pass the bridge can be calculated as shown by Equation D.9: t Bridge span Truck speed (D.9)= The slip rate of the bearing can be determined as shown by Equation D.10: a D t Slip rate 1.33 2 (D.10)1= × θ × × where a = 2, if entire horizontal movement due to girder rota- tion at both ends is taken at one end of the girder (as for a simple-span girder with one end fixed and one end expansion). Otherwise, use 1; q = rotation of the girder end due to truck passage (Rad); D1 = depth of neutral axis measured from the bottom flange (in.); and 1.33 = dynamic impact factor due to truck passage. The following discussion outlines the steps in calculating the slip rate or velocity of horizontal movement at the expan- sion bearing during passage of trucks for the prototype bridge. If it is assumed that the truck passes the bridge’s 200-ft span with a speed of about 50 mph, then time in seconds for the truck to cross the bridge would be as follows: t x t 200 ft 50 mph 5,280 ft mi 3,600 s h 2.73 s= = × × = The horizontal movement at the expansion bearing for the prototype bridge from FE results was 0.146 in.; therefore, in the slip rate equation above (Equation D.10), qD1 = 0.146 in., and a = 1. From the equation for t above, it would take the truck 2.73 s to cross the bridge. During this period the horizontal movement will be twice the 0.146-in. deflection (bridge moving down and then returning to its original undeflected position). Therefore, the slip rate or velocity of horizontal movement during passage of one truck, considering the impact factor, will be as follows: Slip rate of the bearing Bearing movement 0.146 in. 2 1.33 2.73 s 0.142 in. s t = = × × = If the bearing movement from the more general case approximate analysis is used, the slip rate would be as follows: Slip rate of the bearing Bearing movement 0.525 in. 2 1.33 2.73 s 0.51 in. s t = = × × =

183 Table D.4 shows the same calculations for other truck speeds, with unit conversion to inches per minute, for movements rep- resenting FE analysis and the approximate analysis. Sliding Speed due to Bridge Free ViBration aFter truck paSSage Once a truck has crossed the bridge, the bridge continues to vibrate freely until the vibration is damped out. The average sliding speed of horizontal movement after truck passage is equal to the cumulative movement after truck passage divided by the time required for that cumulative movement to occur. This sliding speed due to free vibration is independent of the initial truck speed and depends rather on the dynamic char- acteristics of the bridge (stiffness, mass, and damping). The distance traveled due to free vibration after truck pas- sage can be computed by Equation D.11: Distance traveled during free vibration 1.33 2 (D.11)1 a D f( ) = × × θ × × − where a = 2, if entire horizontal movement due to girder rota- tion at both ends is taken at one end of the girder (as for a simple-span girder with one end fixed and one end expansion). Otherwise, use 1; q = rotation of the girder end due to truck passage (Rad); D1 = depth of neutral axis measured from the bottom flange (in.); 1.33 = dynamic impact factor due to truck passage; and f = factor used for calculating total distance traveled taken from Figure D.10 (this value is 13.7, 10.1, and 7.4 for damping ratios of 5%, 7%, and 10%, respectively). The term (a × q × D1) in Equation D.11 is the initial maxi- mum movement at the expansion bearing, which from the FE analysis of the prototype bridge was 0.146 in. Figure D.11 shows the total horizontal movement with time and how the cycles of movement dissipate over time after the truck leaves the span. The total time t for the free vibration after truck passage can be determined from Equation D.12: t T T 2 0.366 (D.12)= + ξ × where T is the free vibration period of the bridge, which is discussed in the next section on natural frequency, and x is the damping ratio. The term 0.366/x, which is from Equation D.5, is the num- ber of cycles required to reduce the free vibration amplitude to 10% of the first cycle amplitude. For the prototype bridge, the horizontal movement of 0.146 in. was directly extracted from the 3-D model at the actual location of the roller bearings under the bottom flange of the girder. Therefore, the multiplier 2, for the transferred horizontal movement from the other end of the girder, does not apply to this case. From Equation D.11, the distance traveled due to free vibration after truck passage with a 10% damping ratio can be calculated as follows: Distance traveled during free vibration =1.33 0.146 7.4 2 1.05 in.( ) × × − = Time t for this amount of travel can be computed from Equation D.12, and with a 10% damping ratio is given by t 1.26 2 0.366 0.10 1.26 5.2 s= + × = Therefore, the speed of movement is Speed 1.05 5.2 60 12.1 in. min= = This is the average speed of bearing movement during the free vibration phase that occurs after passage of a truck. It is Table D.4. Sliding-Surface Slip Rate for Various Speed Limits During Truck Passage Truck Speed (mph) Slip Rate (in./min) From FE Analysis From Approximate Analysis 40 6.8 18.4 50 8.5 23.0 60 10.3 27.8 Figure D.11. Cycles of horizontal movement with time.

184 based on the initial movement from the FE analysis and assumes a 10% damping ratio. Natural Frequency of the Bridge Structure The natural frequency of the bridge can be determined by considering the structure as a single degree of freedom (SDOF) system, as shown in Figure D.12. The first step is to determine the stiffness of the equivalent SDOF system. Table D.2 shows that the external girder’s mid- span deflection under the weight of an HL-93 truck is 1.28 in. HL-93 truck weight is 72,000 lb. Therefore, the midspan stiffness of the girder can be determined as follows: K Load Deflection 72,000 1.28 56,250 lb in.= = = Now the total mass of the bridge (M) can be determined as follows (bridge span is 200 ft, and bridge width is 42 ft): Area of each girder 20 1 84 0.625 22 1.57 107.04 in.2 = × + × + × = w M w g s 8 12 150 200 42 4 107.04 12 200 486 1,129,008 lb 1,129,008 386.4 2,921.86 lb in. 2 2 ( ) ( )= × × × + × × × = = = = Therefore, the bridge period can be determined from the following equation: 2T K M Kn LM e e = pi where KeLM is load–mass factor for elastic range (0.78 for this case). This factor is used because the entire bridge mass has been concentrated at the midspan of the bridge, and the mid- span stiffness of the bridge is used (the uniform distributed bridge mass has been concentrated as a lump mass). 2 0.78 2,921.86 56,250 1.26 sT = × pi × = Phase 2: Conceptual Design of the Experiment Test Parameters NCHRP Report 432 (Stanton et al. 1999) describes the follow- ing parameters to be significant in the wear of PTFE surfaces: • Contact pressure; • Temperature; • Lubrication; • Type of PTFE; • Sliding speed (speed of travel); • Travel distance; and • The type of backing material to the sliding surface (neo- prene, fabric, steel pads). The literature review revealed that the COF of PTFE decreases when the contact pressure (or the compressive stress on the mating surfaces) increases. The effects of this parameter on MSM material were also studied in this project. Although temperature has been proven to affect the wear behavior of PTFE, this parameter was not studied in this proj- ect because of the limited scope, which was proof of concept testing only. Lubrication has been shown to modify PTFE performance. Stanton et al. (1999) reported that “dimpled lubricated PTFE is the most resistant PTFE to wear.” However, plain unlubricated PTFE was compared with the other materials in this study because this is the material used most often in practice. Actual Bridge Lumped Mass Model Equivalent SDOF System Figure D.12. Modeling of bridge girder using the single degree of freedom (SDOF) system to determine the natural frequency of the structure.

