TSUNAMI IMPACT ON FUEL STORAGE CONTAINERS Hillary Brooker Lehigh University Project PI: Clay Naito Lehigh University NEES REU Program August, 2011
Abstract The Tohoku Japan tsunami in 2011 caused many fuel storage containers in Japan to fail. Failure of these containers resulted in major damage to Japanese cities when fuel leaked out of the containment basins that surrounded most storage areas. Oil damage is difficult to clean up and often precludes rehabilitation of any contaminated structures. There are four major types of failure, crushing of the container, loss of tie down restraints, lateral movement, and debris impact of the receding wave. This paper studies the conceptual hydrostatic, buoyant, and hydrodynamic loads generated by the tsunami inundation that cause the first of these three failure modes. These forces have more of an effect on the containers when they are close to being empty. Calculations of the demands and capacities of a typical container support the failure modes observed. A proposed retrofit is provided to prevent future oil spills from tsunamis. 1 P a g e
Introduction As a consequence of high inundation velocities and depths during tsunami events fuel storage containers can fail and move considerable distances. During the tsunami in 2011 one facility located near sea level exhibited complete loss where all of the containers were moved off their bases. After the event, containers from the facility were found around the region, in one case at a distance of over 1 km from the original location. The second facility located at approximately 16m above sea level was also damaged resulting in movement of the storage containers outside of the containment walls; however due to higher elevation above sea level the containers did not travel as far. This is the location of the standard container used in this report. The movement of storage containers from their walled containment areas can result in significant oil contamination to the buildings in the region. Literature Review Previous fuel storage container failure research has been mostly conducted based on hurricane events such as Katrina and Rita. Field inspection after Hurricane Katrina illustrated a number of failure modes for containment structures. These failure modes can be modeled by FEMA P646 [FEMA 2008] which explains eight types of forces exerted on a rectangular structure. These forces include hydrodynamic, buoyancy, hydrostatic, impulse, uplift, debris damming, debris impact and added gravity due to retained water in the floors of the building. Damage to fuel storage containers caused by extensive flooding and high winds stopped approximately 50 percent of America s oil production and 25 percent of its fuel production. Approximately 7 million gallons of oil were spilled from industrial plants, storage depots and other facilities around southeast Louisiana. Damage was due to buoyancy failure of the containers as many had no tie down restraints and low fuel levels. Also, winds exerted high pressure on the containers which caused them to buckle. [Godoy 2007] Many containers suffered collapse from wind pressure without evidence of being floated off foundations. [NIST 2006] Fuel storage container systems are typically designed as cylindrical thin shell elements to support the lateral pressure demands generated when filled. Under conventional loading the wall of the container is subject to tensile stresses which are readily resisted by the thin steel shell and standard welds used to fabricate the cylinder. In regions of high seismicity and in areas where high wind loads are possible additional design requirements are prescribed. [API 2007] These demands often result in tie down requirements between the base of the container and the supporting foundation to prevent overturning and translation of the container. As a safety precaution in Japan the containers are often attached using flexible piping connections to allow for potential movement during seismic events and are fully surrounded by a reinforced concrete wall to contain loss of oil from the container. Seismic and wind design requirements do not provide adequate protection to fuel storage containers during tsunami events.. Earthquake induced damage can be characterized by elephant foot or diamond bucking of the base of the container, anchorage failures, base sliding, and 2 P a g e
