MORGAN BRIDGE HAERVT-33 (Upper Bridge) VT-33 Spanning Nortli Brancli Lamoille River on IVlorgan Bridge Road Belvidere Junction Lamoille County Vermont PHOTOGRAPHS WRITTEN HISTORICAL AND DESCRIPTIVE DATA REDUCED COPIES OF MEASURED DRAWINGS FIELD RECORDS HISTORIC AMERICAN ENGINEERING RECORD National Park Service U.S. Department of the Interior 1849 C Street NW Washington, DC 20240-0001 HISTORIC AMERICAN ENGINEERING RECORD INDEX TO PHOTOGRAPHS MORGAN BRIDGE HAER VT-33 (Upper Bridge) Spanning North Brmich Lamoille River on Morgmi Bridge Road Belvidere Junction Lamoille County Vermont INDEX TO BLACK AND WHITE PHOTOGRAPHS Jet Lowe, photographer, June 2003 VT-33-1 WEST PORTAL. VT-33-2 PERSPECTIVE FROM SOUTHWEST. VT-33-3 PERSPECTIVE OF EAST PORTAL. VT-33-4 EAST PORTAL. VT-33-5 INTERIOR, VIEWED FROM EAST. VT-33-6 INTERIOR, EAST PORTAL. VT-33-7 NORTH PANEL FROM SOUTHWEST. VT-33-8 WINDOW DETAIL, LOOKING NORTH. VT-33-9 BELOW DECK VIEW OF BRACING SYSTEM. VT-33-10 SOUTH ELEVATION. VT-33-11 NORTH ELEVATION. HISTORIC AMERICAN ENGINEERING RECORD MORGAN BRIDGE HAERNo.VT-33 LOCATION: DATE OF CONSTRUCTION: BUILDER: PRESENT OWNER: PRESENT USE: SIGNFICANCE: AUTHORS: PROJECT INFORMATION: Morgan Bridge Road spanning the North Branch of the Lamoille River, Belvidere, Lamoille County, Vermont UTM: 18.0679866E.4956962N, Johnson, VTQuad. Possibly 1886, rebuilt 1898-99 Attributed to Lewis Robinson, Charles Leonard, and Fred Tracy Town of Belvidere Vehicular bridge A good example of the once common queenpost, Morgan Bridge is most remarkable in that it is one of the few that still functions as a timber load-bearing truss bridge. One of about sixteen queenposts remaining in Vermont, it was entered in the National Register of Historic Places in 1974. Dr. Mark M. Brown, August 2003 Megan Reese, Engineering Technician, and Dario A. Gasparini, Ph.D., Professor of Civil Engineering, Case Western Reserve University, Summer 2003 The National Covered Bridges Recording Project is part of the Historic American Engineering Record (HAER), a long-range program to document historically significant engineering and industrial works in the United States. HAER is administered by the Historic American Buildings Survey/Historic American Engineering Record, a division of the National Park Service, U.S. Department of the Interior. The Federal Highway Administration funded the project. MORGAN BRIDGE HAERVT-33 (Page 2) Description The Morgan Bridge is a subdivided queenpost covered bridge with an overall bottom chord length of 62'-l 1/4", clear span of 54'-10", and a road deck width of ll'-8 5/8" supported by concrete abutments. It connects a back road to State Road 109 along a north northwest-south southwest axis and in doing so rises I'-IO" as it crosses the North Branch of the Lamoille River (sometimes known as Kelley, or Kelly, River). The bottom chords are three 4" x 11" timbers with bolted shear blocks spaced an average of every 4'. Timbers 7-1/2" x 10" x 23'-6" to 24' brace the 9-1/2" x 13-1/2" x 15" queenposts against the 10" x 10" x 12' top chords. Counter braces measuring 5" x 10" and two struts, 4" x 10" each, respectively, support the queenpost braces and the top chord against buckling. Seven vertical rods also resist buckling forces in the queenpost braces and top chord of each truss. The three-centermost rods are 1-1/4" diameter steel while the outermost ones are 1" diameter wrought iron. These metal rods also serve the function of hanging the 9-1/2" x 11-1/2" deck beams against the bottoms of the bottom chords. Three of the seven deck beams have mortises used in earlier lower lateral bracing system and as such are clearly older. The current lower lateral bracing system consists of 4" x 4-1/2" to 5" timbers arranged in a double Warren pattern perpendicular to the longitudinal axis of the bridge. A single Warren pattern—just two diagonals—is used adjacent to the abutments. Stringers, 4" x 7" and 7" x 7", support the 3" thick deck planking, which in turn are topped with 3" thick wheel path runners. Some of these tie beams are supplemented with metal rods. There are also knee braces that augment the tie beams. Most of them are secured with toenails. Notable exceptions are those at the queenposts. Unlike the arrangement elsewhere at Morgan, there are three knee braces at each queenpost. Two of them are mortise and tenon connections, while the third is toenailed. An interior photograph taken by noted covered bridge authority Richard Sanders Allen about 1940 shows this configuration. The fabric of Morgan Bridge has many anomalous features that suggest repairs or rebuilding with previously used material. All of the tie beams, except the two that connect the tops of the queenposts, have step-lapped rafter notches of the sort used in the rafter beam and also common in bams. Were these timbers reused from Morgan Bridge, from another bridge or from a bam? Most of the posts that support the rafter beams have some sort of currently non-functioning notching. Perhaps most striking of all are unused connection points in the upstream queenpost braces. Virtually identical to the framing cuts for the hanger rods and stmts connection on the braces, they are however, oriented in a way that indicated they were never used in their current ^ Terminology in this section follows that in Joseph C. Nelson, Spanning Time: Vermont's Covered Bridges (Shelbume, VT: New England Press, 1997) whenever possible. See especially pp. 230, 240, and 244-245. ^ The measured drawings do not show that at least two of the four connections between the queenpost braces and the lower chords have been strengthened with steel plates. ^ A photocopy of Allen's image is in the field notes. The originals are part of the Allen Collection, National Society for the Preservation of Covered Bridges, Westminster, VT. MORGAN BRIDGE HAERVT-33 (Page 3) location. Are they the resuh of mistakes or are the timbers reused from another structure? Unfortunately, the historical record does little to clarify how this complicated fabric acquired the present configuration. History A Belvidere selectman, Clyde Lanpher, told the writer that [Morgan] bridge, which was posted against trucks the fall of 1971, also may soon have to have steel girders installed to sustain for continued use. -Robert L. Hagerman, Covered Bridges of Lamoille County, Vermont, 1972. The North Branch of the Lamoille River descends through the Town of Belvidere in a roughly