Bursting of a silo

Engineering Failure Analysis, Vol. 4, No. I, pp. 4~55, 1997

~ ) Pergamon

© 1997 Elsevier Science Ltd. All rights reserved Printed in Great Britain 1350~307/97 $17.00 + 0.00

PII:S 1350-6307(96)00026-X

BURSTING

OF

A SILO

R. K I E S E L B A C H Failure Analysis of Metals, Swiss Federal Laboratoriesfor Materials Testing and Research, Uberlandstrasse 129, CH-8600Dfibendorf,Switzerland (Received 30 August 1996) Abstract--This paper describesthe bursting of a large silo on a farm, whichcausedconsiderableenvironmental damage and cost. The cause was misuse of the silo for vegetableslurry instead of for feed for livestock, and overfillingthe silo. © 1997 ElsevierScienceLtd. All rights reserved. Key words: silo, failure, rupture, hydrostaticpressure

1. I N T R O D U C T I O N Most of the accidents in connection with silos are due to suffocation or gas poisoning of the farmers entering a silo. Some are also caused by the explosion of methane, which is produced by fermentation of the forage. Cases of bursting or explosion are, nevertheless, rather rare. In the present case, three identical silos had been built on a farm, each with a diameter of 6 m (20 ft) and a height of nearly 25 m (80 ft). The hull of the vessels was made of steel plates measuring 1.4 x 2.68 m, and the thickness of the sheets varied from 5.7 mm at the bottom to 2.4 mm at the top. All in all, the silo consisted of 16 rings and one base ring. The individual sheets had been protected against corrosion by enamelling, and were joined by bolts and nuts. The joints were protected against corrosion by a special kind of mastic. The total capacity of a silo was approximately 630 m 3. Since the farm no longer had any use for the silos, they were rented to a feed company for the storage of feed for pigs. The slurry was delivered in tank cars and pumped into the silo. The silo was filled up repeatedly in the following months. Finally, a few minutes after a delivery, when the tank car had just left the site, the silo burst and spilled its contents, a slightly sour slurry. The collapse of the top of the silo also damaged the next, still empty silo, which buckled and also collapsed partially. The spilled slurry caused considerable environmental damage in addition to the cost of the silos and the cost of the interruption to service. According to the lorry driver, the silo had been, at that time, approximately three-quarters full, and the manhole lid had not been fastened, but only laid loosely on its flange.

2. I N V E S T I G A T I O N S A N D TESTS P E R F O R M E D 2.1. Visual inspection The site of the accident was visited and the following observations could be made (see Figs 1-3). Silo 3 had failed and was severed above the seventh ring (counted from the bottom), where a reinforcement ring was attached. A zone, four rings high, had been separated, and hung partially on the silo, partially on the ground. The contents of the silo had spilled for approximately 30 m in a semicircle uphill and 200 m downhill. The pasture had been destroyed, the slurry being slightly sour after lactic acid fermentation. The detached rings were separated into several pieces, and were in some places still immersed in a pool of slurry, such that it was difficult to make out where the pieces had belonged. Failure had 49

R. KIESELBACH

50

Fig. I. Damaged silos: view of the site.

'

?

t~

c

Fig. 2. Bolted joints of the sheets used for the silo. In the lower part, a reinforcement ring was attached.

occurred by r u p t u r e of the boltholes of the vessel in circumferential a n d l o n g i t u d i n a l directions (Fig. 4). After the search had been carried out, specimens were taken, as detailed in Table 1. After the accident, the silo was still full up to the seventh ring (counted from the b o t t o m ) , as can be seen from Fig. 5.

51

Failure of sewage silo ring 1

I-------'f-----I I

2

I

,I

3

I

"1---I

-- 1 . . . . I

1-------I I

I

. . . .

r------I I

I

2.4

|

I

I

I

I

I

I

I

/

I

I

,'

I

,'

2.4

I

I

I

I

I

I

2.4

I

2.4 2.4 2.4 2.4

E

3.4 3.4 3.4 4.2

6 m Dia.

'~ 13 14 15 16

~---~---4---~ ~ i

~ i

....

