Design of Crane Runway Beam

with Channel Cap

DUANE S. ELLIFRITT and DUNG-MYAU LUE

INTRODUCTION πCb   2 
Mn =  EI y GJ  B1 + 1 + B2 + B1   (1)
KL  
A common practice in industrial buildings is to weld a
channel, open side down, to the top flange of a standard where
rolled beam for use as a crane runway. In many cases, it is πβ x EI y
not possible to brace a crane runway laterally between B1 = (2)
2 KL GJ
columns, so the channel provides additional lateral stiffness.
There are several interesting structural questions π 2 ECw
associated with the practice, like, what should be the welding B2 = (3)
( KL) 2 GJ
pattern? How are the residual stresses affected? What if the
channel has a yield strength different from the beam?
1
However, the primary question addressed in this paper is,
how does one check such a beam for lateral-torsional
βx =
Ix ∫
A
y( x 2 + y 2 )dA − 2 yo (4)

buckling? L 1 L s
The AISC Specification provides formulas for lateral- Cw = ∫o
Wn2 tds, Wn =
A ∫
o ∫
Wo tds − Wo , Wo = ρ o ds
o
(5)
torsional buckling of monosymmetric sections, but these are
derived for three-plate sections, as shown in Figure 1(a).
Some of these equations require the calculation of section ∫
J = r 2 dA
A
(6)

properties that are not readily available from handbooks, like K = effective length factors (Assume Kx = Ky = Kz = 1.0)
the warping constant, nor do they apply to the real section as
shown in Figure 1(b). To evaluate the elastic nominal moment (Mn) according to
The purpose of this research was to develop an analytical Equations 1, 2, and 3, one has to determine the coefficient of
solution to the problem of lateral-torsional buckling of a
rolled beam with channel cap in terms that are readily
available from standard section property tables. A program
of full-scale testing was carried out to verify the analytical
solution. It should be noted here that the solution proposed in
this paper applies only to light- to medium-duty cranes, those
classified as A, B, or C by the Crane Manufacturers
Association of America (CMAA).

ANALYTICAL SOLUTION FOR ELASTIC

BUCKLING
According to Clark and Hill (1960), Galambos (1968 &
1988), and Johnston (1976), the elastic nominal moment (Mn)
of a monosymmetric beam, including the section of a rolled
beam with a channel cap, can be expressed by the following
equations:

Duane S. Ellifritt is professor of structural design, University of

Florida, Gainesville, FL.
Dung-Myau Lue is associate professor, National Chung-Hsing Fig. 1. Common monosymmetric sections (a) 3-plate section, (b)
University, Taiwan. rolled beam with channel cap.

ENGINEERING JOURNAL / SECOND QUARTER / 1998 41

monosymmetry (β x), the warping constant (Cw), and the  2 I yc   D 
torsional constant (J). β x = 0.9 − 1  D + L  (10)
 yI   2 
For a monosymmetric section made of three dissimilar
plates, Equations 2 and 3 can be simplified into Using the forty-five sections shown in the AISC Manual
(AISC, 1989) as a data base, it was found that Equation 11
  I yc   h  I y gave a very close approximation of Equation 9
B1 = 2.252  − 1  (7)
  I y   Lb  J  D 
β x = 087
. ( R − 1) D + L  (11)
 2 
 I yc   I yc   h 
2

B2 = 25 1 −    (8) where

 I y   J   LB 
2 I yc
R=
In this way, the need to calculate the warping constant is Iy
avoided. Equations 7 and 8 are identical to the equations in
The solution (Kitipornchai and Trahair) and the proposed
the footnote on page 6-114 of the AISC-LRFD Specification
model are plotted in Figure 2. It can be seen that the
(1993).
approximate model (Equation 11) is accurate within ± 4
However, these equations are correct only for the three-
percent.
plate section; they are conservative for the section which is
the subject of this paper. For the beam-and-channel section,
The Cwc Model
one should calculate β x, Cwc, and J according to Equations 4,
5 and 6, but this can be a daunting task for routine design The authors used a computer program originally written in
office use. It would be better to have models for these three BASIC language by Dr. T. V. Galambos, to calculate the
items which make use of known properties, or easy-to-find exact values of warping constants (Cwc) of a beam with a
properties, that are still reasonably accurate. channel cap. The results for the forty-five sections shown in
the AISC manuals (AISC 1989 and AISC 1993) are plotted
PROPOSED MODELS in Figure 3. By applying a multiple linear regression
technique to the data, the following curve was found to be a
The β x Model good fit.
According to Kitipornchai and Trahair (1980), β x for a  Ac 
Cwc = Cw  0.79 + 179
.  (12)
Aw 
section with a lipped top flange is given by

