Tumbling Dice

SOMETIME AFTER MIDNIGHT ON APRIL 8, 1998, A MAN AND A WOMAN near Birmingham, Ala., woke up to hear the freight-train roar of a powerful tornado approaching their mobile home. With only moments to react, the couple placed their three-year-old daughter in the best shelter they could think of: their clothes dryer. It was to be their last decision as parents.

The industry does not like the popular phrase “mobile home” —it reserves that term for units built before the federal government began regulating its product. Until 1976, house trailers were not subject to any building code. Since 1976, the industry has been covered by a federal rule known as the HUD Code, and in 1994, in the aftermath of Hurricane Andrew, the HUD Code was toughened for coastal high-wind zones. Manufacturers call units built after 1976 “manufactured housing” or “HUD-Code homes,” pointing to the national rules as an assurance of quality.

But the 1994 code revision (which an industry association, citing cost factors, tried to block in federal court) does not apply to units in the nation's midsection. And, call it what you want, a lightly framed, narrow box, braced mainly with drywall or thin paneling and resting on stacks of cement blocks, is no place to hide from nature's most violent wind-storm. News accounts of tornadoes routinely mention death and destruction at mobile-home neighborhoods—commonly enough to spawn popular rumors that “trailer parks” somehow actually attract tornadoes. Engineer Tim Reinhold, now vice president for engineering at the insurance industry–sponsored Institute for Business and Home Safety, even recalls an official inquiry from a congressman's office to the government agency where he once worked: The congressman wanted to know whether tornadoes could really seek out mobile-home parks.

“Because it came from a congressman,” Reinhold remembers, “we actually had to treat that as a serious question.” But the truth is more ordinary, he says. Reinhold explains that elevated rates of death and injury in mobile homes stem from the way the structures are built and set: “By any objective standard, mobile homes, particularly the ones that are built in the middle of the country where tornadoes are likely to occur, don't have the kind of strength that you get with a site-built home.”

BD061101128L1.jpgCLICK HERE FOR IMAGE GALLERY

BD061101128L1.jpgCLICK HERE FOR IMAGE GALLERY

SURVIVAL OF THE FITTEST: Above, a conventional house in Newburgh, Ind., had its roof ripped off by the same storm that killed 18 residents of the Eastbrook Mobile Home Park. At right, student volunteers salvage an appliance from a destroyed Newburgh house. Strong tornadoes can destroy even well-constructed site-built homes; however, the survival of even one interior room (or “core element”) can spell the difference between life and death.

TRAIL OF TRAGEDY: Shown here is a portion of the storm track from the September 2005 Evansville tornado. Luck plays a major role in tornado events, as homes in the storm's narrow track may be pulverized while buildings a hundred feet away escape damage. But strong in-flow winds can also suck poorly anchored manufactured units off their pads and into the vortex.

BREAKING LOOSE: Ground anchors failed in multiple ways in the September 2005 Evansville storm. Straps may pull through the anchor ratchet (above), or anchor bolts may shear off, leaving the ratchet attached to the strap (next photo). At bottom right, an earth-auger anchor has been bent and twisted by off-center forces. Earth anchors in particular tend to take up strain at varying rates due to variations in soil strength or anchor depth, forcing some anchors to handle loads beyond their rated capacity.

HOW TIE-DOWNS FAIL: In a typical installation (above), the unit's chassis is set on stacked masonry block with wood leveling shims and tied down with steel straps wrapped around the chassis I-beams. Over-the-top straps to resist overturning are often provided, but are only required to be actually used in Zones II and III near the coast, not in Zone I inland. In this photo, the over-the-top strap is still rolled up in its “factory roll.” Above right, a tie strap that snapped in the Evansville storm lies on the concrete pad. Straps commonly tear at the beam from metal fatigue and eccentric loading (next photo).

WRENCHED APART: When anchors are stressed unevenly, failure of one tie-down may transfer its load to the next, and a domino effect can quickly “unzip” the entire tie-down system and subject the unit to bending and twisting forces not considered in the unit's structural design. Concentrated forces may rip the unit off its chassis, pull the chassis from its anchors, or both. The mangled chassis above has lost its trailer; the chassis at right lies in a pile of debris.

ENVELOPE CHOICES: Sheathing makes a difference in the storm performance of structures. Here, foamboard sheathing (photo #13) and asphalted particleboard sheathing (previous photo) have been stripped off two units at the Eastbrook Mobile Home Park, along with the vinyl siding. In the next photo, a unit that was blown off its foundation still retains some shape because of its structural panel sheathing.

