collapse and the local buckling of structures - critical_buckling

8/3/2019 Collapse and the Local Buckling of Structures - Critical_buckling

1/14

Collapse and the Local Buckling of Structures

Nar Sripadanna, P.E.

Abstract

The collapse of structures usually takes place when vertical members of the construction

suddenly buckle or when the lateral restraining system fails or does not exist. Material

housing structures such as tanks, bins, and silos all have one thing in common; most ofthem are constructed with walls that have large height to thickness ratios. Examples of

structural failures having been documented by the author are presented in this article.

The first case presented is a large rectangular grain storage facility that stored corn

product. The 20-foot high walls of the rectangular structure collapsed during loading ofthe product due to guy wires that were cut to allow forklifts to operate within the

structure. The second case involves a group of 45-foot tall by 16-foot in diameter grain

silos that developed localized buckling at approximately mid height. This type of phenomenon was wrongly attributed to wind loads from a storm event instead of the

negative internal pressures exerted on the silo shells from unbalanced product removal.

The final case discussed here is where a 10,000 gallon horizontal vessel used to separate

water from used engine oil collapsed during the operation of the vessel. Vacuum andtemperature differentials during the operation of the vessel caused the vessel shell to

buckle inward.

Collapse of Structural Frames

A 40-year-old steel structure clad with corrugated metal panels and a 150' x 400' footprintcollapsed after loading corn grain into the storage building. The structure burst open near

the middle of the longitudinal sides spilling the grain out onto the surrounding ground.

The perimeter wall of the structure was roughly 20 tall. The facility had been vacant forsome time prior to the collapse event and had previously been used for soybean storage.

The structure consists of latticed columns and sloping roof girders at every 20 feet on-

centers. Girths run horizontally across the latticed columns at 2 feet on-centers. Twocolumns equally spaced between the two main latticed columns provide resistance to the

lateral grain pressure. The grain storage facility complex contains several structures of

similar size. Some of the structures have guy wires or bracing of the intermediatecolumns along the perimeter. The columns for the structures without the guy wires have

a base plate with four anchor bolt moment connections and the columns for the structureswith guy wires have a base plate with two anchor bolt pinned connections.

Nar Sripadanna, P.E.


2/14

Fig. 1: Grain Storage Facility

The collapsed structure was constructed with guy wires or diagonal braces at the non-load bearing interior columns. The same columns in other structures of the complex are

constructed with moment resisting connections without diagonal braces or guy wires.

The guy wires in the collapsed structure were cut during interior preparations, as they

were obstructing the forklift operation. The warehouse employees made an attempt to re-

install guy wires prior to the filling of the grain structure, although not with the samematerial as the original braces or at the same locations. We observed that these guy wires

were not installed at everynon-load-bearing column as was done in the past, but that onlyevery other column was braced.

The lack of sufficient bracing caused the non-load-bearing perimeter columns to tilt out

thus breaking the simple non-moment resisting connection at the bottom. Once one

column failed then other columns on either side of the failed column either buckled orsnapped their pin connections. It appears that the corners of the structure were still

holding together due to hoop action of the roof-framing members. Only the perimeter

columns in the longitudinal middle appeared to have failed. If all of the non-load-bearingcolumns were braced properly, this failure could have been avoided.

The author became involved with a similar wall collapse incident when the roof of ahistoric building became damaged by fire. The repair contractor, in an attempt to shore

the unsupported walls, inadequately braced them by nailing pre-cast wall shores to the

wooden floor deck. The wall collapsed during a moderate wind event when the shores

became detached from the wood deck.


3/14

Ill. 1: Photo collection of steel frame structural collapse.

Lack of adequate bracing during construction and renovation has resulted in the collapse

of many structures. The photographs above show a warehouse distribution center thatcollapsed during construction when subjected to moderate wind pressures. The collapse

of metal building and steel-framed structures is common during erection. Many times,

heavy loads are placed on the unfinished structures, such as steel decking stacked on topof the steel framing, thus causing an unbalanced loading situation.