sloshing damage to the upper shell and roof [Malhorta, Wenk, and Weiland 2000]. Damage to fuel storage tanks during the 2004 Indian Ocean Tsunami were observed by Goto [Goto 2005]. He discovered instances of failure due to sliding, floating, and buckling. Sakakiyama [Sakakiyama et al. 2009] performed 1/40 scale hydraulic model tests for an idealized case of a single cylindrical tank and for a site-specific power plant with multiple tanks and containment walls to estimate tsunami forces on the tanks. They used finite modeling to study the pressure distribution around the tank and found that buckling occurred near the point of maximum external pressure. Their study recommended that the lowest oil level in the tank should be kept higher than the inundation depths to prevent floating and buckling. This paper discusses the sliding, floating, and buckling in further detail and presents an additional solution to buckling as well as hydrostatic and hydrodynamic failure. Failure Modes The first failure mode observed in the Japan reconnaissance trip was crushing. Crushing occurred when the level of fuel in the container was low and there was a net hydrostatic force inward on the tank was from the tsunami wave. Fuel storage containers are designed to support an outward net hydrostatic force from the fuel, but cannot support this inward force. When the hydrostatic force from the tsunami wave became too large the containers experienced crushing as shown in Figure 1. Figure 1: Crushing due to hydrostatic loading The second failure mode observed was buoyancy failure. Buoyancy failure occurred when the buoyancy force exerted by the rising inundation height overcame the weight of the container and oil and the tie down strength of the bolts. This caused a net upward uplift force which pulled the containers from the ground and caused them to move away from their base. In cases of low elevation, the containers moved over 1000m. Figure 2 depicts a container at a higher elevation that was subject to buoyancy failure at a lower inundation height which caused the container to tip over and move slightly outside of its containment area. The container on the right is the container used later in the case study. 3 P a g e
Figure 2: Tipping and sliding due to buoyancy failure The third failure mode observed was hydrodynamic failure. Hydrodynamic failure occurs when hydrodynamic forces from the tsunami wave are large enough to overcome the friction between the base of the container and the concrete floor and as well as the sheer force of the tie downs if applicable. Hydrodynamic failure occurs when there are high flow velocities and the tank is pushed along by the flowing water in the direction of the wave. Figure 3 shows an example of a container subject to the hydrodynamic failure with no tie downs. Figure 3: Sliding due to hydrodynamic Failure Conceptual Loading A tsunami exerts eight types of loads on a rectangular structure [FEMA P646]. These loads are hydrodynamic, buoyancy, hydrostatic, impulse, uplift, debris damming, debris impact and added gravity due to retained water in the floors of the building. The Tohoku tsunami in 2011 showed five of these forces had a significant effect on fuel storage containers. This section will discuss 4 P a g e
the first three in detail and explain equations used in the case study. Forces from debris impact and damming may have caused some buckling failure observed but will not be covered in detail. The remaining three loadings did not have a significant effect on these containers. The impulse from the initial wave front was reduced significantly by the concrete containment wall surrounding the containers. The uplift force referred to in FEMA occurs due to water rising under the upper floors of the structure and therefore did not affect the fuel containers. Figure 4 shows the conceptual hydrostatic loading exerted on a container in the empty and full case. The container on the left is empty and therefore there is no outward hydrostatic force acting on the container walls. This container would most likely be subjected to a crushing failure. The container on the right is full and therefore does have an outward hydrostatic force. This container would be less susceptible to failure. Figure 4: Hydrostatic forces on an empty and a full tank Hydrostatic forces on a wall panel were determined according to FEMA P646 by the formula shown in Equation 1 where ρ is the density of seawater, g is the gravitational acceleration, b is assumed to be one foot, and h is the inundation height. This is a per foot calculation because it is assumed that a one foot wide strip of the container is approximately flat. F hs = ½ gρbh 2 (Equation 1) Figure 5 shows the conceptual buoyancy loading exerted on a container. The container is subject to an upward buoyancy force from the tsunami wave as well as three downward forces from the weight of the container and oil and the force of the tie downs. When the buoyancy force is stronger than the downward loads, a resulting uplift force is exerted on the container which pulls it from the ground. Tanks are much more susceptible to uplift forces when they are empty since the only force resisting the buoyancy force is the weight of the container. 5 P a g e