NEVER CONNECTED: A system is only as strong as its weakest link. At left, block piers sit where a unit had been placed with no tie-down anchors, sitting unsecured on the piers. Scattered debris in the background is all that remains of the unit—while a better-secured nearby unit shows no damage.

DEGREES OF DAMAGE: Here, photos show several degrees of damage to mobile homes. Tornado wind speed is estimated from the observed damage. Although newer units may perform better than older ones, mobile homes in general tend to suffer worse damage than site-built houses in small tornadoes. Total destruction of a mobile home might only indicate a tornado of strength F2 or F3, while total destruction of a site-built house might indicate a tornado of strength F4 or F5.

Government scientists at the National Weather Service keep track of tornado deaths, noting where and how they occur. The numbers they see prompt them to include specific advice in their tornado warnings: If you're in a mobile home, get out and seek better shelter. Those who stay in their mobile homes, in the words of one Red Cross official, risk “becoming a part of the debris field along with your mobile home.”

But better shelter is not always close at hand, and the advice may feel counterintuitive to citizens—who wants to venture out of a dry trailer in the middle of the night into pelting rain, hail, lightning, and wind? And tornadoes really are rare and concentrated events. Even in a tornado watch, the odds between the chance of total destruction if you stay put and the certainty of being exposed to ordinary rough weather if you go outside to hunt for better protection often fall short of convincing people to leave their homes. For those of us who have never experienced it, it is hard to imagine the overwhelming violence of a killer tornado—and until the storm actually hits, a manufactured house may still feel like adequate refuge.

The Alabama parents who put their pre-school daughter into a clothes dryer that spring night in 1998 made a desperate, last-minute choice that saved the little girl's life. After the storm, rescuers found the child still in the dryer and unhurt, 250 yards from where her home had stood. Choosing to remain in the trailer at all, however, resulted in disaster for the parents. The clothes dryer proved better shelter than the home itself, and the couple received fatal injuries as their mobile home rolled 100 yards and came to rest against another building. Had the unit been equipped with an in-ground shelter, the whole family could have survived, and the girl would likely still have her parents today.

In a world full of risks, every safety choice is a roll of the dice; tornado hazards bring up their own calculation of the odds. And the fact is that violent tornadoes are quite rare compared to the everyday risks of life. Some 1,500 tornadoes are reported each year in the U.S. It sounds like a lot, but America's a big country—so even for Tornado Alley states, such as Oklahoma, the odds are a thousand to one against seeing even a weak tornado in a given square mile in a given year.

Moreover, most tornadoes are weak. The violent tornadoes that cause most of the death and destruction are only about 5 percent of the total. An average square mile in the U.S. heartland could easily go 50,000 years without experiencing a real killer tornado, and that's for the whole square mile—the chances the storm would hit one particular building are even lower. In a typical year, tornadoes might kill 50 to 80 Americans. That puts twisters in the same ballpark as snakebites (about 15 deaths per year) and lightning strikes (about 85 deaths per year). Tornado fatalities are far less commonplace than fatal auto crashes, gun deaths, or even deaths caused by falling down stairs. (As of mid-September, tornadoes have killed 54 Americans in 2006, according to the National Weather Service; the fraction involving mobile homes is not yet published.)

But when tornadoes do hit, your choice of housing sharply affects your risk of death. Meteorologists Harold Brooks and Chuck Doswell, from the National Oceanographic and Atmospheric Administration, have analyzed the historic data on tornado fatalities and found that recent numbers reveal a steep decline. In a paper published in 2002, they report: “The number of deaths per year caused by tornadoes has generally been much less in the last quarter of the 20th century than it was previously. In particular, the number of deaths in the ‘big years' has dropped dramatically.” Brooks and Doswell go on to explain that between 1875 and 1974, there were 54 years wherein at least 100 tornado deaths occurred per year. However, the statistics drastically improve between 1975 and 2000, wherein only two years (1984 and 1998) contained as many as 100 tornado deaths.

The bad news? Mobile homes. Brooks and Doswell write, “Mobile-home safety is the biggest obstacle to reduction of tornado deaths in the U.S.” The May 3, 1999 tornadoes in Oklahoma City, for example, damaged or destroyed several thousand site-built homes and struck fewer than 100 mobile homes—but 11 of the 36 people killed by the storm died in those mobile homes. “The death rate in mobile homes was more than one person per 10 units, while the death rate in permanent homes was less than one person per 200 units,” write Brooks and Doswell—a twentyfold greater risk for the mobile-home dwellers.