4/14

Buckling of Thin Walled Structures

Immediately after a significant wind storm event, the owner of a group of 6-43-6 tall

grain storage silos noticed indentations in silo surfaces. The 6 storage silos were empty

at the time of the windstorm event. The silos were reported to be 20 years old.

Fig. 2: Mapping of thebuckles in one of the group of grain silos.

The official weather reports obtained from the National Oceanic and AtmosphericAdministration, National Climatic Data Center indicate that wind speeds of 70 miles per

hour (61 knots) were recorded in the vicinity of the silos.

A review of literature on the collapse of grain silos yielded the following information.

In an article written by James Skaret, P.E. on the Insurance Canada Website titled Grain

Silo Collapse Wind or Other Phenomenon, Mr. Skaret indicates that what may appearto be wind damage often can be attributed to other phenomena that are not related to high

wind pressure.

Additional information along this line is found in another article, by John W. Carson

titled Silo Failures: Case Histories and Lessons Learned which was presented at the

Third Israeli Conference for Conveying and Handling of Particulate Solids, May 2000.According to this article, silos fail with a frequency which is much higher than almost

any other industrial equipment. Sometimes the failure only involves distortion or


5/14

deformation which, while unsightly, does not pose a safety or operational hazard. In

other cases, failure involves complete collapse of the structure

Silo design requires specialized knowledge. Material flow properties, flow channel

geometry, flow and static pressure development, and dynamic effects should all be taken

into account when designing a silo. Problems such as rat-holing and self-inducedvibration have to be prevented.

One of the most common problems that designers often ignore is the bending of circularwalls caused by eccentric withdrawal.

Ill. 2: View of mid-level skin buckling in vertical grain silos.

Typical Failure Example: A silo storing sodium sulfate consisted of a 14-foot diameter by

50-foot tall cylinder section, below which was a screw feeder. A significant inward dentdeveloped about mid-height in the cylinder section. It extended about a quarter of the

way around the circumference and was centered slightly offset from the long axis of the

screw at the back end. The problem was caused by eccentric withdrawal due to animproperly designed screw feeder.

Similar Example: A blending silo utilized 24 external tubes to withdraw plastic pellets atvarious elevations from the cylinder. Significant wrinkles developed in the cylinder

section above several of the external tubes.

Another factor to consider in the diagnosis of buckling walls is that the walls of outdoor

metal silos can expand during the day and contract at night as the ambient temperatures

fluctuate. If there are no discharges taking place and the material inside the silo is freeflowing, it will settle as the silo expands in the rising temperatures of the day. However,


6/14

as temperatures recede during the night, the stored material cannot be pushed back up as

the silo walls contract, so it resists the contraction, which in turn causes increased tensilestresses in the wall. This phenomenon, which is repeated each day that the material sits

at rest, is called thermal ratcheting.

Furthermore, an unusual loading condition can occur when moisture migrates betweenstagnant particles, or masses of stagnant particles, which expand when moisture is added

to them. If this occurs while material is not being withdrawn, upward expansion is

greatly restrained. Therefore, most of the expansion must occur in the horizontal plane,which will result in significantly increased lateral pressures on, and hoop stresses in, the

silo walls.

A properly designed and properly constructed storage silo should have a long life.

Unfortunately, this is not always the case. Problems can arise when the flow properties

of the stored materials change and/or the structure is changed due to normal wear andtear. If a different bulk material is placed in a silo other than the one for which the silo

was designed, obstructions such as arches and rat-holes may form where the flow patternand loads may be completely different than expected. When a poorly flowing material is

placed in a silo that was not designed to store and handle it, flow stoppages due toarching or rat-holing are likely. Sometimes these obstructions will clear by themselves,

but more often, operators will have to resort to various means to clear them. No matter

which method is used, the resulting dynamic loads when an arch or rat-hole collapses canbuckle or dimple the silo walls.