 2 I yc   I y  
2
D  where
β x = 0.9h − 1 1 −     1 + L 
 (9)
 Iy    I x    2D Ac
0.2 ≤ ≤ 0.95
Aw
Recognizing that (Iy/Ix)2 is a very small number, and letting h
= D, this equation can be reduced to The resulting curve is superimposed on the data of Figure 3.
The model as given by Equation 12 requires no calculation of

Fig. 3. The warping constant model as a function of beam and

Fig. 2. The accuracy of the β x model.
channel area.

42 ENGINEERING JOURNAL / SECOND QUARTER / 1998

section properties and uses only Cw and the ratio Ac / Aw as t2 = flange thickness of the wide flange
the independent variables. These variables Ac, Aw, and Cw are
All the variables shown are listed in the AISC manual.
given in the AISC Manual of Steel Construction. In Figure 4,
Again, the model is applied to the forty-five sections in
it can be observed that the model gives results with errors of
the AISC Manual and the results are plotted in Figure 5. It
–3 percent to +5 percent in most sections. [It should be noted
can be seen that the model overestimates the value of J by
that Equation 12 is slightly different from the formula
+2.1 percent to +8.2 percent.
published by Lue and Ellifritt (Lue, 1993) previously. This
equation is more accurate, while retaining the simplicity of
APPROXIMATE DESIGN USING THE PROPOSED
the one published in 1993.]
MODELS
The J Model The proposed models based on the previous discussions can
be summarized by the following equations.
The torsional constant (J) can be expressed by
πCb 
n Mn = EI y GJ  B1 + 1 + B2 + B12   (16)
∑ KL   
1
∫
J = r dA =2
bi ti3 (13)
A 3 1
where
where
bi and ti = width and thickness, respectively of each πβ x EI y π 2 ECwc
B1 = B2 = (17)
element of the cross-section 2 KL GJ ( KL) 2 GJ
n = number of plate elements
 D  2 I yc
A further modification of Equation 13 has been done for the β x = 0.87(2 R − 1) D + L  , R = (18)
 2  Iy
need of the practical designer. The modified formula of J is
calculated based on the section properties and dimensions of
I yw
wide flanges and channel caps as listed in AISC manual. The I y = I yw + I xc , I yc = + I xc (19)
proposed J is given by the following equations. 2
n  Ac  A
∑ Cwc = Cw  0.79 + 179  , 0.2 ≤ c ≤ 0.95
1 1 1
J= bi ti3 = J w + J c + b f (t1 + t2 ) 3 − b f (t13 + t23 ) (14) . (20)
3 3 3  Aw  Aw
o
= Jw + Jc + bft1t2(t1 + t2) (15) J = Jw + Jc + Bt1t2(t1 + t2) (21)
where
The forty-five AISC sections are used to examine the
Jw and Jc = torsional constants for the wide flange and differences between the exact nominal moments (Mn) and the
the channel, respectively ones based on the proposed models.
bf = flange width of the wide flange The curves of nominal moment (Mn) versus unbraced
tl = web thickness of the channel length (Lb) were examined by using the proposed models. A

Fig. 4. The accuracy of the warping constant model. Fig. 5. The accuracy of the torsion constant model.

