Gurfinkel 1989
Gurfinkel 1989
Gurfinkel 1989
windless, day in the Midwest. Field observations and analysis showed that failure
of the suspended steel hopper triggered the collapse. Attachment of the suspended
hopper to the concrete wall of the silo had been designed awkwardly and was
subject to substantial overstressing in service. The quality of fabrication and weld-
ing of the various components for the hopper-supporting system was poor. The
question of why it took four years of service before the silo collapsed but an ad-
jacent, twin, silo remained standing, is answered. Design of the strengthening re-
quired for the remaining twin silo is also described. Advice to designers of sus-
pended hoppers is also given at length.
INTRODUCTION
FIG. 1. Tall Concrete Silo Collapsed and Damaged Surrounding Structures while
Twin, Adjacent, Silo Remained Standing
Some background on the silos and their design is necessary at this point.
Each silo consisted of three significant components, namely: (1) A rein-
forced-concrete cylindrical shell; (2) a hybrid concrete-steel roof structure;
and (3) a suspended steel hopper in the shape of an inverted truncated cone
(see Fig. 2). The silo rested on a reinforced-concrete ring foundation.
The concrete wall of the silo was built by jumpforming [in 4-ft (1.22-m)
high increments] a cylindrical shell having a 30-ft (9.14-m) inside diameter
to a height of 128 ft (39.01 m). The thickness of the shell was 10 in. (0.25
m) for the lower 24 ft (7.32 m) of wall above the foundation and 6 in. (0.15
m) for the rest of the wall to the top. The abrupt change in wall thickness
of the shell took place at the outside surface; this avoided creating an inside
244
• &
222'
GRAIN
\\ '•'.:.,
30ftI.D.
/ - S i l o Wall,
' Reinforced Concrete
t = 6"
lJ2l.5'(Topof
Compression Hopper)
Ring, RC
•Suspended Hopper,
Steel
t =10-
,102.5'
vlOO' - ^
mmmsg- ••to
7777777/ 777777/
G_ \u-Ring
Ri Foundation
ledge and, thus, left the inside surface of the wall continuous throughout the
full height of the silo. The 10-in. (0.25-m) thick wall was reinforced with
No. 8 hoop and vertical bars, at 24-in. (0.61-m) spacings. The 6-in. (0.15-
m) thick portion of the wall was reinforced with No. 6 hoop bars, at a spac-
ing that increased from 5.5 in. (0.14 m) at the bottom to 12 in. (0.30 m)
245
\ ,.»...^J-£5-6"Well Compacts
Compacted
_
Gravel
\ f&sfH-'s"- or Stone as Req'i
'd
<--#8xlO'-0"at9"O.C.
IO'-6" Wide
—*»"|
Ring Foundation
at the top, and No. 4 vertical bars at 18-in. (0.46-m) spacing. No special
reinforcement was provided for the portion of the wall directly above and
below the level where the abrupt change of thickness takes place.
The roof structure was made of a 4-in. (0.10-m) thick concrete slab at the
perimeter, sloping to a 6-in. (0.15-m) thickness at the center. The slab was
reinforced with 6 X 6-10/10 steel mesh and No. 3 bars at 18 in. (0.46 m)
on centers; it was cast in place on heavy-duty corruform steel sheets sup-
246
The ring plate was attached to the concrete walls of the silo by means of
stirrups made out of 5/8-in. (0.016-m) diameter bars, spaced 8.5 in. (0.22
m) on centers; the stirrups were welded to the ring plate. Although the de-
signer showed closed stirrups in his drawings, the contractor used open stir-
rups instead (see Fig. 4). This departure resulted in only 1 in. (0.025 m) of
each stirrup leg welded to the ring plate, instead of the full 6 in. (0.15 m)
required by the designer.
Where the hopper was attached to the concrete walls the designer created
an ad-hoc composite compression ring [within the 10-in (0.25-m) thickness
of the wall] by placing nine No. 8 hoop bars, 8 in. (0.20 m) apart, sym-
metrically located about the supporting stirrups and opposite to the ring plate
(see Fig. 3).