185 Lubrication is not used much in normal U.S. practice because of the uncertainty of how long it will last. The sliding speed is an important factor in the behavior of PTFE material. There exists a sliding speed beyond which the level of friction in PTFE appears to remain constant. However, there is a large discrepancy between the values of this speed among different researchers. The lowest speed has been iden- tified by Mokha et al. (1990) to be around 240 to 480 in./min. The calculated sliding speed for the prototype bridge ranges from approximately 8 to 13 in./min. Therefore, as can be seen for the prototype bridge, the sliding speed is far below the low- est speed range suggested by Mokha et al. (1990), so COF in real-life bridges will vary. Experimental Approach Two approaches were considered for the type of movement to be investigated in the experimental program: (1) simulat- ing temperature-induced bridge movements, which are low- frequency, high-amplitude movements; and (2) simulating horizontal movements due to truck passage, which are high- frequency, low-amplitude movements. Truck passage, as discussed in previous sections, causes significantly higher movement speeds and cumulative total movements for the expansion bearing sliding surface through- out the life of the structure, although its movement amplitude is smaller than the amplitude of temperature-induced move- ment. Further, the contact pressure during a truck passage is higher than the contact pressure during thermal movements, which increases wear. Considering all this information, it was concluded that selecting the second testing strategy will result in simulating a more severe condition for a sliding surface. Mating Surfaces The LRFD Specifications require the use of stainless steel for flat mating surfaces. This stainless steel should be Type 304, conforming to ASTM A167/A264, and have a surface finish of 0.20 µm RMS or better. The LRFD Specifications further state that “ASTM A240M, Type 304, stainless steel, with a surface finish of 4.0 × 10-4 mm (0.40 µm) RMS or better, is appropriate, but the surface mea- surements are inherently inexact, and hence it is not a specified alternative.” For this study, ASTM A240M was used to be comparable with previous research testing. Contact Pressure The LRFD Specifications state the values in Figure D.13 as the range of contact pressures for PTFE. The contact pressures in the long-term sliding test carried out by Maurer on MSM sliding bearings were 4.35 ksi for dim- pled PTFE and 8.70 ksi for MSM. Campbell and Kong (1987) suggest allowable contact pres- sures of 4.35 ksi under dead load and 6.525 ksi under total load for PTFE in bearings in the United States. In the experimental program reported in NCHRP Report 432, contact pressures of 500 to 6,000 psi were used (Stanton et al. 1999). For their wear rate experiments, a contact pressure of 3,000 psi was used, which was considered an upper bound. For this study, an initial contact pressure of 3,000 psi was chosen to be compatible with previous wear tests. Temperature All tests were carried out at room temperature. To maintain room temperature, a fan was installed to cool the samples during the experiment when heat due to friction was observed or experienced. Lubrication Lubrication has been shown to enhance PTFE performance. NCHRP Report 432 (Stanton et al. 1999) notes that dimpled Source: AASHTO 2010b. Figure D.13. LRFD Specifications maximum contact stress for PTFE at service limit state (ksi).

186 lubricated PTFE is the PTFE most resistant to wear. However, it is uncertain that sliding material lubrication in a bridge bearing will stay in place over the life of the bridge. Because unlubricated, plain PTFE is most commonly used in prac- tice, it was chosen as the base PTFE sliding material to be com- pared against MSM and filled PTFE (Fluorogold) in wear tests. Lubricated PTFE and MSM samples were used, however, to evaluate and compare COF in the lubricated condition. Type of PTFE Flat, unlubricated, unfilled (plain) PTFE exhibits much higher wear than dimpled lubricated, woven, or glass-filled PTFE (Stanton et al. 1999), but because it is the most commonly used sliding material in U.S. practice, it was used as the base material for establishing wear rates and for comparison. Plain PTFE was also tested in previous research, so the results from this study could be compared against previous results for establishing baseline wear rates. Previous studies have shown that plain PTFE will experi- ence significant wear at low cumulative movements when subjected to high sliding speeds; however, much higher cumulative movements were expected for Fluorogold and MSM. Failure of PTFE is based on the level of wear, which occurs within relatively low travel distances. The travel dis- tance to wear out MSM and Fluorogold has not yet been established; however, it was expected that preliminary wear rates would be developed in this study. Sliding Speed (Speed of Travel) Two types of slip rates (sliding speeds) can be considered for the sliding surfaces of bridge bearings. The sliding of a bearing due to the passage of a truck is of a low-amplitude, high-frequency nature, but it results in a very large accumulative movement. The sliding of a bearing due to daily thermal movements is of a high-amplitude, low-frequency nature. As previously discussed, the sliding rate for a simple-span bridge with a 200-ft span and the passage of a truck with a speed of 50 mph is about 0.107 in./s, or about 6.42 in./min. This value for a speed of 70 mph would be 8.99 in./min. The speeds used in the long-term sliding test carried out by Maurer on MSM sliding bearings were 0.0787 in./s (4.72 in./ min) for dimpled PTFE and 0.59055 in./s (35.43 in./min) for MSM (MSM material was developed to be used in high-speed railroad bridge bearings). Based on the report by Campbell and Kong (1987), little information is available for sliding speeds of bearings in real service. In their document they reported thermal movement rates up to 0.000024 in./s (0.0014 in./min) (reinforced con- crete bridge) and 0.000157 in./s (0.0094 in./min) (unsurfaced steel box) for a 200-ft-span bridge. These values are negligible compared with live load movement speeds as high as 0.787 in./s (47 in./min) for a steel railway bridge and 2.756 in./s (165 in./min) for a steel–concrete composite bridge. There- fore, in their report, Campbell and Kong recommended per- forming laboratory tests using maximum speeds of 2.756 in./s (165 in./min). The maximum sliding speed for the PTFE wear tests car- ried out for NCHRP Report 432 (Stanton et al. 1999) was 25 in./min. To be consistent with that study, an initial slid- ing speed of 25 in./min was adopted for this study. The actuators in the structures laboratory of the University of Nebraska–Lincoln have a maximum capacity to apply slip rates of about 60 in./min (for the prescribed travel amplitude). Travel Distance Based on the results of the FE and theoretical studies carried out in the previous sections of this appendix, the average accumulated travel distance at an expansion bearing due to truck load for a common simple-span steel stringer-type bridge, within the parameters studied and in 100 years of ser- vice life, would range between 300 to 1,500 mi depending on the AADT, bridge length, bridge stiffness, and so forth. These values would be different for concrete structures. The travel distances used in the long-term sliding test carried out by Maurer on MSM sliding bearings were approximately 13 mi for dimpled and lubricated PTFE and 31 mi for dimpled and lubricated MSM. Traffic-induced movements range from 476 ft (0.09 mi) to 16,404 ft (3.1 mi) per year (Lebek 1985; Muller-Rochholz et al. 1986; Hakenjos et al. 1985); therefore, for a 100-year service life, the accumulated movement distance discussed in these reports could range from 9 to 313 mi. Extended long-term field measurements and monitoring are necessary to verify the actual amount of bearing movement due to truck load. Types of Backing Material for PTFE Various types of backing materials for sliding surfaces can be used, such as neoprene or fabric pads or steel plates. However, for wear testing, where no rotation is permitted, the simplest method is to use steel plates and chemically bonding and recessing the PTFE. This method was used in this study to be compatible with previous tests by other researchers. Size of Sliding Surface Specimens Discs having a diameter of 3 in. have commonly been used in PTFE testing in Europe and can be an accepted standard (Campbell and Kong 1987). This specimen size was also used in the studies for NCHRP Report 432. The previous MSM