That case illustrates a trend, say the researchers. In the 1970s, only a quarter of tornado deaths occurred in mobile homes. By the late 1990s, half of all tornado victims died in mobile homes. At 1.23 deaths per million people per year, mobile-home occupant tornado death rates are stuck back in the 1920s—while the rate for the population as a whole has dropped to only 0.12 deaths per million. As the proportion of Americans living in mobile homes increases, Doswell and Brooks report, those numbers are skewing the national curve: “The increase in mobile-home residency may have reversed the approximately 70-year-long trend of decreasing death rates in tornadoes.”

Reached by telephone this summer, Brooks said that recent statistics confirm his earlier conclusions. “When we wrote that paper, we had numbers through the year 2000, and now we have data through 2005,” he notes. “And yes, the overall death toll from tornadoes stopped decreasing in 1995. And the fraction of deaths involving mobile homes is now a little higher than it was in 2000.”

HUD-Code home manufacturers have at times pressured Weather Service forecasters to stop targeting mobile-home dwellers in warning messages. But Brooks argues that the specific warnings are justified: “The bottom line to me is, the Weather Service should stop singling out mobile homes for particular call to action when we start seeing the rates of death in mobile homes and site-built homes being roughly comparable.”

To bolster the statistics, Brooks points to decades of post-storm observations by meteorologists that show a clear pattern of greater destruction to mobile homes than to site-built houses when both are hit by the same storm. Wind damage to trees provides a convenient benchmark for comparison, he notes: “What I personally have seen is that there are times when mobile homes are seriously damaged, rolled over or something, and there is just not a whole lot done to the trees right around them. And by the time you start seeing significant damage to permanent homes, you're seeing significant damage to the trees.”

What accounts for the disparity in performance between site-built and manufactured units? Three main elements: A weaker structure in the unit itself; a less substantial foundation; and a weaker connection between the foundation and the unit. It's something of a toss-up, when a tornado strikes a mobile home, whether the dwelling will be ripped off its chassis, or the entire unit, chassis and all, will part company with its supports and tumble downwind.

Even for solidly built units, installation flaws are a common, critical weakness. According to structural engineer Larry Tanner, P.E., many site crews—perhaps most—are failing to install and anchor mobile homes properly, with proper spacing, proper-size piers, and lateral anchorage.

Tanner is a research associate with Texas Tech University's Wind Science & Engineering Research Center. Together with colleague James Waller, P.E. (past president of the National Storm Shelter Association and a tornado shelter manufacturer), Tanner traveled to Evansville, Ind., in November 2005 to analyze damage after a tornado wrecked 110 units and killed 18 people in the city's Eastbrook Mobile Home Park. Key to the extreme destruction, Tanner and Waller reported, were widespread failures of the anchoring systems.

Weak attachments. The wood platform of a mobile home is usually fastened to a steel I-beam chassis, with wheels attached for transporting the unit. To place the unit, installers jack the chassis up, remove the wheels, then rest the I-beams on stacks of dry-laid concrete block, leveling them with wood shims. Flat steel tie-down straps are typically attached to the chassis by hooking the end of each strap onto a flange of the I-beam, wrapping the strap around the beam, then inserting the free end into an anchored ratcheting device used to tension the strap.

Installers at the Eastbrook park had used various methods to anchor the ratchets, Tanner and Waller found. Some straps were attached to rebar loops poured into the concrete pad, some to expansion anchors set in holes that had been drilled into the slab, and some to earth anchors augured into the ground. But “the mechanism of failure was generally the same; the strap broke,” write the engineers. “The failure sequence would begin with the wind forces pushing, lifting, and sliding the unit off its piers, a process facilitated by loose ties, and ultimate failure of the strap at a bend on the frame flange.”

In a telephone interview, Tanner said: “In reality, those flat straps are never going to be in pure tension.” Over time, he explains, occasional winds cause the straps to loosen, and the high buffeting winds of a strong storm cause the mobile homes to rock, twist, and slide. “When the units twist, then those straps fail, much like steel banding on a crate. You pull on that banding, and it's really hard and tough. But you can take a screwdriver or claw hammer and grab ahold and twist it, and it snaps just instantly. And that's exactly what happened in Evansville, and that's what happens everywhere else,” says Tanner. “That tells me that that method is fundamentally flawed, and they need to devise a new method.”

