To: Larry Eblen,
Warning and Coordination Meteorologist
NWS Austin/San Antonio
2090 Airport Road
New Braunfels, TX  78130

Dear Mr. Eblen,

I trust that if you are not the proper recipient of this letter that you will forward it to the proper person or at least let me know the proper address and name for this submittal.

This letter is in response to your email in which you attached a field report of the damage of a tornado event on November 15, 2001, Travis Co in south Austin along William Cannon Road about 1/4 mile east of IH-35. While I know that the National Weather Service (NWS) would not want to mislead anyone and would certainly not advocate that the residents of mobile homes were safe in their homes during severe weather, I'm afraid that some will misconstrue the NWS finding. Some residents familiar with this event will assume that their homes are capable of sustaining at least an F1 tornado, not understanding the full range of an F1 tornado. Others will as well use such information as a measure of the safety of the product they manufacture and sale to the public (marketing gimmick). This could lead to both past and future purchasers of this product making unwarranted assumptions about the safety of their mobile home (manufactured home) during severe weather, which could have serious negative implications.

After reading the following information and analysis and given the history and professionalism of the NWS, if the NWS still determines that a tornado of an F1 magnitude was the cause of the damage at Bluff Springs, I will accept this determination without further question. This is simply a presentation of available information and my observations at the Bluff Springs site and analysis of this event.

This will be broken into two parts, with the first part containing three (3) elements, which the presence of one of any of these elements would greatly reduce the effectiveness of the support and stabilization systems used on the affected manufactured homes at Bluff Springs. The phrases "Manufactured Home" and "Mobile Home" will be used interchangeably and federal law specifies that both are one in the same, even though efforts are made to create the appearance that they are two different types of housing. The most accurate terminology for describing this housing is pre-HUD code construction (before 1976) and post-HUD code construction (post 1976). Of the affected homes, all were post-HUD code, with the exception of possibly two, which I was unable to enter for locating the Manufacturer's data plate. The apparent construction techniques and materials used on the overturned home make it a high probability that this home was post-HUD code. The style of the one home that suffered complete roof assembly loss leaves the potential for this home to be pre-HUD, and I was unable to determine any measure with which to determine pre or post HUD-code for this particular home.

Elements affecting anchor capacity at Bluff Springs

(1) Soil Conditions
A visual inspection of the soil at Bluff Springs indicates that the soil is of a clay classification. Interviews with several residents of the effected homes revealed that their homes, pre-November 15, 2001 experienced structural problems. These problems were described by the residents as windows being difficult to open and close, doors not closing or opening or being difficult to close or open, creaking floors, etc. The severity of these problems varied with the degree of moisture content in the soil (wet to dry climatic conditions). The problems described by the residents are indicative of stress loads transmitted to the home through the foundation system by expansive clay soil. The soil when balled and rolled in the palm of my hand, exhibited cohesive qualities, which is as well indicative of clay soil.

The National Bureau of Standards (now NIST) conducted extensive studies of shallow soil anchors used for anchoring mobile homes (NBS series 142, [1982][7]). In this study, many different soil conditions, soil characteristics, types of anchors, load vectors, were studied.

Three soil types were chosen for testing, sand, silt, and clay. Of these three, anchors of all types used for anchoring mobile homes were found to perform the poorest in clay soil.

Yokel et al. (1982)[7]
Figure 4 compares the load-displacement characteristics of vertical anchors installed on the three test sites and pulled at a 40º angle. Note that the initial anchor stiffness on the sandy site was less than that on the silt and clay sites. However, the stiffness on the sandy site increased rapidly with increasing loads and the peak resistance was reached at a smaller displacement than that in the silt and clay site. It is noteworthy that while on the sandy and silty site the load capacity of the vertical anchors pulled at an angle tended to be higher than that of the axially pulled vertical anchors, a similar increase in load capacity did not occur on the clay site. This can be explained by the fact that the compressive forces exerted in this loading mode on part of the soil mass surrounding the anchor substantially increased the shear strength of the sands and silts, which increases with increasing confining pressures, but not that of the clays, which entirely depend on cohesion and thus tends to be independent of confining pressures.

