Safe Room Wall Design for Tornado Survival

A safe room is a reinforced structure specially designed to meet the Federal Emergency Management Agency (FEMA) criteria and provide protection in extreme weather events, including tornadoes and hurricanes. The materials used for safe room walls are expected to resist loads imposed and tornado debris impact. Tornadoes rated as high as EF-5 can create maximum wind speeds of more than 300 mph, which is enough to demolish any structures in their path. These maximum wind speeds produce forces that are about twice as large as those produced by the strongest hurricanes. One of the common safe room wall designs is made of plywood/steel-plate composite. The main objective of this paper is to use a finite element simulation code, LS-DYNA, to predict the dynamic response of plywood when impacted by tornado missiles such as 2 x 4 wood timbers.


Introduction
The safe room project is part of a continuing FEMA initiative named Project Impact: Building Disaster Resistant Communities designed to urge people and communities to take measures to protect themselves and their property before disasters occur.
Throughout history tornadoes have signified a force of nature that is both admired and feared. Tornadoes tend to strike with little warning, without regard to season, and just as quickly they are gone. The result is often devastating in terms of property damage and the human toll. Until recently, the fundamental knowledge of tornado behavior escaped explanation. As our understanding of tornadoes advances, it is more evident than ever that tornadoes are not magical entities. They are natural phenomena that follow the laws of physics, and as such are excellent subjects of deep research. Most of the research regarding the tornado belongs to the atmospheric sciences, but this status is experiencing a change. With the introduction of engineering into the sphere of tornado research, a new paradigm is changing the way that the world looks at these natural wonders.
While tornadoes occur in different parts of the world, the highest concentration of tornadoes occur in the United States [1]. Approximately 1,200 tornadoes per year strike the United States; more than any other nation in the world. According to the National Climate Data Center from 1953 through 2005, there were 48,632 reported tornadoes in the United States, resulting in 4,388 deaths [2]. The most devastating tornado in recorded history took place in 1925. The path of this tornado (or possibly a family of  tornadoes) extended through Missouri, Illinois, and Indiana. During the 219-mile track (at times 1200 yards wide with a forward speed of up to 72 miles per hour), 695 people were killed, 13,000 injured, and caused $17 million in property damage [3]. This tornado stayed on ground for 3 hours. Though tornadoes do not represent the greatest threat against life in the United States, they embody a particularly disturbing hazard.
Debris is a hidden danger in tornado. People can be easily injured by flying debris. One of the strongest tornadoes, rated an EF-4 ripped though the Union University campus in Jackson Tennessee in 2008. The damage at Union University was severe.
Automobiles became debris and some of the dormitories suffered catastrophic damage (see Figures 1  and 2) [4].
The materials used for safe room walls are expected to resist loads imposed and tornado debris impact. Tornadoes rated as high as EF-5 can create maximum wind speeds of more than 300 mph, which is enough to demolish any structures in their path. These maximum wind speeds produce forces that are about twice as large as those produced by the strongest hurricanes. The main objective of this paper is to investigate the performance of safe room wall assemblies when subjected to the impact and penetration of windborne debris.

Penetration mechanics
Widespread research exists on penetration problems by many researchers to develop fundamental relationships applied to areas such as hypervelocity impact, shaped-charge penetration, long-rod penetration, small arms, ballistic protection, and armor design. The basic understanding of penetration mechanics is as follows. Given a projectile, a target, and details of the initial geometry, kinematics, and materials properties; investigate whether or not target perforation occurs. If perforated, investigate what the residual characteristics of projectile and target will be, and if not, investigate the depth of penetration.
Penetration mechanics is one of the most involved problems in the research field of mechanics, and researchers have been working on the solutions many years. Solution approaches exist on three different levels as follows [5]: • Data correlation  [6, 7, 8, 9, and 10]. Zukas and Anderson reviewed the state of the art of numerical simulations which describes the impact and penetration mechanics including a comprehensive review [11]. They examined the general capabilities and limitations of numerical simulations. They stated that as the size and speed of computers have increased so has the complexity of codes used for these purposes; but certain characteristic difficulties persist [12].

