The Dangers of Modular Construction

BY KEVIN A. GALLAGHER

As if lightweight construction materials in residential occupancies weren’t challenging enough, consider an attic filled with lightweight trusses supporting a roof over floor joists that are attached to the underlying gypsum board by highly flammable glue! Sound alarming? Now consider a concealed space greater than 20 inches high, 48 feet long, and 14 feet deep between the first and second levels of the same home. Oh, by the way, all the ceiling joists in the void are sprayed with the same flammable glue. These homes are often covered with vinyl siding or, what one chief calls, “solidified gasoline wrap.” One more thing: These conditions can be found in thousands of homes across the country.

THE INCIDENT

On January 13, 2008, the Acushnet (MA) Fire/Rescue Department responded to a 911 call reporting a fire at a single-family residence. Four minutes after receipt of the call, a member of the Acushnet Police Department arrived on-scene and reported the structure to be 75 percent involved. Approximately five minutes later, a member of the fire department captured images of the burning structure (photo 1). You could see flames above the tree line from more than one-quarter mile away.


(1) When the fire department arrived, the structure was already 75 percent involved. [Photos courtesy of the Acushnet (MA) Fire/Rescue Department.]

While conducting defensive fire suppression operations, it became clear that the speed with which this fire spread was considerably faster than that at other structure fires I have witnessed over my 22-year career in the fire service. As we investigated the cause and origin of the fire, we discovered certain construction features that we are convinced played a role in this fire. The questions we raised and subsequently asked of local, state, and national fire service leaders convinced us that very little is known about a clear and present danger to firefighter lives and safety. One danger involves the flammable glue used to hold the gypsum ceiling in place and a large void space that contains this glue and structural members. Also, firefighters and fire officers formulating firefighting tactics and strategies do not have a working understanding of these construction features.

The Structure

The residence was a single-family, wood-frame structure of prefabricated or modular construction. The structure was manufactured in 2005 in a factory, shipped over the road, and then assembled on-site on a foundation that had been poured several months earlier. The structure was a two-story Colonial style home featuring a full basement, vinyl exterior siding with plastic house wrap underneath, and a vinyl vented soffit with asphalt shingles on a gable roof of lightweight truss construction. The exterior walls were made of ½-inch CDX plywood. Gypsum board measuring 5⁄8 inch comprised the batten-insulated interior walls. The first floor was 1,736 square feet (sq ft), the second floor 1,376 sq ft, and the basement 1,064 sq ft; a 672-square-foot garage was incorporated into the basement level.

Plans on file at the Acushnet Building Department indicate that this structure consisted of five modular boxes contractors assembled on-site while following industry protocols. The four primary boxes measured 48 × 14 feet. The two first-floor units were placed by crane, front to back, on the foundation and secured by bolts at the marriage walls, per industry protocol. The two second-floor boxes were placed on top and secured to each other, as well as the first-floor boxes, with bolts and metal straps. The fifth modular box measured 24 × 15 feet and was attached to the foundation in such a way as to add a single-level “family room” to the primary structure. A deck was attached to the rear of the structure; it was accessible through a sliding glass door. A porch of approximately 38 feet, with an attached roof commonly referred to as a “farmer’s porch,” was attached to the front of the structure and positioned in such a way that the ends of the porch were an equal distance from the front door.

On the day of the fire, four wire-frame window boxes filled with organic material were stacked on the far left side of side A near the A/B corner of the front porch. The home-owners said the window boxes were in that position, covered by the roof, since the end of the growing season.

The homeowners told investigators that on the evening of Saturday, January 12, 2008, they hosted a party for family and friends to watch the New England Patriots football game. The owners designated the rear deck as the smoking area. During the course of the game, an individual went to the front farmer’s porch, smoked a cigarette, and disposed of the smoking material in one of the window boxes. Early the next day, the female owner of the house was lying awake in the second-floor master bedroom and sensed that the room was getting warm. As this sensation continued, she awoke her husband and got out of the bed. As she walked by the window facing the front of the house, she noted a fire on the roof of the farmer’s porch. She alerted her husband, gathered the four children from their bedrooms (two children lived in the home, two others were visiting and had slept overnight), and exited the home by the front door. The adult occupants told investigators that they noticed no visible smoke during this period and that the smoke alarms had not sounded.