187 studies were performed on samples with diameters of 3.0 and 6.1 in., with most tests performed on 3.0-in.-diameter samples. Previous studies have also shown that the size of the sliding surface specimen has a minimal effect on its behavior, includ- ing its COF. A 3-in.-diameter specimen size was selected for all tests in this experimental program. Thickness of Sample, Backing Plate, and Mating Surface A minimum thickness is required for the PTFE sheets to accommodate wear and ensure a uniform bearing. Accord- ing to the LRFD Specifications, at least 0.0625-in. thickness is required for all PTFE sheets after compression. This value is at least 0.1875 in. for recessed PTFE sheets when the maximum dimension of the PTFE is equal to or less than 24.0 in. For stainless steel mating surfaces, the LRFD Specifications also recommend at least 16-gauge thickness when the maxi- mum dimension of the surface is less than or equal to 12 in. This thickness ensures no buckling or wrinkling. The LRFD Specifications also require that the mating surface for flat slid- ing surfaces be attached to a backing plate by welding in such a way that it remains flat and in full contact with its backing plate throughout its service life. The restrictions on the attachment of the mating surface are primarily intended to ensure that the surface is flat and retains uniform contact with the PTFE at all times without adversely affecting the friction of the surface or gouging or cutting the PTFE. In the studies for NCHRP Report 432, 0.25-in. thickness was used for dimpled PTFE, and 0.125-in. thickness was used for plain PTFE. A 0.06-in.-thick stainless steel plate, highly polished to a No. 8 mirror finish, was used as the mating surface. The most common thickness used for Fluorogold is 0.094 (3⁄32) in.; it is supplied from the factory already bonded to a backer plate. A thickness of 0.125 (1⁄8) in. is also available. The thickness of MSM varies between 5 and 8 mm (0.1969 and 0.3150 in.). Normally, an 8-mm (0.3150-in.) thickness is used. Plan of Tests Because of the more intense wear condition resulting from traffic loads, tests were carried out at high movement speeds to simulate conditions due to truck passage. The significant parameters affecting the wear of sliding sur- faces were selected based on previously undertaken research, most common industry practices, code requirements, and limitations of the available equipment. The stroke length was 1 in. That is, the stainless steel plate moved upward from its central position 1 in., and then it moved back down for 2 in. (i.e., it passed the initial central position by 1 in.), and then back up to the original position. The total movement per complete cycle was 4 in. The flat mating surface was 0.06-in.-thick Type 304 stain- less steel with a finish of 0.20 µm RMS. Steel plates were used as the backing plate material for all the experiments. The sliding surfaces were mechanically recessed in or chemically bonded to (or both) the backing material to ensure that no failure would occur in the sliding surface backing-material plane. The standard size of a 3-in. diameter was selected for all specimens. The thickness of all plain PTFE specimens was 0.25 in. The thickness of the MSM specimens was 0.315 in. (8 mm). A thickness of 0.094 (3⁄32) in. was used for the Fluorogold specimens. Table D.5 and Table D.6 summarize the testing program parameters. Attachment of PTFE to Backing Plate Typically, PTFE is recessed one-half of its thickness into the backing plate to prevent creep. In some cases, it is also chemi- cally bonded. When bonded, PTFE sheets are typically etched on one side to increase bonding capability. Table D.5 identifies the type of attachment between the sliding material specimens and the backing plates for this study. PTFE and MSM specimens were recessed and bonded. Fluorogold specimens were only chemically bonded as typi- cally provided by the manufacturer. The sliding surface specimens were oriented such that the direction of texture on the etched side was perpendicular to the direction of sliding. The stainless steel mating surfaces were aligned such that the direction of sliding was perpendicular to the direction of polishing (to the extent practicable). Design of Test Fixture The test setup used for the sliding surface wear tests is shown in Figure D.14. In this test setup, an MTS cyclic actuator was vertically installed in a large steel frame. The bearing test fix- ture was installed below and connected to the actuator. The sliding surface test fixture consisted of four main components: 1. A moving steel plate with a stainless steel plate mounted on both sides (sliding platform); 2. Two stationary sliding surface specimens connected to backing plates; 3. Two horizontal hydraulic jacks that applied contact pres- sure; and 4. An MTS actuator that provided vertical cyclic movement.

188 Table D.5. Experimental Program: Test Specimen Parameters Type of Specimen Total No. of Samples Sample Diameter (in.) Sample Thickness Mating Surface Backing Material to Sliding Surface and Stainless Steel Surfaces Connection to the Backing Plate 1 Flat unlubricated, unfilled PTFE (plain PTFE) 2 3 ¼ in. Stainless steel Type 304, 0.06-in. thickness, highly polished to No. 8 mirror finish Steel plate 1⁄8 in. Recessed and bonded 2 Dimpled unlubricated, unfilled PTFE 2 3 ¼ in. Stainless steel Type 304, 0.06-in. thickness, highly polished to No. 8 mirror finish Steel plate 1⁄8 in. Recessed and bonded 3 Dimpled unlubricated MSM 2 3 8 mm (5⁄16 in.) Stainless steel Type 304, 0.06-in. thickness, highly polished to No. 8 mirror finish Steel plate 5 mm Recessed and bonded 4 Unlubricated Fluorogold 2 3 3⁄32 in. Stainless Steel Type 304, 0.06-in. thickness, highly polished to No. 8 mirror finish Steel plate Bonded Table D.6. Experimental Program: Test Parameters Type of Specimen Displacement per Cycle (Amplitude) (in.) Contact Pressure (psi) Slip Rate (in./min) Cycling Frequency (Hz) PV (lb/in.2  ft/min) 1 Flat unlubricated, unfilled PTFE (plain PTFE) 2 3,000 25 0.10 6,250 2 Dimpled unlubricated, unfilled PTFE 2 3,000 25 0.10 6,250 1,000 2,080 1,500 3,120 2,000 4,160 3 Dimpled unlubricated MSM 2 3,000 (4,900 effective) 25 0.10 6,250 50 0.20 12,500 5,000 25 0.10 10,410 4 Unlubricated Fluorogold 2 3,000 25 0.10 6,250 50 0.20 12,500 1,000 50 0.20 4,160 5,000 25 0.10 10,410

189 The sliding platform was sandwiched between the two sta- tionary sliding surface specimens mounted on steel backing plates. The two horizontal hydraulic jacks applied horizontal pressure perpendicular to the surface of the sliding specimens to create the required contact pressure. Each backing plate was mounted on another larger plate (a mounting plate for the backing plate) that held the back- ing plate and specimen stationary. A reaction plate was placed on the outside of each hydrau- lic jack, and four Dywidag rods ran from the four corners of the west reaction plate through both specimen mounting plates and then to the east reaction plate. This arrangement resulted in an equal force on both backing plates. This entire system was supported on a bottom railing system. To run the test, the sliding platform, which was connected to the cyclic MTS actuator using T plates, moved upward and downward by the MTS cyclic actuator (see Figure D.15). The bottom rail consisted of two 2-in.-diameter stainless steel rods machined to slip fit the holes on two T assemblies. A camshaft bearing was press-fitted into the hole on each T assembly. The same type of hole was also prepared on the bottom of the two mounting plates through which the stain- less steel rod ran. Thus, the 2-in.-diameter stainless steel rods almost perfectly fit in the holes and at the same time, the plates slid easily on the rods (see Figure D.16 and Figure D.17). A neoprene pad was placed between the horizontal actua- tors’ piston and the mounting plate for the sliding surface. This pad, which is called a packing bearing, worked as a spring and retained the pressure exerted from the actuators to the plates in case the plates moved slightly, or in case the actuators leaked slightly and lost some pressure. Phase 3: Construction of Test Apparatus The frame in which the test fixture and the MTS actuators were installed was assembled on the laboratory strong floor (see Figure D.18). Two MTS actuators were installed with the potential of being able to run two test setups simultaneously; however, it was ultimately decided to use only one. Both the MTS actuators were mounted (posttensioned) on the top girder while the frame was lying on the strong floor. The MTS actuators were restrained against out-of-plane swiveling. Two precast concrete columns were leveled and then posttensioned to the strong floor. The entire frame was lifted afterward and tied to the columns (see Figure D.19). The frame was also posttensioned to the strong floor. * To serve as elastomeric springs Pre-Load Rods23 ft 11 ft Load frame 8" 6' 3" MTS Actuator Backing Plate to PTFE Stainless Steel Mating Surface Splice Plate Linkage Component to Actuator Test Specimen Sliding Platform Connected to Actuator Guide Rail T-Assembly for the Guide Rail Mounting Plate for the Backing Packing Bearings* Figure D.14. Sketch of loading frame and test fixture for wear tests in University of Nebraska–Lincoln structures lab.

190 Backing Plate to PTFE Stainless Steel Mating Surface Splice Plate Linkage Component to Actuator Test Specimen Sliding Platform Connected to Actuator Guide Rail T-Assembly for the Guide Rail Mounting Plate for the Backing Packing Bearings* * To serve as elastomeric springs Pre-Load Rods Figure D.15. Sketch of test fixture. PL 22" x 22" x 1.0" PL 31" x 12" x 1.0" PL 15" x 12" x 0.75" PL 9.5" x 12.5" x 1.5"PL 16" x 18" x 1.5" Packing Bearings* * To serve as elastomeric springs 2- Stainless Steel Shaft 2.12" PL 24" x 12" x 1.0" PL 22" x 10" x 0.75" PL 22" x 8" x 1" Pre-Load Rods PL 18" x 18" x 2.0" Figure D.16. Dimensions of components in test fixture.