southwesterly direction. The narrow and sparsely populated valley is defined by the foothills of Cold Hollow Mountain and dominated by the 2,280-foot Laraway Mountain on the right and left banks, respectively. Only a limited amount of land is suitable for farming and consequently logging and sawmills long dominated the local economy. During the nineteenth-century, however, there was some manufacturing of small-scale wood products such as butter tubs and maple sap buckets. State Route 109 parallels the North Branch and connects the Town's three hamlets, Belvidere Junction, Belvidere Center, and Belvidere Corners. A back road, between Belvidere Junction and Belvidere Center, connects residents and property owners along the right bank to route 109 and the rest of the Town. Two covered bridges, the Mill Bridge (near Belvidere Junction) and Morgan Bridge, increase access to the back road. In 1833, Nathan Morgan purchased a plot of land "supposed to contain 60 acres" in Town Lot 17—probably in the general area of Morgan Bridge. Morgan and any decedents seem the most likely source for the bridge's name as the town records list few Morgans and maps from 1859 and 1878 show no Morgans in the town. Unfortunately, while it is clear that the Morgan crossing is very old, the history of the current bridge and any predecessors is as confused and muddled as the bridge's fabric. In March 1886, -7 Belvidere "voted to build Morgan Bridge & put in stone abutments with sufficient wings." The 4 Robert L. Hagerman, Covered Bridges of Lamoille County, Vermont (Morrisville, VT: by the author, 1972), 20. ^ Historical Records Survey (Vermont), Inventory of the Town, Village and City Archives of Vermont No. 8, Lamoille County, Vol. I, Town of Belvidere (Montpelier, VT: The Survey, 1940), 11. ^ Town of Belvidere, DeedBooks (Town Clerk's Office, Town of Belvidere Offices, Belvidere Center, Vermont) Vol. 2, May 9, 1833, p. 40, Nelson, Spanning Time, 65; Henry Francis Walling, Map of the Counties of Orleans, Lamoille and Essex, Vermont: from actual surveys under the direction ofH.F. Walling (New York: Loomis & Way, 1859); F. W. Beers, Atlas of the Counties of Lamoille and Orleans, Vermont (New York: F. W. Beers & Co., 1878), 18. ^ Hagerman, Covered Bridges of Lamoille County, 19, gives the date for this as 188*7*, which surely is a typographic error. See Town of Belvidere, Town Records, (Town Clerk's Office, Town of Belvidere Offices, MORGAN BRIDGE HAERVT-33 (Page 4) Morrisville and Hyde Park, Vennont, News Citizen's coverage of the Town meeting reports the vote to raise taxes for highway and bridge repair, but makes no mention of the vote to build the bridge. In September 1898, ihsNews Citizen reported, "new abutments have been put under the covered bridge a mile north of the Junction." It has not been possible to confirm or dispute Hagerman's report that in 1899, ihQNews Citizen records that "the Morgan Bridge is being rebuih."^'^ Hagerman does correctly report that Richard Sanders Allen believes that a Lewis Robinson built the bridge, but Hagerman does it in the way that obscures the fact that Allen dates the bridge to 1895. Hagerman concludes his account of Morgan Bridge with an oral source indicating that the bridge's construction as taking place in 1898 by three men, Lewis Robinson, Charles Leonard[,] and Fred Tracy. The day the bridge was raised, ... workers at the nearby Page mill left their jobs to help in the work. This account, it would 1 ¦^ appear, refers to the later rather than the original construction of Morgan Bridge. To Hagerman, then, Morgan Bridge was originally constructed in 188[6] and underwent a "major, and possibly complete, reconstruction" in the late 1890s by Robinson, Leonard, and Tracy. More recently. Nelson misreads Hagerman by stating the bridge "was built by Lewis Robinson, Charles Leonard, and Fred Tracy in 1887." Even carefi accounts might not resolve this confused historiography. Charles Leonard, and Fred Tracy in 1887." Even careful reading of a decade of newspaper In the summer and early fall of 1979, the Vermont Agency of Transportation planed extensive renovations to the Morgan Bridge. Apparently the southeast stone abutment collapsed in the course of giving it a concrete facing. The project was expanded to stabilizing the abutments with concrete facings and providing for deck beams at an average of 8' intervals. In dealing with the latter, the agency's engineers installed five steel vertical rods in each truss. This probably required new holes in the queenpost braces and in the top chord. The new deck beams would place new vertical loads on the braces and top chord, and thus increase the likelihood of buckling. To prevent this, the engineers added new struts between the queenposts as well as Belvidere Center, Vermont), Vol. 2, p. 101. Also of interest is the wording of the article in the warning (on page 99): "Art 10'^ to see if the Town will vote to build the Morgan Bridge so called." ^ ''Belvidere," Morrisville and Hyde Park (Vt.) News Citizen, 13, no. 51 (March 4, 1886): 3. ^ "Belvidere," Morn5v;7/e and Hyde Park (Vt.) News Citizen, 17, no. 46 (September 7, 1898): 5. ^^ Hagerman, Covered Bridges ofLamoille County, 20. ^^ Hagerman, Covered Bridges ofLamoille County, 19; Richard Sanders Allen, Covered Bridges of the Northeast, rev. ed. (Brattleboro, VT: Stephen Greene Press, 1974), 109. ^^ Hagerman, Covered Bridges ofLamoille County, 20. ^^ Nelson, Spanning Time, 64-65. ^'^ Drawings 2, 3, 4, Belvidere, Cov. Br. 48, Proposed Improvement, Bridge Project, Town of Belvidere, County of Lamoille, July 1979, Records of the Vermont Agency of Transportation, Project Development Division, Montpelier, Vermont. ^^ Hugh Tallman, Road Commissioner, Town of Belvidere, Vermont, Personal conversation with author, August 20, 2003. MORGAN BRIDGE HAERVT-33 (Page 5) vertical struts parallel to the outer most rods. While not executed with great elegance, as witnessed by the poor alignment of the mid-span vertical rod and struts on the upstream truss, the work was accomplished without a steel deck. The new deck beam system also required new lower lateral bracing. In 2001, a crack in the counter brace on the northwestern most comer of the bridge prompted the use of a threaded steel bolt to secure the counter brace against the underside of the queenpost brace. A green standing seam roof, one of many placed on covered bridges throughout Vermont was installed