After exiting the structure, the male adult reported that he looked back to see flames engulfing the full length of the farmer’s porch roof with fire working its way up the front of the building toward the roof on the left side of the farmer’s porch. A neighbor informed the owners that he had called 911. While en route, I noted from a considerable distance a large column of heavy black smoke as well as flames above the tree line. Shortly after my arrival at 0643 hours, the roof collapsed.

Given the safe evacuation of the occupants and the large volume of fire observed on arrival, it was immediately apparent to the first-arriving fire officers that fire suppression activities would be defensive. Once the main body of fire was extinguished, a member of the Massachusetts State Police, assisted by local investigators, initiated an investigation. Based on the physical evidence observed and the occupants and neighbors’ statements, the origin of the fire was determined to be the wire-frame flower boxes on the far left side A wall near the B side of the farmer’s porch. The cause of the fire was determined to be the careless disposal of smoking materials. The fire was ruled “unintentional.”

CONSTRUCTED WORKING SCENARIO

Based on the on-scene investigation, we developed a working scenario for this fire as follows. The cigarette smoldered in the flower box for several hours until reaching a sustainable temperature for ignition. The position of the flower box placed the burning materials close to the vinyl siding. Once ignited, the siding fueled the fire, which spread upward to the vinyl soffit material on the porch ceiling. The structural members of the porch roof assembly further fueled the fire. Flames then continued to extend up the front of the building by way of the vinyl siding until it reached the perforated vinyl soffit, which allowed the fire to extend into the attic space. The homeowners told investigators that they did not store items in the attic space because of the small interior attic access opening.

Given the timeline of the incident, the questions remained: How did this fire spread so quickly, and why did it burn so intensely?

Follow the Clues

The following day, we reviewed the building plans of the burned structure. It is important to keep in mind that when these modular units arrive for assembly in your community, the wall, ceiling, and floor assemblies have been completed. All electrical wiring has been run, and all plumbing has been placed. Even the smoke detectors are preinstalled. The traditional role of the local building inspector is limited to an inspection of the foundation and a review of the appropriate labels indicating that the unit has been inspected and deemed acceptable by a third-party inspector. In short, the level of local inspection conducted on modular construction prior to the issuance of a certificate of occupany is limited because of the unit’s having been approved prior to its departure from the factory.

With the assistance of the Acushnet building inspector, we observed markings on the plans that indicated a void space between the first and second levels. As written, these plans—not specific to the home that was built, but generalized to the model purchased—indicated that the farmer’s porch roof was attached above the void area. This led us to conduct a field inspection; we found evidence of a large void space between the ceiling of the first level and the floor decking of the second level. The distance from the top of the 2- × 10-inch floor joist (position of the second-level floor decking) and the bottom of the 2- × 6-inch ceiling joist (position of the first-level gypsum board ceiling) was approximately 22 inches. With the two modular boxes of each level attached at a marriage wall, two voids each measuring 48 × 14 feet were noted during our investigation of the charred remains (photo 2).


(2) The void space between the first and second levels.

In addition, two holes, an equal distance from the sides of the front modular box, were noted on the front of the house. These holes were positioned so that they were lined up with the void space found between levels. It was later learned that the manufacturer makes these holes at the time the boxes are built in the factory and that they are used to run the cables through the floor system to allow the section to be lifted into its appropriate location.

Coincidentally, a few days into our investigation, the Acushnet building inspector was informed that a two-story Colonial of modular type construction was to be assembled in our community. With the permission of the homeowner, the building inspector and I observed the assembly process, which made it possible for us to view each modular box from all six surfaces (front, back, sides, top and bottom) prior to assembly. Each of the boxes arrived on-site complete with ceiling and floor assemblies. Cabinets were in place, fixtures were mounted, and all electric wires were prepositioned.

Not surprisingly, the roof was made of lightweight truss construction of a unique design. During transport, the roof lies flat. Once the upper box is set in place, the roof is lifted, propped up, and supported by a knee wall that is secured to the roof by articulating mechanical fasteners (photo 3). The roof of the front box is then secured to the roof of the rear box.


(3) This view from inside the attic shows the typical roof assembly of modular construction. Note the (black) articulating mechanical fasteners.

