Progress reports on experimental metal spars (Airplane Section report)

              r-,1/ '1- v ~. I I ,
lrJ t 11/2 : 6 /s1o NO
A uburn Un1ve1sity L ihr<1 iiPS
111111111/ll Ill~ 11111 11111111/l lllll lllll lllll lllll lllll lllll 111111111111111111
\.... 3 1706 025 85125 !)
File D 52.331 / 64
AIR CORPS INFORMA T
Vol. VI
(AVIATION)
PUBLISHED BY THE CHIEF OF AIR CORPS, WASHINGTON, D. C.
August 26, 1927 No. 590
PROGRESS REPORTS
ON EXPERIMENTAL METAL SPARS
( AIRPLANE SECTION REPORT )
Prepared by A. S. Niles, jr., and E. C. Friel
Materiel Division, Air Corps
Wright Field, Dayton, Ohio
August 13, 1926
Ralph Brown Onrn~,i'• ~ ·!
LIBRARY
UNITED STATES
GOVERNMENT PRINTING OFFICE
WASHINGTON
1927
MAY 2 8 Z013
Non·Oepoitorv
Auburn University
CERTIFICATE: By direction of the Secretary of War the matter contained
herein is published as administrative information and is required for the proper
tran action of the public business.
PROGRESS REPORTS ON EXPERIMENTAL METAL SPARS
INTRODUCTION
1. Late in 1924 a series of "Standard loading con­ditions"
were formulated by the Engineering Division
for use in comparing the merits of different types of
metal wing-spar construction and published in Serial
Report ro. 2450. Early in 1925 a circular proposal
was sent to the industry calling for bids on designs of
metal spars for the "Single-bay observation " loading
of this report. Several of the designs submitted in
response to this circular were purchased, and these
spars as well as several others have been tested by the
division. In all, 41 spars have been tested, not in­cluding
one or two wood rectangles that were tested to
t ry out the test jig.
2. On account of the number of spars that have now
been tested and the interest that has been shown in
this project by the industry it has been considered
advisable to issue a progress report giving the essential
data obtained from the tests made to date. These
data have not been fully digested, so this report can
not be considered as fin al, even with respect to the
spars tested. Another progress report has been
started which will give the results of t he complete
study of these spa rs, but it will be several months
before it can be issued, and the information that can
be embodied in this present report is considered of
sufficient immediate value to warrant issuing it at
this time.
3. The principal subjects considered in this report
will be the progress made to date, the present status
of the project, proposed developments, and a dis­cussion
of certain auxiliary questions that have arisen
in connection with the tests.
PROGRESS MADE TO DATE
wood spars were tested as examples for the officer's
school.
5. The tests made to date have allowed several inter­esting
conclusions to be made, even though the data
obtained have not been thoroughly studied. These
conclusions will be mentioned in the next few para-
~~~- '
6. The most striking result of the tests is in regard to
the liability of the spars to fail by lateral buckling of the
compression flange. The wood spars did not show the
slightest tendency to buckle, even though supported
laterally only at the center of the test length. Before
the design of these spars was completed their compres­sion
flanges were investigated as pin-ended columns
between the points of lateral support, and the section
designed for the test spar without taking this factor
into consideration was found to be satisfactory for this
criterion. With the metal spars this was not found to
be the case. All of these spars, which were tested with
a single lateral support (as specified in the design data
furnished the designers) , failed by lateral buckling of
the compression flange except those two types in which
this action was specially considered in the design.
Many of the types of metal spars tested failed by buck­ling
of the compression flange, even though the lateral
supports were placed at the third points of the specimen
instead of only at the center. In these cases the load
at failure was increased, though not as much as was
expected.
7. The increase in strength obtained by the use of
additional supports raises some interesting questions.
If the compression flange be regarded as an Euler
strut, reducing the unsupported length by one-third
should increase the allowable load in t he ratio of 9 to
4, or more than double it. The actual percentage in­creases
were 14, 18, 19, and 21. These are in close
4. Forty-three spar tests have been made by the agreement with each other, but far below the 125 indi­Material
Section using the jig devised for the "Single- cated by the theory. In one case the spar was sup­bay
observation" loading. Thirteen of these tests ported at the fifth points. Here the Eul er strut theory
have been made on wood and 30 on metal spars. A sum- would indicate that the strength against lateral buck­mary
of the results of 40 of the e tests is given in the ling would be increased in the ratio of 25 to 4. It was
appended tables. The other three tests are not actually increased only 42 per cent. This was the
recorded in this report, as the weights of the pars in type in which third-point support increased the strength
question are not known, the spars having been re- 19 per cent. This problem has just been touched, and
paired for further tests instead of having the end fit- it is hoped that further study of the test data will per­tings
cut off so t hat the remaining 7-foot section could mit the formulation of a satisfactory theory to account
be weighed. The relatively large number of tests on for the facts. All that can be said definitely at present
wood spars were made for the following reasons: It is that the increase in strength due to an increase in
was desired to test solid rectangular, I , and box spars the number of lateral supports is nowhere near as great
to serve as a standard of comparison with the metal; as was expected. This may be due to the fact that
it was desired to check the action of the jig, using cheap the compression flanges are short struts, and not
wood spars rather than expensive metal spars; and two long ones, as implied in our present theory, or some
(1)
other factors may be found to be responsible for this
result.
8. Comparison of the observed values of EI,.,. for the
different spars with the types of failure and the loads
at which it occurred showed that there is a definite re­lationship
between the lateral stiffness of the spar and
the maximum allowable spacing between lateral sup­port
if lateral buckling of th compression flange is to
be avoided. The precise determination of this rela­tionship
has not yet been achieved, but the indications
are that, unless a test spar bas a value of EIYY of at least
7,000,000 inch units, it will fail in that manner if
tested with a single lateral support at the center of the
span, the remainder of the loading being that used in
all of the te ts recorded in thi report. Advantage was
taken of this knowledge, and in four of the more recent
tests those spars showing lower values of EIYY were
te ted with third-point lateral supports. In two of
these case failure by lateral buckling of the compres­sion
flange was avoided, but in the other two it oc­curred.
It is interesting to note that the two which
did not fail laterally had somewhat higher EIYY values
than the other two. The values of Ely,. were obtained
before the main te t by a bending test with the spar on
its side.
2
9. On the basis of the tests IDllde to date, it is
apparent that the wood box and I are superior to any
of the metal types. This is not necessarily an inherent
advantage, but is certainly due, in part at least, to
the fact that wood construction bas been studied more
carefully and extensively than metal construction in
this country. As a result, American designers are able
to de ign wood spars of the required strength and high
strength-weight ratios with fewer trials and greater
certainty of getting the desired results. Considering
only directly comparable spars, all three conventional
wood boxes and four out of five wood I's gave better
strength-weight ratios than any of the metal types.
The fifth wood I was below only one of the metal
spars in strength-weight ratio. All of the conventional
wood boxes, wood I's, and wood rectangles carried
over 93 per cent of the de ign load, though provided
with only one lateral support. There were five metal
spar that gave strength-weight ratios as high as the
poorest of the wood I 's. Only one of these was tested
under directly comparable conditions and it exceeded
only one of the wood I's and none of the wood boxes
in strength-weight ratio. Three of the others were
tested with two lateral supports, yet none of the three
showed as high a trength-weight ratio as the be t of
the I's or any of the three conventional boxes. The
fifth metal spar showed the highest strength-weight
ratio of al1 the spars tested to date, but that was due
so largely to camber and restraining moments at the
ends of the specimen that the results are not at all
comparable. Of these five metal spars howing high
strength-weight ratios, one carried 104, one 92.5, and
the other three less than 81 per cent of the design load.
Most of the failures of metal spars have been of a
character that indicated that their performances could
be appreciably improved with further study, and it is
important that this study be expedited.
10. European experience regarding the relative
strength-weight ratios to be obtained with wood and
with metal is of no value for indicating what is likely
to be obtained in this country. The development of
the wood spar has not progre sed as far in Europe as
in the United States, while in Europe that of metal
has progressed much further. It is the belief of the
authors that when both materials have been about
equally developed the difference in merit between
individual designs will depend much more on the
abilities of the designers than on the material used,
and a first-class designer will always be able to design
a better spar in wood than a second-cla s designer can
design in metal, and vice versa.
11. Some of the spars te ted had the encl pins offset
from the neutral axis, so that the axial load caused
bending moments of oppo ite sign to those caused by
the ide loads, or the same effect wa obtained by
cambering the specimen. The tests showed that
great increases of both strength and stiffness were to
be obtained by these means, and al o that these features
could be applied just as ea ily and effectively to wood
as to metal construction.
12. The purpose of the division in carrying out this
series of tests on experimental spars is to determine
the relative merits of the different TYPES OF CON­STRUCTION.
In order to do this satisfactorily, it
is es ential that all of the spars tested be designed for
and tested under exactly the same loads. Otherwise,
it becomes necessary to make allowances for the
effect of the variations in load, and it i" impossible to
do this in a really satisfactory manner. Owing to the
looseness of the wording in specifying the loads, some
of the spars had initial eccentricities of the end pins,
or camber, or both, as mentioned aboYe. As the effect
of these features is to reduce the effective external loads
on the test specimen, the specimens having these features
are not considered directly comparable to those with­out
them. The results of the tests on those spars are of
great interest as showing the effect of eccentricities
and camber, but can not be used with the tests on the
other spars to determine the relative merits of the types
of construction used. In many of the table which
follow the e spars are, therefore, segregated in the
group entitled "Eccentrically loaded spars."
13. On testing the spars it wa found that many of
them failed by lateral buckling of the compression flange .
When spars failed in this manner it was not considered
that the test gave a very reliable indication of the
merit of the type of design. If a par in a wing
gets a greater degree of lateral upport from the ribs,
etc., than is represented by a single support in the
center of the test specimen, the te tis decidedly unfair
to the type of construction involved, and the type of
construction should be rated on the basis of te ts of a
spar given more adequate lateral support. On the
other hand, if the test set-up does simulate practical
conditions with reasonable preci ion, it would not be
fair to the type that carried the load with but one
lateral support to make no distinction based upon the
number of lateral supports used. As the question of
bow many lateral support are needed in the test rig
to imulate properly the conditions in a practical wing
i.s unsettled, a number of spars that experience showed
to be weak laterally were tested with two, and in one
case four, lateral supports. In the tables the data on
the pars with but one lateral upport and on tho e
with more than one lateral support are segregated. If
the single lateral support best represents practical condi­tions,
only the r esult on the spar of the first group can be
con idered a representative of the merits of the respec­t
ive types of construction. In this ca e the results of
the tests on the second group are not directly com­parable
with those of the first. Even if two lateral
support prove to give a better representation of the
condition in practice, the results of the two group
are not directly comparable, as everal pars of the
first group could have been designed more efficiently
if two lateral upport c.ould have been a urned in
the design; but the difference i not as great as in the
fiT t case
3
14. The tables giving data on the different te ts in
which the par are arranged in order of merit accord­ing
to some phase of the test results are, therefore,
divided into three section . The first group i beaded,
"Spars tested under the design conditions." These
are the spars without eccentrically located end pins or
camber that were tested with a single lateral support.
The second group is headed, "Spars tested with addi­tional
lateral supports." These are the spars without
eccentrically located end pins or camber that were
tested with more than one lateral support. In all
cases but one there were two lateral support . In this
one case (spar 14C) four lateral upports were used.