in 2002. Despite its many trials and murky history, Morgan Bridge still functions as a working queenpost truss. Morgan Bridge was placed on the National Register of Historic Places in 1974. The Queenpost Truss The queenpost truss, a horizontal top chord between two vertical posts supported by two diagonal braces that are tied together by a horizontal bottom chord that creates an overall 1 7 trapezoidal shape, dates back at least to the Italian Renaissance. Compared to the kingpost truss with its single vertical and two inclined end-braces, the queenpost can span somewhat greater distances. As the span of a queenpost truss is increased, the distance between the bottom chord panel points also increases. At some point the longer and longer bottom chords start to sag from dead loads (self and bridge deck weight). One simple solution to this phenomenon that also retains the overall truss configuration is to add additional members to the truss — a process called subdividing — to support the middles of the long bottom chord members. There are limits both to subdivision and to the maximum length of a particular truss type; the longest queenpost in 1 Q Vermont measures 88'. Essentially, the twentieth-century repairs and alterations to the Morgan Bridge represent subdivisions of the original queenpost configuration by the addition of kingposts. That is, the struts and the vertical rod added to the panel between the two queenposts is a kingpost. The struts and hanger rods between the braces and the queenposts is one-half a kingpost with the balance of the kingpost form provided by the existing brace. Tom Peters has observed that the overlaying of familiar truss types, as seen at Morgan, in order to create longer and stronger bridges was common in at least the eighteenth and nineteenth centuries. Conclusion ^^ Hugh Tallman, August 20, 2003. ^^See any of the numerous reprints of Andrea Pahadio's / Quattro Lihri dell'Architettura of 1570. ^^ Nelson, Spanning Time, 245. ^^ Tom F. Peters, "Bridge Technology and Historical Scholarship," Proceedings of the First International Congress on Construction History, ed. Santiago Huerta (Madrid: Instituto Juan de Herrera, 2003), Vol. 1, 62-64; Likewise, a reverse procedure of peeling off simple truss forms was used in the nineteenth century to analyze certain statically indeterminate forms. See Historic American Engineering Record (HAER), National Park Service, U.S. Department of the Interior, "Structural Study of Pennsylvania Historic Bridges," HAER No. PA-488, 11-13. MORGAN BRIDGE HAERVT-33 (Page 6) While it may have been academically trained engineers who planned the 1980 renovations, and while the execution of the plan does not show the care associated with the craft tradition, the forms selected for the work were very consistent with the craft tradition. More importantly, these changes avoided the use of steel beams such as found immediately down stream at the Mill Bridge while keeping the trusses of an increasingly rare type ftally ftanctional. ^^ According to information compiled by David W. Wright and Joseph Conwill of the National Society for the Preservation of Covered Bridges, Personal conversations with author, August 6 and 19, 2003, there are sixteen old queenposts left in Vermont plus two others that are hard to classify. MORGAN BRIDGE HAERVT-33 (Page 7) APPENDIX A, Engineering Report Introduction The Morgan Bridge is currently one of the few remaining covered wooden bridges making use of the queenpost truss form. This report investigates the engineering aspects of queenpost bridges in general, and the Morgan Bridge in particular, including the effect of modifications made to the structure in 1979. It will also compare the use of experimental load testing with the more common practice of computational modeling as a means of analyzing the structural behavior of a unique, timber bridge. Historical Perspective The history of structural engineering is not only an interesting topic to engineers, it is also essential for a complete understanding of structures. Engineers must understand the reliability of materials, design details, and construction techniques used in the past so that they can make appropriate, informed decisions about ageing structures in the future. Ever changing building codes, methods of construction, and material strengths add difficulty and confusion to projects involving structures that were built generations ago. The ability to make accurate assessments of the condition of existing structures may help prevent unnecessary replacements, and also assist engineers to make appropriate decisions about rehabilitation. To fully appreciate and therefore fully understand any structure, engineers must be able to identify not only its structural behavior, but also its place in society, both socially and symbolically. There are 101 known queenpost covered wooden bridges left today, which account for a little over 10 percent of all remaining covered bridges in the United States. Similar, but generally shorter, kingpost bridges account for a little over 3 percent. These are fairly low numbers, considering the relative ease with which both types could be constructed, and the consequently large number that were likely once in use. It is not surprising, however, that the simplest bridges were not among the first to be protected, as they were short spans that broke no new technical ground, and typically treasured only locally. Nevertheless, this study of the most basic truss forms should prove valuable, since they established many of the principles that made the more-complex designs possible. These two forms date back to at least the sixteenth century. The kingpost is unarguably the simpler form of the two, consisting of a bottom chord, a center vertical member (the kingpost) and two diagonal bracing members that extend from the top of the kingpost to the ends of the bottom chord. The bottom chord and kingpost act in tension, and the diagonal braces are in compression under typical loading conditions. The forces in the structural members are illustrated in Figure 1, where compression members are labeled "C," and tension members are labeled "T." World Guide to Covered Wooden Bridges, 1989. MORGAN BRIDGE HAERVT-33 (Page 8) Figure 1. Typical kingpost configuration The queenpost is a variation of the kingpost form that was most likely developed to span longer distances. The queenpost form contains