191 3.0" 3.0" Ø 1316" Typ. PL 22" x 8" x 1" PL 22" x 10" x 0.75" 2.0" 9 " 2-Ø 2.1200" 2.75" 4 " Web of T-Assembly Flange of T-Assembly T-Assembly for the Guide Rail Guide Rail (Will be supplied) 22 " D D DD D: T-Assembly Mounting Holes C C C: Guide Rail Holes Sized for supplied camshaft bearings to be press-fit Shall be machined down for supplied camshaft bearings to be slip-fit 16.00" 2.00" 22 " Stainless Steel Shaft 2.12" L=18" PL 22" x 10" x 0.75" PL 22" x 8" x 1" 10 " 8 " To be fabricated by DS Brown To be fabricated by DS Brown A A A A A: Backing plate Mounting Holes B B B B B: Pre-load Rod Holes C C C: Guide Rail Holes Mounting Plate for the Backing E E: Mounting Holes E EE F F: Splice Holes F FF 3 " Typ. 4.0" Sized for supplied camshaft bearings to be press-fit 16.00" 4.5" 4.5" 3.0" 3.0" 3.0" 3.0" 2.0" Ø1316" Typ. PL 22" x 22" x 1.0" Backing Plate to Sliding Surface Specimen 22 " 22 " 1" 9.5 " 12.5 "1.5" PL 9" x 12" x 1.5" PL 24" x 12" x 1.0" Sliding Platform Connected to Actuator 24 " 12 "1.0" Stainless Steel Mating Surface 3" Diameter Recess PL 7" x 7" x 0.06" Ø 1 34" Typ.6.0" 3.0" 3.5" Ø 1316" Holes; W/ Ø 2 12" Chamfer; 34" Depth @ Center 2.5" 2.5" 3.00" Ø1316" Typ. 2-Ø 2.1200" Figure D.17. Sketches of components of test fixture. The bearing test fixture was then assembled on the bottom element of the frame (see Figure D.20). Figure D.21 and Figure D.22 show the assembled test fixture. Phase 4: Testing Plain PTFE specimens were tested first to establish a baseline against which to compare MSM and Fluorogold results. This was followed by MSM (dimpled), Fluorogold (plain), and dimpled PTFE. MSM and the second set of PTFE specimens were first tested for a prescribed number of cycles in a lubri- cated condition. Testing was then stopped, the specimens were cleaned, and the testing resumed in the unlubricated condition. All tests were conducted at room temperature. In all tests, the cyclic displacement was applied on a sine wave with a stroke length of 1 in. (total cycle length of 4 in.) at an initial sliding speed of 25 in./min (0.41667 in./s). The

192 Figure D.18. Test frame and MTS actuators being assembled on the strong floor. Precast Columns Figure D.19. Lifting operation of test frame with MTS actuators. Figure D.20. Assembly of test fixture on bottom element of frame.

193 cycle frequency was initially set to 0.1 Hz (25 in./min) for all tests. This frequency was maintained for PTFE until the entire thickness was worn down. However, because of the time limi- tation, the frequency was increased for MSM and Fluorogold specimens to 0.2 Hz (50 in./min) after about 116,000 cycles. The sliding surface samples were cycled under the pre- scribed pressure and cycle rate for various numbers of cycles, and the thickness change after various cyclic intervals was measured using a feeler gauge. The cycling intervals were 50, 100, 200, 250, 500, 1,000, 2,000, 4,000, and 8,000 at the rate of 0.1 Hz, and 16,000, which was applied at the rate of 0.2 Hz. The pressure on the samples was constantly recorded, moni- tored, and reset to the prescribed pressure if needed. The spacing between the stainless steel backing plates and the sliding-material backing plates was measured at four cor- ners, and the averages were used to determine the change in sliding material thickness. The tests were terminated when there was close to steel-on- steel contact for the plain PTFE samples (i.e., when the entire thickness of the testing samples was nearly worn away). How- ever, achieving this level of erosion required a much longer cyclic testing time for the high-performance sliding surfaces, and the tests had to be terminated well before the thicknesses of the samples were worn out. The following is the step-by-step sequence of the experi- ment for each sliding material test: 1. The two sliding material specimens, which were attached to steel backing plates, were connected to the mounting plates (see Figures D.15, D.16, and D.17), and the bolts were torqued. 2. The steel plate with the stainless steel mating sheets on each side was attached to the MTS actuator using splice plates, and the bolts were torqued. 3. The MTS actuator was moved vertically until the center- line of the stainless steel mating sheet was located on the centerline of the backing plates to the sliding material specimens. Figure D.21. Completed test fixture. Figure D.22. Completed test fixture in test frame.

194 4. The mounting plates for the sliding material specimens were pushed toward each other until the samples touched the stainless steel mating sheet. 5. The packing bearings (neoprene pads) were placed. 6. The horizontal jacks were placed on a wooden support bench on both sides of the fixture. 7. The Dywidag rods were placed through the holes, and the reaction plates were placed behind the jacks on the rods. The nuts were turned on the Dywidag rods but were not tightened. 8. The fixture was leveled and adjusted using the nuts on the Dywidag rods. The nuts were slightly tightened to prevent the system from moving. 9. The thicknesses of the samples were measured using a feeler gauge to measure the spaces between the backing plate to the sliding surface and the backing plate to the stainless steel. These measurements were carried out on four corners of each plate (eight measurements total). 10. The jacks were loaded until the required pressure on the sliding material was reached. This pressure was set by measuring the force in the jacks. For example, if 3,000-psi pressure was required on the sliding surfaces, the jacks were loaded to 21,000 lbs of force (knowing the samples are 3 in. in diameter, their area is about 7 in.2). 11. The thicknesses of the samples were measured again. 12. The MTS machine was set to the desired numbers of cycles (with the wave type and amplitude also set), and cycling was started. The MTS system continuously recorded the force required to cycle the actuators (10 measure- ments per second). 13. During the cycles, the force in the horizontal jacks was monitored and reset to the desired value if there was a drop in the force. 14. After the MTS actuator was stopped, the thicknesses were measured again. 15. Steps 10 to 14 were repeated. 16. After the test on a sliding material was finished, the force in the horizontal jacks was released, the steel backing plates with the test specimens were spread apart, and the backing plate with the stainless steel mating sheets was removed. Finally, the backing plates with the sliding material specimens were removed. The process returned to Step 1 for the next sliding material to be tested. Test Results COF, the amount of accumulated wear, and the wear rate were evaluated and compared in the experimental program. Coefficient of Friction COF is an important parameter for a sliding surface because it controls the level of restraint at the bearing and the amount of force transferred between the bridge superstructure and substructure. It can also affect the wear rate of the sliding surface material. PRoCeSS FoR meASuRiNg CoeFFiCieNt oF FRiCtioN COF was determined by measuring the force needed to move the center sliding plate with the stainless steel mating surfaces in between the stationary material test specimens. This force, which was recorded by the MTS machine during the tests, would typically include the following three components: 1. The net force needed to move the center steel plate with the stainless steel mating surfaces against the two station- ary material specimens (friction force); 2. The inertial force needed to keep the motion going; and 3. The weight of the attachments to the actuator. The measured force was set equal to zero at the beginning of the tests so the forces measured by the MTS machine were only the friction forces plus the inertial forces. Once the inertial force is known, the friction force can be determined by subtracting the inertial forces from the total measured forces. For the sine displacement shape and for a stroke length of 1 in. at a 25-in./min pace, it can be shown that the maxi- mum inertial force is about 0.43 times the mass that the MTS machine is moving, which is negligible compared with the actual forces that the machine is experiencing. Therefore, the inertial force was neglected, and the mea- sured force from the actuator was the actual friction force. The cross-sectional area of each test specimen was 7.07 in.2. With a surface pressure of 3,000 psi, this results in 21.2-kip force. COF can now be determined as follows: Friction force N= µ × where µ is COF, and N is equal to 21.2 kips. For the actual calculations, the real values of the perpen- dicular force that were recorded during the tests, and had minor variations, were used. meASuRed ReSultS Figure D.23 shows the variation in calculated COF for the three sliding surface materials (PTFE, Fluorogold, and MSM) against the number of cycles and accumulated travel distance for the baseline test parameters (P = 3,000 psi; V = 25 in./min). Static and dynamic COF values were experienced for all three materials. As shown in Figure D.23, the COF values experi- enced an initial spike (static value) that gradually decreased to a more stable value as the cycles increased (dynamic value). The spikes occurred when cyclic testing restarted after thick- ness measurements were taken.