ENGINEERING JOURNAL / SECOND QUARTER / 1998 43

straight line, which is adopted by the current LRFD method, Table 1.
is used for the inelastic nominal moment between the plastic Percent Error in Proposed Elastic Buckling
moment (Mp) and the limiting moment (Mr). The results of Moment Model for Combination Beam
and Channel Sections Shown in the
the twenty-eight sections in the AISC-LRFD Manual are AISC-LRFD Manual (1993)
recorded in Table 1. It can be seen that the model gives Beam Channel Ac / A w % Error
errors of –2.5 percent to +1.8 percent. 920×223 (36×150) 380×50 (15×33.9) 0.225 –0.8 to –0.1
The proposed formulation, shown in Equations 16 840×210 (33×141) 380×50 (15×33.9) 0.239 –0.9 to –0.2
through 21, makes use of only dimensions and properties 610×125 (24×84) 310×31 (12×20.7) 0.247 –1.8 to –0.7
readily available from the AISC Manual and produces results 920×223 (36×150) 460×64 (18×42.7) 0.285 –1.4 to –0.4
840×176 (33×118) 380×50 (15×33.9) 0.287 –1.1 to –0.3
only ±2.5 percent from the exact solution.
760×173 (30×116) 380×50 (15×33.9) 0.291 –1.3 to –0.4
Bear in mind that the foregoing is only an elastic solution 840×210 (33×131) 460×64 (18×42.7) 0.303 –1.3 to –0.7
and the increased lateral strength of adding a channel cap 610×101 (24×68) 310×31 (12×20.7) 0.303 –2.1 to –1.0
may mean that lateral-torsional buckling will be inelastic. 530×101 (21×68) 310×31 (12×20.7) 0.305 –1.9 to –0.5
However, the inelastic transition curve will still be affected 530×92 (21×62) 310×31 (12×20.7) 0.333 –2.0 to –0.5
by the elastic curve. In this research, the AISC-LRFD 760×147 (30×99) 380×50 (15×33.9) 0.342 –1.4 to –0.4
approach of using a straight line inelastic transition curve 690×140 (27×94) 380×50 (15×33.9) 0.360 –1.2 to –0.5
840×176 (33×118) 460×64 (18×42.7) 0.363 –1.3 to –0.5
between Mp and Mr was followed. 760×173 (30×116) 460×64 (18×42.7) 0.368 –1.2 to –0.5
Experimental Study 690×125 (27×84) 380×50 (15×33.9) 0.402 –1.2 to –0.3
610×125 (24×84) 380×50 (15×33.9) 0.403 –1.2 to –0.5
Full-scale tests were performed on the beams shown in Table 460×74 (18×50) 310×31 (12×20.7) 0.414 –2.0 to –0.8
3. The testing apparatus is shown in Figure 6. The beams 760×147 (30×99) 460×64 (18×42.7) 0.433 –1.0 to –0.2
were tested with a single concentrated load applied to the top 610×101 (24×68) 380×50 (15×33.9) 0.496 –1.1 to –0.2
flange through a ball joint either at mid-span or at the one 530×101 (21×68) 380×50 (15×33.9) 0.498 –0.6 to +0.1
third-point. A gravity load simulator, as shown in Figure 6, 360×44 (14×30) 250×23 (10×15.3) 0.507 –1.2 to +0.2
530×92 (21×62) 380×50 (15×33.9) 0.544 –0.5 to +0.1
was used to ensure that the load would remain vertical during 410×53 (16×36) 310×31 (12×20.7) 0.575 –0.4 to +1.4
lateral buckling of the beam. An independent stand held two 310×39 (12×26) 250×23 (10×15.3) 0.587 –2.5 to –0.8
LVDT's for measuring both vertical and horizontal 460×74 (18×50) 380×50 (15×33.9) 0.678 +0.1 to +0.6
displacements. The ends of the beams rested on rollers and 360×44 (14×30) 310×31 (12×20.7) 0.688 –0.4 to +0.6
could rotate about both horizontal and vertical axes, but were 310×39 (12×26) 310×31 (12×20.7) 0.796 –1.6 to –1.3
restrained from torsional rotation about a longitudinal axis. 410×53 (16×36) 380×50 (15×33.9) 0.940 +0.4 to +1.8
Notes: Ac = Area of Channel Aw = Area of Beam
Fy = 350 MPa (50 ksi)
Unbraced Length = 19 to 23 m (60 to 70 ft.)
Mn1 = Theoretical Nominal Moment
Mn2 = Nominal Moment by Proposed Model
% Error = (Mn2 – Mn1) / Mn1 * 100

In selecting the beam and channel sizes to be tested, an

attempt was made to devise combined specimens that
represented low, intermediate and high values of the area
ratios (See Table 1 for representative values of Ac / Aw).
In order to get the most out of the material, each beam
was tested twice: once without the channel cap and once with
it. In the first test, the beam was made long enough to ensure
buckling in the elastic range. Then it was unloaded, taken
down and a channel was welded to the top flange. The
composite beam was then put back in the fixture and tested
Fig. 6. Experimental loading apparatus. again. In the second stage, failure was either by inelastic