three panels, separated by two vertical tension members (queenposts). As in the kingpost, the bottom chord is under tension and the diagonals are in compression, as shown in Figure 2. The center panel, between the two queenposts, is an open box formed with the top and bottom chords. Having no diagonal braces, it would be unstable, except that the truss must have an effectively continuous bottom chord capable of resisting moment loads at the bottom of the queenposts. Figure 2. Typical queenpost configuration The Morgan Bridge An inspection of the Morgan Bridge made it clear that, while its design was carefully planned, its purpose was purely utilitarian. Many of the timber members exhibited unused notches, indicating that the bridge may have been built of reclaimed timbers. As detailed in the HAER historical report, records clearly indicate the rebuilding of the bridge during 1898-99, but the original bridge may well have been constructed at this location as early as 1886. Its design may have differed from the 1899 form, and original timbers could have been reused, leaving the notches as remnants from its original construction. When the Morgan Bridge was rebuilt in 1899, it most likely resembled the structure illustrated in Figure 3. In typical queenpost form, there were no members in the middle panel, but both diagonal braces and counterbraces in the side panels. Only three floor beams were used MORGAN BRIDGE HAERVT-33 (Page 9) to transfer loads from the deck to the truss. Metal rods in the side panels doubled as tension members (posts) in the truss and as a means to transfer load from the floor beams. Since the bridge had no post at its center, short rods were used to connect the floor beam to the lower chords at mid-span. For reasons lost to time, an odd post was inserted in the north side panel, but not quite at the intersection of the two diagonal members. The south side panel had no counterpart. Figure 3. 1899 Configuration of the Morgan Bridge In 1979, the Vermont Agency of Transportation rehabilitated the Morgan Bridge to the configuration illustrated in Figure 4. Four floor beams were added to bring the total to seven, and as in the original construction, metal rods were used to attach the new floor beams to the truss. It appears that the engineers' primary concern at the time was the floor, rather than the truss itself. Except for the addition of diagonals in the center panel to create a new panel point for the center floor beam support rod, there is no indication that individual truss members were strengthened. One area of particular interest is whether these new diagonals significantly altered the truss's performance. Figure 4. Current (post-1979) Configuration of Morgan Bridge The 1979 rehabilitation opened up another interesting area for exploration as well. Comparing experimental load tests with computer models of the current configuration should provide evidence and feedback to determine if the selected models can be used to evaluate other configurations of timber bridges—in this case, the 1899 design—with confidence, and to understand the limits of the models. Since rehabilitation is often necessary if these structures are to remain in service, assessing the positive and negative impacts of alterations can help engineers and owners to make better decisions when questions arise about the condition of other existing bridges. MORGAN BRIDGE HAERVT-33 (Page 10) EXPERIMENTAL LOAD TESTING OF THE MORGAN BRIDGE Testing Procedure The vehicle used to load test the Morgan Bridge was a Chevrolet Astro Van. Figure 5 is a photograph of the vehicle and Figure 6 illustrates the assumed axle weights, based on the curb weight data obtained from the manufacturer's specifications. Figure 5. Chevrolet Astro van used in experimental load testing 2150 lb. 2150 lb. Figure 6. Assumed axle weights of load test vehicle Instrumentation used to measure the bridge's response to live loading was placed on the bridge prior to testing. The truck was driven across the bridge from north to south several times, stopping at predetermined locations along the span so that data could be recorded from the instrumentation. These selected stopping points were chosen based on preliminary computer modeling done prior to field work. Tape was used to mark the exact axle locations during the first run of the truck across the bridge so that the same locations could accurately be found in each subsequent run. The arrows in Figure 7 indicate the locations at which data were recorded. MORGAN BRIDGE HAERVT-33 (Page 11) FB#1 FB#2 FB#7 Figure 7. Typical front axle locations along the west truss (view from inside bridge) Instrumentation Field testing of the Morgan Bridge involved the measurement of displacements, or structural movement under load, and strains in selected members. Several types of instrumentation were employed to acquire this data, including: Linear Variable Differential Transformer (LVDT) Position Transducers Extensometer Strain Gages Surveying Equipment An LVDT may be used to measure the relative movement between any two points. Using an LVDT alone is somewhat limiting in fieldwork, so a linear spring-cable system was introduced to the assembly. The cable allowed measurements between two points spaced far apart, and the spring allowed the LVDT to capture both positive and negative movements. The LVDT assembly shown in Figure 8 had one end attached to the bridge at mid-span and other end fixed to a point in the stream bottom. Figure 8. Linear variable differential transformer ( jVDT) assembly MORGAN BRIDGE HAERVT-33 (Page 12) Two Position Transducers were used in the load-testing program. Their assembly also employed a linear spring-cable assembly similar to that used with the LVDT, as shown in Figure 9. The difference in the two assemblies is that the position transducers capture relative movement through a different mechanism than the LVDT. Additionally, the position transducer's range of motion is only 4 centimeters, as opposed to 6 centimeters for the LVDT used. In Figure 9, the position transducer is mounted underneath the top chord of the west truss, adjacent to the north queenpost, and the cable runs diagonally to a position on the east truss, adjacent to the south queenpost. This was done to examine out-of-plane effects of live loading on the bridge. Figure 9. Position transducer assembly An extensometer also measures the relative movement between two