As can be seen, even when better performing anchors are used in clay, the anchors still only achieve about 500 pounds of resistance at 4" horizontal displacement, which is consistent with a later study by Person et al.[6]. Note the large horizontal displacements necessary to achieve 3000 lbs of resistance. This far exceeds the lateral displacement necessary to assure pier stability. 3000 lbs at a 4" horizontal displacement is the industry standard design load used for the type of support and stabilization system (dry stack concrete block, soil anchors and strapping) used on the affected mobile homes. Helix soil anchors installed in moderately wet clay soil have been consistently found to produce only about 17% of the required resistance (3000 lbs) to lateral displacement at 4" lateral displacement that is necessary to assure pier stability. This means that the affected homes at Bluff Springs would have been primarily relying on their dead load weight for their resistance to wind storms. Singlewide mobile homes are inherently more susceptible to overturning and lateral displacement than multi-section (double wide) mobile homes for obvious reasons [8][9].

(2) Anchor Type
It was found by both Yokel et al. [7] and Person et al. [6] that the poorest performing anchors were double helix soil anchors, which were used exclusively on the affected homes at Bluff Springs (30" double helix, 5/8" shafts - Bluff Springs).

Yokel et al. (1982) [7]
There was a considerable difference in stiffness between anchors ST 91 (4-in, double helix) and ST89, namely the 6-in. single helix anchor had smaller lateral displacement than the 4-in double helix anchor. This difference was consistently observed in all the anchor tests and was not anticipated, since it was thought that the long slender shaft of the 6-in single helix anchor would provide less resistance to lateral displacement.

Pearson et al. (1991) [6]
It will be noted that Fig. 2 contains four sets of results for the 30-in. anchor with two 4-in. diameter auger disks. The anchor was installed at approximately 15 degrees with the vertical axis and pulled initially at 45 degrees with the vertical axis. No stabilizing plate or concrete collar was used at the surface.

In our opinion, the last two loading curves are not representative load-deflection curves for this soil anchor. For curves labeled "3rd" & "4th" loading, the loading rod bound against the horizontal member of the frame and, therefore the measured load was not applied to the anchor. This problem was rectified by modifying the load frame. Several other anchors were pulled using the same procedure. The results were very similar to those reported in Figure 2. It was clear from the above description and results in Fig 2, i.e. large deflections at relatively low loads, that these test results were totally unexpected.

Yokel et al. (1982) [7]
Several factors combined to produce the difference in performance: 1. The embedded depth of the helix of the 6-in single helix anchor is deeper (3.7 ft vs. 2.6 ft): 2. The 6-in single helix plate has a larger area (2.25 times the area) and   3. The 4-in double helix anchor has two helixes. Many authors claim that anchor capacity is proportional to the area of the anchor plate.

The maximum horizontal displacement of the home is well recognized to be 4" [3][4][5][6][8] [9]. HUD has been unwilling to answer as to whether this 4-inch lateral displacement is permissible at the design load (3000 lbs) or at the ultimate load (4750 lbs). It is argued by some in the Engineering community (neutral parties) that the standards do not allow for any movement at the design load, and only 4" of lateral movement at the ultimate load capacity (4750 lbs). Irregardless of the answer, as can be seen from the Figure 2, at 4" lateral displacement, the anchors only achieved from between 400 and 500 lbs, assuming that 4" lateral displacement is permissible at 3000 pounds of resistance. The value of 3000 pounds at 4" displacement is used as the standard by the manufactured housing industry [1]. The type of anchor used exclusively on the affected homes at Bluff Springs have typically only produced about 17% of the required anchor capacity of 3000 pounds even at 4 inches lateral displacement, yet they continue to be the most commonly used anchors because they are simply the easiest anchors to install. This means that the affected homes at Bluff Springs would have been primarily relying on their dead load weight for their resistance to wind storms.

(3) Anchor embedded depth
It was observed that almost none of the anchors at the Bluff Springs site on the affected homes were installed to their full-intended depth. It was observed that most had at least 6 inches of exposed anchor shaft. It was observed that of the anchors that were pulled from the ground that their failure appeared to have resulted from primarily lateral loading and not vertical loading. Given that the shafts of many of the anchors not pulled from the ground were still in their installed position (suffered connecting hardware failure) of 15 degrees of vertical away from the vertical wall of the home, it can be safely assumed that the anchors were not fully embedded at the time of their installation. The practice of not fully embedding anchors is common and has been observed on hundreds of new installations in Central Texas. The same condition existed on the home that was overturned. Both Yokel et al. [7] and Person et al. [6] noted a significant reduction in anchor capacity for anchors not fully embedded.