44 ENGINEERING JOURNAL / SECOND QUARTER / 1998

 Table 2.
Beam Test Results (without Channels)
Span Pu Pe
Test No. Beam m (ft) Load @ kN (k) kN (k) Pu/Pe
W-1 310×28 (W12×19) 7.3 (24) ½-pt. 15.6(3.50) 10.8(2.44) 1.43
W-2 310×33 (W12×22) 5.5 (18) -pt. 46.7(10.5) 32.4(7.29) 1.44
W-3 250×22 (W10×15) 5.5 (18) -pt. 18.6(4.20) 15.1(3.40) 1.24
W-4 310×28 (W12×19) 5.5 (18) -pt. 44.8(10.0) 23.5(5.28) 1.89
W-5 310×28 (W12×19) 3.7 (12) ½-pt. 70.3(15.8) 56.9(12.8) 1.23
W-6 250×22 (W10×15) 3.7 (12) ½-pt. 40.0(9.00) 36.3(8.15) 1.10
W-7 200×9.7 (M8×6.5) 3.7 (12) ½-pt. 8.0(1.80) 4.8(1.08) 1.67
W-8 200×9.7 (M8×6.5) 3.7 (12) ½-pt. 6.7(1.50) 4.8(1.08) 1.39
Notes:
Pu is the elastic buckling load from test
Pe is the theoretical elastic buckling load based onCb = 1

buckling or yielding and plastic hinge formation at the load elastically, while most of those with a channel cap either
point. Because of the limitations in our laboratory, it was buckled inelastically or formed a plastic hinge.
difficult to get lengths of combined specimens long enough to Note the test load/predicted load in Table 2. The
produce elastic behavior. The results of all tests are shown in predicted load was calculated using a moment diagram
Tables 2 and 3 and graphically in Figures 7 and 8. Tensile modifier, (called Cb in AISC) of 1.0. The Pu / Pe values then
coupons were taken from each member and those results are should be somewhat representative of the proper Cb values.
shown in Table 4. The loading was done with a single concentrated load at
Figure 9 shows a beam being loaded. Note the end either the mid-span or point. Calculated Cb values vary
fixtures that prevent rotation about the longitudinal axis but from 1.32 to 1.41. With a couple of exceptions, most tests
allows rotation about the x and y axes. Figure 10 shows a fall into this range. The average for all tests is 1.42.
beam with channel cap after buckling. Note that the load at The Pu / Pe values for the tests with channel, however,
mid-span is applied uniformly across the flange, rather than are much lower, ranging from 0.93 to 1.23 with an average of
at a ball-joint as in Figure 9. This was to see if the loading 1.10. This indicates that the current formula for Cb in the
device had any restraining effect on lateral-torsional 1993 AISC-LRFD Specification is accurate enough for
buckling. bisymmetric sections, but overestimates the value of the
moment coefficient for monosymmetric sections.
Evaluation of Test Results Nethercot and Rockey (1971) proposed a moment
modifier that was dependent on where on a cross section the
In Figures 7 and 8, the test results are plotted against the
load was
theoretical solution described in the early pages of this paper.
Note that all rolled beams without a channel cap buckled

Fig. 7. Tests of rolled beams without channels. Fig. 8. Tests of beams with channel caps.

ENGINEERING JOURNAL / SECOND QUARTER / 1998 45

 Table 3.
Test Results, Beam with Channel Cap
Span Pu Pe
Test No. Beam Channel m (ft) Load @ kN (k) kN (k) Pu / P e
WC-1 310×28 (W12×19) 150×12 (C6×8.2) 7.3(24) ½ 53.8(12.0) 57.8(12.9) 0.93
WC-1A 310×28 (W12×19) 150×12 (C6×8.2) 5.5(18) ½ 84.5(19.0) N/A
WC-2 310×33 (W12×22) 150×12 (C6×8.2) 5.5(18) 136(30.5) 148(33.3) 0.93
WC-2A 310×33 (W12×22) 150×12 (C6×8.2) 5.5(18) 173(39.0) N/A
WC-3 250×22 (W10×15) 150×12 (C6×8.2) 5.5(18) 102(22.9) 81.8(18.4) 1.23
WC-4 310×28 (W12×19) 150×12 (C6×8.2) 5.5(18) 158(35.5) 128(28.8) 1.23
WC-5 310×28 (W12×19) 150×12 (C6×8.2) 3.7(12) ½ 220(49.5) 222(49.9) 0.99
WC-6 250×22 (W10×15) 150×12 (C6×8.2) 3.7(12) ½ 145(32.5) 125(28.2) 1.15
WC-7 200×9.7 (M8×6.5) 100×8 (C4×5.4) 3.7(12) ½ 44.5(10.0) 36.3(8.16) 1.23
WC-8 200×9.7 (M8×6.5) 100×8 (C4×5.4) 3.7(12) ½ 40.0(9.0) 36.3(8.16) 1.10
Notes:
Pe in this table is either the plastic bending load or the inelastic buckling load, based on the LRFD straight line transition.
The "A" subscript refers to those tests in which the load was applied through a flat surface in contact with the entire compression flange,
as opposed to a ball joint.