points. It is a small instrument with only a %-inch range of motion, but it is highly accurate and a good instrument for measuring movements across connections. For these tests, an extensometer was mounted at the connection between one of the new diagonal members in the center panel and a queenpost, as shown in Figure 10. MORGAN BRIDGE HAERVT-33 (Page 13) Figure 10. Extensometer spanning the joint between a diagonal and queen post Strain gages were used to measure the axial strain in bridge members when subjected to live loading. While it is possible to mount them to wood members, they are more accurate when mounted to metal. Figure 11 shows one of the strain gages mounted to a metal rod. The gages used were capable of capturing up to 5 percent strain. Figure 11. Strain gage mounted to a metal rod on the west truss Surveying equipment, normally used for land surveying, was used to collect overall deflection data. The equipment included a total station and several reflective prism targets. MORGAN BRIDGE HAERVT-33 (Page 14) These are shown in Figures 12 and 13, respectively. Prisms were securely attached to the structure using custom-built brackets, and their initial positions were measured and recorded by the total station. These positions corresponded to the bridge in its dead load condition. The positions of these prisms were again measured and recorded by the total station under the various live load conditions to allow calculation of the deflection at each target location. Figure 12. View of total station with respect to the bridge Figure 13. Typical reflective prism target and bracket STATIC ANALYSIS OF THE MORGAN BRIDGE MORGAN BRIDGE HAERVT-33 (Page 15) Geometry and model construction Based on infonnation collected in the field, a computer model of the Morgan Bridge truss was developed using the MASTAN2 structural analysis software. A three-dimensional model will be presented briefly later in the report, but a majority of the analysis, both in the field and using the model, was performed in two dimensions, and on the west truss alone. Since the two trusses appeared to be alike, it was believed that they would perform the same way. Figure 14. Single-line diagram of west truss model As the diagram in Figure 14 shows, the model of the bridge used had forty-six nodes connecting sixty-six beam elements. The elements represent the centerlines of the actual member dimensions, as measured in the field. While a typical truss model assumes pinned connections at all nodes, in this case only the ends of certain members are pinned, while others are modeled as fixed connections. A fixed connection allows the transfer of moment between members, but a pinned connection is capable of transferring only axial forces. The circles shown in the model indicate the connections modeled as pins. (The actual bridge has fixed connections wherever a member, such as the top or bottom chord, is continuous across a connection, but there is limited rotational flexibility between the chords and the posts, so pin connections were used.) One location where all members were modeled as fixed was at the north (right) end of the west truss, where a triangular metal plate is securely fastened between the bottom chord and top diagonal bracing. The plate was modeled as a series of vertical steel elements, fixed at the top and bottom where they connected to the timber truss members. This plate can be seen in the Figure 15 below. MASTAN2, version 2.0, developed by R. D. Ziemian and W. McGuire, 2000. MORGAN BRIDGE HAERVT-33 (Page 16) Figure 15. Triangular metal plate in northwest corner of west truss In addition to capturing the varying joint stiffnesses between members, an attempt was also made to capture any significant eccentricities that might affect the structural behavior of the bridge. One of these eccentricities was at the connection between the new (1979) bracing in the center panel and the original top chord. The eccentricity of this connection can be seen in Figures 4, 14, and 16. / "IX.jX{{{{v!HH!5^ .: * ': ¦ iiiSiiii:-" "i^^ils-lJ Figure 16. Eccentricity of connection at mid-span of top chord. MORGAN BRIDGE HAERVT-33 (Page 17) Another variable thought to influence the bridge was the composition of the bottom chord. Since the bridge is approximately 60' long, the builders used three parallel pieces of wood, staggered and spliced along their lengths, to act as one continuous piece with a varying cross section. The full cross section showing all three pieces is shown in Figure 17, and Figure 18 is a plan view that shows the scarf joints and member end locations. The highlighted areas along the chord illustrate individual members that were included in the model. The basis for this selection is the conservative assumption that anywhere an individual member terminates, whether at a butt joint or scarf joint, it is not likely to contribute significantly to the overall strength of the chord. Figure 17. Cross section of lower chord Figure 18. Plan view of lower chord with contributing members highlighted The external boundary conditions at the ends of the bridge were modeled as simple supports, with the exact point of support located at the intersection of the bottom chord and queenpost brace centerlines. If it is assumed that the reaction force from the abutment beam seat is distributed evenly over the portion of the bolster beam supported by the abutment, then the resultant of the distributed reaction may be assumed to act at the midpoint of this supported length. Mechanical Properties of Structural Members One of the most challenging aspects of modeling the structural behavior of a nineteenth- century timber bridge is determining how best to quantify the mechanical properties of its members, especially its wooden members. Experimental means exist for determining certain mechanical properties, but they are expensive and, consequently, unavailable in most instances. The only definitive information regarding the species of wood used in the Morgan Bridge came from the U. S. Forest Products Laboratory, which tested a small piece of wood from one queenpost and determined that it was spruce. The laboratory can usually identify the genus of a sample through microscopic examination and its knowledge of the anatomical characteristics of different woods. So, while the exact species of spruce could not be provided, it was assumed for the purposes of this MORGAN BRIDGE HAERVT-33 (Page 18) analysis that it is eastern spruce, the only species that would have been readily available in New England. For such a small bridge, it is unlikely that wood would have been shipped from the southern states, or the Rockies, where other varieties of spruce are prevalent. Table 1 lists the mechanical properties assumed for the timber members in Morgan Bridge, as well as the mechanical properties assumed for the metal rods. Table 1. Mechanical Properties used in Computer Model Element Material E (psi) V Density (pcf) Wood members Eastern spruce 1,400,000 0.4 27.9 Metal rods Mild steel 29,000,000 0.3 490 Panel Point Loads due to Live Loads MASTAN2 computes dead loads internally using user inputs for cross-sectional area and material density. For vehicular live loads, it was desired to apply the wheel loads to the computer model in such a way that appropriate comparisons could be made between the model results and the field data. Using known axle weights and distances, it was possible to determine the distribution of loads from the tires to the deck, thence to the stringers, floor beams, and, finally, the truss. It was assumed for these calculations that the stringers were simply supported between the floor beams, and that the floor beams were simply supported between the east and west trusses. Table 2 summarizes the panel-point loads resulting from the truck located at the positions shown in Figure 7. Table 2. Panel Point Loads Resulting from Live Load Positions Floor Beam Front Axle Location Reaction A B C D E F G FB No. 1 586 420 FBNo. 2 903 90 i FB No. 3 65') 827 1075 142 FB No. 4 894 932 993 554 FB No. 5 180 1075 924 1206 155 FB No. 6 232 389 9!9 FB No. 7 1075 MORGAN BRIDGE HAERVT-33 (Page 19) Influence Lines Engineers often use graphical methods to describe the behavior of a structure. Common methods include shear, moment, and axial force diagrams, which are often used to demonstrate the effect of fixed or dead loads. Bridge live-load analysis requires the use of another graphical method due to the moving live loads. The curves generated are called an influence lines. Influence lines differ from the other diagrams in that they represent the effect of a moving load at one particular location on the bridge, rather than the effect of all loads at all locations on the bridge. Since the field test data on the Morgan Bridge were taken at specific points while the vehicle was moved to several different locations, influence lines provide a useful way to present and compare test data and computed values. Figures 19 and 20 form an elementary example showing the influence line of quarter- point deflection on a simply supported span. t 5(2.1) &C2,2) t S(2,3) 6(2,4) t 5(2,5) Figure 19. Series of loading conditions on one beam used to develop an influence line MORGAN BRIDGE HAERVT-33 (Page 20) ¦(2,1) ¦(2,5) CM "(3 c ,o |5 (2,4) 5 (2,2) 5 (2,3) 2 3 4 distance along span Figure 20. Influence line for quarter point deflection of a beam with simple supports TRUSS BEHAVIOR UNDER DEAD LOAD The dead load of each structural member was computed using member sizes obtained from field measurements and the weight densities given in Table 1. The dead load is the sum of the weights of all truss members plus the weights of all floor and roof components. The existing floor system was well documented so its weight could be computed and applied to the appropriate panel points. With the exception of fewer floor beams and panel points, it was assumed that the original floor system was similar to the existing one. Detailed measurements were not made of the roof, but it is most likely similar to the original structure, so a weight comparable to the floor system was used and applied to the top chord where it connects to the roof truss. The only dead load neglected in the analysis was that of the siding. Figures 21 and 22 show the axial force diagrams for the existing and original configurations, respectively. The thickness of each member in these diagrams is indicative of its stress, with the shaded members in compression and the unshaded ones in tension. While the diagrams are not labeled for simplicity, they are drawn to the same scale. The compression forces in the top chord and tension forces in the bottom chord are similar for both configurations. Specifically, the maximum compression in the top chord is approximately 15,200 pounds for the existing configuration (Figure 21) and 12,500 pounds for the original configuration (Figure 22). The maximum tension in the bottom chord is 14,000 pounds for the existing configuration and 11,300 pounds for the original configuration. The forces due to dead loads in the existing configuration are approximately 20 percent higher than those in the original configuration, which is due to the addition of four floor beams in 1979, as well as the significant change in the floor system load distribution that resulted from this change. Prior to the rehabilitation, the bolster beams likely distributed more load from the floor system to the abutments, but with addition of floor beams so close to the bolster beams, their role has likely been reduced significantly. The forces in both configurations are typical for the queenpost form, with only the bottom chord and queenposts in tension (plus the floor beam hanger rods in the existing configuration). MORGAN BRIDGE HAERVT-33 (Page 21) 7 -I O O Pn \ \ PH \: 1^ 1^ S MORGAN BRIDGE HAERVT-33 (Page 22) TRUSS BEHAVIOR UNDER LIVE LOAD Mid-Span Deflection and Overall Bridge Stiffness Due to the considerable variability in geometry and materials, it is difficult to draw comparisons among different covered wooden bridges. Stiffness can be a convenient, normalizing tool with which to make this comparison, as it reflects the behavior of the bridge system, rather than that of individual members. The formula for stiffness is: F k= — 5 where: k = flexural stiffness of the system F = force applied to the system S = overall displacement of the system In the field, an LVDT was used to take mid-span deflection measurements of the bridge under vehicular live loading. Figure 23 shows two photographs of this test assembly. On the left is the position of the LVDT at mid-span, as seen from inside the bridge. It was firmly connected to the lower chord, near the center vertical rod. The cable ran down through a space between the chord and deck, and was securely fastened to a tripod that sat in the water, held