Even if the straps had performed at their rated strength, Tanner and Waller found few trailers at Evansville with enough straps to hold a mobile home down securely. Rebar-loop anchor points provided in the concrete slab pads were spaced much too far apart, they report (12 or even 24 feet, instead of the recommended 6 or 8). Lateral end straps, designed to keep the units from shifting in the long dimension, were provided on the mobile homes, but almost never attached to anchors by the installers, who had also ignored over-the-top straps provided on some trailers to counteract wind uplift forces. The engineers saw many straps still rolled up and hanging from trailer undercarriages, just as they had left the factory. In one case, a unit had been set with no fastening straps at all; nothing was left on that pad but the concrete block piers.

Fragile framing. But mobile homes can still fail even if their foundation straps hold. Government reports are replete with cases where a unit's floor has separated from its chassis, the walls have detached from the floor, or the whole structure has simply blown apart.

In the Evansville disaster, says Tanner, sheathing materials mattered. “All these units appeared to be within about 10 years old,” he says, “but we found various types of construction. Some just had 2x3 studs with housewrap on the outside and had vinyl siding attached over that; some had fiberboard and insulating foam sheathing; and some had OSB wood panel sheathing. And the OSB sheathing really kept those units together. Those were the ones that survived the very best.”

Steel framing connectors are another variable. “Zone II and III units [in coastal areas] have a requirement for steel straps that connect the roof structure to the stud walls, and connect the studs to the floor diaphragm,” Tanner says. “Zone 1 units have no strapping requirement. And you really need that connection, because the industry has minimized the size of the structural members to the point of just achieving the load requirement in a controlled testing environment. If you put it into the real world, there's just not the redundancy that allows you to withstand these extra high winds or swirling winds.” Nails and sheathing aren't enough to do the trick, Tanner maintains: “You don't have enough surface area to make that connection with toe nails and stuff. You really need straps so that you can put your fasteners in shear, rather than in tension.”

While there's a clear pattern of risk for manufactured home dwellers, not all units are the same. Units built after the introduction of the HUD Code in the 1970s are stronger than previously built units, and post-1994 units in the Zone II and Zone III coastal regions are even stronger.

Moreover, some manufacturers choose to build more conservatively than they have to. One example would be Dallas-based Palm Harbor Homes. Bert Kessler, P.E., company vice president for engineering, says, “We're in the upper end of the market, and I don't engineer our units down to the bare nitty-gritty.” For one thing, Palm Harbor aims to exceed energy code requirements, framing walls with 2x6 studs to accommodate R-19 insulation—which also creates a stiffer wall. And although the company's Zone I homes don't have the same shearwall details as its Zone II or III homes, Kessler does specify the same framing connections in plants that supply more than one zone, “so as not to confuse people on the shop floor.”

And Kessler does his own research into storm performance. “Any time we have a major storm event, you'll find me there within a day or two,” he says. “I did this after Andrew; I did it after Charley, and also after we had some tornadoes go through Central Florida back in the late 1990s. I go there just to see what performed and what did not.”

Stronger connectors. Kessler, like Tanner, has noticed the problem with tearing of attachment straps. “Our testing showed that if the strap was stressed as little as 9 degrees off from perpendicular, it failed at half the rated load,” he says. “So in 1994 we developed a bolted pivoting connector that can handle that kind of loading, and we supply that for all our units.” While manufacturers don't control the critical issue of installation quality—local authorities are in charge of that—at least the improved connector makes it easier to do installations right. “Now the alignment of the anchor is less critical,” observes Kessler.

Kessler complains that press reports about tornadoes don't seem to acknowledge the improvements his industry has already made. “When there's a storm,” he says, “the news media does not take you to newer construction. They take you to units that were built 15 or 20 years ago. The February 1998 tornadoes that went through Central Florida, one of them happened to go through a manufactured home community that had two clearly distinct areas, one where we had post-1994 units, and then the older section with pre-1994 units—and the difference in the damages between the two is significant. What that showed me was that what we were doing was good. What the state of Florida was doing as far as tighter control over installation helped a lot. And in the newer side, I was able to clearly see the differences in performance of conventional straps versus the pivoting attachment we had developed. It was very, very clear.” In the aftermath of Hurricane Charley in 2004, FEMA damage assessment teams found that newer HUD-Code units that were well-anchored performed as well as IRC-compliant site-built housing, in general.