195 Figure D.24 and Figure D.25, respectively, show the max- imum (static) and minimum (dynamic) values of COF for all three test materials for the baseline test parameters (P = 3,000 psi; V = 25 in./min). In each figure, the lowest curves represent plain PTFE, the middle curves represent Fluorogold, and the highest curves represent MSM. The static values were taken from the first cycle of each cyclic testing interval, and the dynamic values were taken from the last cycles of each cyclic testing interval. As expected, plain PTFE exhibited the lowest values, with static values ranging from about 7% to 8% and dynamic values ranging from about 5% to 6%. (The results from the second Figure D.23. Variation of COF for sliding surfaces versus the total numbers of cycles and total travel distance (P 5 3,000 psi; V 5 25 in./min). Figure D.24. Variation of static COF for sliding surfaces versus number of cycles (P 5 3,000 psi; V 5 25 in./min).

196 set of PTFE specimens, which are not shown in Figures D.23 through D.25, showed lower COF values, of about 3% to 3.5% at this pressure and speed [see Figure D.28].) Fluorogold showed relatively good performance and had static values somewhat comparable with the first PTFE specimen tests, with values typically ranging from about 7% to 9%, and had dynamic values in the 6% to 7% range. These tests showed that Fluorogold had a COF only slightly higher than that of plain PTFE. Unlubricated MSM, however, exhibited much higher COFs, with wide fluctuation in static values ranging from 10% to 17%. Dynamic values were also high but more stable, ranging from about 9% to 10%. Overall, the values found for MSM in this study were nearly twice the values found for plain PTFE at the initial pressures (3,000 psi) and testing speed. (It should be noted that the literature for MSM indicates that COF is lower for higher contact pressures, and it has greater performance than plain PTFE in low tempera- tures.) MSM tested in this study at a higher contact pressure (5,000 psi) exhibited a lower COF. See the section titled “Effect of Contact Pressure Variation.” Test results for COF in this study were somewhat compa- rable with values reported in NCHRP Report 432, Appen- dix C. At 3,000-psi pressure, 25-in./min sliding speed, and room temperature, NCHRP Report 432 showed plain PTFE to have an initial spike of about 11%, which dropped to about 4.5% to 5% with subsequent cycles, but then gradu- ally increased to about 8% with a greater number of cycles. Glass-filled samples had initial spikes up to the 14% range, but subsequent values dropped to the 5% to 7% range; these values were comparable with the ones obtained in this study. The NCHRP Report 432 tests showed that higher sliding speeds (i.e., going from 2.5 to 25 in./min) had somewhat higher COFs, given the same contact pressure and tempera- ture (3,000 psi at room temperature). Figure D.26 shows the LRFD Specifications recommended values for COF. Fluorogold test results were comparable with glass-filled recommended values, but plain PTFE tests in this study showed higher values. However, results from this study were comparable with NCHRP Report 432 results when con- sidering sliding speeds of 25 in./min. Figure D.25. Variation of dynamic COF for sliding surfaces versus number of cycles (P 5 3,000 psi; V 5 25 in./min). Source: AASHTO (2010b). Figure D.26. AASHTO-recommended design COFs based on 0.2-µm-finish mating surface.

197 Lubricated Specimen reSuLtS The MSM specimens and the second set of PTFE specimens, which were both dimpled, were initially tested using silicone grease lubrication that was shipped together with the test specimens from the manufacturer. The samples were cycled for 50- and 150-cycle intervals. The baseline test parameters (P = 3,000 psi; V = 25 in./min) were also used for these tests. This was done to determine the lubricated COF of the MSM sliding surfaces, which would correspond to the way this material was tested previously by Maurer and the way this material has typically been used in practice. As shown in Figure D.27, the lubricated COFs for the MSM and PTFE materials were significantly lower than the unlubricated cases (see Figure D.23). The COF of each material seemed to behave oppositely from one another in the early cycles, with PTFE having a low value at the start that rose sharply before stabilizing to a more consis- tent value. MSM started with a high value and dropped sharply before stabilizing to a more consistent value. Ultimately, the lubricated PTFE COF was comparable for MSM and PTFE, with PTFE having only slightly lower values than MSM. Results of lubricated PTFE tests from this study were comparable to (a) (b) Figure D.27. Coefficients of friction for dimpled, lubricated (a) PTFE and (b) MSM sliding surfaces (P 5 3,000 psi; V 5 25 in./min).

198 results in NCHRP Report 432, which reported COFs of about 1% for lubricated PTFE at this pressure and sliding speed. effect of contact preSSure Variation In the second set of PTFE samples (dimpled, unlubricated), the contact pressure was changed at various intervals, and COFs were recorded. Figure D.28 shows the comparison of COFs of these samples for various contact pressures (V = 25 in./min). The sequence of applied pressures was 3,000, 1,000, 1,500, and 2,000 psi. As shown in Figure D.28, there was little difference in COF when the pressure was reduced from the initial 3,000-psi pres- sure to 1,000-psi pressure. The coefficient increased, however, with progressive increases in pressure from 1,000 to 1,500 psi and ultimately to 2,000 psi. Previous research showed that higher pressures resulted in lower COFs. The results shown in Figure D.28, however, show a somewhat opposite effect. This may have been due to continued use of the same samples with varying pressures at various stages of testing, instead of using new samples for each new pressure test. In this research, the condition of the sliding surfaces was likely very different at the start of each new phase because of continuing wear and deg- radation of the material surface. Previous research has also shown that at a 25-in./min speed, COF will increase with increasing cycles because of material surface wear. Therefore, it is likely that the behavior shown in Figure D.28 was influ- enced by continuing surface wear of the sliding material. Figure D.29 and Figure D.30 show COFs at a contact pres- sure of 5,000 psi for unlubricated MSM and Fluorogold sam- ples, respectively. At this higher pressure, MSM experienced a significant reduction in dynamic COF from about 9% to 10% to about 7% to 7.5%. This reduction is consistent with manu- facturer data that indicates this trend and recommends use of this material at higher pressures. Fluorogold experienced an increase in COF from about 6.5% to 9% at this higher pres- sure. It should be noted, however, that Fluorogold literature recommends limiting the contact pressure to 2,500 psi, which results in a lower COF. Thickness Reduction The average thicknesses of the specimens were measured using a caliper at three locations on the periphery of each specimen before mounting it in the test fixture. The average thickness change for the specimens was also determined at various stages of the cyclic testing using a feeler gauge at four corners between the sliding-surface backing plates and the stainless steel backing plate. The feeler-gauge measurements were carried out at various cycling intervals, which were more frequent at the beginning of the tests and less frequent toward the end. Figure D.31 shows the percentage thickness reductions of the various samples with respect to the travel distance and number of cycles. The circled numbers depicted on the graph and desig- nated in the text as PV1 through PV4 represent the various combinations of pressure and testing velocity (PV zones) that were evaluated. In Figure D.31, the bottom lines (black and green) are for PTFE, the middle lines (blue) are for MSM, and the top lines (red) are for Fluorogold. The two lines shown for MSM and Figure D.28. Comparison of COFs of dimpled PTFE samples for various contact pressures (V 5 25 in./min).

199 Figure D.29. COF of MSM samples at high pressure of 5,000 psi (V 5 25 in./min). Figure D.30. COF of Fluorogold samples at high pressure of 5,000 psi (V 5 25 in./min). Fluorogold represent the values for the two simultaneously tested specimens. Initially (PV1), all materials were tested at a 25.0-in./min travel speed and 3,000-psi contact pressure. As expected, the plain PTFE specimens exhibited severe wear at this initial PV combination, and were substantially worn with 80% reduction after about 1.5 mi of accumulated travel. The rate of wear was initially higher, but became more constant after about half the travel length. The unlubricated MSM specimens performed significantly better than PTFE, but they had about a 10% initial thickness reduction at the start of testing, possibly due to material com- pression. Thereafter, the specimens exhibited another 10% thickness reduction within the first 3 mi of accumulated travel. The material then became somewhat stabilized with little or no additional thickness reduction throughout the remainder of the PV zone, which was carried out to about 8 mi of accumulated