This difference in te t conditions is indicated in the
tables by an asteri k after the reference number of
this spar. The third group, headed "Eccentrically
loaded spars," includes all of those spars with eccen­trically
located end pins or with camber. In this group
no division is made according to the number of lateral
supports employed in the test , but an asterisk is
placed after the reference numbers of those (31B and
40A) for which two supports were used. The a terisk,
therefore, indicates that the spar in question had more
lateral supports than the other spars in that group.
15. All of the spars tested are designated by a mun­ber
and a letter. The number refers to the general
type and the letter to the serial number of the par­ticular
spar in that type. Thus spar 31B is the second
spar of type 31. Numbers below 10 are reserved for
type of wood construction, from 10 to 29 for dural­umin,
from 30 to 39 for steel, and from 40 to 49 for
combination steel and duralumin.
16. The following table is a list of the spars tested
with a brief description of each.
TABLE !.- List of spars tested
SPRUCE SPARS
lA and lB. Two plain rectangles of spruce. For
small depths thi section shows up very well
on account of the large form factor, and it
was desired to learn its relative efficiency
when the allowable depth was a great as is
allowed in this series of tests. The failures
of these spars are shown in Figure 4.
2A. A conventional wood box with unequal spruce
flanges and two-ply spruce webs. As with
the other spars of type 2, the end pin was
placed at the neutral axis of the section.
This type of spar is shown at N in Figure 2.
2B and 2C. Spar 2A showed so much excess strength
that two more conventional boxes were
designed and tested. More care was taken
in design in order to prevent excess strength,
but there was no change in type; only a
change in detail dimensions.
3A. This was the first of a series of spruce " T box"
spars tested to determine the effect of camber­ing
a spruce box. It differs from the con­ventional
box in that strips of spruce nick­named
"ears" are glued to the plywood webs
on each side of the compression flange. These
ears were tapered from the points of applica­tion
of the side loads to a point about 3 inches
from the end fittings, where they were cut off.
As the pin through which the axial load was
applied were located on the neutral axis of the
end cross ection, the result of this arrange­ment
was to introduce a beneficial eccentricity
of application of the axial load in that part of
the spar in which the stresses were large. As
the top and bottom urfaces of the spar were
parallel planes but the locus of the neutral
axes of the cross sections was a curved sur­face,
the arrangement might be described as a
hidden camber. On account of this feature,
the spars of type 3 are classified as " Eccentri­cally
loaded spars." The results of the tests
on them are not considered as representative
of what can be clone in practice in box-spar
design so much as illustrations of the practica­bility
and effectiveness of the principle of
camber as applied to box spars. This type
of spar might be described as a double I
spar of the general type used in the Curtiss
0-1 and other designs, except that the ear
are omitted on the tension flange. one of
the type 3 spars are shown in the figures,
but it is believed that their construction
can be visualized from the description given
above.
3B. Spar 3A failed under a relatively low load by
crushing of that part of the compression
flange located between the plywood webs about
midway between one end and the adjacent
side load. As the failure was believed to be
due to failure of the glue joint to carry load
into the ears, spar 3B was made with the ears
extending to the end fittings instead of stop­ping
3 inches short of that point. In other
respects it was the same as 3A.
3C. This spar differed very little from 3B, the detail
dimensions being varied a little in the hope
of obtaining a spar of this type that would
carry the required load without failure of
the plywood web to transmit load into the
ears.
4A, 4B, 4C, and 4D. These are all conventional spruce
I beams differing slightly in proportion , but
all routed out of a single stick of wood.
Two of them are shown at Hand I in Figure 2;
the failure of spar 4A is shown in Figure 4.
4E. A conventional I beam differing from the other
four only in that, instead of being routed
out of a single stick of wood, it wa made up
of three pieces glued together. One piece
formed the web and the other two the flanges
DURAL UMIN SPARS
IOA. A duralumin box composed of two flat plates
for the cover plates and two channels formed
from sheet material for the webs. The
webs were tiffened by Yertical channels
formed from sheet, one leg of the e channels
being riveted to each web. This spar is
shown at G in Figure 1.
lOB. This spar differed from lOA only in that the
rivets connecting the webs to the cover plates
were spaced more closely.
IOC. In this spar an attempt was made to improve
the efficiency of the design by increasing the
thickness of the cover plates and decreasing
that of the webs. The tiffeners were vertical
but lightened by flanged holes. This spar
is shown in Figure 4.
lOD. The dimen ions of this spar were the same as
those of lOC, except that the web stiffeners
were arranged in the form of the web of a
Warren truss. It is shown in Figure 4.
IOE. This spar was like lOC, except that flanged
lightening holes were put in the webs. It
also is shown in Figure 4.
4
llA. The chords of this spar were round duralumin
tubes. The webs were rather heavy fiat
plates with rather large lightening holes un­fianged.
The connection between webs and
chords was made by small bolts passing
through the chord tube and both web plates.
Vertical channels were used between adjacent
lightening holes as web stiffenners. The
lightening holes were circular near the ends
of the specimen where the shear is large and
oblong in the central portion where the shear
is small. The failure of this spar is shown
in Figure 4.
llB. This spar differed from llA only in that screws
similar to wood screws were used to connect
the webs to the chords. It is shown at B in
Figure 1.
12A and 12B. Extruded duralumin I beams with
curved flanges and bulbs, as shown at E in
Figure 1. The failure of 12A is shown in
Figure 5.
13A and 13B. Extruded duralumin I beams with
beveled flanges, as shown at D in Figure 1.
Both type 12 and type 13 were shallower than
the 634 inches allowed by the specification.
This was due to the fact that at the time they
were ordered manufacturing facilities did not
permit the extrusion of a deeper section.
14A. This spar was a duralumin plate girder made up
of a duralumin sheet web stiffened by small
vertical angles and "bulb T " extruded flanges.
The vertical leg of the T was slotted so the
web plate could be placed on the plane of
symmetry. This spar is shown at F in Fig­ure
1.
14B. This spar differed from 14A in that, instead of
slotting the extruded T section to receive the
web plate, that plate was riveted to one side
of the vertical leg. In order to keep the
weights of these two spars as nearly the same
as possible, the side of the T was planed down
an amount equal to the thickness of the web.
This spar is shown at A in Figure 1 and its
failure in Figure 5.
14C. This spar was of the more conventional plate
girder type, the flanges being composed of
two angles. The two angles used, when
taken together, have the same cross section
as the bulb T used for spars 14A and 14B.
Spar 14C is shown at Kin Figure 2. ·
14D. A plate girder similar to 14B with its eccentrically
located web plate, except that the amount
of material in the bulbs of the extruded T
flanges was increased.
15A. A duralumin channel Warren truss. The chord
members were single channels with spacers
tying the free edges together to prevent local
failure. Each web member was composed of
two channels facing each other. About at
midheight of the beam the free edges of these
web channels were connected by small plates
to prevent local buckling. All of the channels
used in this spar had a ratio of width of back to
length of leg of about unity. The connections
were riveted. Probably through oversight
the depth of this spar was made 6% inches
instead of the 634 specified. This difference
resulted in the spar giving better test results
than if this error bad not been made, and,
therefore, the results for this spar are not
directly comparable with those of the others.
This spar is not shown in any of the photo­graphs.
16A. Another duralumin channel Warren truss differing
in many details from 15A. Single channels
were used for both chord and web members,
and the ratio of width of back to length of leg
was about 2 in all cases. On account of the
short legs, no spacers were used to tie the free
edges of the channels together and prevent
local buckling. The failure of this spar is
shown in Figure 5.
17A. A duralumin "trussed web dumb-bell" spar. It
is shown at S in Figure 3, and in Figure 5.
It should be noted that the web of Warren
truss type is composed of members of stamped
sheet.
I 7B. This spar was like I 7 A, except that the web
members were composed of duralumin tubes
flattened at the ends. It is shown at T in
Figure 3, and in Figure 5.
ISA. A duralumin "hourglass" type spar similar to
some of the steel spars developed in England.
It is shown at M in Figure 2.
I SB. As ISA failed by lateral buckling at a relatively
low load, the second hourglass spar, ISB,
wa made wider, but otherwise the same.
Its failure is shown in Figure 5.
I9A. A framework of duralumin tubing. The com­pression
chord was composed of two round
tubes connected by a shallow channel so the
whole chord would act as a unit in resistance
to lateral buckling. The tension member
was composed of a single round t ube. The
webs were composed of a number of small
round tubes pinned to the chord tubes.
These web members were in fo ur planes,
each compression chord tube being connnected
to the tension chord by two sets of web mem­bers.
The construction of this spar is shown
at J in Figure 2. As this spar was cambered
by assembling it in a slight ly arched shape,
and a lso had eccentrically located pins in
the end fitting, it is classed among the eccen­trically
loaded spa rs and the test r esults are
not comparable with those in the other two
groups. In the first test the compression
web members near one end fitting failed under
a low load and the spar was returned to the
manufacturer for repair. Heavier web mem­bers
were put in and the spar tested again. The
spar in the form first tested is r eferred to in
the tables as I9AI, and the repaired spar as
I9A2. The photograph of this spar in Figure
5 was taken directly after the first test.
STEEL SPARS.
30A. A steel channel Warren truss. This spar was
very similar to spar 16A, but constructed of
heat-treated alloy steel sheet. On account
of the greater density of steel, this spar was
much narrower than 16A and had such a low
value of Eln that it was tested with two
lateral supports. It is shown at R in Figure 3.
30B. T his spar was like 30A, except that the joints
were made by welding after heat treatment
instead of being riveted. It is shown atQ
in Figure 3 and its failure in Figure 5.
5
31A. A welded chrome-molybdenum steel tube War
ren truss. On account of the small allowable
total depth, the chord members were made of
elliptical tubing, though the web members
were made of round t ubing. This spar is
shown at C in Figure 1.
31B. In manufacturing spar 31A the manufacturer did
not make sufficient allowance for shrinkage
of the spar after welding, and the first spar
constructed was found to be more than an
eighth of an inch below the specified total
depth. In spite of this defect, it was tested,
being designated 3IB.
COMBINATION STEEL AND DURALUMIN SPARS
40A. A spar similar to type 31, the chief difference
being that the chords were made of elliptical
duralumin tubing. As dural chords can not
be welded to steel tube webs, the web mem­bers
were welded to sheet steel "saddles"
and the chord members pinned to these sad­dles.
This spar is shown at L in Figure 2
and also at 0 in Figure 3.
17. In Tables II and III there is a column headed
"Figure and letter,'' which gives a reference to the
figure showing the spar in question, or one very similar
to it, or, where the reference is to one of the pictures
in the first three figures, this column gives the figure
number with the letter designation of the picture in
question. Thus 2M is a reference to picture M in
Figure 2. Only the figure number is given in refer­ence
to Figures 4 and 5, as the spar numbers are given
on those figures.
I S. T able II gives the observed data on the spars
described in Table I. The column headings are self­explanatory,
with the possible exceptions of columns
11 and I2, which are, respectively, the vertical and
lateral EI values obtained from preliminary bending
tests. The moment of inertia about the horizontal
axis is ·denoted by I .. and that about the vertical
axis by I ,y. Thus EIYY is indicative of the stiffness of
the spar in the lateral direction. By comparing the
EIYY values of column 12 with the t ypes of failures
of column 4 it may readily be seen that all the spars
that failed by lateral buckling had very low Ein
values; that is, below 7,000,000-inch units. It is
interesting to note that in a few cases even the use
of additional supports did not prevent this type of
failure.