Figure 5: Conceptual Uplift Forces Equation 2 was used to determine the uplift force on the container where U is the uplift force, B is the buoyancy force, W f is the weight of the fuel, W c is the weight of the container, and T is the tie down force. In order for buoyancy failure to prevented, the uplift force must equal 0. U = B - W f - W c T = 0 (Equation 2) Calculations of the uplift force assumed T to be zero in order to calculate the tie down force necessary to support the container. In T = B - W f - W c ; U = 0 (Equation 3, T is the tie down force necessary to hold the tank to the ground making the uplift force 0. T = B - W f - W c ; U = 0 (Equation 3) The upward buoyancy force is calculated using B = ρv d (Equation 4 where ρ is the density of the seawater and V d is the volume of the water displaced by the tank. The calculations assumed that the container remained attached to the ground until failure; therefore the volume displaced was the area of the base of the container multiplied by inundation depth until the inundation height reached the container height. The volume displaced was the area of the base times the height of the container for all inundation heights above the height of the container. B = ρv d (Equation 4) 6 P a g e
Figure 6 shows the conceptual hydrodynamic loading on a container from a plan and elevation view. The hydrodynamic force is stronger at the center of the container because of the way in which the water flows around it. At the edges, there is less resistance because the water can change direction more easily to avoid the container. The hydrostatic force is opposed by the friction between the container and the base of the container as well as the shear forces of the tie down bolts if they are used to support the container. Figure 6: Conceptual Hydrodynamic Loading Equation 5 is used to determine the hydrodynamic forces on a container based on standard fluid mechanics for a cylinder in a flow field [Roberson, & Crowe, 1990] where ρ s is the density of seawater, Cd is the drag coefficient, A is the diameter of the container multiplied by the inundation height and V is the wave velocity. Cd for this formula was determined using the Reynolds number calculated by Equation 6 where V is the flow velocity, d is the diameter of the container, and ν is the dynamic viscosity of seawater. F hd = ½ρ s C d AV 2 (Equation 5) Re = Vd/ν (Equation 6) Case Study A case study was performed on a container in Onagawa for three set situations at different inundation depths. The standard tank was made of ASTM A36 steel with a diameter of 14m (45.93ft) and a thickness of 3mm (0.0098ft). The height of the tank was estimated to be 15.24m (50ft) because it was not specifically measured. The tank also had 22 ASTM F1554 Grade 36 anchorage bolts each with a diameter of 2cm (0.79 in). Calculations were performed to determine the conceptual net hydrostatic, uplift, and hydrodynamic loads on this tank. Figure 7 displays the net hydrostatic forces between the oil and tsunami wave. The outward hydrostatic force from the oil is assigned to be negative and the inward force from the water is positive. In this case, any point above the x-axis has a net inward force and is subject to failure. It is beyond the scope of this paper to determine at which specific point the tank will fail, but it 7 P a g e
Uplift force and capacities [kips] Hydrostatic Load [kips] can be seen that a full tank has a much higher chance of withstanding crushing failure than an empty one. 200 150 100 50 0-50 -100 Empty Half Full Full 0 10 20 30 40 50 60 Inundation Depth [ft] Figure 7: Hydrostatic load vs. inundation height Figure 8 displays the amount of force that the tie downs must exert in order for the tank to remain grounded. Lines level off after 50 feet because at that point the inundation depth is greater than the tank height and the volume of seawater displaced always equals the total volume of the container, so the buoyancy force becomes constant. The horizontal line labeled bolt failure is the tie down capacity of the standard container in Onagawa. If the container was half full of oil at the time of the event, the inundation height of the wave would have to reach approximately 22 feet for the tank to be subject to buoyance failure. The horizontal line labeled container failure is the point where the walls of the container would rip away from its base. This failure would never occur in reality because the tank fails due to bolt fracture before the uplift force reaches this level. Also, the buoyancy force levels off after 50 feet at which point the inundation height is not high enough to create this much force. 12000 10000 8000 6000 4000 2000 0-2000 -4000-6000 Empty Half Full Full Container Failure Bolt Filure 0 20 40 60 Inundation Height [ft] Figure 8: Uplift vs. inundation height 8 P a g e