applied. It consisted of two terms A and B, and Cb was A SUGGESTED DESIGN PROCEDURE
computed as:
The standard unbraced length vs. moment capacity curve for
Cb = A / B if load is applied to top flange doubly symmetric sections as published in AISC is shown in
Cb = A if load is applied at centroid Figure 11. It is completely defined by Mp, Mr, Lp, Lr, and the
Cb = A • B if load is applied to bottom flange elastic buckling curve, all of which are easily calculated for
rolled shapes whose properties are found in tables.
They derived values for A and B for the case of a
For monosymmetric sections, everything about the curve
concentrated load at the mid-span of a doubly symmetric
becomes more complicated. Mr is the smaller of FLSxc and
section. While this does not exactly fit the beams described
FySxt; some vital section properties have to be calculated,
in this paper, it seems to offer a plausible explanation for the
values of Pu / Pe from Table 3.

Fig. 9. Channel-capped beam under load.

Note the end restraints against rotation
and the top-flange, ball-joint load. Fig. 10. Test beam after loading, showing buckled shape.

46 ENGINEERING JOURNAL / SECOND QUARTER / 1998

 Table 4.
Mechanical Properties of Test Specimens
Beams Fy Channels Fy
310×28 (W12×19) 432 MPa (62.1 ksi) 150×12 (C6×8.2) 435 MPa (62.6 ksi)
310×33 (W12×22) 431 (64.0) 150×12 (C6×8.2) 334 (48.0)
250×22 (W10×15) 370 (53.3) 150×12 (C6×8.2) 350 (50.3)
200×9.7 (M8×6.5) 278 (40.0) 100×8 (C4×5.4) 316 (45.4)

with no small effort; Lr cannot be determined directly, but is W27×94 with a C15×33.9 cap on a 30-foot simple span. It is
the result of an iterative process. assumed that the entire span length is unbraced laterally.
One can help things a bit by initializing the curve, The first example makes use of the curve-fit properties of
dividing Mn by Mp and changing the horizontal axis from Lb β x, Cwc, and J (Equations 11, 12, and 13) of this report. This
to λ = M p Me , where Me is the elastic lateral-torsional provides, of course, only an elastic buckling solution. For the
buckling moment. Using the forms of β x, Cwc, and J presented inelastic case, it is assumed that the AISC straight-line
in this paper (Equations 11, 12, and 13), and developing transition between Lp and Lr is appropriate.
curves for all 48 combination sections from AISC LRFD (pp. In calculating Mr, there is a fundamental question that
1-106, 107) and AISC ASD (pp. 1-83, 85), it was found that, must be addressed: What residual stress is appropriate? The
with small error, "one curve fits all." There is a narrow band Specification requires 10 ksi for rolled sections and 16.5 for
between Lp and Lr, but the elastic curves were practically welded shapes, but does not say what to use if you weld two
identical. This led to the approximate curve shown in Figure rolled shapes together. The welding of a channel to the top
12. flange of a crane beam is usually intermittent—it only being
This eliminates the iteration to find Lr, but also obscures required to transmit horizontal shear, which is low. A 2- or
the unbraced length a bit. With this curve, you will have to 3-in. weld every 4 feet is usually adequate. In the author's
calculate Me for a given unbraced length, determine λ, then opinion, this is not a great amount of heat input to the beam
Mn is either Mp, Me, or and it seems reasonable to use 10 ksi for the residual stresses
in the Mr calculation.
 λ − 0.49  The determination of Lr in Example 1 is still
Mn = M p − ( M p − Mr )  (22) troublesome. There is no way to calculate it directly: It must
 115
. − 0.49 
be iterated, until the unbraced length used in the elastic
buckling formulation produces an Me = Mr. That length is
then Lr.
DESIGN EXAMPLE Example 2 uses the generalized approach discussed
which plots Mn / Mp against M p Me . This has the
There are two design examples attached to this report. The advantage of eliminating the annoying iterations for Lr, but
beam, channel cap, and span are the same as the example in the disadvantage of somewhat camouflaging the unbraced
the AISC Steel Design Guide Series #7—Industrial length.
Buildings: Roofs to Column Anchorage, (p. 52). It is a The procedure is: calculate Me as described herein for

Fig. 11. Nominal moment capacity curve according to AISC-LRFD Fig. 12. Generalized M-λ curve for rolled beams with channel cap
Specification. (A36 steel).