7
67830-27-2
8
9
Flo. 4
ET-0£
01
11
TABLE IL-Observed data on tests of experimental spars
Failure Deflection and EI
Figure Num- Ulti-
~~~ and Type ber mate Weight
letter sup- load 7 feet Ulti- Type Location ports mate 10,000 pounds Eixx Ein
load
-- -----------
1 2 3 4 5 6 7 8 9 10 11 12
WOOD SPARS
lA 4 Rectangle __________ Horizontal shear ______ -------------------- 20, 6251 16. 3 !. J05 0. 465 69, 000, 000 8, 440, 000
IB 4 _____ do _____________ Compression followed -------------------- 20, 625 J6. 5 1. 280 . 455 60, 200,000 7, 290,_000
by tension.
2A 2N Box_-------------- Crushing compression Center. ___________ 24, 690 J5. 0 1. JOO . 353 85, 700,000 11, 770, 000
fiange.
2B -------- ____ _ do ___________ __ _____ do ________________ _____ do ____________ 21, 700 13. 8 1.100 . 380 85,800,000 JO, 300, 000
2C -------- _____ do _____________ _____ do ________________ _____ do _______ __ ___ 22. 380 J3. 5 I. 550 .382 78, 330, 000 8, 790, ()()()
3A -------- Box T ____________ _____ do _______________ _ End of ears _______ 15, 400 J0.0 .600 . 288 64, 100, 000 J2, 250, 000
3B _____ do _____________ 0 Jue shear followed by J7~" from end 14, 070 JO. 5 . 430 . 275 68, 650, 000 11, 900, 000
crushing com pres- pin.
sion flange.
3C -------- ___ __ do _____ ________ Crushing compression 16" from end pin __
flange.
18, 110 10. 7 .600 . 270 67, 500, 000 J2, 250, 000
4A 4 I beam ____________ Horizontal shear ______ ---------- -- -------- J9, 450 J3. 7 . 287 73, 500, 000 JO, 220,000
4B 2H ____ .do _______ ------ ____ .do. __ ------------- ----------------- --- 19, 750 13. 3 . 3J2 74. 300, 000 JO, 680, 000
4C 2I ____ .do _______ - ----- Crushing compression Side load point ____ 2J,320 J3. 3 !. 300 . 465 75, 100, 000 9, 010, 000
flange.
4D -------- _____ do _____________ _____ do ________________ Center ___________ 18, 650 12. 6 1. 200 . 525 66, 700, 000 7,830,000
4E -------- _____ do _____________ Horizontal shear. _____ -------------------- 20, 750 14. J .327 84, 700, 000 J0, 100, 000
DURAL SPARS
JOA 2G Box_ ________ : ______ Tensile rivets of com- JO" from center ___ 19, 375 14. 5 0. 685 o. 292 75, 200, 000 8, 180, 000
pression flange.
lOB -------- ___ __ do ___ __ ____ ____ _____ do ________________ 8~" from center._ 20, J85 14.6 . 950 . 280 91, 200, 000 11, 970, 000
lOC 4 _____ do_ , ___________ ____ . do. _-------------- 10" inside load J5, 900 13. .296 86, 600, 000 JO, 970, 000
point.
IOD 4 ____ .do ____ --- ------ ----.do._---------- -- -- 19" inside load 17, 010 13. 8 . 650 . 287 88, 200, 000 11, 710, 000
point.
JOE 4 _____ do _____________ _____ do ______________ __ 11" inside load 16, 790 12. 7 .332 90, 600, 000 9, 310, 000
point.
HA 4 Tube and plate ___ L ateral buckling ______ ----- --------------- 1 15, 860 13. 4 1.400 . 540 38, 004, 000 4, 875, 000
JIB lB _____ do _____________ Buckling plane of load_ 20" inside _________ 2 16, 250 12.4 1. 750 . 537 54, 600, 000 5, 340, 000
12A lE,5 Bulb r_ ____________ Lateral buckling ______ -------------------- 1 J5, 875 15. 9 . 600 . 287 89, 000, 000 4, 390, 000
l2B IE ____ .do ____ --------- ____ _ do ________________ -------------------- 2 J9, 240 J6. 0 . 317 83, 450, 000 4, 500, 000
l3A lD Bevel r_ _______ ____ _____ do ________________ -------------------- l 14, 375 J4. 5 .580 .350 72, 800, 000 4, JOO, 000
J3B ID _____ do _____________ _____ do ________________ -------------------- 2 J6, 425 15. 0 .800 .370 76, 000, 000 3, 875, 000
J4A lF Plate girder ________ _____ do _____ ______ _____ -------- ------------ l 15, 500 13.8 . 550 . 264 78, 300, 000 4, 755, 000
14B JA,5 ____ .do ____ --------- _____ do ___________ ____ _ Center __________ __ 2 JS, 500 J3. 6 . 750 .283 96,000, 000 4, 755, 000
l4C* 2K ____ .do ____ --------- ___ __ do ________________ ----------- --------- 4 22, JOO 16. 2 .800 . 273 106, 700, 000 5, 870, 000
l5A -------- C hannel truss ____ _ Lateral buckling in 13" inside _________ I 19, 290 13. 6 1. 000 .340 81, 100, 000 8, 460, 000
J6A 3P, 5 _____ do _____________ compression fl ange.
Crushing compression 22" inside _________ 18. 500 12. 7 . 950 .312 77, 500, 000 J6, 920, 000
flange.
17A 38, 5 Dumb-bell trussed L ateral buckling in Near center_ ______ 2 16, J70 10.4 . 950 .446 60, 650, 000 5, 525, 000
web. compression flange.
17B 3T,5 _____ do _________ ____ _____ do ________________ _____ do ___ --------- 2 15, 300 10.1 1.000 . 500 61, 300, 000 5, 490, 000
l8A 2M Hourglass ___ ------- Lateral buckling m id- JO~" inside _____ __ 1 15, 320 12. J . 950 . 437 69, 800, 000 4, 742, 000
span, followed by
compression.
18B _____ do _____ ________ Crushing compression Near center _______ J9, 050 13. 6 1. 000 . 343 80, JOO, 000 9, 135, 000
flange.
19Al Tube framework ___ Compression diago nals Between end and 14, 250 8. 5 . 500 . 242 31, 950, 000 8, 320, 000
side loads.
19A2 2J _____ do ____ _________ Lateral buckling ______ 8~" inside ________ 14, 580 8. 6 . 750 . 350 33, 360, 000 8, 260, GOO
STEEL SPARS
30A 3R Channel truss ______ Crushing compression 8" inside_--------- 20, 800 J3. 7 1. 050 0.320 89,000, 000 5, 780, 000
flange.
30B 3Q, 5 _____ do _____________ _____ do ________________ At weld
side.
5~" in- 2 17, 230 J2. 3 . 315 93, 100, 000 5, 820, 000
3JA lC Tube truss _________ L ateral buckling of 26" inside _________ 14, 125 12. 7 .900 . 260 72, 400,000 5, 280, 000
compression flange.
3IB* l C _____ do _____ ___ _____ Lateral buckling ______ ------ -------------- J6, 700 12. 4 .600 .328 74, JOO, 000 5, 380, 000
COMBINATION DURAL AND STEEL
40A· J 2L,30 I •rube truss _________ .! Crushing compression I Center_ -----------1 2 J
J5, 270 I 11. J I 1.250 I o. 472145, 500, 000 I 5, 010, 000
flange.
67330-27-3
12
19. Table III shows the test beams arranged in order
of their strength-weight ratios. In the fir t division
of pars (that is, tho e tested under design <'ondition )
tbe first seven spars in the order of merit were of the
conventional wood box and wood I designs (types 2
and 4). Among the metal spars the dural channels,
Nos. 16A and 15A, rated highest, followed in order of
merit by the dural hourglass ~o. 1 B and the dural
boxes of type 10. In the second division of spars
(those tested with additional lateral supports) the dural
trussed web dumb-bell spars, Nos. 17 A and 17B, and
the steel channel tru s spars, Nos. 30A and 30B,
ranked highest. The first in order of merit was No.
17 A, the dumb-bell with tamped webs; the second,
No. 30A, the steel channel truss with ri\·eted joint .
Although spar 17 A and 17B were supported at the
t hird points, they failed by lateral buckling. The
high strength-weight ratios of both these spars, as well
as 30A, however, denote sufficient merit to warrant
further development, e pecially toward insurance
against lateral failure. It is interesting to note that,
if the effect of the number of lateral supports used i
d isregarded (that is, if the second diYision be rated
on the ame scale as the first), spars Nos. 17 A, 30A,
and 17B will rank very high in order of merit, the
exact order being fifth, sixth, and seventh, respectively.
In the third divi ion of spars (those eccentrically
loaded) the dural tube framework spar Ko. 19A2 and
the wood box T spar No. 30 are approximately equal
and have the highest strength-weight ratios of all the
spars te ted. Because of the effect of eccentricity,
howeYer, as previously mentioned, these spars are not
at all comparable with those of either of the first two
groups. The future articles of types 31 and 40 \\'ill
be in accordance with the general design conditions
of t he majority of par and, hence, the re. ult ing
data will b more comparable.
T ABLE III.- Tesl beams arranged in order of slrenglh-weighl ratio
- ~ - - -----
U llimnte Weight mumate Percentage I - UILirnate Weight of
mtimate load in load, of best Deflection load, l,OOOX 7 reetx
Spar No. Figure and letter
1 2 I
~::::::: :J ~~; ~;:::::::::: ::: 2B ___________ 2N ______ __________
4B ___________
4D __________ 2H, 2L------------
2lI, 2L .. ----- - ----
4l6EA-- _-_-_-_-_-__-_-_-_- 2H, 21. .. ----------
4A __________ 3P, 5- ----- ----- ---
4----- - ------------
l5A---------- ------- ------------ -
18B . .. - - -----
5 _________________
JOB __________
2G -- - - --- ------ ___
lJOOAE ._._._- _-_-__-_-_-_- 2G ------ -- -- - - - ---
4------------------
IA ... -------- h;c:::::::::::::' 18A .. . -------
!lOBD .. _._ -_-__- -_-_-_-__- 4----- --- -- ---- ---- 4------ --- ---------
llA ..•.. .• . .. 4- - - --- ---- -------·1
IOC ... . . - - --. 4----- --- --- ----- __
14A ___ ___ ____ tF ---- -- - --- - -- - - -
12A ______ __ _ IE,5 . .. . .. • . __ __ __
13A ... -------l 1D .. . ...... . • ... . .