A pressurized cylinder is more resistant to compressive buckling than an unpressurizedone. In addition, if a bulk solid causes this pressure it is even more resistant. The reason

is as follows: Gas or liquid pressure is constant around a silos circumference andremains unchanged as the silo starts to deform. On the other hand, the pressure exerted

by a bulk solid against a silos wall increases in areas where the wall is deforming

inward, and decreases where the wall is expanding. This provides a significantrestraining effect once the buckling begins. However, if an arch forms across a silos

cylinder section and material below it is withdrawn, not only is the restraining effect of

the bulk solid lost, but the full weight of the silos contents above the arch is transferred

to the now unsupported region of the silo wall. Buckling failure is likely when thisoccurs.

Fig. 3: Low pressure on the side of the withdrawal.


7/14

Fig. 4: The computer model on the left shows the (exaggerated) deflected shape of anempty silo under a 70 mile per hour wind speed. The computer model in the middle

shows a localized buckled shape due to a loss of grain support during unloading

operations. The illustration on the right shows a close-up view of the localized, buckledshape of the silo wall due to a sudden loss of lateral grain support.

Based on the literature review and documentation of the distress to the group of six silos,

it can be concluded that the silos were not damaged due to the reported wind forces.

There was no distress at the base of the silos. Over-turning forces from wind, if

significant, should have caused distress at the base of the structure around the anchor boltarea. Fig. 5 below shows a typical deflected shape of the silo structure under 70 mile per

hour wind loading. Because the plate thickness of the structure is very thin, the walls of

the silo will simply deflect inward with the value of deflection starting at zero toward thetop and bottom and reaching a maximum value around the middle of the silo. Wind

forces cannot cause localized buckling of the silo wall.


8/14

Fig. 5: Shows stress distribution in the silo from a 70 mile per hour wind force. Note thatthe maximum stress is at the anchor bolts on the windward side.

The 70 mile per hour wind loading did not cause the stresses in the empty silos to exceed

the allowable steel plate stresses. The actual stress in the wall is 1,062 pounds per squareinch. The yield stress of A36 steel is 36,000 pounds per square inch. The maximum

deflection at the middle of the silo due to the wind load is 0.067 inches. Therefore, the

wind loading did not cause visible or obvious distress to these silos.

The localized buckling reported by the owner had been present prior to the wind event

and the owner probably happened to notice it while inspecting the property for damage

after the wind storm event occurred.

Collapse of a Pressure Vessel

This case history describes a collapsed pressure vessel while being used to remove water

from used automobile engine oil.


9/14

Ill. 3: Oil/Water separation unit collapses after shutdown due to an equipment failure.

Specifications and operation of the oil/water separation unit:

The vessel was approximately 9 years old. Vessel capacity was roughly 10,000/12,000 gallons. The vessel had a 2 in diameter pipe at the top front side spraying 200 F oil to be

recycled.

At the bottom of the vessel, coils would circulate oil at roughly 350 F to heat the oiland water mixture that was pumped into the vessel at the bottom front side.

A vacuum line then removed the water vapors from the top middle portion of thevessel.

There was a shutoff valve on the vessel and a vacuum relief valve on the line thatcreates vacuum in the vessel.

Reportedly, the vacuum relief valve should trigger if the vacuum in the vessel exceeds 15inches of mercury (Hg). There was a vacuum gauge next to the vacuum relief valve and

a vacuum gauge on the vessel itself. A water pump that circulates water from an

underground tank creates the vacuum.


10/14

Just before the collapse of the vessel, a pump that removes the oil from the bottom of the

vessel failed. The operator of the vessel informed us that everything was shut down atthat time. The oil pump was shut down at 10:00 AM and the vessel collapsed at 1:30

PM. The temperature gauge on the vessel read 170 F and the vacuum gauge on the vessel

read 12 of mercury immediately after the collapse. The operator also reported that

during operations of the past, the vessels vacuum gauge had indicated a vacuum of ashigh as 22 of mercury. Ideally, this was not possible since the vacuum relief valve

should have triggered at 15 of mercury. This would have meant that the relief valve on

this vessel was not functioning properly.