The concrete was specified at a 4-ksi (282-kg/cm 2 ) 28-day standard-cyl-
inder strength, and the reinforcing steel at 60-ksi (4,227-kg/cm 2 ) yield strength.
Presumably, ASTM A615 Gr. 60 rebars were procured by the contractor,
although this fact is not certain. No specification could be found for the steel
used for the hopper structure.
FIELD OBSERVATIONS
1. The upper portion of the silo, i.e., the 6-in, (0.15-m) thick reinforced-
concrete shell and roof, toppled over in one piece and collapsed against adjacent
structures (see Fig. 5).
2. The lower portion of the silo, i..e, the 10-in. (0.25-m) thick reinforced-
concrete shell, split open into two portions which separated from the base and
each other and moved away laterally and outwardly. Most of the wall eventually
toppled over, as the corn rushed out, but a portion did not because it came to
rest against the adjacent twin silo (see Fig. 6).
3. There was a horizontal fracture surface, just above the level of the stirrups,
and a vertical fracture surface, where the ring plate used to be. Both fracture
surfaces developed along the circumferential direction (see Figs. 6 and 7).
4. During the first inspection, made shortly after the collapse, the vertical ring
plate and the suspended hopper were conspicuous by their absence. Actually,
the fracture surfaces indicated that they had sheared off (from their stirrup at-
tachments to the concrete wall) and laid under the corn. Further observations
disclosed a number of fractured stirrup legs.
5. Removal of the corn and the concrete debris exposed the remnants of the
hopper and ring plate completely flattened out (see Fig. 8). Additional obser-
vations disclosed the failure of all (calculated at 133) stirrup attachments of the
ring plate to the concrete wall.
6. All stirrups fractured at either their welded attachments to the ring plate or
at a section just away from it (see Figs. 9 and 10). An incomplete survey showed
248
that, of 106 stirrup legs examined, 82 had fractured while only 24 had failed at
the welds. Of the latter, 22 failures had taken place in the upper stirrup legs and
only 2 in the lower legs.
7. The hopper ring plate was fabricated in segments that were welded together
Fractured Stirrup
Downloaded from ascelibrary.org by New York University on 05/12/15. Copyright ASCE. For personal use only; all rights reserved.
Leg
Fractured Concrete
Surface
upon installation in the field. The evidence indicated that welding was highly
defective and mostly cosmetic in nature (see Figs. 11 and 12).
8. Instances of insufficient or nonexistent amount of staggering, for lap splices
of hoop bars, were found in the remnants of the concrete wall (see Fig. 13).
FIG. 8. Portion of Vertical Ring Plate and Flattened-Out Hopper; Note Remnants
of Welded Stirrup Legs on Ring Plate Face
250
FIG. 9. Ring plate at Stirrup Connection; Stirrup Upper Leg Fractured, Note Welded
Remnant; Stirrup Lower Leg Failed at Welding, Note Absence of Remnant
9. The inside surface of the concrete wall was covered by a smooth coating
that had considerably reduced its initial roughness.
SILO W A L L
The silo wall was designed for hoop tensions generated by the radial pres-
sures exerted by the grain; the current provisions of Standard 313-77 (Amer.
Concrete Inst. 1977) were used to check compliance. For this purpose, the
following values were used: unit weight of grain, y = 50 pcf (801 kg/m 3 );
angle of internal friction, p = 23°; and coefficient of friction between grain
and wall, u,' = 0.29. Note that the values selected for p and p/ are on the
low side of the usual ranges recommended for grain, namely: 23° =£ p =£
37° and 0.29 « p,' =£ 0.47.