1
-
17A ... - --- -- -
3S, 5 ___ ____ _______
3l70BA .__• -_-__-_-_-·_-__- 33TR, . 5. -_-_-_-_-__-_--_-__-_--_-__-
30B _______ ___ 3Q, 5 _ ___ _________
14C• ___ _____ 2K ____ __ __________
l4B ____ ___ __ _I
IA, 5-- -- ----- -----
IJB ... ----·--1 lB _________ _______
l2B __________ n-: ______ __________
13B ... - ------
JD ______ __________
19A2--------- 2J .-- - - ------------
3C ___________ --- ----- -- - ------ -- -
19AL .. ------ 5 .. -- - - ---- - -------
3A ... ------ -- ----- ---- - - - - ------ -
40A • ------- -- 2L, 30- - ----------·
31B• ________ I C---------·------
38 •...•. --- -- -- ------- - --- - ------
3IA ___ ______ JC • •. • ....... .....
load percentage weight I load, a 10,000
of design 7 feet I foot of 7 feet weight pounds I load i ratio
3 4 5 6 I 7 I 9 I
SPARS 'r ESTED UNDER D ESIGN co -DITIONS
22, 380 111. 9 13. 5 1 ). 93 1,658 100.0 0.382
24, 690 123. 4 15. 0 2.14 1, 648 99. 4 .353
21 , 320 106. 6 13. 3 I. 90 I, 603 96. 7 .465
21, 700 108. ,5 13.8 L 97 1,572 94.. 8 .380
19, 750 9 . 7 13.3 I. 90 l, 48.'; 9.6 .312
1 ,650 93.2 12. 6 1.80 1,480 9. 3 .525
20, 750 103. 7
14.1 I 2.02 1,472 .8 .327
l ,500
I
92. 5 J2. 7 1. 81 1,457 87. 9 . 312
19, 4.50 97. 7 13. 7 I. 96 1,420 8.5. 6 . 287
19, 290 96.4 13. 6 I. 94 l, 419 8.5.5 .340
19, 050 95.3 J3. 6 I. 94 l , 401 84. 5 . 343
20, 18.5 100. 9 14. 6 2.0 J, 383 83. 4 . 280
19, 375 96. 9 14. 5 2. 07 l,336 0. 6 . 292
16,790 83. 95 12. 7 1. l !,322 79. 7 . 332
20, 625 J03. l 16. 3 2. 33 l, 282 77. 3 .465
15,320 76. 6 12. J 1. 73 l , 266 76. 4 . 437
20, 625 103. l 16.5 2.3f> I, 250 75. 4 . 455
l7, OJO 8.5. 1 13. 1. 97 1, 233 74. 4 . 287
15, 860 79. 3 J3. 4 1. 91 ! , J82 71. 3 . 540
15, 900 79. 5 13.8 1. 97 1, 152 69. 5 .296
15, 500 77.5 13. l. 97 l , J23 67. 7 . 264
15, 75 79.4 11;. 9 2. 27 1 999 60.3 • 287
14, 375 71. 9 14. 5 2. 07 990 59. 7 . 350
PARS TES'l'ED WITH ADDITIOXAT, LAT ERAL S"CPPORTS
-
16, 170
20,800
15,300
17, 230
22, lOC
18,500
16, 250
19, 240
16,425
14, 580
18, !JO
14, 250
15,100
J5, 270
16, 700
14, 070
14, 125
-
80. 9 10. 1 1.49 l, 555 1 104.0 13. 7 J. 96 1, 518
76. 5 10.1 1.44 1,515
86. 2 J2. 3 I. 76 1,401
Jl0.5 16. 2 2. 32 1,363
92. 5 13. 6 1. 94
81. 2 1~.4 l. 77 11,,33J6O1 1
96. 2 16.0 2.28 1, 202
82. l 15.o I 2.14 l,095
ECCENTRICALLY LOADED SPARS
72. 9
90.6
71. 2
77. 0
76. 3
83. 5
70.4
70. 6
8. 6
JO. 7
. 5
10. 0
JI. 1
12. 4
JO. 5
12. 7
I. 23
1. 53
l. 21
I. 43
I. 59
1. 77
l. 50
1. 81
1, 695
l,693
l,677
1, 540
1,376
l, 348
1,340
1,112 1
100.0
97. 6
97.4
90. l
7. 7
7. 5
84. 2
77. 3
70. 4
100.0
99. 9
9 9
90.9
1.2
79. 5
79.1
65. 6
0.446
. 320
.500
. 315
. 273
.283
• 537
. 3J7
. 370
o. 350
. 270
.242
. 288
. 472
.328
. 275
• 260
deilection
at 10,000
pounds
--
JO
58. 6
70.0
4o.8
57. l
fl3.3
35. 5
63. 5
59.3
67.8
56. 7
55. 5
72. 0
66. 3
50. 5
44. 4
35. 1
45. 4
c.9. 3
29.4
53. 7
.'i8. 7
55. 3
4!. 0
36.3
65. 0
30. 6
54. 7
81.0
f-i.5. 4
3600.. 37 1
44.'1
41. 6
67. l
59. 9
53. 5
32.3
50. 9
51. 2
54.3
deilectiou
at J0,000
pounds
11
5.157
5. 295
6.1
5. 24~
4. 150
6. 615
4. 611
3. 962
3. 932
4. 624
4. 665
4.
4. 234
4,216
7.5~
5.
7. 50
3. 961
7. 236
4. 08:;
3. 643
4.56.1
5. 075
--
4.638
4. 38'l
5.050
3. 75
4.423
3. 8.\9
6.6.)9
5.072
5. 550
3. 010
2. 9
2. 057
2. 80
5. 239
4. 067
2. 8
3.302
13
TABLE IV.-Test beams arranged in order of idtimate the other spars. The dural box No. l OB, the dural
load, weight of 7 feet X per cent of design load carried channel trusses r os. 15A and 16A, and the dural
Spar No.
. -- \strength-
. weight Relatiw
Ultimate ratioX strength-
Ultimate ~o~d, 11er cent weightXper
load "<:1P:ht design c<'nt of design
of 1 feet I load load carried
cai ried
2 3
tan cling
in Table
III
SPARS TESTED UXDER DESIGN CONDITIONS
20 __ _____ __ ___ I 22.~0 I 1,658 1 1, 658** 100. o (100. 0) 1 1
2A ___________ __ 4C _____ ________ 24,690 1, 648 1, 648** 99. 4 (99. 4) 2 21,320 1,603 1, 603** 96. 7 (96. 7) I 3
2B _______ ______ 21. 700 1, 572 1,572** 94. 8 (94. 8) 4
4E _____ ______ __ 20, 750 1, 472 1 1, 4 72** 89. 6 (89. 6) 7 4B ______ __ ____ 19. 750 l, 485 1, 466 88.4 (88. 4) 5
4A ___ ___ _______
19,450 I 1,420 1,387 83. 7 (83. 7) I 9
lOB ____________ 20, 185
1,383 1
1,383** 83. 4 (83. 4) 12
4D------------ 18, 650 1,480 1, 379 83. 2 (83. 2) 6
15A _______ ____ 19,290 1,419 1. 368 82. 5 (82. 5) IO
16 1\_ ___ ________ 1 ,500 1,457 1. 343•• 81.8 (81. ) 8
l8B ______ ______ 19, 050 l,401 1, 335 80. 5 (80. 5) 11
lOA ____________ 19,375 1,336 1, 295 78. l (78. 1) 13
l~:::::::::::::I 20, 625 1,282 1,282° 77. 3 (77. 3) 15
20, 625 1,250 1,250 .. 75. 4 ~75. 4) 17
!OE ____________ 16, 790 1. 322 1, 110 66. 9 66. 9) 14
lOD ___ ________ 17, 010 1,233 1, 049 63. 3 (63. 3) 18
!SA ___________ 15,320 1,266 970 58. 5 (58. 5) 16
llA ___________ 15,860 1, 182 937 56. 5 (56. 5) 19
lOC ____________ 15, 900 1, 152 916 55. 2 (55. 2) 20
14A __ _____ ____ 15, 500 1, 123 870 52. 5 (52. 5) 21
12A ____ ________ 15,875 999 793 47. 8 (47.8) 22
13A ___________ 14, 375 990 712 42. 9 (42. 9) I 23
SP_\.RS TESTED WITH ADDITIONAL LATERAL SUPPORTS
30A ____________
20,800 I 1,518 1, 518** 100.0 (91. 6)
l4C ___ ------- -- 22, 100 1,363 1 353•• 89.8 (82. 2)
14B ____________ 18, 500 1, 361 i;259 82. 9 (75. 9)
l7A _____ ______ 16, 170 1, 555 l,258 82. 8 (75. 8)
:ioB ____ ______ __ 17, 230 1,401 1,208 79. 6 (72. 9)
17B ______ ______ 15,300 1, 515 l , 159 76.4 (69. 9)
l2B __________ __ 19,240 1,202 1, 156 76. 2 (69. 7) llB ___________ 16, 250 l,310 1,064 70.1 (64. 2)
13Tl _______ _____ 16,425 1,095 899 59.-2 (54. 2) I
ECCENTRICALLY LOADED SPARS
3C ______ ___ ____ ,
19A2 _____ _____ _
19AL __ ___ ___ _ ,
33lLB- _-__-_-_-_-_-_-_-_-__- I
1
43B0 A__*_-_-_-_-_-_-_-_-_-_- -_
31A _____ ______ _
18, 110 I H,580
14, 250 I 15,400
16, 700
15, 270
14,070
14, 125
I
1,693
1, 695
1,677
1, 51-0
1,348
l, 376
l ,340
1, 112
1,534
1,236
1, 194
1, 186
1,126
1,050
913
785
100. 0 (92. 5)
80. 6 (74. 5)
77. 8 (72. 0)
77.3 (71.5)
73. 4 (67. 9)
68. 4 (63. 3)
61. 5 (56. 8)
51.2 (47.3)
---- ----'-----'---'
2
5
fi
l
4 a
8
7
9
2
1
3
4
6
5
7
8
hourglass types are next in order of merit, with very
little difference between them. In the second division
the riveted steel channel truss No. 30A, the dural
plate girders Nos. 14C and 14B, and the dural trussed
web dumb-bell o. 17A are the best. Spar No. 30A
has a large margin over any of the others, due, of course,
to the fact that it carried the design load. In the
third division the wood box spar No. 3C is much
superior to any of the others, due to the greater per­centage
of design load carried. If all of the spars are
considered as one group, the wood spars, the conven­tional
boxes, wood I , and box T rank first in order of
merit. The riveted steel channel truss ranks next,.
being very nearly equal to the wood box T No. 3C.
The dural tube frameworks, Nos. 19Al and 19A2,
rate very poorly because of the low loads carried.
21. Table V gives the test beams arranged in order
of weight. It has little interest, merely showing that
there is no correlation between the total weight and
the strength-weight ratio.