On a typical day of operation, the oil to be recycled is sprayed at the top while the

recycled oil is removed at the bottom. After a few hours of this process, a vacuumbuildup in the vessel occurs due to the constant removal of water vapors. To remove any

excess oil in the vessel, a 1 diameter bleed valve at the back of the vessel needs to be

opened. Typically, the vessel operator will also open the top hatch to allow the oil pumpat the bottom to drain the oil from the bottom of the vessel. In this case, the vessel

operator also mentioned that he normally tries to keep the maximum operationaltemperature at 250 F.

On the day of the vessel collapse, the system became operational at 7:00 AM in the

morning. Then at 10:00AM a problem developed in the oil pump that drains the recycled

oil at the bottom of the vessel. The vessel operator then shut the system down, whichincluded the pump that sprays the oil to be recycled at the top, the oil pumps in the

heating coils, and the vacuum pump. However, the vessel bleed valve at the back of the

vessel was not released to eliminate the vacuum in the vessel. The temperature in thevessel cooled down from a 250 F maximum to 170 F. This drop in temperature caused

vapors in the vessel to reduce in volume, which resulted in more suction in the vessel.The oil level in the bottom of the vessel increased as the oil pump that pumps the oil out

of the vessel failed. At the time of the vessel collapse, there was 3-6 to 4-0 of oil in

the vessel. Because there is oil at the bottom, a partial loading condition was created.Usually, circular shapes exhibit higher load resistance if they are loaded uniformly rather

than partially.

This phenomenon can also be explained by creating a computer model and comparing thestresses in the vessel skin between full and partial loading conditions. The system

operator suggested that the vacuum relief should trigger at 15 of mercury. It is possible

that the vacuum in the vessel was at or below 15 of mercury. However, it still failedbecause of the fair amount of oil at the bottom that created the partial loading condition

thus contributing to the vessels collapse.

Another factor to be considered is differential temperature in the vessel skin. As there is

hot oil at the bottom, the top surface cools more rapidly when compared to the bottom

area where there is oil. This results in thermal stresses that would exacerbate thepartialloadingproblem.It is also quite possible that the valve on the vacuum line was shut or became stuck thus


11/14

preventing the vacuum relief valve from functioning. Cooling of the vapors created more

vacuum in the vessel than was already there. This factor could also have contributed tothe vessels collapse.

Theory Behind Critical Buckling

Most engineers are familiar with the Euler buckling formula. Presented below is an

example column with dimensions where the Euler buckling formula is then applied.

Fig. 6: Example with column dimensions.

Euler buckling formula applied to the example above. In most cases, this formula is part

of the code checking process.

Pcr

229000 ksi .844 in

4

120in( )2

:=

Pcr 16775.586lb=


12/14

However, most engineers do not realize that this buckling phenomenon can occur not

only to an individual member, but that the entire structure could buckle. Some times thebuckling mode shape of the structure is not very intuitive. By usage of the finite element

program, the critical buckling load and the buckling mode shape can be determined.

Fig. 7: Shows finite element results of a critical buckling analysis. The buckling load

multiplier on a 1 pound load is 16765.3 lbs. This is close to the Euler buckling formula

calculated value above.

While finite element analysis is not required to calculate the critical buckling load for acolumn, a complex structure can buckle in ways we cannot easily envision.

Figure 8 below shows the buckling mode of an un-braced bottom truss chord. Note that

the critical buckling load multiplier is a negative value. This means that the truss bottom

chord will buckle if the loading is reversed.


13/14

Fig. 8: Shows the buckling mode of un-braced bottom chords of a truss system.

Conclusion

Most engineers design structures for strength. However, some thought process needs to

go into the structural stability during the service life or during repair and retrofit of thestructure.


14/14

References

Load Development and Structural Consideration in Silo Design by J. W. Carson Ph.D.,

R.T. Jenkyn, P.Eng. A web article. Original Source: Carson, J. W. and R. T. Jenkyn:

Load Development and Structural Considerations in Silo Design. Presented at ReliableFlow of Particulate Solids II, Oslo, Norway, August 1993.

Article from the Insurance Canada Website titled Grain Silo Collapse Wind or OtherPhenomenon by James Skaret, P.E.

collapse and the local buckling of structures - critical_buckling

Documents