For corn and wheat it would not be unusual for designers to take p = 28°
and p/ = 0.38. However, a low value of p,' was in order because the silo
had contained soybeans, which could smoothen the concrete wall surface by
coating it with an oily residue. Combined with a low value of p this produced
the largest hoop tensions on the wall and the largest vertical load on the
suspended hopper. Because of the pronounced height-to-diameter ratio of the
silos, i.e., H/D = 3.5, overpressure factors, Cd, ranging between 1.8 at
hopper level to 1.5 at the top were used to increase the hoop tensions due
to service conditions; the Cd factors account for additional effects generated
during grain withdrawal. For the determination of the largest downdrag forces
251
FIG. 10. Ring Plate at Stirrup Connection: Both Stirrup Legs Fractured, Note
Welded Remnants
FIG. 11. Fractured Ring Plate at Improperly Welded Connection between Seg-
ments
252
FIG. 12. Fracture Surface of Ring Plate Welded Connection; Note Minimal Weld
Penetration at Less than 1/8 in.
in the wall which, although unimportant for concrete silos, is vital for the
design of steel silos against buckling (Gurfinkel 1988) the greatest possible
value of u>' was used. Pertinent equations that are helpful in understanding
the relationship between these parameters are given in the current literature
(Amer. Concrete Inst. 1977; Gaylord and Gaylord 1984; Gurfinkel 1979;
Safarian and Harris 1985).
The presence of two 14 in. X 14 in. (0.36 X 0.36 m) outlets in the silo
wall, that were situated about 135° apart and 21 ft (6.40 m) above the silo
floor, could cause a 15-ft (4.57-m) eccentricity in the flow of grain because
of side withdrawal; this effect could generate horizontal bending moments
in the silo wall (Deutsches Inst, fiir Normung 1984). Design would be better
served if these moments were calculated and the wall were reinforced using
FIG. 13. Splicing of Hoop Bars in Concrete Wall Was Not Staggered
253
210' i i
200'- ' n
^Design (Effective)
i i
190'
^j I ^ D e s i g n (Specified)
180'-
170'
l.
-Required (ACI Std.313-77)
160'
-Deficiency in Hoop-Steel
Reinforcement
150'
140'-
130'-
I O I K5,(Topof
l21
,
- Hopper)
120'
0.5 1.0 1.5 2.0
2
Hoop Steel, A s , in. /ft
254
tors, Cd.
To determine ultimate hoop tensions, F,„ a 1.7 load factor was used. The
required hoop steel reinforcement, As, was determined from As = T„/<\>fy,
where fy = 60 ksi (4,227 kg/cm2) and cj> = 0.9 is the capacity reduction
factor. The variation of As with wall height above hopper level is shown in
Fig. 14. This must be compared to the designer's specification for As with
which, assumedly, the contractor complied, although this fact could not be
ascertained in the field. Observation of Fig. 14 would indicate that the amount
specified exceeded the amount required for hoop steel throughout the height,
except at a small portion of the silo wall above the hopper, where a slight
deficiency took place.
The situation would be different, however, if one compared the required
and the effective areas of hoop steel. Because of necessary splicing the hoop
steel would not be 100% effective unless the length of splice were sufficient
to develop the yield strength of the steel. The designer specified 30- and 46-
in. (0.76- and 1.17-m) splice lengths for Nos. 6 and 8 bars, respectively;
compare these to the 43- and 71-in. (1.09- and 1.80-m) required values (Amer.
Concrete Inst. 1983), respectively. The latter were obtained by calculation
using/; = 4 ksi (282 k g / c m 2 ) , / = 60 ksi (4,227 kg/cm 2 ), and factors of
1.4 and 1.7 for top reinforcement and class C splices (no staggering), re-
spectively. In view of the shorter splice lengths specified by the designer,
effectiveness factors of 70% and 65% were used for Nos. 6 and 8 bars,
respectively, in obtaining the variation labeled design (effective) in Fig. 14.
A comparison between the effective and required areas of hoop steel indi-
cates that there is a deficiency in the first 55 ft (16.76 m) of wall height
above hopper level; note the shaded area in Fig. 14.
The deficiency in hoop-steel reinforcement of the silo wall could have
been even larger if any of the following departures from the specifications
had taken place; (1) Shorter lap splices; (2) larger spacing of hoop bars; (3)
lower concrete strength; (4) less concrete cover for hoop bars; and (5) lower
yield strength of steel. These items were not actually verified because of the
major amount of field and laboratory work required; they quickly became
irrelevant as progress in the investigation of the suspended hopper pointed
to the actual cause of the collapse.