TABLE V.- Tesl beams arranged in order of weight
I
Ultimate Standin
Spar 0. \W7 efiegehtt , P1o fuonodt s, Ulltoiamda te wloeaigdh, t in Tab!~ of 7 feet III
g
l I 2 3 4 5 6 -
SPARS TESTED UNDER DESIGN CONDITIONS
18A ____________ - ---- 12. l 1. 73 15, 320 1,266 16
4lLDA- -_-_-_-_-_-_-_-_-_-_-_-__-_-_-_-_- 12. 6 1.80 18, 650 1,480 6 !OE __________________ 12. 7 1.81 18, 500 1,457 8 4C ___________________ 12. 7 1. 81 16, 790 1,322 14 4B ___________________ 13. ;j 1. 90 21,3W 1,603 3 llA ______ ___________ 13. 3 1.90 19, 750 1,485 5 2c ___________________ 13.4 1. 91 15, 860 1, 182 19 15A _________________ 13. 5 1. 93 22,380 1,658 1 13. 6 1.94 19, 290 1,419 10
418AB_ ___________ _-_-_-_-_-_-_______- _-_- 13. 6 1. 94 19,050 1,401 11 2B _______ ____________ 13. 7 1. 96 19,450 1,420 9 !OD __________ ________ 13.8 1. 97 21 , 700 1,572 4 IOC __________________ 13. 8 1. 97 17,010 1,233 18 14A _________________ 13. 8 1. 97 15, 900 1, 152 20 4E ___________________ 13.8 1. 97 15, 500 1, 123 21 IOA __________________ 14.1 2.02 20, 750 1,472 7 13A _________________ 14. 5 2.07 19, 375 1,336 13 lOB __________________ H.5 2.07 14, 375 990 23 2A ___________________ 14. 6 2. 08 20, 185 1,383 12 12A _________________ 15.0 2. 14 24, 690 1, 648 2 IA ________________ •_ __ 15. 9 2. 27 15, 875 999 22 lB __________________ 16. 3 2. 33 20, 625 1,282 15 16. 5 2. 36 20, 625 1,250 17
20. Table IV shows the spars arranged in order of
the ultimate load multiplied by the percentage of
design load carried divided by the weight of a 7-foot
length. This derived quantity is the strength-weight
ratio penalized by a factor equal to the amount per SPARS TESTED WITH ADDITIONAL LATERAL SUPPORTS
17B _______________ ___ 10.1 1. 44 15, 300 1, 515 17A ________________ _
30B ________________ _
llB _________________ _
10.4 1. 49 16, 170 I 1,555
12. 3 1. 76 17, 230 1, 401
14B ________________ _ 12.4 1. 77 16, 250 1, 310 30A ________________ _ 13. 6 1. 94 18, 500 1,361
13B ___ ______________ _ 13. 7 1. 96 20, 800 1, 518
l2B _________________ _ 15. 0 2.14 16,425 1,095 14C _________________ _ 16.0 2. 28 19, 240 1,202 16. 2 2. 32 22, 100 1,363
ECCENTRICALLY LOADED SPARS
a
1
4
7
(i
2
9
8
5
cent by which the ultimate load is less than the design
load. It is a combined rating of the designer's ability
to obtain precise design with the type of construction
used and to obtain high strength-weight ratios. Since
the criterion is based upon the ability to design to
meet rather than to exceed the design load require­ments,
it was considered desirable in the cases of those
spars which did exceed the requirements to assume the j
percentage of design load carried as 100 in computing ------------------------
the criterion quantity. Column 5 of the table gives 19AL ___ __________ _ 14, 250 I
the relative values of the strength-weight ratio multi- ~~~~~~:::::::::::::::
8.5 1. 21
8. 6 1. 23
10. o I 1. 43
plied by the percentage of design load carried. The 3B ____ ______________ _ 10. 5 1. 50
10. 7 1. 53
11. 1 1. 59
12. 4 1. 77
12. 7 1. 81
14, 580
15,400
14, 070
18, 110
15,270
16, 700
14, 125
1, 677
1,695
1,540
1, 340
1,693
1, 376
1,348
1, 112
3
1
4
7
2
5
6
8
left side of the column gives the values based upon I ~fa•:::::::::::::::::
the maximum of each division. In the first division ~~r:::::::::::::::::
the wood box and wood I show up much better than -------'----'----'------''------'----
14
22. Table VI gives the spar in the order of the
ultimate load divided by the deflection at one-half
the design Joad. This criterion gives an indication
of the tiffnes of the spars, although it is somewhat
of a combined rating in that it penalizes for failure to
carry the de ign load. In the first division the dural
boxes Nos. lOB and lOA, the wood box No. 2A, the
wood I's No . 4A, 4E, and 4B, the dural channel
truss No. 16A, and the durnl plate girder No. 14A
showed up well, the dural box being especially good.
In the second division the dural plate girders Nos.
14 and 14B, the steel channel truss o. 30A, and the
dural bulb I o. 12B ranked highest. In the third
division the wood box T spar No. 3C, the dural tube
framework No. 19Al, and the tee! tube truss o. 31A
were good. The difference in rating between them
was about 10 per cent. The value shown for spar
19A2 is unreliable, as the pin at one end of the speci­men
bent during the test, resulting in a fictitiously
high deflection reading. In fact, the deflection wa
probably less than that for 19Al, as the specimen wa
the spar of test 19Al with the web members increased
23. o one of the criteria used for arranging the
spars in Tables III to VI, inclusive, can be considered
as a complete criterion of merit. In deciding upon the
spar to be used there are a number of nonquantitative
factors to be considered, such as reliability, availa­bility
of material, experience of the workmen avail­able,
etc., and each engineer has his own ideas as to
the r elative importance of the vaTious factors, both
quantitative and qualitative. For this reason no
attempt has been made in this report to devise a single
over-all rating of merit; the authors have been content
to make only ratings that were facilitated by the
numerical character of the data and that were, more­over,
based upon factors that an engineer choosing a
design would undoubtedly want to consider. Each
engineer will have to decide for himself the relative
importance of the various criteria. A further reason
for not attempting to devi e a single quantitative
measure of merit was the f1,1ct that so many types
failed by lateral buckling of the compression flange.
If, as is quite possible, . uch types should be rated only
on the basis of tests in which the specimen had ade­quate
lateral support, it would be decidedly unfair to
rate them finally on the basis of the present tests.
in size.
TABLE VI.-Test beams arranged in order of tdtirnate
load deflection under 0.5 design load 24. In addition to the criteria on which the tables
mentioned above were based, the spars were studied
in the light of other criteria that gave promise of being
useful in deciding upon relative merit. One of the most
obvious criteria is the unit stress developed at failure.
ffitimate
load, 1,000 Relative Ultimate
Spar No. Xd eflcc- load- load,
tion at deflection weight of
10, 000 ratio 7feet
pounds
Standfag
in Table
Ill
--------- -1------------ Th.is was found to be worthless as a measure of merit
in comparing different types of spars, and not as useful
as the trength-weight ratio in comparing pars of the
same type. This was shown by the fact that, of the
spars investigated, the poore t type showed the highest
2 4
SPAR TE TED UNDER DE IGN CONDITIONS
IOB ___ __ _________________ ___ _
2A ---------- ------------ ---- -
4A . ___ --- -- --- - ----- -- _ ----- -
4lO EA __. ______ _--_-_-_-__-_-_-_-_- -_-_-_-_-_-_-_-_-__--_-_
4B .------------ --- --- --- -----
16A __ ----- - -- -- - --- - - -- - - - ---
10 D- - -- - - --- ---- --- -- -- ---- --
14A . ---- --- -- ----- ----- - - -- - -
20 _ - --- --- --- - --- - --- --------
2B _ -- -- -- -- -- ---- ----- - - - - - -­J5A
_ - - ------ - -- ------- - --- - -­l8B
.. ------ - - -- - - --- - - - --- - --
1120A0 __- __-_-_--_-__-_-_--_-_-_-_-__- -_-_-__- -__-_--_-_-_
JOE _______ ______________ ____ _
4 C. _____ . ___ __ -- -__ __ - - -__ --_
Il AB _____- -_-__-_-____ _-_--_-_-_-___________ _-_-_-_-_-_- __
413DA _·_ _- -__- -----_-__--__-_- -_-- -------_-__-_- -_-__-_- -_
I A_------------- -----------­JJ
A.- -- ------ -'--------- ----- -
72. 0
70. 0
67.
66. 3
63. 5
63.3
59. 3
59. 3
58. 7
. 6
57. 1
56. 7
55. 5
55. 3
53. 7
50.5
45.
45. 4
44. 4
41.0
35. 5
35. 1
29.4
100.0
97. 2
9-1. 2
92.1
88.2
87. 9
82. 4
2. 4
81. 5
81. 4
79. 3
78.
77. l
76. 9
74.. 6
70. 1
63. 6
63.1
61. 7
56.9
49.3
48. 8
40.8
1,383
I, 648
1,420
1, 336
1,472
I, 485
I , 457
l , 233
1, 123
1,658
l, 572
1, 419
J, 401
999
1, 152
1,322
1,603
1, 250
1,282
990
1,480
1, 266
l, 182
5
1 ~ unit stresses. This phenomenon indicated that there
9 is a form factor for each type of metal construction
1 ~ similar to the form factor for wood. The more mate-
5 rial that is located inefficiently near the neutral axis
1 of the spar the higher will be the form factor due to
2i the amount f ligh tly loaded material that can relieve
1i the heavily loaded portions of the cross section near
11 its extremit ies. Th.is condition will be particularly
ro marked when there is relatively little heavily loaded
1~ material, · but a beam with relatively little heavily
11 loaded and relatively much lightly loaded material
~~ will clearly be inefficient. In other words, the factors
6 that tend to increase the form factor tend to decrease
}8 the efficiency. As soon as this fact was clearly realized
no further attempt was made to rate the spars on the
SPARS 'l' E STED WlTH A DDITIO ' .. \L LATERAL SUPPORT basi of maximum computed unit stress at failure.
J4C• - --- - --- -- --- ------ -- ----
14 B _ - ---- - ---- - - - - - - - -- -- -- - -
3102AB __- _-_-_-__--_-__- -_-_-_-_-__-_-_-_-_--_-_-__-_--_- 30B _ - --------- --------- - ---- -
13B - ---- - -- --- ----- - - -- - - - --·
17 A.----- --- --- ----- -- -- ----­1J17
BB_ _ _-_-_-_-_-_--_-_-_-__-_-_-_--__-_--__-_-_-_-_-_- -_
1.0
65. 4.
65.0
60. 7
54. 7
44. 4
36. 3
30. 6
30.3
100. 0
80. 7
0. 2
74. 9
67. 5
54. 8
44.8
37.
37.4
1, 363
1, 361
1, 51
1,202
1,401
l , 095
1, 555
1, 515
I, 310
25. As it was thought that the criteria of strength
~ divided by deflection at half-de ign load penalized
2 exce sively the spar that failed at low loads, the first
4 28 spars tested were rated on the basis of strength i divided by deilection under half-ultimate load. It was
3 found that when this criterion was used the second
7 spar of a pair, though obviously better than the first,
_____E_ C_C_E_<,N''l'R_I_C~A-L_L_Y_7_L_O,_A_D_E_<D -~SPA_R_S_~--- usually showed up as the poorer. tudy of the typical
30. ----- --------- - - --- - - -- -- -
19AL - --- ---- ------ --- -- - -- - - 31A __ ___ __ __ ~ --- __ __ - -- -- -- - -
3A _ ----- - ------------------ - -
3B .. -- - --- - - - - - ---- - - - - - - ----
31B* ------ - - - - - ------ ----- --­l9A2_
------ ---- -- - -- - --- - - - --
40A • -- - -- ---- -- - ------ -- -----
67. l
59. 9
54. 3
53. 5
51. 2
50.9
41. 6
32. 3
100.0
89. 3
80. 9
79. 7
76. 3
75. 9
62.0
48. l
l, 693
1, 677
I, 112
I, 540
1,34.0
J, 348
1,695
1,376
load-deflection curves showed that this was to be
2 expected, as that cu rve is not a straight line but one
3 of decreasing slope. This criterion has, tqerefore, been i dropped also.
o 26. The fir t 2 tests were also studied by the
~ criterion of weight multiplied by the deflection under
half the design load, but no valuable results were
15
obtained from it, and this criterioll is no longer being
used.
27. On the whole, the wood box shows up as the
best of all the types tested, followed closely by the
wood I. Of the metal types, the type 16 dural channel
truss appears to be the best. If it were not for the
problem of lateral support, the best metal types would
be the steel channel truss, type 30, and the duralumin
tru sed web dumb-bell, type 17. These types showed
up so well that they hould be developed further.
Comments Oll each of the types a rc made in the follow­illg
section of this report.