J i

~ l

~----~----I'---~'

....

,origin of ~ruplure i i

- r - - - ' -rl ~ i

..~ o ,,,

i

~'--4----~----,'

i

i

i

J

i

i

,

i

i

i

i

i

i

i

i

,

..,5 = ._o

4.2 5.0 5.7 5.7 5.7

Fig. 3. Schematic di-awing of the silo, showing the location of the rupture: (I) level theoretically necessary for bursting by hydrostatic pressure; (II) level for filling with 407 t; (III) permissible filling for density of 1.05 kg 1-~; (IV) level after bursting.

4 4 4 Fig. 4. Longitudinal bolted joint, presumably at location of start of rupture.

Table I. Specimens and samples taken A B C D E F G

One sheet/plate with a failed circumferential bolted connection One plate containing an intact circumferential joint One plate containing an intact longitudinal joint Samples of slurry on site Samples of slurry retained at the manufacturer of the slurry Textile fibres from an airbag at the top of the silo Two safety valves from the top of the silo

52

R. KlESELBACH

Fig. 5. Top view of failed silo. showing remaining filling level.

2.2. Tests for traces of an explosion Specimens F were subjected to laboratory tests to detect possible traces of heat influence by fire or explosion. No such traces could be found. Thus, one can conclude that failure was not caused by the explosion of methane or any other gas produced in the silo by fermentation. 2.3. Tensile tests The sheet metal was tested using specimens BR, CR, BP and CP, as shown in Fig. 6 and Table 2.

"-I BP

BR

"top"

G Fig. 6. Specimens for mechanical tests.

Table 2. Results of tensile tests on sheet material from silo

Specimen

Orientation

Yield strength (N mm 2)

BR CR BP

• to joint L to joint II to joint IIto joint

288 267 300 278

CP

Tensile strength ( N m m 2)

Reduction of area (%)

Elongation (5 diameters) (%)

Uniform elongation (%)

347 316 345 313

78 77 75 70

44.5 47.5 43 46.3

25 24 24 29

Failure of sewage silo

53

F [kN 30 •

°

"~ %

c'.2 i.2

20

10

/,

c.1

" - - - B.1 i

i

i

i

I

5

10

15

20

25

AI [mm]

Fig. 7. Behaviour of bolted joints in tension tests.

Specimens B. 1, B.2, C.1 and C.2 were tested for the strength o f the bolted joints. F r o m Fig. 7, it can be seen that they started to yield between loads o f 15 and 25 kN, and that the d e f o r m a t i o n before fracture was in m o s t cases m o r e than 25 m m .

2.4. Determination of the density of the slurry F o u r samples E gave an average density of 1.035kgl -~. Sample D taken f r o m the site had a density o f 1.05 kg 1-1.

3. N U M E R I C A L

EVALUATIONS

3.1. Determination of fillin9 level from records of the user After the accident, the user o f the silo supplied notes o f deliveries, f r o m which the theoretical filling level at the time o f the accident could be calculated (Table 3). This height corresponds to filling up to the u p p e r edge of ring 7 (counted f r o m the top).

3.2. Stress analysis of bolted joint T h e average values for yield and ultimate strength are provided by the tensile tests on the sheet material, with Rp02 = 283 M P a and R m = 330 M P a . This gives for shear: Rm Z0.2,pe~m-- V/~

163 M P a ,

(1)

Rm Zm,p~rm-- X//~ -- 191 M P a .

(2)

F o r one bolt, at a distance of 25.4 m m from the edge o f the sheet, one obtains Fo.2 = z0.Z,p~rmx 2.4 x 25.4 x 2 = 19.9 kN,

Table 3. Calculation of filling level from records Contents according to bookkeeping notes 407,790 kg Density according to tests, maximum 1.05 kg 1 Base area of silo (inside) 27.98 m2 Theoretical level 13.9 m

(3)

R. KIESELBACH

54

(4)

Fm=rm.p ...... x 2 . 4 x 2 5 . 4 x 2 = 2 3 . 3 k N . 3.3. Assessment oJ'the theoretical bursting pressure

In the test, the lower bound for the strength of the bolted joint was measured as Fvr~cture= 22 kN. This also corresponds to the mean value of the forces calculated from Eqns (3) and (4). From the spacing of the bolts (108 mm), one obtains the force at fracture per unit length: 22 x 103

T=

108

-204Nmm

~.