SUSPENDED HOPPER
The vertical forces acting on the suspended hopper were its selfweight,
Ws, the weight of the actual grain contained therein, Wh, and the load gen-
erated by the vertical pressure of the grain above, Wq. Calculations yielded
the following service loads: Wg = 10.4 k (4.7 Mg), Wh = 176.7 k (80.3
Mg), and W„ = 1,933 k (879 Mg). The latter value is the vertical force on
the hopper induced by the grain above it and.is equivalent to the effect of
a 55-ft (16.76-m) head of grain. The total vertical load on the hopper, W,
would be 2,120 k (964 Mg).
255
Structural analysis and field observations showed that failure initiated nei-
ther in the upper 6-in. (0.015-m) thick concrete walls of the silo (which,
although overstressed in hoop tension, were not close to their limit strength)
nor in the suspended hopper. Instead, the evidence points out to failure ini-
tiation at the connection of the suspended hopper to the 10-in. (0.25-m) thick
concrete wall of the silo.
Field observations showed that a large number of the 5/8-in. (0.016-m)
diameter stirrups used for the hopper-silo connection had fractured; analysis
verified that the stresses generated in the stirrup legs were large enough to
cause generalized yielding of the steel. In addition, it is likely that the sharp
bending of the stirrup legs, for attachment to the hopper ring plate, caused
prefracturing of the hard-grade low-ductility steel bars.
Fracture of the stirrup legs was followed by separation of the hopper ring
plate from the concrete wall. This caused the suspended hopper to lose sup-
port and fall down instantly as the full mass of grain overhead was set in
motion. Upon reaching the floor the grain mass flattened the conical hopper
and then, by expanding laterally, split open the concrete walls of the silo.
Failure of the lower silo walls occurred instantly and spread vertically up-
ward. As the grain spilled outwards, pushing the lower walls into adjacent
structures, the upper portion of the silo severed off and toppled over.
It may be concluded that the silos had been designed and erected with an
inherent propensity for collapse. The eventual collapse of one of them, four
years after being placed in service, was an incident waiting to happen. The
pending questions on why hadn't the silo failed sooner, why had its twin
remained standing, and whether or not the latter required strengthening, may
now be answered. One must only recognize that the initial surface roughness
of the concrete walls diminished gradually with the passage of time. As a
result, reduced friction of the grain with the tank wall increased lateral pres-
sures against the wall, reduced downdrag friction forces on the wall and
increased vertical pressures on the grain. Because the silo had been provided
with a suspended hopper, the increased vertical pressures on the grain had
particular significance in the collapse; they substantially increased the load
on the hopper and its attachment to the silo wall. The process was faster in
the collapsed silo than in its twin, probably because of a shorter "turn," and
the fact that the former was used for a long time to store soybeans while
the latter was not.
It took four years of wall smoothening, and the heaviest corn in storage
in the last two years, to increase the vertical load on the suspended hopper
257
Repair of the remaining silo focused on the need to prevent the meridional
X
<
LLI Exist. Wall Opening
^
lOin. Thick RC
Wall, Exist.
•Exist. Wall
Opening
Plan
FIG. 15. Additional Steel Columns Supporting Suspended Hopper Are Placed
around Inside-Wall Perimeter of Twin Silo
258
-Detail
Downloaded from ascelibrary.org by New York University on 05/12/15. Copyright ASCE. For personal use only; all rights reserved.
A36 Steel
-Detail C
#8at24"O.C.Horiz, Exist.
3"_
Clear
-Detail D
*-Fill as Req'd
FIG. 16. Steel Columns Support Suspended Hopper and Are Attached to Con-
crete Wall and Foundation
forces exerted by the hopper from being supported by the ring plate and set
of stiiTups. (Direct strengthening of either one was hardly feasible and def-
initely impractical.) Thus, to avoid eventual repetition of the fracture mech-
anism that triggered collapse of the silo, it was only reasonable to bypass
the existing hopper-support system and provide a new one instead.