200 travel. In PV2, the travel speed was increased to 50 in./min to accelerate testing. Some additional thickness reduction was observed in this PV zone, but the material again stabilized with no additional loss after about 2 mi of travel. In PV3 the pressure was increased to 5,000 psi and the speed was reduced to the original 25 in./min. The material experienced some immediate compression under the added pressure and then exhibited about 5% of additional uniform wear until the test was stopped after about 11 mi of additional travel. The over- all MSM test was carried out to about 24 mi of accumulated travel, and the overall average thickness reduction due to combined compression and wear at the end of testing was about 32%. About one-half the thickness reduction may have been due to material compression. The Fluorogold specimens performed exceptionally well, with no measured thickness loss within the first zone of test- ing (about 8 mi of travel). In PV2, the travel velocity was increased to 50 in./min, and the material showed a consistent rate of wear that resulted in about 30% thickness reduction over about 28 mi of travel. In PV4, the pressure was reduced to 1,000 psi with the same 50-in./min velocity, and no additional thickness reduction was observed. Finally, in PV3, with 5,000-psi pressure and 25-in./min velocity, the material again started to show wear, with about 3% loss over about an additional 2 mi of travel before the testing was terminated. The material ultimately had about 34% reduction in thickness at the end of testing, which covered about 47 mi of accumulated travel. Figure D.32 shows the percentage thickness reductions for the two separate PTFE tests on a larger horizontal-axis scale. Again, the circled numbers represent various PV zones. The first PTFE test was conducted at a constant pressure and veloc- ity, PV1, which was similar to the initial PV for the MSM and Fluorogold tests, and resulted in significant consistent wear. In the second PTFE test, the pressure and velocity were modified to evaluate wear under various conditions, and the two speci- mens exhibited some inconsistent behavior. In PV2 with only 1,000-psi pressure, both specimens exhibited only minor wear. In PV3 with 1,500 psi, one of the specimens showed no wear, but the other exhibited very heavy wear. In PV4 with 2,000 psi and 50-in./min speed, both specimens again showed consistent heavy wear. The following conclusions can be derived from the test results shown in Figure D.31 and Figure D.32: • The wear behavior of plain PTFE and Fluorogold (a rein- forced PTFE) is similar in that both materials experience very consistent rates of wear depending on the combination of contact pressure and velocity (PV). Fluorogold exhibited significantly greater wear resistance than plain PTFE, but under certain PV levels, both materials exhibited very con- sistent, albeit different, wear rates. However, MSM, which is a completely different type of material, behaved differently. It had significantly greater wear resistance than plain PTFE, but it also exhibited a certain amount of compressibility, which in some cases resulted in nonuniform thickness loss or wear. A faster thickness reduction was observed at the beginning of the cycles in each test zone. Figure D.31. Thickness reductions for sliding surfaces versus travel distance and numbers of cycles.

201 • For PTFE and Fluorogold, there were combinations of con- tact pressure and velocity in which the wear rate was virtu- ally zero. This behavior was not observed for the MSM material, possibly because of material compressibility. Even though there may have been no real material wear, thick- ness reduction was still observed. • Regardless of material wear behavior, the study showed that high-performance sliding materials (Fluorogold and MSM) demonstrated significantly lower wear rates than plain PTFE, which is the most used material in the United States. • The thickness–travel distance curves for PTFE-based materi- als (plain PTFE and Fluorogold) were almost linear for all samples at all levels of contact pressure and velocity, whereas the MSM curves were not linear. • Fluorogold did not experience creep. The tests showed that Fluorogold has a slight increase in COF, but it does not have creep and has very good wear characteristics. To establish more reliable wear rates for the samples used, additional testing with various contact pressures and velocity combinations are required. For each combination, a new slid- ing sample should be used. In addition, the numbers of data points (total cumulative travel distance) should be adequate to represent the real behavior of the material being tested. In the current study, the experiment on Fluorogold was termi- nated after only a few cycles in PV Zone 3 (see Figure D.31); as a result, material wear characteristics may not be totally representative for that PV combination. Figure D.33 shows the stainless steel mating surface and the sliding surface specimens before and after cyclic testing. The PTFE showed considerable wear and loss of thickness in the form of flaking and shedding of the material. The MSM specimens showed signs of rough surface abrasion, but the dimples were still somewhat evident. The Fluorogold speci- mens showed some surface abrasion, but they did not exhibit the type of material deterioration and shredding that was found with plain PTFE. Wear Rate meaSured Wear rateS Previous research has shown a wide range in wear rates for PTFE sliding surfaces depending on material type, lubrication, sliding speed, contact pressure, and temperature (Stanton et al. 1999). In this study, at the initial high pressure and fast sliding speed, plain PTFE was shown to have a very high wear rate that was consistent with previous studies. High-performance slid- ing materials such as MSM and Fluorogold showed signifi- cantly lower rates of wear. Wear rates for the various materials tested in this study were developed from the measured thick- ness reductions that occurred over the accumulated lengths of travel by determining the slopes of the thickness reduction curves, as shown in Figure D.31. A moving-window averaging technique was used to determine the wear rate based on the thickness changes for various stages of the experiment. Plain PTFE and Fluorogold (filled PTFE) exhibited rather constant rates of wear that varied for different combinations of pressure and velocity. The wear rate for Fluorogold was very close to zero in the initial PV zone, but exhibited a con- stant wear rate in PV zones with higher sliding speeds or higher pressures. The wear rate for MSM samples exhibited a different trend than that of PTFE. MSM samples did not exhibit constant wear rates within PV zones, but appeared to lose thickness more rapidly at the beginning of each zone. The rates tended to slow or stabilize as the tests proceeded. Based on discussions Figure D.32. PTFE thickness reduction versus travel distance.

202 with MSM manufacturer representatives, it was concluded that MSM experiences some thickness reduction from compression and creep while not necessarily losing thickness due to wear. Because of this, it is difficult to determine consistent wear rates for MSM, and PV may not be a suitable wear characteristic, as it is for PTFE-based materials. Table D.7 summarizes the wear rate values that were deter- mined in this study for all samples under various contact pres- sures and travel speeds. Table D.8 shows wear results from NCHRP Report 432 (Stanton et al. 1999) for various PTFE-based sliding materials based on constant pressure, but with variable sliding speed, temperature, and lubrication. Each of these factors was shown to have considerable effect on wear rate. These earlier results confirmed the significant increase in wear due to high slid- ing speeds and low temperatures, and also confirmed that braided and glass-filled PTFE materials showed greater wear resistance. WeAR RAte FoR PRediCtiNg SeRViCe liFe Campbell and Kong (1987) studied the relationship between contact pressure (P) and sliding speed or velocity (V) on the wear of PTFE sliding surfaces. Figure D.34 shows curves devel- oped from tests for sliding speed, pressure, and PV. From this, they developed a general mathematical model for predicting the wear of plain PTFE sliding surfaces as a function of the PV factor, as shown in Equation D.13: (D.13)h KPVt= where h = reduction in thickness (in.); K = wear factor (in.3-min/lb-ft-h); P = contact pressure (psi); V = sliding velocity (ft/min); and t = total time under load (h). Their research also indicated that two distinct values exist for the wear factors: one that is associated with mild wear regimes Stainless steel PTFE MSM Fluorogold Figure D.33. Stainless steel mating surface and sliding surface specimens (left) before and (right) after testing.