(5)

The burst pressure can be calculated fi'om this, using the diameter of the silo (6 m), as 2 x 204 6000

Pburst-

-

0.068 N mm 2

(6)

The corresponding level over the ruptured ring is AH

-

Pb ....

0.068

9P

1.05 x 9.81

x 103 = 6.6m.

(7)

This is equivalent to the height of 4.7 rings of the silo, and would mean that the level of the slurry was approximately in the middle of the third ring (counted from the top). 3.4. Spurtin9 distance From the visual inspection at the site of the accident, the approximate spurting distance of the slurry of 30 m is known. Since this was not a simple parabolical throw, but the jet was dispersed further after hitting the ground, the process can only be calculated approximately. The intention of such an assessment is, of course, to determine the filling height of the silo. The horizontal velocity of the jet is given by v = ~ 0 x ~ A H , and from the distance the jet travelled one obtains

t' =

2(HL-- Ah) "

(8)

Thus, the height of the liquid above the leak is D2 AH =

302 =

( H e _ ;h)4~o2

(9)

{HL(l+d*)2-7(l+d*)}4~°2"

d* = 6/D is the portion of the distance that the jet travels after hitting the ground, ~0 is the factor of constriction of the jet (normally ~0< t), and the other symbols are explained by Fig. 8. If different

H F ,H k

7m T

D -30 m Fig. 8. Schematic view of the spurting of the slurry from the silo.

Failure of sewage silo

55

40 35

•E•-• 30

25

-.- .-.:-

7_-.-_

....

-_-

,,c

._~ 20

--HL=12.22

@ t,=

_

\silo height

rn

HL=13.62 m

15 _= 10

0

......

HL=15.024 m

.........

HL=16.428 m

.......

HL=17.832 rn

I 0.2

0.25

......

~-

--

0.3

÷

t

0.35

0.4

sloshing ratio d* Fig. 9. Filling height of the silo necessary to produce the observed spurting distance, assuming different sloshing ratios d* and differentlevelsHL for the initial leakage. levels HL for the first leak and different ratios of sloshing (d* = 6/D) are assumed, it can be seen from Fig. 9 that the silo must have been filled to the top, and the liquid must have sloshed relatively far after hitting the ground to have produced the observed pattern on the site.

4. C O N C L U S I O N S

(a)

(b) (c) (d) (e) (f) (g)

(h) (i)

The visual inspection at the site of the accident showed the typical picture of failure by overpressure. Indications of an explosion or a chemical reaction, which could have produced the overpressure, were not found. A pressure above atmospheric pressure can also be excluded because the necessary safety devices were installed and operative. According to the manufacturer of the silo, it was permissible to fill the silo with liquid up to the seventh ring (counted from the bottom), i.e. ca 10 m high. The amount of 407 t, admitted by the user, corresponds to a filling height of ca 14 m. The assessment of the filling height from the observed spurting distance also points to a filling level practically at the top of the silo. The design, manufacture and assembly of the silo can be judged as proper, suitable and according to normal engineering practice. Tests on the material also indicate a higher level than the seventh ring (counted from the bottom). This is supported by the observed deformations in the failed bolted joint of the silo. An additional argument against the statement of the user related to the filling level is the fact that the silo was still filled to the middle of the tenth ring (counted from the top), although the whole neighbourhood was covered with slurry from the silo. Based on these findings it can be said that failure of this silo was caused by filling it to too high a level with liquid instead of forage. It cannot be completely excluded that a mix-up in the way of counting the rings has contributed to the failure. Whereas one would normally count the rings starting from bottom, as for buildings, the manufacturer of the silo counts the rings starting from top, because the silo is erected that way, assembling first the top, then putting rings under the top ring until the intended height of the silo is reached.

Acknowledgement--The calculations were performed by R. Primas, Section Materials and Structural Mechanics/Joining

Technologyof EMPA.