2 . It has been the policy of the division to consideJ·
the fir t spar of any type as a "sight ing shot" and to
avoid coming to any defin ite coDclusions as to t he
merit of a type until at least two spar have been
te ted. Final conclusions in regard to t he relative
merit of a type will not be made until it is believed
t hat it has been developed to the utmost of its capabili­t
ies or until some fundamental defect appears that
precludes its further consideration. Readers of this
report should take the same attitude and be very
careful in the judgments t hey make on a type when
only one spar has been test ed or when all of the spars
te ted were manufactured before the first of the type
has been tested. The types in which this is the case
a re Nos. 15, 16, 17, 19, 3Q, 31, and 40.
PROPOSED DEVELOPMENTS
29. Wood types, 1, 2, 3, and 4.-No furth er develop­ment
of types 1, 2, and 4 is contemplated at present,
as it is considered that the data already obtained are
satisfactory for use as standards for rating the variou·s
de i,gns of metal spars. The development of type 3,
the T box, is being continued, as the design of the
three spars already tested bas brought to light some
interesting facts about the design of T box and double
I spars that should be investigated further. Although
t his type, as designed for the current series of tests, is
con idered impracticable, it has several features that
can profitably be applied to special design problems,
and is, therefore, well worth development.
30. Duralumin boxes, type 10.- The first two dura­lu.
min boxes, spars lOA and lOB, showed up very well,
and it was thought that by lightening the web and
increasing the amount of material of the cover plates it
could be improved. Spars lOC, lOD, and lOE were
built to test this conclusion and also to study the
efficiency of different types of web stiffening. These
three spars did not show up as well as the first two, and
it was concluded that if any appreciable improvement
over spar lOB were to be obtained it would be necessary
to t urn down the edges of the cover plates, making
them shallow channels, or to provide stiffening in some
other manner. Three pairs of boxes with turned-down
cover plates and different types of web s tiffening have
been ordered, and also a pair of spars with the cover
plates made of two pieces of sheet arranged as a hex­agonal
cell. As the last three spars tested failed by
local buckling of the cover plates, they do not give
very much light on the problem of web stiffening.
Apparently there is little difference between the use
of vertical and diagonal stiffeners, except in regard to
ease of manufacture, though the vertical stiffeners used
in connection with flanged lightening holes gave the
best strength-weigh t ratio, but it is not believed that
this would be the case if the cover plates had not
failed locally. The results of this pha e of the tests
might be expressed by the statement that the flanged
lighteniDg holes increased the ability of the web to
stiffen the cover plates again t local failure, but whether
they would stiffen the web itself more efficiently if the
cover plates were stiffened by other means is doubtful.
31. Duralumin boxes with shallow channel cover
plates have been ordered with three types of web
stiffening. The first type has t he webs stiffened with
vertical corrugations spaced at about 6 inches center
to center. The second type will have vertical channels
riveted to both webs, like spars lOA, lOB, and lOC.
The third type will be like the second, except that
alternate channel stiffeners will be replaced by small
vertical angles riveted to the outsides of the webs. In
practice these angles would be used for connecting the
ribs to the spar. Two types of web stiffening will be
tried out on the spar s with hexagonal cell chord .
The first type will have the webs made of corrugated
sheet with the corrugations vertical. The second type
will be similar except that the corrugations will be
horizontal.
32. Duralumin tube and plate spars, type 11.- This
type showed up very poorly in test on account of the
inefficient location of the material of the round tube
chord members, too little material being highly tressed
and too much being used near the neutral axis. It
might be improved by flanging or omitting the lighten­ing
holes or by flanging over the extreme edges of the
web plates, but it is not believed that these measures
would be sufficient to give good r esults as long as the
round tubes are used for the chords. Development
of this type has, therefore, been dropped.
33. Extruded ditralumin I beams, types 12 and 13.­When
the spars of types 12 and 13 were purchased it
was impossible to have them manufactured with a
depth of 67;1'. inches, so those tested were only about 5%;
inches deep. When this disadvanta.ge is considered
it will be seen that they showed up very well in respect
to stiffness and reasonably well in regard to strength.
The bulb flanges of type 12 showed up to much better
advantage than the beveled flanges of type 13. No
further development of type 13 is cont emplat ed, but
study of type 12 will very likely be resumed when it is
possible to obtain spars of full depth 1l.nd when study
of the plate-girder type has indicated the cross section
that is desired.
34. Duralumin plate girders, type 14.- In the earlier
series of tests to develop a suitable metal ·spar for use
in large airplanes one of the best spars submitted was
a duralumin plate girder furnished by t he Aeromarine
Co. This plate girder differed from standard struc­tural
steel girders only in material and the fact that
bulb angles were used for the flanges instead of ordinary
angles. As the chief defect of the Aeromarine spar was
its lack of depth (it was less than 7 inches deep, while
the others of the eries were 15 inches deep), it was be­lieved
that this type would show up very well under
the conditions of these tests. At the same time it was
16
believed that an advantage would be gained if each
chord were made of a single bulb T instead of two bulb
angles. Three plate girders were ordered, one with the
vertical leg of the T slotted to receive the web, one
with the web placed slightly off center so that it could
be ri veted to one ide of the T , and one with the chord
made up of two angles. It was hoped that this would
allow a determination to be made of the relative merits
of the two types of chord members and the two methods
of locating the web. In the first test the spar failed by
lateral buckling of the compression flange. The sec­ond
spar, that with the eccentric web, was, therefore,
tested with two lateral supports instead of only one.
Tested this way it carried a larger load but still failed
by buckling laterally. The t hird spar, the one with
the two angle chords, was tested with four lateral sup­port
. Thi spar carried the design load and failed
suddenly, forcing aside the lateral braces. The result
of the tests was to show the lack of torsional rigidity in
this design and the effect of increasing the amount of
lateral support on the strength of the spar. The fac­tors
on which it was hoped to obtain data did not
affect the failure so far as could be noticed.
35. After testing spars 14A, 14B, and 14C, a fourth
plate girder, 14D, was ordered. In this spar the size
of the bulbs was increased to obtain greater lateral
and torsional stiffness. The eccentric location of the
web was used for simplicity of manufacture. This
spar has been tested twice, but each time the connec­t
ion between the end fittings and the spar web has
failed , and so the tests have not been considered repre­sentative.
This spar is now being repaired for a final
test. After a satisfactory test has been obtained from
it another spa r of t he same general type will be pur­chased
and the development of this type continued.
It does not seem possible to obtain much torsional
strength with a plate girder, but it can be developed
to have adequate lateral strength, the only question
being whether or not this lateral strength can be ob­tained
without excessive weight. The I section is so
much used, either in the form of an integral beam or as
a plate girder in structural steelwork, and has shown
up so well on a few wing designs in which it has been
used, that it is considered desirable to continue the
study of the type until its relative merits have been
thoroughly determined.
36. Duralumin channel trusses, types 15 and 16.-
0f the two type of duralumin trusses already tested,
type 16 has given t he best results. One of the chief
diffi culties of any such design is the likelihood of failure
by local buckling of the free edges of the channels. In
type 15 this was guarded against by mall spacers
tying the channel legs together. In type 16 an
attempt was made to so proportion the channels that
they would not fail by such local buckling before the
spar would fail as a whole. This attempt was suc­cessful
with respect to spar 16A, so it is thought that
the greater simplicity of the resulting design makes
this definitely the better type. Six more spars of type
16 are being designed to determine the effects of small
detail changes. On the whole, this is considered the
best type for practical design that has yet been tested,
and considerable study is contemplated for the de-termination
of the allowable stresses and the proper
proportions for the channels. Another type very
similar to type 16 has been ordered and one spar tested.
In this type, instead of making each web member a
single channel, tension web members were made up of
two small channels. This spar failed by buckling of
the free edges of some of the compression web members.
These members are being reinforced, and t he par will
then be retested. Whether this type will turn out to
be better than type 16 or not remains to be determi ned.
37. Duralumin trussed web dumb-bells, type 17.-This
type is the result of an attempt to make a spar with a
single web out of sheet material. The two spars
tested differed only in the character of the web mem­bers
used . Both had little lateral stiffness, and failed
by lateral buckling of the compression flange, although
tested with third-p oint support. Both showed excel­lent
strength-weight ratios for metal de igns, indicat­ing
that the type is well worth development. Though
no new spars of t his type have yet been ordered, it is
expected that this will be done as soon as the method
for increasing the lateral stiffness has been agreed upon
with the designer. Of the two types of web, that
made of stamped sheet appeared to be more efficient
than that made of tubes with flatten.ed ends. Two
similar types have been ordered from another designer.
In these the shape of the chord members will be some­what
different, and both will have webs of a continuous
sheet of duralumin. In one case the web will be
reinforced by small D-shaped members arranged like
the web members of a Warren tru s; in the other the
web sheet will be stiffened by horizontal corrugations.
38. Diiralumin hourglasses, type 18.-The first hour­glass
failed at a low load by lateral buckling. The
second article, spar 18B, was made "ider and de­veloped
a good strength-weight ratio. It failed by
collapse of the compression chord under more than
90 per cent of the design load. It is proposed to test a
revised design in which the two webs would be tied
together at midheight by some kind of spacers instead
of bringing the webs together. The type of spacer
to be used is still under discussion.
39. Duralumin tube triangular frameworks, type 19.­The
first spar of this t ype, 19A, failed by buckling
of the compression web member in the first test.
In the second test, when those members had been
replaced with heavier ones, t he failure was caused
by lateral buckling of the compression flange. A
second article has been received and is now awaiting
test. This article differs from the first in that the
main t ubes a re la rger and that the channels which
connected the compression tubes of the first spar have
been replaced by a latticing of small tubes. What
further development of this type will be attempted
will depend on the results of the tests of this second
spar.
40. Steel channel trusses, type 30.- Two steel Warren
type channel trusses have been tested. Spar 30A had
riveted joints, while the joints of spar 30B were welded
after heat treatment. AB t he welded spar had much
less strength than t he riveted one, no further study of
welded spars of heat-treated material is contemplated.
The welded spar was expected to be the weaker, but it
17
was built and tested in order to determine the relative
strength quantitatively. Not only is the probable
weakening due to welding too great to be allowed in
airplane construction but the exact amount is so uncer­tain
that such construction is very undesirable. The
riveted type will be further developed. Spar 30A was
so lacking in lateral stiffness that it was tested with two
lateral supports. The first step contemplated is to
widen the spar by increasing the width of the backs of
the channels forming the chords. If this is to be done
without unduly weakening the spar, it will be necessary
to stiffen the backs of these channels in some manner,
arrd this phase of the problem will be next to be studied.
41. Welded steel tube trusses, type 31.- The welded
truss of chrome-molybdenum steel tubing gave excellent
results in the tests on spars of bomber size, and it is
desired to determine its merit for the observation size.
The first spar of this type, 31A, was constructed with
elliptical tubes instead of round tubes for the chords on
account of the restricted depth and the necessity of
locating the material as far as possible from the neutral
axis. This spar did not show up well, as it failed by
lateral buckling of the compression flange. Spar 31B
was somewhat shallower than 31A, due to insufficient
allowance for shrinkage in manufacture. This spar
was tested with third-point lateral support, and it also
failed by lateral buckling. A new spar is on order that
will have wider flanges, and it is believed that it will
be able to carry the load, though supported laterally
at the center only. The compression chords of both
spars 31A and 31B were larger than the tension chords,
so that the neutral axis was not at midheight. The
pins of the end fittings were at midheight, so there
was a helpful eccentricity of application of the axial
load. In future spars of this type the pin will be placed
at the neutral axis so that the tests will be comparable
with those on the other type.