ENGINEERING JOURNAL / SECOND QUARTER / 1998 47

whatever unbraced length you have, then calculate λ. If λ is Journal of the Structural Division, ASCE, Vol. 86, No. ST7,
between 1.15 and 0.49, use the straight line transition shown 1960, pp. 175–196.
in Equation 22; if it is less than 0.49, Mn = Mp; if it is greater Galambos, T. V., Structural Members and Frames, Prentice-
than 1.15, Mn = Me. Note that the results of the two examples Hall, Englewood Cliffs, NJ, 1968, pp. 80–158.
differ by only 1 percent. Galambos, T. V., Guide to Stability Design Criteria for
Metal Structures, Wiley, NY, 1988, pp. 155–188.
SUMMARY AND CONCLUSIONS Johnston, B. G., Guide to Stability Design Criteria for Metal
Structures (SSRC), Wiley, NY, 1976, pp. 95–115.
A mathematical model was developed to predict the elastic
Kitipornchai, S., and Trahair, N. S., "Buckling Properties of
buckling capacity of a hot-rolled wide flange beam with a
Monosymmetric I-beams," Journal of the Structural
channel cap, using only those section properties that can be
Division, ASCE, Vol. 106, No. ST5, 1980, pp. 941–957.
found in steel manuals. The cumbersome calculation of a
Lue, D. M., and Ellifritt, D. S., "The Warping Constant for
warping constant has been reduced to a simple function of
the W-Section with a Channel Cap," Engineering Journal,
handbook properties.
AISC, 1st Qtr., 1993, pp. 31–33.
Full scale tests were made on beams without channels.
Nethercot, D. A., and Rockey, K. C., "A Unified Approach to
These specimens were chosen with unbraced lengths and
the Elastic Lateral Buckling of Beams," The Structural
cross-sections designed to ensure elastic buckling. The
Engineer, Vol. 49, No. 7, London, UK, 1971, pp. 321–330.
average of test load/calculated load was 1.42. After testing,
channels were welded to the top flanges of all these beams
NOMENCLATURE
and tested again. This time the average test/calculated load
was 1.10. E = Young's Modulus—200 MPa (29,000 ksi)
The calculated buckling loads assumed Cb = 1, so the G = Shear Modulus—80 MPa (11,300 ksi)
test/calculated values roughly correspond to the correct Cb Lb = Unbraced Length mm (in.)
values. Because the shear center and centroid do not coincide I x, Iy = Moments of inertia about x and y, axis,
in the monosymmetric shapes, the Cb indicated is somewhat respectively, mm4 (in.4)
lower than would be calculated by the current AISC Iyc = Moment of inertia of compression flange about
Specification. This suggests that, in the absence of more y-axis
testing, it might be appropriate to take Cb as 1.0 for D = Depth of Rolled Beam, mm (in.)
monosymmetric sections. DL = Depth of Lip on Channel Cap, mm (in.)
Cw = Warping Constant of Rolled Beam, mm6 (in.6)
ADDITIONAL COMMENTS Cwc = Warping Constant of Rolled Beam with Channel
Cap
Another aspect of this problem, (not a part of the research
Ac = Area of Channel, mm2 (in.2)
reported herein) is the restraining effect of the crane runway
Aw = Area of Rolled Beam, mm2 (in.2)
on the other side of the aisle. The maximum wheel load
J = Torsional Constant, mm4 (in.4)
occurs when the trolley and lifted load is all the way to one
Iyw = Moment of inertia of rolled beam about y-axis,
side. Because most cranes have double-flanged wheels, the
mm4 (in.4)
runway beam on the other side, being very lightly loaded,
Ixc = Moment of inertia of channel about its local x-
will add its buckling resistance in the same way a stable
axis, mm4, (in.4)
column supports a "leaning" column. This was not a part of
βx = Monosymmetry Parameter
the University of Florida research. In fact, this author knows
K = Effective Length Factors
of no physical tests of this phenomenon, but it makes one
Mp = Plastic Moment
wonder if lateral-torsional buckling of a crane runway beam
Me = Elastic Buckling Moment
can really occur in a normally-proportioned girder, given the
Mn = Nominal Moment Capacity
lateral support from the other side.
Mr = (Fy – Fr)Sxc
REFERENCES Mp
λ =
American Institute of Steel Construction, Manual of Steel Me
Construction, Allowable Stress Design, 9th Ed., Chicago, IL,
APPENDIX
1989.
American Institute of Steel Construction, Manual of Steel
Example 1
Construction, Load and Resistance Factor Design, 2nd Ed.,
Chicago, IL, 1993. W27×94: Aw = 27.7 in.2, Iyw = 124 in.4 Cw = 21,300 in.6
Clark J. W., and Hill, H. N., "Lateral Buckling of Beams," C15×33.9: Ac = 9.96 in.2, Ixc = 315 in.4 bf = DL = 3.4 in.