Hydrodynamic Load and Capacities [kips] Container Height Figure 9 is a graph of the hydrodynamic loading vs. the inundation height. The first four data series display the hydrodynamic force on a container at velocities of 5, 10 15, and 20 mph calculated using Equations 5 and 6. The last three data series are the resistances of friction and the bolt shear strength in the empty, half full and full case. The friction changes with the inundation height because an increasing inundation height has an increasing buoyancy force which in turn lightens the weight of the oil and reduces the friction force. The horizontal line at which the friction forces level off represents the bolts shear strength. 2000 1500 1000 Standard 5 mph Standard 10 mph Standard 15 mph Standard 20 mph Bolt Strength Friction Empty & Bolt Friction Half & Bolt Friction Full & Bolt 500 0 0 20 40 60 Inundation Height [ft] Figure 9: Hydrodynamic loading vs. inundation height Proposed Retrofit To remediate failure of the fuel storage containers during tsunami (and flood events) a cable tie system is proposed. Tie down failure of an empty container is very costly due to the large amount of tie downs necessary to support it. Instead the system allows failure to occur and relies on a cable tether system to hold the container in place within the containment area. The cables would initially be slack and would tighten as the inundation height increased. They would be designed for a maximum inundation height based on tsunami models and previous tsunami occurrences so that they would not be subjected to the large uplift forces that the tie downs are subjected to. They would also be slightly elastic to allow for some fluctuation in the inundation height. Future Work 9 P a g e
In the future, the proposed retrofit must be developed in more detail. Calculations need to be performed to determine the strength of the cable supports. Specifications for how to attach the cables to the container and the ground must be designed. A maximum inundation depth must be determined in order to calculate the length of rope needed to support the containers. Contact Information Hillary Brooker hmb213@lehigh.edu Clay Naito cjn3@lehigh.edu Acknowledgements The project was partially funded by NSF through the George E. Brown Network for Earthquake Engineering Program (CMMI-1041666 and EEC-1005054). Special thanks to my advisor and mentor Clay Naito for helping me with the loading diagrams and designing the solution. Photographs for figures 1 through 3 were taken by Clay Naito in Onagawa Japan. Also thank you to the following for support on this project. Daniel Cox, Payam Aghl, Pat Trasborg, Kent Yu, Kristen Peterson, John Archibald. References Roberson, and Crowe. (1990). Engineering Fluid Mechanics, 4nd Ed., Boston, Massachusetts. National Institute of Standards and Technology (NIST), Performance of Physical Structures in Hurricane Katrina and Rita: A Reconnaissance Report, NIST Technical Note 1476, Gaithersburg, MD, June, 2006. American Petroleum Institute (API), Welded Steel Tanks for Oil Storage, API Standard 650, Eleventh 22 Edition, June 2007. Godoy, Luis A. (2007) Performance of Storage Tanks in Oil Facilities Damaged by Hurricanes Katrina and Rita, Journal of Performance of Constructed Facilities Federal Emergency Management Agency (FEMA), Guidelines for Design of Structures for Vertical Evacuation from Tsunamis, P646, June 2008. Y. Goto, Reconnaissance on Off-Sumatra Earthquake Indian Ocean Tsunami Disaster of December 26, 2004 Damage to Oil Storage Facilities of Cement Factory at Lho nga and Petroleum Delivery Terminal at Krueng Raya in Aceh Province, Sumatra, Indonesia, National Research Institute for Earth Science and Disaster Prevention Earthquake Disaster Mitigation Research Center, 27p, 2005. Malhorta, P., Wenk, T., Wieland, M., Simple Procedure for Seismic Analysis of Liquid-Storage Tanks, Structural Engineering International, Vol.10, No. 3, August, 2000, pp. 197-201. 10 P a g e
Sakakiyama, T., Matsuura, S., Matsuyama, M., Tsunami Force Acting on Oil Tanks and Buckling Analysis for Tsunami Pressure, Journal of Disaster Research, Vol.4, No.6, 2009, pp.427-434. 11 P a g e