Modeling
Composite materials have been used widely in many applications such as in the oil industry, space structures, pressure vessels, and automobile industries. Many researchers have investigated a solution to the impact on composite laminates using finite element method [13,14,15,16].
Plywood is a layered, cross-ply, unidirectional, fiber-reinforced composite. Plywood strength and progressive failure resistance are important for the survivability of house frames when earthquakes, tornados, and/or hurricanes occur. Plywood is made of thin sheets of timber called plies. The plies are cut by rotating the trunks then stacking them together with the fiber direction of each ply perpendicular to the fiber direction of its adjacent ones. The plies are bonded using strong adhesive under heat and pressure. In general, plywood has an odd number of plies [17].
One of the common safe room wall designs is made of plywood/steel-plate composite. In this research, numerical simulation LS-DYNA analysis code predicts the dynamic response of plywood when impacted by tornado derbis such as 2 x 4 wood timbers will be used. Due to the differences in speed between the types and classes of wind events, specific criteria were developed to simulate the results of such an impact event.
In this paper, the LS-DYNA analysis code is used to identify the damage of the composite plywood impacted by a rigid projectile. Because of the symmetry of the problem, only one quarter of the geometry is modeled. In this case, the plywood is assumed to be five layers of composite material. In order to reduce the computational time while keeping the inertia of projectile same as 15 lb 12 ft 2x4 wood timber, a 1.5 inch projectile was used. The projectile was assumed to be rigid and the target (3/4 plywood) assumed to be elastic-plastic material. In modeling, solid elements were used because of the ease of use and better stability of contact problems (Chatiri, Gull and Matzenmiller n.d.). Each layer of the target is 8" x 8" x 0.15" and the projectile dimensions are 1.75" x 0.75 x 1.5". The density of the projectile arranged such that the projectile part has a weight of 15 lbs.
The projectile and the target are discredited by 8-node hexahedron solid elements, using only one integration point. The target has simulated 5 layers through the thickness. The impact region has a fine mesh with element size of 2.12mm x 2.12mm x 3.81mm for each layer. The transition region has an element size of 2.12mm x 4.23mm x 3.81mm, and the transition region's coarse mesh size is 4.23mm x 4.23mm x 3.81mm. The element size for the pro-jectile is finer than that of the impact zone of the target which is 1.27 mm x 1.27mm x 3.81mm. Each layer of the target has 3933 brick elements therefore target has total 19665 brick elements. Projectile has 1818 brick elements.
As shown in Figure 3, two different kinds of boundary conditions are used because of the symmetry of the problem-fixed-boundary condition and symmetry-boundary condition. The symmetry boundary condition is applied on the bottom and right face of both the target and the projectile. The fixed boundary condition is applied on the model on the upper and left face of the target. The bonding between the plies is modeled with "contact tie break surfaces," which has the fracture-mechanics-based delimitation capability. The contact was defined between each ply, allowing them to delaminate from each other. Moreover, eroding contact surfaces are modeled to control the impact force between the projectile and the target. The most important factor of the contact surfaces is that it helps to redefine itself after an element fails and is removed from the model.
To clarify the characteristics, side and back views of the simulation at 2. In some cases there was no deformation found after the impact and the projectile returns to the direction from which it came.
Identical simulation results are obtained until we reach the velocity of 13.9 m/s, meaning, deformation was elastic and no penetration was observed. However, as seen in Figure 5, penetration has occurred at velocity 13.9 m/s. It is clear from this figure that the penetrating projectile has extricated a solid block in the positive x-direction.
Penetration phenomenon is also evident from   velocity). This shows that the target absorbed most of the kinetic energy. As shown in this figure, the velocity remained in the positive region. As shown in Figure 6(a), the projectile travels at 4 m/sec in the positive x-direction and made contact with the target at which point the projectile and the target exchange energy -the projectile is deflected and travels in the negative x-direction with the rebound velocity of 3.75 m/sec. This makes the rebound velocity 94% of the initial impact velocity. This illustrates that the target absorbed very little of the kinetic energy from the impact. As shown in this figure, the transition of the velocity from positive to negative is exposed. Figures 7 and 8 show the impact velocity profiles for all the velocities mentioned earlier. Figure 7 illustrates the first set of example. All the projectile impact velocities in this set fails to penetrate. As shown in this figure, rebound velocity occurs between 4 m/s and 13.8 m/s with the projectile deflecting in the x-direction (below zero) because of the energy exchange between the projectile and  the target. However, as shown in Figure 8, all the projectile impact velocities penetrate the targetresidual velocity occurs between 13.9 m/s and 45 m/s with the projectile penetrating the target. Simulation at an impact velocity of 13.9 m/s is important as it predicts the limiting velocity for penetration. Texas Tech University test results indicate that the range of limiting impact velocity for penetration is between 13.2 and 14 m/s (29 and 31 mph). Note that since the simulation result at this impact veloc-ity indicates that the remaining (residual) velocity is 2.75 m /s, the limiting velocity for penetration is somewhere below 14 m/s (31.3 mph). This is in reasonable agreement with the TTU experimental results. Figures 9 and 10 illustrate the kinetic energy changes between projectile and the target (plywood). As shown in Figure 9, for example, when the projectile impact velocity is 4 m/s, the kinetic energy of the projectile decreased rapidly after the impact. Since the projectile did not penetrate but changed direction, the velocity went to zero as the direction changed and then kinetic energy increased rapidly. However as shown in Figure 10, the kinetic energy of the projectile traveling at 13.9 m/sec decreased rapidly after the impact of the system. Since there is penetration, the target absorbed most of the energy -after the penetration projectile has almost zero velocity Penetration limit velocity, V L as mentioned earlier, is the minimum velocity level for penetrationof the projectile to embed in the target. In other words, when the projectile passes through the thickness of the target, remaining velocity becomes zero. For impacting velocities smaller than the penetration limit velocity, the projectile rebounds from the target. In this study, V L -13.9 m/s is the limit velocity to penetrate one plywood. This value can also be verified by using following equation [1]: Where R t is the strength of the plywood (taget), w is the thickness of the target, and Where m is the mass of the penetrator (projectile), A is the penetrator average cross-sectional area, and ρ t is the density of the target. Using the data given in this paper, Eqn (1) provides limit velocity of approximately 14.3 m/s. Figure 11 shows penetration limit velocity with respect to moisture content of the plywood. As seen from the figure, fully saturated wood is weaker than the less saturated wood. The penetration velocity is 13.6 m/sec for 30%, 13.73m/sec for 20%, 13.8 m/sec for 10%, and 14 m/sec for 1%. Thus, as the moisture content decreases, the material becomes stronger.

Conclusion
A 15 lb. 2"×4" rigid projectile impact on plywood composite panels has been studied by LS-DYNA for modeling the progressive failure behavior of a one single plywood composite layers. The Texas Tech University Wind Science and Engineering Research Center's study on the wind-generated missile impact on a composite wall was a major motivation on this study. Existing TTU experimental test data of 1 layer of 3/4 in plywood impacted by a 15lb 2"×4" board is used to guide and judge the finite element model development. Dr. Atila Ertas is a Professor of Mechanical Engineering and Director of Transdisciplinary Ph.D. Program on Design, Process and Systems. Ertas is author/co-author of four books and co-editor of more than 35 proceedings. He has been teaching capstone design classes for more than 25 years. He is a Senior Research Fellow of the IC 2 Institute at the University of Texas Austin, a Fellow of American Society of Mechanical Engineers(ASME), a Fellow of Society for Design and Process Science (SDPS) and a Fellow and honorary member of The Academy of Transdisciplinary Learning & Advanced Studies (TheAT-LAS). He is also an honorary member of International Center for Transdisciplinary Research (CIRET), France. He has published over 150 scientific papers that cover many engineering technical fields. He has been PI or Co-PI on over 40 funded research projects. Under his supervision more than 180 MS and Ph.D. graduate students have received degrees.