Of particular interest was that solid foam was attached on one side of the visible ceiling joists. A contractor indicated that the foam substance was a structural adhesive used to affix the gypsum board to all ceiling joists and interior partitions and that this construction technique was common in prefabricated homes.

We conducted an Internet search on structural adhesives used in prefabricated housing. The information we found indicated that glue is used in wall and ceiling assemblies to prevent the mechanical fasteners from popping through during transport of the modular boxes to the location of assembly. The glue makes it possible to use fewer mechanical fasteners in construction. The International Code Council’s (ICC) International Residential Code (IRC) approves this practice. We have since learned that the Federal Emergency Management Agency applauded the structural integrity of modular homes in Florida; they withstood fierce wind forces during Hurricane Andrew. We found information on two brands of structural adhesives available only for use in prefabricated housing. Both are described as a two-part, polyurethane foam plastic structural adhesive. Evaluation reports and product listing specifications available on the Internet do not indicate the flammable nature of these products.

Lacking information on the flammability of the structural adhesive, we contacted officials at Worcester Polytechnic Institute’s Center for Fire Safety Studies in Worcester, Massachusetts. We discussed the scenario of the modular building fire in our community and asked for information relative to the flammable nature of several items collected during the investigation. The researchers agreed to conduct Cone Calorimeter testing on the following samples of items from the fire scene and a newly assembled modular home in our community:

  • unburned vinyl siding (fire scene),
  • plastic house wrap (fire scene),
  • vinyl soffit cover (fire scene), and
  • polyurethane foam plastic structural adhesive (newly assembled modular home).

Test Results

The researchers explained that the information that can be determined by Cone Calorimeter testing includes the speed at which an item ignites when exposed to heat, the amount of heat released, the amount of mass the sample loses when subjected to heat, and the optical density of the smoke generated.

All four samples were exposed to the same amount of heat energy (35 kW/m2). In addition, the specimens were rated for smoke obscuration by establishing the Specific Extinction Area (SEA). A rating of 1.0 indicates thick black smoke (difficult to see through). A rating of 0.1 is tolerable smoke that can be seen through for a good distance. The data attached on each sample indicated the following:

  • The vinyl siding ignited after 30 seconds of exposure to the heat energy and had a peak heat release of 132 kW/m2. A peak SEA of 0.68 indicates the production of thick smoke.
  • The vinyl soffit material ignited after 75 seconds of exposure to the heat energy and had a peak heat release of 141 kW/m2.
  • The house wrap material ignited after 14 seconds of exposure to the heat energy and had a peak heat release of 130 kW/m2.
  • The polyurethane foam plastic structural adhesive ignited after seven seconds of exposure to the heat energy and had a peak heat release of 241 kW/m2. A peak SEA of 1.199 indicates the production of thick black smoke.

In addition, the test results indicated that the adhesive sample lost a significant amount of its mass during testing. The sample weight prior to the test was 7.7 grams. The sample’s end weight was 1.9 grams, for a loss of 5.8 grams.

Finally, researchers noted that the “specimen ignited rapidly and achieved full ignition over the entire specimen surface.”

Researchers familiar with Cone Calorimeter testing claim that peak heat releases measuring above 200 kW/m2 are hazardous. The sample of the polyurethane foam plastic structural adhesive released nearly seven times the amount of heat energy it took to ignite it. By comparison, Cone Calorimeter testing on samples of the polyurethane foam taken from The Station nightclub following the tragedy of February 2003 indicates that the foam sample was also exposed to 35 kW/m2 of heat energy, ignited in seven seconds, and released significantly more heat energy (605 kW/m2). Information on the testing conducted on The Station polyurethane foam can be found in the “Report of the Technical Investigation of The Station Nightclub Fire” compiled by the National Institute of Standards and Technology.

Void Space

As found on the plans on file for all the modular homes in our community and confirmed by visual inspection on-scene following the fire, a void space is created when the second-level modular box is placed on top of the first-level box. This void is the result of each box’s being made at the factory with a completed ceiling and floor assembly. When attached on-site, a void in excess of 20 inches from ceiling to floor can be created.