down by rocks and laboratory weights. An initial displacement was induced into the assembly by stretching the spring, to allow for both positive and negative deflections under live load. The cable in Figure 23 is darkened for clarity. Eleven trials were performed with this instrumentation in place. Figure 24 shows the results of the mid-span deflection tests. The independent (horizontal) axis demonstrates the location of the vehicle's center of gravity along the west truss, with zero at the south end, and the dependent (vertical) axis is the mid-span bridge deflection measured by the LVDT. A negative deflection value indicates that the bridge was moving downward. The dashed line shown on the plot is an interpolation of the likely influence line for mid-span deflection. Mid-span deflections obtained from the computer model are provided in Table 3, along with the mean and standard deviation of the experimental data plotted in Figure 24. MORGAN BRIDGE HAERVT-33 (Page 23) cable from LVDT '¦imvwjv.vj ^f "¦ :: 'i i cable from LVDT Figure 23. Measuring mid-span bridge deflections with LVDT O -0 03 ¦ 20 30 40 50 distance along west truss (ft) Figure 24. Influence line for mid-span deflection as a result of live load MORGAN BRIDGE HAERVT-33 (Page 24) Table 3. Comparison of mid-span deflection data Truck Position Field Data Computer Model Description Location (ft.) Hs (in.) as (in.) S(in 1.) A 11.05 -0.0315 0.0025 -0.0360 C 22.80 -0.0506 0.0026 -0.0523 D 29.80 -0.0499 0.0027 -0.0525 F 34.80 -0.0412 0.0035 -0.0485 G 45.80 -0.0119 0.0014 -0.0292 fi-s = mean as = standard deviation Generally, the model predicted the data within one standard deviation. The only exception occurred at the north end of the bridge, vv'here the prediction -was nearly 50 percent greater than the actual deflection. This, the quarter-point truck location, -was, hovv'ever, close to the triangular steel gusset plate. Given such small deflections, it is possible that the plate -was providing more strength in that corner of the bridge than the model, vv'hich simulated the plate -with five vertical, fixed-connection members, predicted. The foUovv'ing calculations compare the overall stiffness of Morgan Bridge, first as calculated using the computational model results, and then using the field results. For these calculations, it -was assumed that the overall stiffness -was equal to the mid-span stiffness -with the live load at location D. 4300 kips Ko,ei = ^^.^.. =81,900 k/in. 0.0525 in. versus 4300 kips 4300 kips (0.0499-F0.0027) in. '^p'^'"''""^^ (0.0499-0.0027) in. 81,750 k/in.<^__,,<91,100k/in. The predicted stiffness -was vv'ithin the possible range of stiffness determined from field measurements (considering instrument error). Given the practical assumptions needed for the model, this -was good agreement. Lower Chord Scarf Joint Behavior Scarf joints have been used in many timber bridges as a solution to the problem of transferring tension forces across a connection betvv'een t-wo pieces of vv'ood. The Morgan Bridge MORGAN BRIDGE HAERVT-33 (Page 25) is no exception, and it employs three scarf joints on each bottom chord to "splice" together two pieces of wood. The bottom chord appears to be part of the original structure, as evidenced by the scarf joints that are worn down and cracked from years of use. The following diagram illustrates how scarf joints would have originally worked. Figure 25. Scarf joint mechanism Field tests were performed using an extensometer in an attempt to determine the extent to which the scarf joints were still functioning. The scarf joints in both the interior lower chord member near the quarter point and the exterior lower chord member at mid-span were tested. The extensometer was positioned to span the top of the joint. Figure 26 is an illustration of the lower chord, showing the positions of the scarf joints. Three trials were performed for each joint location, and for each trial data was recorded with the truck at six different positions along the span: A, B, C, D, F and G (see Figure 8). QUARTER POINT MID-SPAN SCARF JOINT Symmetrical about CL Figure 26. Plan view of southern half of lower chord showing locations of scarf joints MORGAN BRIDGE HAERVT-33 (Page 26) Figures 27 and 28 show the extensometers in place at the two scarf joints. Figure 27. Extensometer spanning the scarf joint at mid-span Figure 28. Extensometer spanning the scarf joint at quarter point MORGAN BRIDGE HAERVT-33 (Page 27) Results of the data collected from the extensometer are provided in Figures 29 and 30. For both plots, the independent axis demonstrates the location of the vehicle's center of gravity along the west truss, with zero at the south end. The movement of the scarf joint is shown on the dependent axis. 20 30 40 location of load along west truss (ft.) 60 Figure 29. Results of extensometer placed at the mid-span scarf joint 0.0006 0.0005 - .E 0.0004 - = 0.0003 - E g 0.0002 - 0.0001 - 0.0000 £ -0.0001 I- O (A * -0.0003 - -0.0004 0 20 30 40 location of load along west truss (ft.) 50 60 Figure 30. Results of extensometer placed at the quarter-point scarf joint MORGAN BRIDGE HAERVT-33 (Page 28) Initially, it was thought that the scarf joint would act as a "pointer," such that a positive result would indicate the two members were pointing upward ( a ) and a negative result would indicate that they were pointing downward (v). According to the influence line in Figure 29, however, this would mean that the mid-span scarf joint consistently pointed up, even though the deflected shape at mid-span suggested just the opposite. Rather, it seems the results indicate that the scarf joint was not rotating at all, but was simply opening up as a result of tension in the bottom chord. Still, it is not clear why the mid-span joint would open more with the truck near the south end than it did with the truck near the north end. Nor is it clear why the scarf joint at the quarter point closed with the truck near either end of the bridge, since the bottom chord was always in tension, regardless of the truck's position. One thing is clear from the findings—^the assumption in the model that the areas around scarf joints did not contribute to the strength of the bottom chord was justified. If a stronger correlation between magnitudes of force in the bottom chord and the magnitude or direction of scarf joint movements existed, it would have been plausible to assume that some force was being