Challenges in Designing for Tornadoes

As previously mentioned, the research goals related to tornado-based structural and building envelope design consists of building understanding of tornado climatology and near-surface flow characteristics. These include the horizontal and vertical components of motion and the spectrum of scales of velocities within a tornado flow, which contribute in various ways to tornado loading. Most current understanding of these quantities relies on laboratory and/or numerical simulations with few observations of real tornadoes providing verification. Addressing deficiencies in these approaches (e.g., lack of realistic turbulence, boundary layer depth and structure, etc.) demands additional data collection of real tornado flow fields, which are often difficult to capture.

Direct observations of tornadoes offer numerous logistical difficulties. From a forecasting perspective, pinpointing the location of a tornado in advance of storm formation remains nearly impossible. As such, deployment of instrumentation, particularly such that requires being deployed on the order of minutes (or longer) before the storm and/or formation of a tornado is difficult. For in situ measurements, deployment of such instrumentation in the path of the tornado requires doing so at far enough distances from the storm that operators can safely evacuate, and the farther the associated distance, the more difficult it becomes to pinpoint a tornado's path. Remote sensing techniques (e.g., mobile radar) provide an easier means of observation logistically but offer additional difficulties in terms of collecting the desired data, including the inability to sample near the ground except at very close distance due to beam geometry; volumetric sampling, which often limits the number of samples possible in a tornado; and the inability to sample in close proximity to surface obstacles (e.g., structures, terrain, etc.) due to beam blockage and other interference. For these reasons, expanding the useful tornado dataset requires continued attempts to sample tornadoes, especially at close distance to maximize the potential for near-surface observations. In addition, other techniques, such as dust devil sampling, may provide realistic depictions of full-scale tornadoes but at a higher frequency and without safety concerns.

From a climatological perspective, the techniques that are commonly employed for wind speed estimation of tornadoes are conducted post-event (i.e., via ground or aerial damage survey) and do not pose the same data collection difficulties as direct observations. However, the long-term vision for research in this realm involves unifying and improving the tools used for data collection and analysis among all interested parties. Given the heavy involvement of the National Weather Service in both tornado damage survey operations and the creation and maintenance of the current data storage tools, modification of these techniques and tools may require considerable time from proposal to implementation due to governmental procedures. In addition, a strong desire exists to further unify engineering, meteorology, and social science partners in data collection efforts. Given the large variance in the typical goals of these groups, converging on a set of universal goals and methods that sufficiently satisfies their unique research objectives may prove difficult.

Wind load measurements in tornado simulators pose many challenges well beyond those associated with boundary layer wind tunnel tests for straight-line winds upon which all the wind loads in ASCE/SEI 7-22 are based. These challenges relate to the fact that each time a nominally identical moving tornado is simulated, the exact path the tornado takes differs owing to “wandering” of the tornado. This wandering results in the need to replicate each experiment numerous times. Currently no guidelines specify the number of simulations required to achieve reliable results, with different investigators using different numbers of simulations. Real-world tornadoes can travel much faster than the translation speeds that can be modeled in existing tornado simulators.

Furthermore, because tornadoes wander, the location where the minimum atmospheric pressure difference occurs differs each time. Consequently, when the results from each experiment are averaged together the result is a single tornado with an apparently larger core radius and a lower atmospheric pressure drop (i.e., a broader and weaker tornado).

Whether the tornado design philosophy in calculating external tornadic loads on a structure needs to be shifted (where the effects of the APC and the wind-induced (GC_p)’s are combined), or whether the current approach is maintained (where the two are separated) needs to be addressed. Each approach has advantages and disadvantages. The first approach automatically includes differences in tornado-induced loads relative to atmospheric boundary wind-induced loads caused by the atmospheric pressure drop and changes in the characteristics of the (GC_p)’s due to differences in the flow field, but the approach may be limited by the number of combinations of tornado characteristics and building configurations needed to develop a standard. The second approach, which removes the effects of the atmospheric pressure drop and develops factors to adjust the (GC_p)’s for tornado loads, clearly requires approximations but the methodology can be applied to a wider range of building geometries.

In calculating the tornado-induced internal pressure on a structure, the effects of the atmospheric pressure drop relative to the total loads acting on a building are lessened due to the porosity of the building envelope in many buildings. The effective internal pressure will take on a weighted average of the overall openings in the building envelope, including leakage around overhead doors, garage doors, other fenestrations, HVAC, and so on. The porosity of many building types is expected to decrease with changes in energy codes. Obtaining information on the porosity in real buildings is beyond the scope of experimental studies and must be treated probabilistically in a reliability study, which could be coupled with either of the two previously described methods for estimating the effective external tornadic loads.