Recently, a modular home was assembled in our community. This structure is also a single-family, two-story Colonial measuring 46 × 28 feet. Part of the first-level ceiling assembly was missing; this allowed photographs to be taken of the inside of the void space found between the first and second levels. One photo, looking down the length of the ceiling joists toward the exterior wall, shows where two-inch rigid insulation (also flammable) can be seen. Also pictured is the polyurethane foam structural adhesive (photo 4). This glue was found along the joists used to support the second-level ceiling as viewed from the attic.


(4) This photo was taken from inside the void space. Note the adhesive on the ceiling joist and the rigid insulation on the far wall.


As a firefighter, I am particularly concerned about the amount of concealed void space created by this construction method. Fire, as we know, requires the presence of heat, fuel, and oxygen to propagate. The void space found in this modular home has all three of those ingredients in large quantities. In addition, this void space, as well as the void space in the residence described in the incident above, contains a structural adhesive placed along each structural member that we now know is easy to ignite, burns readily, and loses the vast majority of its mass. A fire in this void space would easily spread horizontally by the movement of superheated gases and the available fuel in the forms of the wood subfloor and the floor and ceiling joists, with the glue acting as an accelerant.

Building codes attempt to minimize the spread of fire within a structure by blocking the upward movement in wall spaces by mandating fire blocking. Draft-stopping is also used to compartmentalize a space so that in a fire the movement of the superheated gases and other products of combustion are slowed. Yet, after reviewing the plans of all the modular homes in our community, a significant void space—on one plan labeled as a “plumbing chase”—is incorporated into each design.

A review of the IRC finds that in certain circumstances draft-stopping is required in concealed spaces. Section R502.12 reads as follows:

When there is usable space both above and below the concealed space of a floor/ceiling assembly, draft-stops shall be installed so that the area of the concealed space does not exceed 1,000 square feet. Draft-stopping shall divide the concealed space into approximately equal areas. Where the assembly is enclosed by a floor membrane above and a ceiling membrane below, draft-stopping shall be provided in floor/ceiling assemblies under the following circumstances:

1. Ceiling is suspended under the floor framing.
2. Floor framing is constructed of truss type open web or perforated members.

Clearly, the intent of the code is to create smaller-sized compartments to limit the spread of fire and the products of combustion. However, given the measurements of the modular boxes used in the homes in our community (typically 48 × 14 feet), the area measures 672 square feet, well below the 1,000 square-foot trigger found in the IRC. It would be interesting to ask those who authored the code if they considered the presence of polyurethane foam along each structural member when the code was developed.

Given the realities found in typical modular construction, I question the code’s use of area—that being length × width—as an indicator of an allowable size for the concealed space. I would ask if a void space of 1,001 square feet with a height of one inch is really that more troublesome than a void of 672 square feet (48 × 14 feet) in an area with a height of 22 inches. Could a more applicable standard be the use of volume (length × width × height) to ascertain the number of cubic feet of space found in the void?

Regardless of how it is measured, the presence of highly flammable glue in a large concealed space in typical modular residential construction is information that every fire official in the country should be made aware of so they can adjust their strategy and tactics at fires involving this type of construction. Imagine the hazards of pulling a ceiling above which is a polyurethane foam-fed fire in a tightly sealed void of considerable volume! The potential for a backdraft is very real.

REVISED CONCLUSION

Based on all the information we have gathered, and given our new understanding of modular construction, we have revised our initial scenario of the fire as follows. As stated earlier, the fire was initially contained under the farmer’s porch roof, where it burned the structural members and vinyl covering. We believe that the fire communicated into the void space between levels through a weakness in the point of connection between the first and second stories. The holes used to run the lifting cables are in this general location. Once inside the void space, the fire consumed the flammable glue, which caused the collapse of the first-floor ceiling. Similarly, once the fire extended into the empty attic, it burned away the glue, resulting in the collapse of the second-floor ceiling and eventually the truss roof. This revised scenario has been shared with fire officers, fire investigators, and fire researchers. All viewed the sequence of events, given the presence of the glue and the void space, as highly plausible.


(5) Modular boxes being assembled.

I view this fire incident as a systematic failure of several of the systems that made up this structure. The domino effect caused by the burning of the vinyl siding, vinyl soffit covering, and polyurethane foam adhesive resulted in the flame impingement of the structural members of the floor and ceiling assemblies. The survivability of a structure is improved by slowing the spread of fire. The presence of flammable glue on each ceiling joist, as well as the presence of a large concealed space of considerable volume, adds to the speed with which fire can spread and increases the possibility of structural collapse.

Several of the officials with whom I spoke regarding this fire share my concern regarding the presence of flammable glue in partition walls in close proximity to electrical heat sources. In light of our experience, the flammable glue must now be viewed as a readily identified hazard that can burn quickly when exposed to heat energy. Would an arcing wire act as a possible source of heat in terms of energy and duration, possibly igniting a fire in a partition wall space? Are firefighters familiar with the layout of the electrical system in typical modular homes that have the wires placed along the exterior side of the wall joists, fitted into place by notches, covered over with steel nail plates, and then covered by the exterior sheathing?


(6) Modular boxes are lifted by cables and put into place.

Our experience raises serious concerns as to the overall safety—to occupants and firefighters—with respect to certain modular construction techniques. Further testing and research may be appropriate. Can draft-stopping or fire-blocking be used to minimize the size of the concealed space? Can the void space be filled with noncombustible insulation to slow the fire? Most importantly, are there alternative methods for attaching gypsum board to structural members by structural adhesive that does not incorporate the use of polyurethane foam glue? The fire service has known for years the dangers polyurethane foam poses. Upholstered furniture and foam mattresses have long been known to ignite readily, burn hot, and create copious amounts of toxic smoke. In February 2003, we all learned of the role polyurethane foam played in the tragedy at The Station nightclub. The foam in The Station nightclub was installed as an interior finish material on some of the interior wall surfaces; it was intended to reduce noise.

According to an article in the Boston Globe1: “Untreated polyurethane foam is highly flammable, emitting a thick, black, toxic smoke. If the Derderians (the club owners) were unaware of its hazards when they put it on their club walls in 2000, Jeffrey Derderian knew about the dangers of similar products a year later. In February 2001, as a reporter for Boston’s WHDH-TV, he told viewers, ‘Fire safety experts call this stuff solid gasoline.’”


(7) An interior view of preassembled boxes.

Solidified gasoline on the outside and solid gasoline on the inside! Welcome to the ever-changing realities confronting the fire service in the 21st century.

LACK OF DATA COLLECTION

As a result of the fire that destroyed a single-family home of modular construction in Acushnet, Massachusetts, local fire officials have asked if their experience was an isolated incident or part of a national trend.

Officials have not been able to answer this question because of a lack of data collected by state and national fire data collection centers. Although the National Fire Incident Reporting System (NFIRS) requires information relative to the structure type (enclosed building, fixed, portable, or mobile structure, etc.) and the building status (occupied and operating, vacant, etc.), it does not require information on the methods (stick built or modular) used to construct the building. Individual state reporting requirements often reflect the NFIRS data collection fields. In Massachusetts, these data are not collected.

Because of the lack of available data on the types and extent of fires in residences of modular construction, we were forced to post inquiries on Web sites seeking experiences with these types of fires. These sites proved useful for anecdotal information, but no clear analysis or conclusions can be made.

The construction type should be included as a collectible field for data gathered on building fires from departments across the country. We can’t act until we know the extent of the problem. Based on my experiences, I’d say the problem is very real.

This past October, an informational session on fire safety and the use of structural adhesives, sponsored by the Fire Protection Research Foundation, was held at the National Fire Protection Association headquarters in Quincy, Massachusetts. State regulators from New England and representatives from the modular home industry and manufacturers of structural adhesives attended. Attendees agreed to continue the dialog on several important items that were discussed.

The United States Fire Administration recently included the basics of modular construction, including the existence of large void spaces and the use of structural adhesives, in a six-part “Coffee Break Training” series. These documents can be downloaded at www.usfa.dhs.gov/nfa/coffee-break/. This information should be part of every firefighter’s toolbox.

Endnote

1. “Deception, missteps sparked a tragedy,” Boston Globe, June 8, 2003.

KEVIN A. GALLAGHER, a member of the Acushnet (MA) Fire/Rescue Department since 1986, has served as chief since 2004. Previously, he was a call firefighter and a career firefighter/paramedic. He is an adjunct instructor in the fire science technology program at Bristol Community College. He represents the Fire Chiefs Association of Massachusetts on the Fire Prevention/Fire Protection Advisory Committee for the Commonwealth of Massachusetts.

No posts to display