The design was based on the largest possible load on the suspended hop-
per. The load was determined using the grain-withdrawal condition, and the
lowest possible values of the parameters jo-' and p, to account for the effects
259
2-|l/2"Diam. A325
Bolts, 6"Apart
Use E60I8
Weld Electrodes
(JLCol.
FIG. 17. Anchor Bolts Attach Steel Column to Compression Ring in Concrete
Wall
RC W a l l ~ %- No. 4 at
12" - f
(Exist.)
-Embed WI2's
in Concrete to
Floor Level ~
2-ll/2"Diam.xl,IO"A325
-18" Thick RC K Anchor Bolts Set in POR-
ROK in 2"Diam. Holes
Ring Foundation • Drilled into Existing Concrete
(Exist.) Spaced 12 in. Apart.
J_tLJ
FIG. 18. Steel Columns Bear Directly on Existing Ring Foundation and Are At-
tached to It
are given in Fig. 17. Note that a steel plate 24 X 16 x 3/4 in. (0.61 X
0.41 x 0.019 m) was welded to both the hopper and the flanges of each
W12 X 53 column in order to improve the transfer of hopper meridional
forces to the new steel columns. The horizontal component of the force is
transferred to the original compression ring in the 10-in. (0.25-m) thick wall
by means of two 1 1/2-in. (0.038-m) diameter A325 bolts that attach each
steel column to the wall at hopper level (see Fig. 17).
Because of the discrete number of columns supporting the hopper, hori-
zontal bending moments are generated in the composite compression ring at
the wall. The combined action of these moments and the hoop-compression
force was checked against the thrust-moment interaction diagram of the com-
posite compression ring and was found acceptable. Finally, lateral stability
of the steel columns was secured by bolting them to the silo walls at mid-
height (see Fig. 19).
261
o
0)
m «
Downloaded from ascelibrary.org by New York University on 05/12/15. Copyright ASCE. For personal use only; all rights reserved.
*-.2>
*-
0 1
o
o o
z 2-%"Diam.
A325 Bolts ~ Inside
Silo -
6 in. Apart
•Outside
Silo ~
~IOin.Thick
RC Wall, ~ W 12x53
Existing ~ A36 Steel
Column ~
FIG. 19. Steel Columns Are Attached to Silo Wall at Midheight for Lateral Sta-
bility
APPENDIX. REFERENCES
"Building code requirements for reinforced concrete." (1977). ACI 313-83, Amer-
ican Concrete Institute, P.O. Box 19150, Redford Station, Detroit, Mich.
Gaylord, E. H., and Gaylord, C. N. (1984). Design of steel bins for storage of bulk
solids, Prentice-Hall Inc., Englewood Cliffs, N.J.
Gurfinkel, G. R. (1979). "Reinforced concrete silos and bunkers." Structural en-
gineering handbook, E. H. Gaylord and C. N. Gaylord, eds., 2nd Ed. McGraw-
Hill Book Co., Inc., New York, N.Y., 221-237.
Gurfinkel, G. R. (1988). "Tall steel tanks: Failure design and repair." J. Perf. Constr.
Fac, ASCE, 2(2), 99-110.
Hodge, P. G. (1959). Plastic analysis of structures. McGraw Hill Book Co., Inc.,
New York, N.Y.
"Lastannahmen fur Bauten, Lasten in Silozellen, Entwurf und Erlauterungen." (1984).
DIN 1055, Sheet 6, Deutsches Inst, fur Normung, ev. Burggrafenstrasse 4-10,
1000 Berlin 30, West Germany, Feb.
263
for storing granular materials (ACI 313-77) and commentary." (1977). Revision—
ACI 313-R77, ACI J. May-June 1983, 250-252.
"Reinforcing steel welding code." (1975). AWSD12.1-75, Table 5.2 by Amer. Welding
Soc, Inc., 2501 NW 7th St., Miami, Fla.
Safarian, S. S., and Harris, E. C. (1985). Design and construction of silos and
bunkers. Van Nostrand Reinhold Co., New York, N.Y.
264