Yokel et al. (1982) [7]

Table 2. Comparison of the Strength of Coaxially Loaded Anchors Installed at Various Depths

Test No.	Depth, ft	Inclination α2	UQ, lb	ŪQ, lb	v
ST 40	1	90º (vertical)
			700
ST 41	1	90º	800	840	0.19
ST 42	1	90º	1020
ST 43	2	90º	3120
ST 44	2	90º	3300	3200	0.03
ST 45	2	90º	3180
ST 46	3	90º	5250
ST 47	3	90º	5600	5217	0.08
ST 48	3	90º	4800

ŪQ = average load capacity, lb
v = coefficient of variation of the sample

Pearson et al. (1991) [6]
Almost immediately, difficulties were encountered with installing the anchors. During the installation process, the four-foot (48-in) anchors typically stalled after being driven two to three feet into the sand.

In all cases, the only obstacle encountered was the frictional resistance of the sand acting on the auger disks. This condition was verified in several instances by digging up the anchor and examining the subsoil for rock, boulders, or other obstacles in the vicinity of the auger. No obstacles were found. Because of these difficulties, less than half of the ninety test program anchors in the field were installed to their full-intended embedded length.

Pearson et al. (1991)
The majority of 48-in anchors were unable to be driven to their full depth at the test site. Because of the type of soil, it was expected the soil be the most favorable for installing these types of anchors. This is very significant as anchors, which were not completely embedded carry very low load levels and exhibited large deformations at these low load levels.

Pearson et al. (1991)
The failure criteria suggested in the proposed ASTM Standard was IGNORED in subsequent testing. For subsequent testing, it was decided in consultation with the HUD GTR, to test anchors that were fully embedded and record load and deflection until one of the following conditions existed: a. the applied load began to decrease with increase in deflection or b. the horizontal deflection exceeds approximately 18-in. The test was stopped if one of these two conditions was exceeded.

As can be seen in table 2, there is a significant reduction in anchor performance when anchors are not fully embedded. These results were reinforced by Pearson et al [1991] during their studies. The anchors being tested by Yokel et al. [7] (48" single helix) were found to be superior in capacity to the 30" double helix anchors present at the Bluff Springs site. The results of testing for the double helix anchors studied in both the Yokel (NBS) [7] and Pearson (HUD) [6] studies indicate that double helix anchors only achieved about 23% of the required 3000 pounds, at 4 inches lateral displacement, even when fully embedded. This means that the affected homes at Bluff Springs would have been primarily relying on their dead load weight for their resistance to wind storm events.

The proceeding three elements, which each in themselves would significantly reduce anchor capacity, were present on all of the mobile homes affected by the tornado at Bluff Springs. (1) Clay soil, (2) anchor type used, and (3) anchors not fully embedded.

In a letter written February 06, 1998 by the Director of Manufactured Housing for The Texas Department of Housing and Community Affairs, which was addressed to all anchor manufacturers and distributors; all manufactured home manufacturers, retailers, and installers; all manufactured housing division inspectors, and all cities with contracts for installation inspections, the Director state.. "For your information, we have begun to check various sites to evaluate the extent to which installers are selecting the proper anchor based on the torque values.  I want to advise you that our early testing using the torque probe indicates serious deficiencies.  For example, along the Texas coast, tests reveal torque values of under 250 inch pounds yet installers were still using 30 inch double helix anchors.  In these soil conditions, much longer and a different type of anchor is required to provide the required restraint."

In later correspondence by the Director, the Director stat.. "Based on the 1998 soil test results over 50 % of the manufactured homes installed in Texas do not meet the federal windstorm protection standards".

Given the presence of the three aforementioned elements and that the problem of selecting and installing the wrong anchor for the soil conditions is an acknowledged problem in the State of Texas, it would seem wise and prudent to exclude the fact that the affected homes were anchored as an indicator of the magnitude of the tornado that caused the damage at Bluff Springs.