An overarching issue in designing residential structures for tornadoes that remains difficult to overcome is how to incorporate the results of research into the residential building stock in meaningful ways. This challenge is interdisciplinary, posing arguably more sociological research questions than engineering research questions. One of the primary reasons for this adoption gap is the increase in upfront cost to homeowners in the United States. More public engagement and investigation into incentivization from a social science perspective is needed to determine how to shift the conversation from upfront cost to lifetime cost analysis to better justify new types of benefit-cost analysis. This shift, although difficult, would allow home builders to use more durable building materials and adopt construction techniques with greater documented resistance to tornadic loading. This challenge has two subpoints.

First, a correlation exists between ability-to-pay and willingness-to-pay, meaning that impoverished households may seem less risk-averse when in fact they just have no other option. This equity concern is important to consider when weighing possible incentive structures or cost-sharing programs that promote housing justice. For an individual home buyer, whom these upfront costs ultimately impact, it is a matter of qualifying for a home mortgage. Ultimately, can a home buyer still qualify for the mortgage and find their monthly payments palatable when resilience measures are integrated into the design or retrofit of a house? How could policy shifts incentivize resilience measures either through insurance or property tax reductions when these resilience measures have been taken? Similar to how rebates are offered for energy efficiency, there could be rebates for building fortification efforts.

Second is the low probability of a tornado hitting a given structure. The 700-year mean recurrence interval maps provided in ASCE/SEI 7-22 do not account for tornadoes. Therefore, to design for tornadoes, residential design needs to shift to scenario analyses so that individual homeowners and community stakeholders can make informed decisions that will allow for optimized resilience at a given price point. At the individual housing unit level this is as simple as installing a metal roof to prevent water infiltration through the roof during an EF1-level event. At the community level, this may mean setting acceptable thresholds for population dislocation and economic impact following an EF4-level event.

Another important element in the conversation about residential design challenges is the fact that a very small portion of nationwide building stock is new construction, and society cannot wait for changes recommended here to impact only that small portion of the housing market until a critical mass of residential construction has been built using these new and improved methods. Instead, action must be taken to generate guidance on retrofitting standards for tornadoes, similar to what has been done for earthquake design of commercial construction with ASCE/SEI 41 (ASCE 2023).

A unique feature of residential design when viewed in the broader discussion of this report is the sheer number of residential structures in the nation and the lack of consistent and effective oversight for these structures due at least in part to their omnipresence. This has several secondary effects: first, the average end user of the building (i.e., the homeowner) knows almost nothing about the structure within which they live. This understanding is reduced even further for renters. A false sense of confidence exists that comes from believing that your home was “built to code” and thus can withstand a tornadic event. Lastly, in many areas where building codes have not been adopted or are not adequately enforced, the structural integrity of a building and its capacity to resist tornadic loading is largely left to the discretion of the home builder and the local jurisdiction's building inspector.

Challenges in designing for tornadoes fall into two broad categories. The first category is education, and the second is a gap in technical knowledge to ensure the correct loads are calculated.

Education covers the gamut of the population. Structural engineers need to be informed on the design differences between straight-line wind design and tornadic design. The methodologies differ while the calculation procedure is roughly the same. Historically, engineers have not had to consider the effective plan area, or footprint, of the building to calculate wind loads, but the building size and effective plan area play a key role in determining tornado design wind speeds. Internal pressure coefficients differ for enclosed buildings between straight-line wind design and tornadic design. Ensuring practitioners understand the differences between designing for straight-line wind versus tornadic effects is critical.

Once the tornado provisions and requirements are understood, structural engineers and design professionals must play a part in educating building officials and their communities about what designing for tornadoes means in engineering and societal terms. A disconnect usually exists between the public and engineers. The layperson has a general understanding of the EF scale, because it correlates to the amount of damage seen after an event. This scale has been used in news reports for years, and the public has a comfortable feeling for what this scale means. However, the EF scale does not readily translate to tornado design wind speeds, which can cause confusion, particularly when a tornado design wind speed is significantly less than the observed EF category.

Education includes consideration of types of structures, or the purpose of the structure. ASCE/SEI 7-22 currently requires only Risk Category III and IV buildings, those with a significant impact on human life and/or are required to be operational following an event, to be considered for tornado loads. Structural engineers and design professionals must take on the challenge of educating their clients about potential business interruptions and offer more robust designs. For example, large facilities may be deemed Risk Category II but could contain stocks of life-saving medication that may be required just after an event. A loss of this type of facility could be catastrophic for the resilience of the community following an event.

Once structural engineers and design professionals have educated building officials and the community at large on what tornado design means, expectations of building performance will be more in line with the observed damage after a strike.

Education of all parties involved is important, and ensuring structural engineers have the right tools to calculate these loads is arguably more important. Closing the gap in technical knowledge could be narrowed with questions such as “Are the MRIs right for tornadoes? Should they be the same for straight-line winds?” and “Is the positive internal pressure coefficient correct?” need to be satisfactorily answered. Changing design parameters creates confusion among practitioners and potentially mistrust against structural engineers. Bridging the research gaps will help solidify the understanding of impacts of tornadoes on buildings and structures to create a stronger platform for design.

Many gaps currently exist in the design of tornado shelters and safe rooms due to the complexity of the wind-structure interaction and the current lack of understanding of those interactions. For example, load combinations are difficult to determine due to the lack of knowledge of loading effects. Currently, APC is applied as a static load although it is well-known that it should be applied dynamically to structures. ASCE/SEI 7-22 committees have discussed this issue, but a lack of data describing the APC loading condition prevents action. Other open issues are how to properly separate storm shelters from host buildings, whether any items connected to shelters should be prohibited, and prohibition of canopies and parapets in shelters. More guidance is required for engineers to determine how to treat cases where shelters are connected to host buildings. Better science is needed to characterize tornado wind fields to verify current design tornadic wind speeds for shelters.

Committees have considered but not properly addressed criteria for laydown hazards on shelter walls being impacted by falling debris from nearby structures. Additionally, design standards do not explicitly address collapse loads of host structures although lateral energy is expected to be indirectly accounted for in the design of walls based on empirical examination, observations, and assumptions.

To support industry and implementation, new testing is needed for nonproprietary wall and roof section assemblies. Previous tornado missile impact testing is now out of date and could be improved to support engineers designing shelters. To do this, testing methods would need to be updated, where to carry out testing and where to locate the data would need to be determined, and the community should determine whether to create prescriptive guidance and acceptance criteria for nonproprietary systems.

On the implementation side, special inspection requirements beyond those in the IBC are needed. The community must determine whether more inspections are required, especially in areas where local building codes do not exist or are outdated, and whether ICC 500 should include inspection tables. For example, in cases where no local code is adopted but FEMA is funding safe rooms, the IBC design provisions would be required but the local jurisdiction most likely would not inspect and FEMA does not perform inspections, so design and construction flaws may be missed. Additionally, older structures intended as shelters but not designed to current standards exist but are often not labeled, which could lead to a false sense of security for those sheltering in these structures. In addition to shelter design, a methodology for identification and assessment of Best Available Refuge Areas is needed in cases where shelters do not exist. Significant challenges surround education of the public, design professionals, and building owners on the differences between design for tornado shelters and ASCE/SEI 7-22 tornado load design.

The principal challenges to design for tornadoes in non-building structures focus on costs and the intertwined nature of structural and system reliability. NIST Technical Note 2214, “Economic Analysis of ASCE 7-22 Tornado Load Requirements” (Kneifel et al. 2022), indicates conventional buildings that are designed to the tornado requirements of ASCE/SEI 7-22 may have increased costs to the roof framing and decking; however, the cost impact to the overall project is minimal when including mechanical, electrical, and plumbing trades and architectural finishes. For substations, transmission and distribution structures, telecommunication towers, and solar PV structures, the structural elements themselves make up a significant portion of costs, and increasing design requirements could lead to significant cost effects overall.

In addition, the operating structures of many utilities and companies in these industries adhere to strict budgets and typically will not exceed minimum design requirements unless mandated. These industries, as demonstrated by the fact that many still lack formal or enforced design standards, are also typically slow to adopt new codes and may react slower to structural advancements.

Furthermore, the concept of community reliability becomes increasingly complex for non-building life-line-type structures as other redundancies are typically built into the network. Planning for community resilience with these systems will require a better understanding of not only the environmental hazards but also how the system operates as a whole, which is currently beyond the scope of ASCE/SEI 7-22.