48 ENGINEERING JOURNAL / SECOND QUARTER / 1998

(This is the example used by Fisher in the AISC Steel Design πCb   2 
Me =  EJ y GJ  B1 + 1 + B2 + B1  
Guide Series #7, Industrial Buildings, p. 53) KL  
I y J  B1 + 1 + B2 + B12 
Lb = 30 ft 56,600
=
L
Warping Constant, Cwc (from Equation 12)
439 × 8.46  0.904 + 1 + 0.925 + (0.904) 2 
56,600
=
 Ac  360
Cwc = Cw  0.79 + 179
. 
 Aw  = 24,530 in - k or 2044 ft - k (2773 kN - m)

 9.96  M r = FL S xc or Fy S xt
= 21,300 0.79 + 179
.  436
 27.7  = (36 − 10)
12
= 39,690 in.6 (1066
. × 1013 mm6 ) = 945 ft - k (1282 kN - m) or
36
Monosymmetry Parameter, β x (from Equation 11) × 268 = 804 ft - k (1091 kN - m)
12
Lr = 51.6 ft (15.8 m) (from iteration)
I yw 124
I yc = + I xc = + 315 = 377 in.4 (157
. × 108 mm4 ) Mp =
36
× 357 = 1071 ft - k (1453 kN - m)
2 2 12
I yc 377
= = 0859
. ryc =
377
= 4.65 in. (118 mm)
Iy 439 9.96 + 10 × 0.745
R = 2 × 0.859 = 1.718 300ryc 300 × 4.65
Lp = = = 19.4 ft (5.9 m)
β x = 0.87(R – 1)(D + DL/2) Fy 12 36

 3.4   Lb − Lp 
= 0.87(1.718 − 1) 26.92 +  = 17.88 in. (454 mm) Mn = M p − ( M p − Mr ) 
 2   Lr − Lp 
Torsion Constant, J (from Equation 15)  30 − 19.4 
= 1071 − (1071 − 804) 
J = Jw + Jc + bf t1 t2(t1 + t2)  516
. − 19.4 
bf = flange width = 9.99 in. (255 mm)
= 983 ft - kips (1334 kN - m)
t1 = flange thk = 0.745 in. (19 mm)
t2 = channel web thk = 0.40 in. (10 mm) Example 2
K = 1.0
Using Me from Example 1, calculate λ
J = 4.03 + 1.02 + 9.99 × 0.745 × 0.40 (0.745 + 0.40)
= 8.46 in.4 (3.52 × 106 mm) Mp 1071
λ= = =.724
Me 2044
πβ x EI y 2.527β x Iy
B1 = = =
2 KL GJ L J 0.49 < λ < 1.15, ∴ use Equation 22
 λ − 0.49 
=
2.527
× 17.88
439
= 0.9042 Mn = M p − ( M p − Mr ) 
360 8.46  115
. − 0.49 

π 2 ECwc 2556  0.724 − 0.49 

B2 = =
. Cwc = 1071 − (1071 − 804) 
2
L GJ L2 J  115
. − 0.49 

. × 39,690
2556 = 976 ft - k (1324 kN - m)
= = 0.925
(360) 2 × 8.459

ENGINEERING JOURNAL / SECOND QUARTER / 1998 49