42. Combination duralumin and steel tube Warren
trusses, type 40.-Spar 40A was of the same type as
the spar of the bomber series that gave the highest
strength-weight ratio. The chief defect developed in
the earlier series of tests was the excessivley large de­flection.
Spar 40A, like spars 31A and 31B, had
chords of elliptical tubes instead of round tubes. It
also failed by lateral buckling of the compression flange,
though provided with third-point lateral supports.
This spar had the axial load applied eccentrically in
the same way as the spars of type 31, and, in this case
also, future spars will have this defect corrected. One
such spar is now on order.
43. Work is already well under way on the prepara­tion
of a progress r eport in which the spars already
tested will be carefully studied and conclusions drawn
from the test data. No serious attempt is made in
this report to do more than mention the more obvious
results. Although more spars will have been tested
before this second report is completed, its scope will be
confined to the tests already made, the results of later
tests being put in other progress reports. It is prob­able
that two series of progress reports will be found
advisable, one series giving only the more obvious
results, as was done in this report, and the other series
going into detail, as in the report in preparation.
AUXILIARY QUESTIONS
44. Some question has been raised as to the validity
of the method being followed to determine the best
type of metal wing construction. Some argue that,
while the method may be satisfactory so far as two-spar
construction is concerned, we should consider the
multispar type of Junker or the single-box type of
Rohrbach as the model for our development. Oc­casionally
an advocate for one of these types of con­struction
talks much about how, when we use metal
construction, we should use construction suitable for
metal rather than one suited only for wood. If the
conventional two-spar framework were suited only for
use with wood, that argument would be valid. But that
is not the case, and the two-spar framework is just as
well suited to metal construction as it is to wood. In
fact, its use in wood was taken from analogy to metal
structures like steel-truss railroad bridges. In the
early days, particularly in England, many railroad
bridges were made of tubular construction, but that
type of design has been wholly supersecled by the open
framework similar to our conventional wing construc­tion.
1 aturally, there will be differences in detail
design to take proper account of the different properties
of the materials, but it will not affect the general layout
to any great extent.
45. For certain conditions it might prove desirable
to use a metal covering and design the structure in
such a way that the covering would carry a large portion
of the load. In such cases it might well prove advan­tageous
to use one of the unconventional types of con­struction.
Occasional wings of these types hould be
built for comparison with the conventional designs,
but until they c,an show a superiority they have not
yet proved the major effort should be along the line
of the open framework which has proved so satisfactory
in the past, not only for aircraft but also for railroad
bridges, skyscrapers, etc.
46. Some engineers who favor the two-spar type of
construction claim that the division's method of testing
is unfair, as it puts too great a premium upon ability
to carry load without lateral support. The ribs, they
say, will give all the support needed, so the test spar
should be either rigidly supported at intervals equiva­lent
to the rib spacing or at least supported by mem­bers
equivalent to ribs. The division has been work­ing
on the basis that the lateral support afforded by
the ribs is too small, if indeed it exists, and should be
neglected, only the drag struts being counted upon for
lateral bracing. Here is a fundamental difference of
opinion.
47. Tests made about a year or more ago on some
panels from DH4 wings showed that if the ribs are not
loaded and an axial load were put on the rear spar the
ribs would cause the front and rear spars to have the
same deflection in the plane of the wing panel. The
tests were inconclusive on the main point, as the observ­ers
of the tests felt that if the ribs ha.cl been loaded at
the same time, as they would be in practice, they would
have been unable to cause this unity of action of the
spars unless they were greatly overweight, and any such
extra material would do more good in the spars than
18
in the ribs. It is hoped to make some more tests in
the near future in which the ribs will be loaded.
4 . The fir t tests are being made on the Huff
Daland XHBl wing panel now being static tested.
As it has carried the required loads, t he intermediate
d rag struts have been removed and the wing is being
r te ted, first with t he ribs present and then with the
rib removed. Further tests can probably be made
on uninjured panels of the metal wings submitted by
Thomas-Morse for the DH. These tests should give a
good idea of the effect of the lateral bracing of the rib in
increasing the total load that can be carried by the spar.
49. If it is decided that only the drag struts should
be considered to give lateral support to the spar,
a distance of 4 inche between supports is very
reasonable. The average di tance between drag struts
on the 01, 02, and C05 i a little more than 48 in ches.
The spars of these design · are al o about 6~ inches in
total depth, and tho maximum loads a re comparable
to those developed in our te t jig.
50. Although the abili ty of the ribs to give ade­quate
lateral support to · the spars may be doubted,
there is little question but that tho front spar is given
considerable support by t he loading edge construction.
It would be reasonable to as ume t hat the front spar
was supported at such hort distances that it wotLld be
un likely to fa il by lateral buck ling, though the rear
par were as urned to be o upported only at the drag
strut . This might result in using. a type of construc­tion
like the plate girder for the front spar and one
like the dural box for the rear. It is believed, however,
that the be t re ult can be obtained if the pars are
fir t developed so they will carry the design load with
only one lateral support, and then lightened and tested
with more supports if it is hoped t hat any advantage
could be obtained by so doing. Of two spars with
the ame strength-w ight and t rength-d eflection
ratio , the one that will carry the load with the lea t
lateral upport is the better.
51. It should be noted that the increase in t rength
obtained by using more lateral supports was not nearly
a great as had been expected. When two supports
were used instead of one, the increa e, instead of being
125 per cent averaged only about 20 per cent. With
four supports, the increase in strength over one sup­port
was 42 per cent, but the spar was much heavier.
The increase in strength-weight ratio was only 21 per
cent. When four support are compared with two
support , the increase in . trength-weight ratio is less
than 1 per cent. The only type of spar tested with
four supports was the dural plate girder, and there
were differences in detail design between the different
article which prevent the compari on from being
exact but not enough to prevent their being approxi­mately
accurate. If anything, the fou r-support speci­men
should have shown up relatively better t han it
did, a it wa slightly wider than the other two. The
indications of the spar te ts mad so far is t hat the
effect of additional lateral . upport has been over­rated,
though the t sts a lready completed on the
XHBl wing tend t a d ifferent conclusion . Some
further light may be throw11 on this matter by a com­parison
of the action of the test spar submitted by
Thomas-Morse for the DH4 metal wings and the same
section when used in the wings. This will be investi­gated
a a part of the next report on the e spar tests.
52. Various engineers have made several criticisms
regarding the test loading. Some of these criticisms
were the result of rather sloppy fits in the jig. In
future tests every effort will be made to eliminate these,
although some of them are not considered as serious
as these engineers have claimed. In every produc­tion
airplane there are loose fits similar to some of
the defects of the test jig, and for a par to be fully
practicable it should be able to stand a certain amount
of play in the joints. On the other hand, it is impos­sible
to have exactly the same amount of play in each
test, and the existence of play adds another variable
of uncertain amount. Therefore, play in the joints
should be eliminated as far as pas ible without getting
too much friction in order to make the tests strictly
comparable.
53. Others of the criticisms of the loading used are
not s well founded. One is regarding the use of a
pin at the neutral ax is at each end of the specimen, it
being claimed that t he pin shou ld be eccentric so there
would be a bending moment at each end of the speci­men.
There is no real advantage in having the
moment at the specimen ends, a the on ly important
resu lt would be to increase the length of test spar
needed to get the desired bending moment at its cen­ter.
It is difficult to p ictu re t he kin d of spar that
would show up any better in stri ctly comparable tests
if moments were applied at its ends than with no
moment applied at the encl .
54. Another crit icism on :imewhat imilar grounds
is that the lateral load is applied at only t \,-o points,
whereas it should be applied at about eight points,
roughly a foot apart. If this were done it would be
possible to do more in t he way of tapering the chord
members of certain types of constru ction than with
the present loading, but t he po sibility and gain to
be derived from tapering can be determined quite
readily by means of computation. Experience with
the 15-inch spars indicated the desirability of having
the maximum load on the spar extend over a fairly
long section which included no fittings. This is
obtained by the present loading, and t he results in
the tests made so far have borne out the wisdom of t he
earlier deduction. None of the spar s of the 6~-inch
series have failed at a fitting, so each test has shown
the merit of the spar construction unconfused by con­siderations
of the strength or weakness of the fittings
used. In the 15-inch spars practically all of the fail­ures
came at a fitting, and in many cases it was impos­s
ible to separate the effect of the fitting from that of
the spar construction .
CONCLUSIONS
55. It is the opinion of the division that this project
is of very great importance, that it is being carried
out a long the proper lines, and the work a lready clone
i. of great value. It will be prosecuted as energetically
a financia l considerations permit, and t he industry
will be notified of the results being obtained by progre s
reports of which this is the first.
APPENDIX I
DATA ON MORE RECENT TESTS
1. While the main body of this report was being of round t ubing. The chief detail of interest in their
written nine spars were tested. It is not considered design was that the tension-web members were carried
practicable to incorporate the results of these tests in through the chord member tubes and were welded to
the body of the report, so the essential data from these both surfaces of the latter . Spar 32A was made of
more recent tests are given below. Each spar is first tubing as received from the mill without heat treat­described
and the type of failure noted. The numeri- ment. It failed by crushing of the compression chord
cal data are then given in Table A for all nine spars. near a weld about 3Y2 inches from the center of the
2. Spar 3D was another T box. The chief feature length of the spar . The tubes for spar 32B were heat
of detail design of this spar was that a direct connec- treated to 150,000 pounds per square inch tension
tion through a glued joint was provided between the before welding. This spar failed by crushing of the
"ears" and the central portion of the compression compression chord adjacent to the end fitting. Both
flange. This feature made it possible to impose a fair of these spars were tested with two lateral supports.
share of the load on the ears, and the spar failed by An attempt was made to test spar 32A with a single
horizontal shear on the webs, between the end fitting lateral support, but the compression flange began to
and the side load. buckle at about 12,000 pounds axial load, and the test
3. Spar 14D was a plate girder, and is described conditions were changed.
above in paragraph 35. The data in the table below 9. Spar 21B was of the same type as 21A, the chief
refer to the third test, in which the spar failed by modifications being an increase in the pitch of the
lateral buckling of the compression chord. rivets connecting the webs with the cover plates and a
4. Spar 19B was the second triangular framework decrease in the pitch of the vertical corrugations in the
referred to in paragraph 39 above. The spar failed web. While the beam was carrying the maximum
by shearing of rivets in the side pull fittings. load, but before complete failure, it was noted that the
5. Spar 20A was a duralumin box with the chords webs were badly wrinkled. After the load bad been
made of two sheets forming a hexagonal cell, as noted removed these wrinkles disappeared. The complete
in paragraph 30 above. The webs were of corrugated failure was due to buckling of the compression flange
duralumin sheet and the corrugations were vertical. near the section of maximum moment. As this spar
In the first test the rivets in the end fittings failed. developed the highest strength-weight ratio of all the
These rivets were replaced by machine screws, and in metal spars tested under design conditions, the box
a second test the spar failed in the webs near the encl type must be considered as of approidmately equal
fittings. The data given below are from the second test. promise to the channel-truss type. Before this test it
6. Spar 21A was a cluralumin box of the first type had been supposed that the channel truss was defi­described
in paragraph 31 above. The failure in test nitely, though not very greatly, the superior in respect
was from buckling of the compression cover plate and to strength-weight ratio. It should be remembered ,
tension in the rivets connecting the cover p late to the however, that even these two t ypes, which are the
webs. The location of this failure was about SY2 inches most successful of the metal types tested, are well
inside the side load. below the conventional spruce box in strength-weight
7. Spars 22A and 22B were duralumin channel ratio.
trusses of the type described in p aragraph 36 above. 10. Spar 3F was a T box, differing from 3D mainly
The first failure of spar 22A is described in that para- in that web stiffeners were provided where the shear
graph. The compression web members were rein- was large. It failerl by a combination of shear and
forced by riveting small angles to their free edges, and bending at the end fitting. The strength-weight ratio
in the second test rivets connecting the web members developed, 1925, is the iargest yet attained. The
to the chords failed. The data given below for this results of the test on this type of spar show that such
spar are from the second test. Spar 22B differed from design refinements as camber can be applied to wood
22A in that the free edges oi the compression web construction just as easily, if not more easily, than to
members were turned in to stiffen them and make the metal construction, and that the resulting increases in
use of angles unnecessary . The weight was also r e- strenght-weight ratio will be as great.