transferred through that connection. For instance, if the mid-span scarf joint had opened its maximum amount with the truck at mid-span and less with the truck near the ends, it might well have indicated that the two sides of the spliced bottom chord member were reacting to the increased force in the chord, engaging more as the tension force increased and less as it decreased. As the results show, there was no such correlation to suggest that the movements are in any way related to the magnitudes of the tension forces. Wood Joint Behavior For covered wooden bridge builders, one of the main design goals was to minimize the number of tension connections. Unlike compression joints, wooden joints that functioned efficiently and reliably in tension were difficult to fabricate and maintain. This goal was apparent from the two earliest truss designs, the kingpost and queenpost. As discussed above, tension connections occur only where the vertical posts meet the top and bottom chords, and between adjacent bottom chord members. In the Morgan Bridge, the bottoms of the queen posts were notched to transfer tension force in the form of bearing pressure to the bottom chord, while mortise-and-tenon joints were employed forthe same purpose at the tops of the posts. Where adjacent timbers were needed to make up the length of the bottom chord, the scarf joints just discussed were used. These joints also transfer tension by way of bearing pressure between the interlocking knuckles. If the application of live load causes a member in compression under dead load to see tension instead, its compression joints cannot transfer this tensile load to the rest of the bridge, rendering the member useless. For early bridges—^those built before prestressing became common—the usual way to avoid this kind of stress reversal was to be sure the compression force from the dead load was great enough to maintain a net compressive force, even when live loads tended to reduce it. Thus, the joints would remain tight. This was the case for the original configuration of Morgan Bridge. When Morgan Bridge was rehabilitated in 1979, two counterbracing timber members were added to the center panel. While apparently installed to give better support to the center MORGAN BRIDGE HAERVT-33 (Page 29) floor beam rods, their orientation is such that under dead load they are in compression. Interestingly, these members were toe-nailed in place, something that should not have been necessary in compression joints. (Exactly when this was done is not known.) When the live load was positioned with a significant portion within the exterior panels, however, the model predicted, and field testing confirmed, that these new counterbracing members experienced tension. Closer inspection of the structure revealed that one of these members has recently been replaced, possibly indicating excessive stress as it at some point. This behavior is apparent in the axial force diagrams of Figure 31. The first two diagrams show the axial forces in the original bridge, first with live load alone (Figure 31a), then with the combined dead and live load (Figure 3 lb). The second two diagrams show the axial forces in the existing bridge, first with live load alone (Figure 31c), then with the combination of dead and live load (Figure 3 Id). The live load for all four diagrams in the model was placed at location A, where the maximum effect was observed in the field (see Figure 7). Note, however, that due to the difference in the number of floor beams between the original and existing bridges, the load distribution varied between the two models. (a) (b) (c) (d) Figure 31. Axial force diagrams for (a) live load at A in original configuration, (b) dead and live load at A in original configuration, (c) live load at A in existing configuration, and (d) dead and live load at A in existing configuration Experimental field tests were performed to determine the validity of the computer model with respect to the existing configuration. With the difficulties of measuring strain in timber, it was decided instead to measure displacement across the joint between the south counterbrace in the center panel and the south queenpost. An extensometer was positioned along the MORGAN BRIDGE HAERVT-33 (Page 30) longitudinal axis of the counterbrace, spanning its connection with the queenpost (see Figure 32). Three trials were performed, stopping the vehicle at five locations along the span each time to collect data. Figure 32. Extensometer in place at the connection The results of the field test and computational model are shown in Table 4 and Figure 33. Table 4 provides a comparison of the observed field data with the values predicted by the computer model. Figure 33 illustrates the data with the x-axis denoting the vehicle's location along the west truss (zero represents the south end), and the y-axis showing the amount of movement recorded in the joint along the longitudinal axis of the counterbracing member. A positive movement correlates to the joint opening up due to a net tensile force, while a negative movement correlates to its closing under net compression. The dashed line provides an approximate solution of the influence line for joint movement. MORGAN BRIDGE HAERVT-33 (Page 31) Table 4. Comparison of experimental results and predicted results for joint movement at queenpost - counterbrace connection Truck Position Field Data Computer Model Description Location (ft) ^15 (in) CT6(in) 5 (in) A 11.05 0.0106 0.0006 0.0014 C 22.80 0.0005 0.0001 -0.0001 D 29.80 -0.0030 0.0001 -0.0020 F 34.80 -0.0038 0.0001 -0.0021 G 45.80 -0.0023 0.0000 -0.0010 1^5 = mean cfs = standard deviation While the calculated and experimental magnitudes differ, their matching trends are clearly evident in Figure 33. The greatest variation between the experimental and computational results was when the live load was very close to the south abutment. The errors in the computed values were due to the model having a pinned joint at this location instead of one that suddenly became disconnected. In the field, the extensometer was observed to widen steadily as the truck moved past the queenpost location. It was apparent from both the experimental and computational analysis that the addition of these center panel diagonals added nothing to the performance of the truss system. In fact, the bridge is doing its best to remove them! 0.012 0.010 - 0.008 - 0.006 - I 0.004 -