203 Table D.8. PTFE Wear Rates with Constant 3,000-psi Pressure Material Lubrication V (in./min) T (F) PV (lb/in.2 ft/min) Wear Rate (mil/mi) Unfilled PTFE Dimpled lubricated 2.5 68 625 0.3 25 68 6,250 0.5 Flat unlubricated 2.5 68 625 0.7 25 68 6,250 189 2.5 -13 625 10 25 -13 6,250 259 Woven PTFE Flat unlubricated 2.5 68 625 0.3 25 68 6,250 17 2.5 -13 625 27 25 -13 6,250 24 15% Glass filled Flat unlubricated 2.5 68 625 -1 25 68 6,250 -0.5 2.5 -13 625 no result 25 -13 6,250 6 25% Glass filled Flat unlubricated 2.5 68 625 -0.3 25 68 6,250 2 2.5 -13 625 4 25 -13 6,250 46 Source: Stanton et al. (1999). Table D.7. Calculated Average Wear Rates Based on Slope of Thickness Reduction Graphs Sample Effective Area (in.2) P (lb/in.2) P Effective (lb/in.2) V (in./min) PV (lb/in.2 ft/min) PV Effective (lb/in.2 ft/min) Wear Rate Sample 1 (mil/mi) Wear Rate Sample 2 (mil/mi) Average Wear Rate (mil/mi) Unlubricated PTFE 7.07 3,000 3,000 25 6,250 6,250 81.21 80.32 80.78 Dimpled unlubricated PTFE 4.19 1,000 1,600 25 2,083 3,333 9.89 2.98 6.43 1,500 2,500 3,125 5,208 110.35 18.48 64.42 2,000 3,300 4,167 6,875 152.32 151.05 151.68 3,000 5,000 6,250 10,417 121.18 49.18 85.18 MSM 4.19 3,000 5,000 25 6,250 10,417 0.88 1.17 1.01 5,000 8,400 10,417 17,500 0.75 0.68 0.71 3,000 5,000 50 12,500 20,833 0.6 0.32 0.48 Fluorogold 7.07 3,000 3,000 25 6,250 6,250 0.06 0.03 0.04 5,000 5,000 10,417 10,417 0.89 1.68 1.29 3,000 3,000 50 12,500 12,500 0.95 0.92 0.94 1,000 1,000 50 4,167 4,167 0.01 0.12 0.07

204 (low K), and one associated with severe wear regimes (high K). There was an abrupt transition between the two regimes at a so-called PV limit. The results of the SHRP 2 study combined with the results in NCHRP Report 432 also indicated the potential of service life prediction based on the PV factor and further confirmed the potential of the PV limit transition between zones of low wear and severe wear. Figure D.35 shows the wear rate versus PV data for plain PTFE tests performed in this study combined with NCHRP Report 432 tests. From this plot, there appears to be a PV limit beyond which the wear rate increases significantly. There fur- ther appears to be the potential for a service life design curve for wear rate, which would be a function of PV as shown on the figure. It is recognized that the results shown are from very limited test data and that further testing would be required to establish more reliable curves and a true PV limit. As with plain PTFE, the Fluorogold (glass-reinforced PTFE) test results also indicated the potential of the PV factor as a means of predicting service life. Figure D.36 shows test results for wear rate versus PV factor over the range of PV zones tested. Test results from NCHRP Report 432 for glass-filled samples are also plotted. Similar behavior to plain PTFE is shown in that there appear to be a PV limit differentiating zones of higher wear and very low wear, albeit the wear rates are significantly lower than plain PTFE. As with plain PTFE, further testing is needed to establish a more reliable curve and the real PV limit. ProPosed Wear rate Methodology In addition to the PV factor, two other factors can contribute to predicting overall wear rate for a PTFE-based sliding material: temperature and lubrication. Considering all these factors, Equation D.14 can be consid- ered as a general equation for estimating the wear rate for a sliding surface as a function of material type and other contrib- uting parameters: wear rate base wear rate (D.14)CT CL= × × Source: Campbell and Kong (1987). (a) (b) (c) Figure D.34. Variation of wear rate with (a) sliding speed, (b) pressure, and (c) PV. Figure D.35. Variation of wear rate with PV for plain PTFE. Figure D.36. Variation of wear rate with PV for Fluorogold sliding material.

205 where wear rate = total thickness reduction (mil) per mile of travel distance; base wear rate = defined as a function of material type, con- tact pressure, and velocity from PV curves based on experimental tests for plain PTFE, glass-filled PTFE, or braided PTFE; CT = modification factor for the effects of low temperature (function of material type); CL = modification factors for the effects of lubri- cation (function of material type); P = contact pressure acting vertical to the slid- ing surface; and V = travel speed of the sliding bearing. The base wear rate defined in this procedure is the wear rate determined from tests conducted at various combinations of speed and contact pressure at room temperature, without lubrication. The effects of low temperature and lubrication can be included by the factors CT and CL, respectively, which are added to the equation for the sake of being complete. In low- temperature regions, CT would be greater than one, which results in greater wear. With lubrication, CL would be consid- erably less than one, which results in less wear. These factors are also a function of material type and must be determined from tests. At this time, there are insufficient data to develop these factors accurately for final service life design, but esti- mates can be drawn from Table D.8. The travel speed of the sliding bearing can be determined from the methods described in the subsection “Speed of Bearing Movement for Truck Load.” Once the wear rate is known, the total amount of wear can be determined over the design life of the bearing from the expected cumulative amount of bearing travel as described in “Phase 1: Determination of Bridge Movements.” SlidiNg SuRFACe deSigN PRoCeduRe FoR SeRViCe liFe The following steps summarize the proposed procedure for design of sliding surfaces for service life. The procedure involves first determining demand requirements, which are based on bridge loads, traffic, and temperature data. Supply require- ments are then determined based on selected material proper- ties for the proposed sliding surface. The first major part of the proposed procedure for design of sliding surfaces, determining demand requirements, involves three steps: Step D1. Calculate pressure (P; an unfactored average con- tact stress measured in pounds per square inch) on trial sliding surface due to all loads and due to permanent loads, based on the LRFD Specifications design criteria. Maximum allowable contact stress will vary based on the type of sliding surface material. Step D2. Calculate sliding movement velocity (V; measured in inches per minute) of the sliding surface for vertical truck load, vibration, and thermal movement. Step D3. Calculate total accumulated movement per year (M) computed for truck passage and vibration based on ADTT and computed for thermal movement based on daily and yearly thermal variations. The second major part of the proposed procedure, deter- mining supply requirements, involves three main steps: Step S1. Calculate PV factors and determine base wear rates from PV curves for the sliding material under consideration (Figure D.35 for plain PTFE). Determine wear rate as a function of base wear rate by using Equation D.14, with factors for tem- perature zone effect (CT) and use of lubrication (CL). At this time, there are insufficient data to develop these factors accu- rately for final service life design. Until further tests are con- ducted, these factors are assumed to be equal to one. Step S2. Determine wear in mils due to 1 year of accumu- lated sliding movement based on truck passage and vibration and for thermal movement, as follows: wear (mils/year) = [wear rate (mils/mi) × accumulated sliding movement (mi)]. Step S3. Considering the standard thicknesses available for the sliding material under consideration, determine the ser- vice life of the sliding surface material for a trial thickness and compare it with the design service life of the bridge system. If the service life of the sliding surface is less than the design service life of the bridge system, there are three options: a. Consider the required replacement schedule; b. Consider increased thickness of the sliding surface material, or increased area of the sliding surface, which reduces contact pressure and reduces wear rate; or c. Consider higher-performing sliding material with greater wear resistance. Depending on the options chosen, various demand and supply steps will be repeated with new parameter values to arrive at a final solution. Summary and Conclusions Many bearing types currently use PTFE as a sliding material to accommodate horizontal bridge movements. Previous research investigating the behavior of sliding materials for bearings has shown that plain PTFE can experience consider- able wear, particularly when subjected to high rates of horizon- tal movement. This type of movement, which is due mostly to girder end rotations associated with cyclic truck loads, is low- amplitude, high-cycle movement with fast movement speed.