duced by putting flanged lightening holes in the backs 11. Spar 3E was similar to 3F, but the webs were
of t he compression web members. made of inferior material. It was not as stiff as 3F
8. Spars 32A and 32B were welded chrome-molyb- and failed under 17, 740 pounds by shearing of the
denum steel tube trusses, very similar to those of type webs and crushing of the compression flange about 15
31. The chords were of elliptical tubing and the webs inches from the end pin. As the weight was 10.07
(19)
20
pounds the strength-weight ratio was only 1, 760. The I the previous test of spar 3F indicating that the original
EI values of this spar are unknown, as material was flange would be stronger than the inferior plywood
planed off the compression flange just before the test, webs.
TABLE A.-Data from recent tests
b b b £ ~'"O
X ~-o Deflection 8 o""'
3l 3l .,
.a-a ·~
., o'O
.!'1 'O "' oo- gs: ~~§
8.
..., ..., "' ..., "o ""_ ....
'O 0 0 £ 8 .8~ 8.
.,._ "' ·;p~~ 'O ~~ ., 0 "' _..., ,,,
"' ~1:2 .8 Spar No. ~ ;!.~ ~~ ~ """" "' 'O ., "'"' '"' 0. £ ..!.s:I
., ....
£ " ~ QCI ~'0§ ;;; """~ '"'"' .E 0 ..., 8"" .cl...,O .E 0
..;-~ .§ ~ 8. 8 ..... .,-" :;;·~ ...,"_ ca 0 .:.X e "' .a ·-..., ..... ""8"' "' 0.
8 -" ...,-" :.3'G) ~ ~ 8 ~ '§ s1~ .... -§,ci.-1 "" .. S-~ Cl ""'10 § El~ ., -:;; 6 2! ~ ~~~ ·a; ·a;
<> § h~a 6 ..., .
..:i !::: !::: !::: p:; "' s r°"1"' ["1 ...
--------- ------ --------- ---
- -12--,--1-3-
--- ---
1 2 3 4 5 6 7 8 9 10 11 14 15
SPARS TESTED UNDER DESIGN CONDITIONS
14D . -------- -- - - --- - 1 19, 690 15. 81 2.26 1, 245 75.1 98. 45 1, 226 0. 70 o. 245 133, 500 \ 6,860 80.4 3. 873
20A .. ----------- --- - 1 19, 740 14. 66 2.09 1,347 81. 2 98. 70 1, 329 1.15 .308 93, 900 9,310 64.1 4. 515
21A .. -------------- - 1 20, 180 14. 84 2.12 1,360 2. 0 100. 90 1,360 . 85 . 292 93,400 9,500 69. l 4.333
22A .. --- ------------ 1 18, 450 13. 81 1. 97 1,336 80.6 92. 25 1, 232 -------- .404 74, 350 13, 270 45. 7 5. 579
22B .. - -------------- 1 19, 800 13. 57 1. 94 1, 458 87. 9 99.00 1, 443 1. 75 . 366 2,900 13,080 54. 0 4. 967
21B .. -- - ------ -- -- -- 1 20, 980 14. 25 2.04 1,472 88.8 104. 90 1, 472 1. 30 .307 87,400 9,550 68. 3 4. 375
SPARS TESTED WITH ADDITIONAL LATERAL SUPPORT S
32A. -- --------------\ ~ 1 17. 250 I 14. 28 1 2.04 1 1,207 j 77. 6 1 86. 25 1 !: g4J 1·-·1:10·\ 0. 365 1 89.000 I 6,090 I 47. 3 1 5. 212
32B. -- ---- ----- - - --- 19, 660 14.40 2.06 1, 365 87. 8 98. 30 .364 87, 000 6,390 54. 0 5. 242
ECCENTRICALLY LOADED SPARS
3D------ -- - --- ------1 t I 18, 680 I 10. 64 1 1. 521 1, 7561 103. 6 I 93. 40 I 1, 640 1-- - -----1 0. 2281 61,600 I 10, 180 I 81.9 1
2. 426
19B: ... .. . . . . ....... 18, 440 9. 98 1. 43 1, 848 109. 0 92. 20 1, 704 1. 00 .202 31,420 13, 650 91. 3 2. 016
3F ....... . .. : • ... . . .. 20, 720 10.77 1. 54 1, 925 113. 7 103. 60 1, 925 . 90 .240 59, 200 9,680 86.4 2.585
. 1 Based on best strength-weight ratio of spar in same group in Table Ill.
APPENDIX II
DATA ON SPARS TESTED JANUARY 1 TO FEBRUARY 14,
1927
1. Since January 1, 1927, 14 experimental metal
spars have been tested. Pictures of these spars are
hown in Figures 6 and 7 and the more important
test results in Table B.
2. Spar 16F failed by buckling of the free edge of
the compression-web members. As all 7 of the serie
16 spars had practically the same size web members,
the compression-web members of all 7 were reinforce·d
by small angles. The rivets used for the purpose can
be easily seen in the photograph. Only the test of
spar 16F made after the web had been reinforced is
reported here, the designation used being 16F2. The
first test of spar 23B was made with a single lateral
support; the second test was made with two lateral
supports. The end fittings of spar 31C failed in the
first test. These fittings were reinforced and the spar
retested.
3. The most interesting of the spars tested was 20B.
This spar gave the highest strength-weight ratio of all
the metal spars tested under the design condition or
with additional lateral supports. It did not give as
high a ratio as the type 19 spars, which had camber
and restraining end moments, and are not, therefore,
directly comparable. Spar 20B also showed a higher
strength-weight ratio than the wood I beams, though
not as good as two of the three wood boxes. The
total load carried by this spar wa 17 per cent in excess
of the design load, and it would be very desirable to
see if the strength could be brought down to the design
strength without loss in strength-weight ratio. It is of
interest in this connection that the two wood boxes
showing higher strength-weight ratios carried 12 and
23 per cent overloads. On the whole, it can be said
that this spar has been by a good margin the most
successful of the metal spars tested and is the only
one tested so far that appears to be of practically
equal merit with the wood box. It is the opinion of
this branch that this type of spar deserves further
and more intensive study.
4. At the time t he series 16 spars were designed
that was believed to be the most promising type.
Though it has since been equaled by the box type as
represented in series 20 and 21, the difference is not
great, and it is believed that for depths any greater
than 6'4 inches the channel truss would prove the best.
The spars of series 16 were designed to bring out the
effects of certain points in detail design. Though some
of the points on which information was desired had
little light thrown upon them, the results of the tests
do give valuable information on others. The principle
results of these te t were to show the importance of
careful study of the design, of the connection of the
web members to the chords, and the advantages of
corrugating the backs of the channels forming the chord
members. The relatively poor results obtaii.1ed from
the spars with unequal chord members, 16D and 16E,
were due primarily to the low-strength properties of
the material used in them. Between the tests on the
spars as units and tests now under way on sections of
the chord members, it is hoped to determine the value
of the restraint coefficient that may be used in design
and also to check the validity of the curves of "form
factor" for duralumin channels t hat have been worked
out from recent tests or single channels at the division.
5. The series 23 spars modeled on the type being
used by Breguet gave rather poor results. Whether
this is an indication that the Breguet design is inferior
to the dural boxes and channels represented by series
16, 21, etc., or that we have not found out the propor­tions
used by Breguet, can not be determined yet,
due to lack of pertinent data.
6. Spar 31C (a welded steel truss) and spar 40B (a
combination steel and duralumin truss) gave very poor
results, and it is the opinion of t he division, as the
result of tests on these two series and also on series 32,
that neither of these types is suitable for such shallow
spars and that their further consideration should be
dropped.
(21)
23
24
TABLE B.-Spar tests-January-February, 1927
Spar Type
16B ________________ Dural channel truss ______________________ _
160 ____________________ do ____________________________________ _
16D _______________ _____ do _______________________________ ____ _
16E ____ _________ ________ do _____ ________ _______________ _______ _
16F2 _ ___ ____ ______ ..... do ___________________________________ _
16G -- -·--- _____ . __ • __ ._.do ...... __ ------- _____ ------- _________ _
16H. ___________ _______ .do _________________ ----- ______ . __ . ___ _
20B .... ______ __ _ ___ Dural box _________ _ ------- ___ ________ __ __
210 __ _____ . ______ ______ .do ....... _______ __________ • ______ ____ _ 21D .. ___ _________ _____ _ do __________________________________ __
23A ..... __ _ _ __ __ __ Dural dum b-beJL _ --- __ __ ____ • ___ ----- ___ _
23BL .... __________ ..... do ...... ------- ---_ -- - --- • --- ___ _____ __
23B2 .... _______________ .do .......... _______ ---- ____ ---- ______ __
3311C0L2. _._._._ __. _________________ __S_t_e e. dl-otu_ _b_e_ _t_r_u_s_s_ .. ._ _-_-_-_-_-_-_--_-__-_- -• -_-_-_-_-_-_-_______ _. ____
40B ..... __ _ __ ___ __ Combination truss.--------- ____ . __ ______ _
Maximum Weight Load
load 7-feet weight
18, 050
18, 750
14, 850
14, 960
16, 740
20, 150
19, 900
23,510
20,450
21, 920
17, 580
15, 600
15, 760
20, 480
22, 120
18,300
13. 06
13. 42
12. 07
11. 97
12. 58
13.46
13. 72
14. 56
14. 50
14. 62
12. 94
11. 73
11. 73
18. 52
18. 52
18. 21
1,382
1, 397
1, 230
1, 250
1, 331
1, 497
1,450
I, 615
1,410
1,499
1, 359
1, 330
I, 344
I, 106
l, 194
1, 005
Deflection at-
EI~
10,000 Maximum
pounds load
0. 257
• 278
. 332
.342
.361
• 281
. 297
. 375
.268
• 213
.380
. 430
.405
.252
.280
.445
L 05 83. 7Xl0'
LIO 87.l
1. 15 79. 9
L30 72.3
.65 78.0
L2ii 84.6
L 10 87.4
L30 91.4
• 85 100.1
. 95 97. 5
1. 00 68. 1
. 90 62. 7
.90 62. 7
. 75 97.1
. 90 97. 1
1.80 54.3
EI,,
18. 58XIO'
20. 7
13. 13
13. 62
18. 35
20. 2
19. 53
8. 75
10.30
10. 55
7.09
6. 705
6. 705
12. 09
12. 09
18. 70