You are on page 1of 5
FINITE-ELEMENT ANALYSIS OF STEEL GIRDER HIGHWAY BRIDGES By Mounir E. Mabsout,’ Kassim M. Tarhini,’ Member, ASCE, Gerald R. Frederick,’ and Charbel Tayar,' PE, Anstaact: This paper compares the performance of four finite-element modeling techniques reported inthe literature used in evaluating the whee! loud distribution factors of stel girder Bridges. A typeal onespan, simply supportd, two-lane, composite bridge superstructure was relected for this study, American Association of State Highssay and Transportation Officials (AASHTO) HS20-4 design truck loads were positioned to produce mi mam moments in the girder. Two finite-element programs, SAP8O and ICES-STRUDL, were used to perform the analysis along with their pre-and postprocessing capabilites. The results ofthese modeling techniques were ‘compared with AASHTO whee! load distribution factors (in 1996 and in 1994) and published expedimental ‘esults. The four finite-element modeling techniques yielded similar load distribution factors. Furher parametric ‘study, varying the span length and girder spacing, was conducted and the distribution factors obtained using Wo (ofthe Tour finite-element modeling techniques correlated well with AASHTO"s 1994 load and resistant factor ‘design (LRFD) based formula, but not with AASHTO's 1996 simple formula (SIS8). The AASHTO LRFD formula also correlated well with prototype experimental results reporced in the literature by various researchers. ‘This paper will assis bridge engineers in gaining confidence using the new AASHTO wheel load disteibution factors and adopting a fnte-lement modeling technique When 2 carrying capacity of existing bridges. INTRODUCTION ‘A common type of highway bridge superstructure is the concrete slab placed on steel girders. The analysis of these bridges is complicated by the geometric boundaries and load- ing conditions. ‘The American Association of Stale Highway land Transportation Officials (AASHTO 1996) standard spec- Ifications for Highway Bridges contain simple procedures used in the analysis and design of highway bridges. The typical AASHTO procedure is used to calculate the maximum bend- ing moment based on a single line of wheel loads from the S20 design truck or lane loading. This bending moment is then multiplied by the load distribution factor (S/5.5 where S is girder spacing in ft) to obtain the moment in an individual girder. The load distribution factor is @ key element in analyz= ing existing bridges and designing new ones. Field testing has shown these factors to be conservative. Therefore, improving the load distribution factors or adopting a finite-element mod- cling technique will permit more realistic calculation of the actual bridge capacity. This may permit mote bridges to re- ‘main in service with oF without minor repairs Recently, the specifications outlined in LRFD Bridge Design ‘Specifications (AASHTO 1994b) were adopted. This code in- Uwoduced new load distribution factors based on a comprehen- sive research project, National Cooperative Highway Research Program (NCHRP) 12-26, which resulted in the Guide Spec- {fication for Distribution of Loads for Highway Bridges (AASHTO 1994a). This guide recommends the use of simpli- fied formulas, graphical and simplified computer analysis, and ‘or detailed finite-element analysis (FEA) in calculating the ac- ‘wal distribution of loads in. highway bridges. The new {Formulas are generally more complex than those recommended by Standard Specifications for Highway Bridges (AASHTO "prof, Dept. of Cis: Engrg. Univ. of Neva, Las Vegas, NV BD1S4- ae Al-Handasah, Shar and Pater, Bei, Lebanon Nove. Dicinion pen unl January, 998. To exe the closing fae one month, writen request ost he Bled wit the ASCE Manager ‘of Journals. The manasrpt for this paper was submited fr review ahd Possible publcason om Janaary 23,1996. This paper Is pr ofthe ure ‘al of Bridge Enaineering, Vol.2, No.3, Auge 957 ASCE, ISSN Toetbreasiic-ooss-ioeiss00) +330 per page Paper ring new bridges or evaluating the load 1996), but their use is associated with a greater degree of ac curacy. These formulas were developed for typical AASHTO design loads and typical bridge types and dimensions. For ex- ample, the lateral load distribution factor for bending moment in interior girders of concrete slab on steel girder bridge su- perstructures developed by NCHRP 12-26 project is os 6)'@ Ge) ‘wheel load distribution factor; S = girder spacing (ty 35 << 16); L= span length of beam (A, 20. shell elements FIG. 4, Typical Cross Section through Part of Fiite-Element Model (case) ‘ments and the girders as concentric space frame elements (case ‘The other FEA cases could be adopted to accurately ide- Alize the actual boundary conditions; however, they will r- ‘uire more time to prepare the input due tothe increase in the ‘number of nodes and elements, more computation time, larger computer memory, and ealculation of girder bending moments Using stress values at critical sections. Bridge engineers must also understand the basic difference between the various fnite- element models when comparing the reported maximum de- Actions. “Tarhini and Frederick (1992) performed a parametic study using the case d FEA modeling technique. The computer pro- gram, ICES STRUDL Il, was used in their study. Various bridge geometric parameters investigated were chosen to vary within practical ranges. Span length and girder spacing were found t0 be related directly to whee! load distribution factors. However, the complexity in implementing and extracting sig- nificant results from this FEA modeling technique might deter engineers from considering it in day-to-day practice. There fore, a duplication of the previous FEA study was necessary Using the FEA modeling technique outlined in case a since it is simple to implement and manipulate the results. The various simple-span straight bridges had steel beams (W 36 160) spaced at 6, 8, and 12 ft (1.83, 2.44, and 3.66 m) with span | sowwr ‘oss ow vw _ | Pe aac | xr 20 SAPO FIG. 5. FEAModel Case (L=56 ttand 5=81t) FIG. 6, FEA Model Case d(L=55 and $= 61) TABLE 1. Results of Four FEA Models Maximus ‘girder | Ovbuton| AASHTO | NCHRP 12.25 Fea | moment | tector | (08) caso | nap) | (FEA) [OF = 555 © | Ve o @ a) 3396 120 tas & | S386 the 1s ae & | abe cia a5 oe a | $206 19 as ey lengths of 9, 45.5, 56,77, 98, and 119 f (10.67, 13.87, 17.07, 23,47, and 36.27 m). AASHTO wheel loads were positioned on the bridge inorder to produce maximum bending moments ‘and calculate the maximum wheel load distribution. Table 2 ‘and Fig. 7 show a comparison of FEA cases a and d as they are compared with AASHTO (1996) and NCHRP 12-26 (Eq 1) load distribution factors. In genera, the two finite-element ‘modeling techniques a and d produce load distribution factors similar to the NCHRP 12-26 equation but all are less than ‘AASHTO (1996) factor. EXPERIMENTAL EVALUATION ‘The previous section demonstrated that NCHRP 12-26 load Aistribotion factor for steel girder bridges correlates well with the two finite-element modeling techniques cases a and d. Nu- merous field testing programs has been reported in the liter ture by various investigators, The maximum experimental ‘wheel load distribution factors for seven bridges are compared with NCHRP 12-26 and AASHTO (1996) distribution factors as shown in Table 3. The tested bridges were atleast two lanes, they were straight, and they were loaded by placing trucks on all lanes. A summary of each bridge test program is presented inthis section Hindman and Vandegrft (1945) performed a field test on the Ohio Bridge type A structure, which was continuous with ‘equal span lengths of 88 fe (26.82 m). The bridge deck was 32 ft (9.75 m) wide and consisted of an 8.25 in. (21 em) thick reinforced concrete slab carried by five W 36 X 230 steel girders spaced at 6.93 f(2.11 m). Influence lines for deflection Were drawn due 0 concentrated loads. The maximum load distribution factor was determined, using deflection readings, to be 1.05, Linger (1963) reported the static load distribution factors of| two continuous I-girder, four-span, highway bridges. The two structures located around Des Moines, Towa, were continuous aluminum and steel girder bridges. Both bridges caried two lanes of trafic and were instrumented with SR-4 strain gauges at critical locations inthe outer and inner spans as well 3s the TABLE 2._Comparison of Wheel Load Dletribution Factors (DF) rom FEA Mesh Modeling wth AASHTO Formulas Fite Element L s | DF | OF | oFeq | or=-s | it | eeasea | casea] cy | 55 wm | @ | o@ | ow | @ | © 3s [tae [tas P1208] 1.09 s | ras | ras | te | as a fan S| los | ate 455 s | to |} og7 | ras | 109 e | te foie | oa fous | tm | tes | 190 | zie 56 6 | or | 02 | 110 | 109 B ] te | ons | ise | as [ores | tse | tho | ate n ¢ | os | ose | ior | 10 e fos | ue | i | us | ts | tas] gs | ate oe 6 | om | oss | oss | 109 s {0a | to | oie | is a | tae | uae | ase | ate 1 6 | om | os | 09 | 1.09 8 | ta | tos fone | tas a [tw | io | ise | ate ‘Rote FEA = Hate clereat anajlsi DF = disebaton factor ‘Note = yan length and = der spacing JOURNAL OF BRIDGE ENGINEERING / AUGUST 1907/85

You might also like