206 Movement due to thermal load, however, is high-amplitude, low-cycle movement with slow movement speed, and is less likely to cause wear. Fast sliding speed combined with high contact pressure and low temperatures all contribute to PTFE wear. Currently there are minimal data to confirm actual bearing movements due to truck load or test data to determine a life prediction model for sliding surfaces that use PTFE. This research program, as part of SHRP 2 Project R19A, studied the feasibility of achieving increased service life of sliding bearings that are subject to high sliding speeds through the use of alternative high-performing materials that have greater wear resistance than conventional plain PTFE. Two potential high-performance sliding materials were studied: (1) MSM, an ultrahigh-molecular-weight polyethylene devel- oped in Germany; and (2) Fluorogold, a special glass fiber– reinforced PTFE. Plain PTFE was also studied to use as a benchmark for high-performance material comparison. The research program conducted both analytical studies and testing. The analytical studies first investigated potential bearing movements and movement speeds due to truck load and thermal load. A proof of concept testing program studied relative performance characteristics between plain PTFE, MSM, and Fluorogold, specifically with regard to COF and wear under similar conditions. The test program evaluated material wear associated with combined high movement speed and high contact pressure because it would serve as an upper bound on the main parameters affecting wear. Analytical Study The analytical studies determined potential bearing movement speeds under truck loading and thermal loading and evaluated potential accumulated sliding movement over a bearing design life. For simplicity, the analytical studies considered a 200-ft simple-span, multisteel girder prototype bridge with one end fixed and one end expanded. Girder deflections and end bear- ing movements were computed using both a 3-D FE model and an approximate single line-girder analysis with a maximum girder deflection of L/800. The study included a dynamic analy- sis that investigated the potential bearing movement due to free vibration and damping of the span after passage of a truck, which greatly increases the total accumulated movement. The following conclusions were determined from the ana- lytical studies: 1. The FE analysis calculated the horizontal one-direction slid- ing for a single truck load at the expansion bearing to be about 0.15 in. Considering LL + I, this amount was 0.20 in., which resulted in a cycle of 0.40 in. per truck passage consid- ering rebound to the original position after truck passage. 2. The approximate method, which considered a single line girder and assumed a maximum girder deflection of L/800 under truck LL + I, determined the maximum one-direction horizontal movement to be approximately 0.50 in., with a total cycle of about 1.0 in. per truck passage. This value was considered an upper bound. 3. The considerably lower value obtained from the FE analy- sis was attributed to the presence of the other girders in the bridge system, which increased the bending stiffness of the external girder. 4. The dynamic analysis showed that the total movement, which includes the amount of movement due to truck passage and following free vibration, would be approxi- mately 7.4 times the maximum one direction of move- ment due to truck passage for a 10% damping ratio. 5. Considering the FE analysis, with a maximum LL + I one- direction movement of 0.15 in., the accumulated total expansion end movement due to truck load with vibra- tion (10% damping ratio) over a 100-year service life with an ADTT of 1,000 would be approximately 800 mi. 6. Without considering additional movement due to free vibration, FE analysis showed that the accumulated total expansion end movement due to truck load for only a single cycle of movement per each truck passage would be approximately 200 mi over the same service life and truck volume as Number 5 above. 7. For the prototype bridge, the total accumulated move- ment due to daily and annual thermal variation was only about 0.8 mi over a 100-year service life. 8. Computed bearing sliding speeds for the initial full cycle of movement were based on the time for a truck to pass the bridge, which varies depending on truck speed. For the pro- totype bridge with movements computed from FE analysis, the expansion bearing slip rates were computed to be about 7 in./min for a 40 mph truck speed and about 10 in./min for a 60 mph truck speed. Using movements computed from the approximate analysis, the corresponding slip rates were about 18 and 28 in./min for 40 and 60 mph truck speeds, respectively. (These slip rates were considered to be high- end limits.) Little field data are available to confirm slip rates for actual sliding bearings in service. 9. The sliding speed due to free vibration is independent of the initial truck speed and depends rather on the dynamic characteristics of the bridge (stiffness, mass, and damping). For the prototype bridge, the computed bearing sliding speed during the period of free vibration was 12.1 in./min. Test Program A proof of concept test program was conducted at the Uni- versity of Nebraska–Lincoln structures laboratory. Sliding

207 material specimen sizes (3-in. diameter), initial testing speed (25 in./min), and initial contact pressure (3,000 psi) were established to be consistent with tests reported in NCHRP Report 432. The test setup used for the sliding surface wear tests used an MTS cyclic actuator that was vertically installed in a large steel frame. A specially designed sliding material test fixture was installed below and connected to the actuator. The test fixture included a center moving plate (attached to the MTS actuator) with stainless steel surfaces attached to each side. This moving plate was sandwiched between two station- ary material test specimens mounted on steel backing plates. Two horizontal hydraulic jacks applied horizontal pressure perpendicular to the surface of the sliding specimens to create the required contact pressure. In all tests, the cyclic displacement was applied on a sine wave with a stroke length of 1 in. In other words, the center plate with stainless steel surfaces was moved upward from its initial central position 1 in. It was then moved back down for 2 in. (past the initial central position by 1 in.), and then back up 1 in. to the original position. The total movement per complete cycle was 4 in. Plain PTFE specimens were tested first to establish a base- line with which to compare the MSM and Fluorogold results. This was followed by MSM (dimpled), Fluorogold (plain), and dimpled PTFE. MSM specimens and the dimpled set of PTFE specimens were first tested for a prescribed number of cycles in a lubricated condition. Testing was then stopped, the specimens were cleaned, and testing was resumed in the unlubricated condition. All tests were conducted at room temperature. The cycle frequency was initially set to 0.1 Hz (25 in./min) for all tests and was maintained for initial PTFE testing until the entire thickness was worn down. However, because of time limitations, this frequency was increased to 0.2 Hz (50 in./min) after about 116,000 cycles for the MSM and Fluorogold specimens. Based on the results of this limited experimental investiga- tion, the following conclusions were drawn: 1. Tests confirmed that both MSM and Fluorogold can pro- vide considerably greater wear resistance than conven- tional plain PTFE and can be used to increase service life when sliding surfaces are subject to high movement speed and high contact pressure. 2. Plain PTFE was shown to wear at a very high consistent rate under the initial combination of high sliding speed and contact pressure, resulting in 80% thickness loss within less than 2 mi of accumulated sliding length. 3. Fluorogold exhibited the best wear resistance of all materials tested, with no material thickness loss over about 8 mi of accumulated sliding length within the initial high values of pressure and velocity. It started to show some wear at a much higher travel speed (50 in./min). 4. Plain PTFE and Fluorogold (a PTFE-based material) both exhibited rather constant rates of wear that varied for different combinations of pressure and velocity. The wear rate for Fluorogold was very close to zero in the initial PV zone, which exhibited a constant wear rate (albeit very low) in higher PV zones. Plain PTFE exhibited consistent wear behavior and showed very low wear in a relatively low PV zone (about one third of the initial PV). This finding indicated that a correlation occurred between wear rate and PV factor for PTFE- based materials. 5. MSM exhibited some initial thickness reduction of about 10% (described in the next item), but showed only about 10% additional loss over 13 mi (20.9 km) of accumulated travel. 6. MSM exhibited different behavior with respect to thick- ness reduction than plain PTFE or Fluorogold. MSM samples did not exhibit constant wear rates within PV zones, but appeared to exhibit initial rapid thickness reduction at the beginning of each zone. The wear rates tended to slow or stabilize as the tests proceeded. The ini- tial thickness reduction was attributed to compressive deformation and creep, and was not necessarily due to wear. This behavior made determining consistent wear rates for MSM difficult. Because of inconsistent wear rates, it was concluded that the PV factor may not be a suitable wear characteristic for MSM, as it is with PTFE- based materials. 7. The results of the SHRP 2 study combined with NCHRP Report 432 results indicated the potential of service life prediction based on the PV factor and further confirmed the potential of the PV limit transition between zones of low wear and severe wear. It was concluded that the PV factor could be used in a service life design method for PTFE-based sliding materials based on wear over an accu- mulated length of travel. 8. Fluorogold exhibited a COF of about 2% higher than plain PTFE at the initial contact pressure of 3,000 psi. Unlubricated MSM, however, had a COF that was consid- erably higher. The COF of MSM was found to reduce con- siderably with increased contact pressure. At 5,000-psi contact pressure, the MSM COF was comparable with the Fluorogold COF at 3,000 psi. Lubricated MSM had a rela- tively low COF that was comparable with lubricated PTFE. 9. Based on the limited proof of concept testing, Fluorogold (a glass-reinforced PTFE sliding material) exhibited the best overall high performance with a combination of high wear resistance and relatively low unlubricated COF.

208 Recommendations for Future Work Only proof of concept testing was performed in this study. To establish more reliable wear rates for the materials sampled, additional testing with various contact pressures and velocity combinations is required. From this, a more statistically reli- able model of PV versus wear rate for various types of PTFE- based sliding materials could be developed for actual service life design. Test data developed in this study combined with those reported in NCHRP Report 432 are adequate for con- firming a trend, but they provide a limited basis for develop- ing reliable design curves. The effects of temperature were not considered in this study, but they were evaluated in the NCHRP Report 432 study. Fur- ther testing is required to properly evaluate temperature as part of the wear rate model for service life design along with pres- sure and velocity. Field testing is needed to determine actual bridge move- ments and movement speeds at sliding expansion bearings for different types of girders (steel and concrete) under truck load and thermal load. There are little current data to sub- stantiate these movements and movement speeds, and the results of analytical studies need to be validated against actual conditions.