But those were still Zone III units, not the Zone I units found in most of tornado country. Upgrading the whole country to the Zone III standard (which some have suggested) might cost his company $2,500 or $3,000 per unit for a typical double-wide, Kessler estimates: Roof trusses would have to be spaced at 16 inches on center, walls would need continuous structural sheathing, and many framing connections would have to be reinforced with steel connectors. For other companies, whose Zone I units may not be built as well as Palm Harbor's to begin with, the up-charge could be even greater. And that's not counting improved installation—the requirement for vertical straps to prevent overturning, with diagonal straps also installed at the same locations, which local inspectors would have to enforce. “I'm seeing a price of $75 to $150 per connector,” says Kessler.

Is it worth it? At a guess, the safety improvements might only raise the monthly housing cost for manufactured-home residents by $50 to $100 per month. And that cost might be offset in many cases by improved financing terms. But even so, given the relatively small and sporadic risk of an actual tornado strike, it remains debatable whether the cost of safety upgrades is worth the benefit. Even after last year's deadly strike in Eastbrook, Ind., Gov. Mitch Daniels remarked, “When you see the randomness of an event like this, you have to reckon with the fact that we cannot expunge every vestige of bad luck from life.” (Daniels did, however, support a move to require licensing of manufactured-home installers and also improve foundation tie-down connections.)

Mobile-home residents themselves may not choose safety over economy. For example, retired police officer Jerry West Sr. barely escaped a tornado that destroyed his mobile home in Madison County, Tenn., by crawling out the window into a makeshift shelter he had built himself. He then reportedly moved back into a new unit placed on the same site. He was considering upgrading his shelter, reported The Jackson Sun, but not moving to a more substantial building. The 63-year-old West told the paper, “I don't think a mobile home is near as safe as a house on a permanent foundation. But I guess that's the way I choose to live.”

And even an expert who is fully aware of the risks and the structural issues, such as engineer Reinhold, is hard pressed to put tornado shelters at the top of his list of priorities—even after a Clemson University colleague experienced a near-miss from a strong tornado that could have killed the man's whole family. Rein-hold lives in a conventional site-built house, not a HUD-Code home, but he has considered an in-house tornado shelter. Says Reinhold, “If you look at the statistics—really deal with the risks—and you are going to spend $4,000 on a tornado shelter, well, you know—I've got teenage kids. I looked at that choice and I said, ‘I'm going to put that money into the safest automobile that I can afford for them to drive,' because that is the greatest risk.”

Texas Tech engineer Tanner acknowledges the cost issue: “You're living in a mobile home because you do have limited means, and so every nickel is important to you.” But Tanner's not ready to give up on improving the industry's safety performance. “America has been built on doing things better and smarter,” he argues, “and I think we can do that in the manufactured-housing industry—and without taking a significant quantity of people out of owning their home.”

Weather scientists grade tornadoes on the Fujita Scale (or F Scale), named for the late tornado expert Theodore “Ted” Fujita. Tornado winds are hard to measure directly. When they do happen to strike a weather station, they break the wind gauges. So instead, the Fujita Scale rates tornadoes from F0 to F5 based on destruction of natural and man-made objects. “In its rawest form, the Fujita Scale is not a wind-speed scale; it's a damage scale,” explains engineer Tim Reinhold.

In an attempt at rough precision, the F Scale takes into account the difference between various construction types—and it assumes a variance between mobile-home and stick-house survivability. Signposts for an F1 or “moderate” tornado, toward the bottom of the scale, include the criteria “peels surface off roofs; mobile homes pushed off foundations or overturned.” An F2 or “significant” tornado, according to the scale, would cause “roofs torn off frame houses; mobile homes demolished.” Not until F4 does the scale expect “well-constructed houses leveled.”

A new “Enhanced Fujita Scale,” developed at Texas Tech, lowers the wind-speed estimates associated with various degrees of damage, but provides a spread of possible wind speeds for particular damage indicators. By the enhanced scale, a stick-built house showing “total destruction of the entire building” would indicate winds of about 170 mph (within a broad range from 142 to 198 mph). Manufactured housing, by contrast, would experience “destruction of roof and walls, leaving floor and undercarriage in place” around 105 mph. “Unit rolls or vaults; roof and walls separate from floor and undercarriage” around 109 mph. “Undercarriage separates from unit; rolls, tumbles, and is badly bent” around 118 mph. And “complete destruction of unit; debris blown away” occurs at 127 mph (the range is 110 to 148 mph). In other words, the new scale, which is based on both field observation and engineering analysis, assumes that weak tornadoes, peripheral winds from stronger storms, or even straight-line winds from ordinary thunderstorms—winds which might do only non-structural damage to well-constructed stick homes—are likely to cause major structural damage to many mobile homes.