Pier type used at Bluff Springs

While no one is advocating that dry stacked concrete block provides adequate support against wind loads, in this case, they performed better than the metal jack stands used for the support of the overturned home at Bluff Springs. Manufactured homes subjected to horizontal or lateral loading experience uplift on the windward side of the home, which decreases windward pier loading and increases pier load on the leeward side (NBS PB81180978 [1981] [8], NBS PB81182552 [1981] [9]). This phenomenon was observed on the affected homes at Bluff Springs where the leeward side piers were visible for inspection. It was observed that piers on the leeward side of homes not overturned were no longer level, but tilted from the bottom of the pier to the frame member of the home on the leeward side of the home, away from the direction of loading or in the direction of lateral movement. This loading affect, once the leeward piers fail, coupled with the windward side lift, causes the leeward side to drop to the ground and the windward side to rise, exposing the bottom of the home to horizontal wind loads, hence the home is overturned [8][9]. The mere fact that all but the overturned home had a larger surface area (concrete block) to slide on, and a more structurally sound footing (16 x 16 x 4 concrete block) before pier failure occurred (the home falls from its leeward piers) was the difference in this case to the event not being more catastrophic.

The overturned home was resting on metal jack stands, with only a piece of plywood between the ground and the bottom of each jack stand. It is doubtful that plywood in direct contact with the ground for any period would have provided much if any stability when needed for the metal jack stands. There is relatively no surface area for the frame of the home to slide on when metal jack stands are used, and the frame and metal jack stands of the overturned home had no mechanical connection, meaning the home reached its overturning moment at approximately 1-inch of lateral displacement.

In the final analysis, given the low resistance that would have been provided by the homes anchoring system and the improper installation of the metal jack stands, the home that overturned had a significantly higher probability of overturning at much lower wind speeds than did the other homes affected at Bluff Springs. It would seem prudent to exclude the mere fact that this home was overturned as a means for determining the magnitude of the tornado that affected the Bluff Springs area.

Roof separation at Bluff Springs

The roof assemblies for these two units had been removed from the site and the rest was too badly damaged to assess making it impossible to determine the method of attachment to the wall structures of the home. It was observed on the two units with roof separation that the top plates were almost completely smooth (virtually no splintering noted), which is indicative of a roof assembly that was poorly attached.

Excluding one home that suffered a partial roof assembly failure, the only homes that suffered significant damage, which did not result from projectile damage, were oriented with their largest vertical surface exposed to the direction of the horizontal wind load during this event.

Add on structures at Bluff Springs

Add on structures, such as porches and covers of the affected mobile homes survived in relatively good shape. Most of these add on structures were displaced with the home due to their attachment to the home. Had they not been attached, it is doubtful if the add on structures on the windward side of the homes would have moved at all. It should be noted that the porch on the overturned home was apparently not attached to the mobile home and was still in its original location and suffered relatively minor damage considering it had been hit by an F1 tornado. Add on structures on the leeward side of the homes that were affected were pushed with the home, but were still usable. It is generally asserted by the manufactured housing industry and HUD, that add on structures are destroyed, resulting in failures in the mobile home envelope and subsequent failure in the mobile home box. My research prior to the Bluff Springs event brought this assertion into question. The Bluff Springs tornado event makes the industry and HUD assertion even more highly questionable. Add on structures of the type generally added to mobile homes in generally provide a much smaller profile to wind loading than do the homes themselves, making it more likely that the home moves first. It is the inadequate support and stabilization systems used for mobile homes that permit the home to move, causing damage to both the add on structure and the mobile home envelope. Had the mobile home not failed structurally, no significant damage would occur to the home or add on structures. At least in this case, it appears that add on structures were more structurally sound than the mobile homes, with the mobile homes causing what little damage occurred to the add on structures.

The trees that were uprooted were noted to have shallow root systems, and all fell in the same direction, although I must rely on the expertise of the NWS in this area.

If the NWS relied in part or in whole on the requirement for these homes to resist 80 mph winds without failure (HUD-code)(Wind Zone 1 or non-Hurricane) in their determination, the available scientific studies and conditions at the Bluff Springs site provide sufficient evidence that the use of mobile home damage as an indicator would not provide an accurate measure for determining tornado velocity. While I am certain that a downgrade in the intensity of this tornado at Bluff Springs will not be received with open arms by some given its implications, we trust the NWS to make a fair an accurate determination of this tornado's magnitude.

Sincerely,

John Taylor
Founder - The American Internet Society of Manufactured Home Owners

TAISMHO

References: