Fire walls

3.4.1 GENERAL CRITERIA FOR FIRE WALLS

Fire walls are intended to prevent fire spread from one side to the other for a specified period of time. The following three criteria must be satisfied to achieve this function:

1. The wall should have a fire resistance rating at least equal to the specified period of fire confinement.

2. The wall should prevent fire spread around or over it as well as through it.

3. The wall should remain standing despite the collapse of the building roof or framing as a result of fire exposure on one side.

The fire resistance criterion can be satisfied by a standard fire exposure test as described in Section 3.3.1. A 4-hour fire resistance is called for in the Factory Mutual Data Sheet for Maximum Foreseeable Loss Fire Walls (FM Data Sheet 1-22, 1985), which is intended to provide fire confinement despite the loss of automatic sprinkler protection. Since the entire wall cannot be tested in the furnace, it is important that a representative section be tested with representative structural loading.

The second criterion may entail the use of wall extensions such as parapets and the use of fire resistant construction on the roof and exterior wall sections adjacent to the fire wall. Factory Mutual recommendations in Data Sheet 1-22 (1985) call for a parapet height of at least 76 cm (30 in) above the top surface of the roof, while British Building Regulations cited in Cooke (1985) specify one-half this value (15 inches or 38 cm). FM Data Sheet 1-22 (1985) also recommends that at least 7.6 m (25 ft) of the roof surface on either side of the fire wall be covered with gravel or slag or an outdoor fire resistant coating. Other recommendations in the FM Data Sheet 1-22 (1985) cover the location of combustible structures and heat and smoke vents on the roof near a fire wall, and the maximum elevation (3 ft or 0.94 m) of wall penetrations such as pipes, conduits, cables, and ducts. Duct penetrations should incorporate breakaway connections and fusible link actuated dampers. Fire door designs and performance are discussed in Section 3.5.

The stability criterion for an effective fire wall requires special construction techniques to achieve stability despite the deformation and eventual failure of building structures on the fire side of the wall. Typical designs and design basis loads are described in Section 3.4.2.

3.4.2 FIRE WALL DESIGN

Figure 3.6 shows three types of fire walls constructed between columns on a double column line. The walls differ only in the connection between the wall and the adjacent structural steel sitting on the columns. If the wall is fastened to the building framework on both sides of the wall, it is called a tied fire wall. If there are no ties to the adjacent framework, and the wall is self-supporting, it is called a cantilevered fire wall. If the wall is fastened to the framework on one side and independent of the framework on the other side, it is called a one-way fire wall.

Tied fire walls

In order to maintain support during a fire, the columns and steel framework adjacent to tied fire walls should have a fire resistance rating at least equal to that of the wall. The forces exerted on a wall tied to a purlin heated by a fire are illustrated in Figure 3.7. The top sketch in Figure 3.7 shows the heated beam pushing the fire wall that is restraining it from thermal expansion. It can be shown (Cooke, 1985) that the horizontal force, P1 (lb_f), is

P1= αsAbEs(Tb− T0) [3.4.1]

Roof deck H ≥ 30'

Flashing

I beam

I beam

I beam

Section A-A

Longitudinal I beam

Reinforced masonry block wall

Protected column

A A

FLOOR

FOOTING Reinforcing

bar

Tied fire wall

Cantilevered fire wall

One way tied fire wall

Figure 3.6. Types of fire walls (shown for purlin parallel to fire wall)

where α_s is the coefficient of thermal expansion of steel (^◦F⁻¹), A_b is the beam cross-sectional area (in²), Es is the Young’s modulus of elasticity for steel (psi), and Tb− T0 is the temperature rise of fire heated beam (^◦F).

As the steel weakens from heating, the roof loading causes the purlin to sag and pull the fire wall toward the fire as illustrated in the middle sketch in Figure 3.7. FM Data Sheet 1-22 (1985) suggests that the pulling force can be estimated by treating the sagging beam as a cable subjected to a vertical force per unit length, W . The parabolic approximation to the catenary curve of a sagging cable is

P2= W Lb2

8δ_c [3.4.2]

where P₂ is the tension in collapsed beam an pull on fire wall (lb_f), L_b is the beam span perpendicular to fire wall (in), δc is the sag of midpoint of collapsed beam (in), and W is the roof weight per unit beam length (lbf/in).

T_b P₁

P₁ = a_aA_bE_s (T_b − Tq)

Fire Tied

fire wall

Heated beam

Fire W₌ wS_b

WL_b² 8 d_c

Tied fire wall

P₂

P₂ = Collapsed beam

d_c L_b

Beam Beam Beam

Fire wall

L_b S_b

Figure 3.7. Forces on tied walls due to heated/collapsed beams perpendicular to wall

The roof loading, W , is equal to wSB, where w is the weight per unit roof area (psi) and SB

is the beam spacing (in) as shown in the bottom sketch in Figure 3.7. According to FM Data Sheet 1-22 (1985), the sag, δ_c, is approximately equal to 0.07 L_b for open web joists, and equal to 0.09 Lb for heavy trusses and wide flange beams. A Similar equation is given in the NFPA firewall standard (NFPA 221-1994) with a safety factor of 1.25 and additional guidance for wood trusses. Alternatively, the sag can be estimated from the midspan deflection of a uniformly loaded simply supported beam. Before the elastic limit is exceeded, this deflection is

δc = W Lb4

384EsI [3.4.3]

where I is the moment of inertia of the beam.

If there is continuous steel framework through the wall, the horizontal forces calculated from equations [3.4.1] and [3.4.2] should be resisted by the lateral strength of the steel on the other side. If it is a one-way fire wall, the wall itself will have to provide the resistance. FM Data Sheet 1-22 (1985) shows the recommended form of the ties for one-way fire walls and discontinuous roof trusses with purlins parallel or perpendicular to the fire wall.

Cantilevered fire walls

In the case of a cantilever fire wall, the spacing between the wall and adjacent steel framing should allow for steel expansion and/or wall deflection during the fire. The beam elongation, δb, is

δb= αsLb(Tb− T0) [3.4.4]

while the deflection at the top of a wall heated from one side is (Cooke, 1985) δ_w = α_cH²(T2T1)

2d [3.4.5]

where δ_w is the horizontal deflection of top of wall (in), H is the wall height (in), d is the wall thickness (in), and T₂− T1 is the temperature difference across heated wall (^◦F).

Figure 3.8 illustrates these deflections when the wall bows away from the heated steel beam and when the fire occurs on the other side of the wall such that the wall bows toward the beam. When the fire is under the beam but not close enough to the wall to heat it, δw is negligible and the gap between the wall and the end of the beam should be greater than the beam displacement given by equation [3.4.4]. If T_b− T0= 1000^◦F (538^◦C), and α_s is given by equation [3.1.1], δ_b= 0.008 Lb. The spacing recommended in FM Data Sheet 1-22 (1985) is approximately 0.0105 Lb. This displacement corresponds to a temperature rise of 1240^◦F (671^◦C).

In the case of a fire on the other side of the wall, as illustrated in the bottom sketch in Figure 3.8, the wall-to-beam gap should exceed the deflection given by equation [3.4.5]. If T2− T¹= 500^◦F (260^◦C) (T₁should be less than 250^◦F or 121^◦C to pass the fire resistance test), δ_w= 0.002 H²/d.

FM Data Sheet 1-22 (1985) specifies that H /d should not exceed 20 for hollow masonry walls and 30 for masonry walls that are at least 75% solid. At the upper limit for solid masonry walls, δw= 0.060 H . This requirement can dictate larger gaps than that based on δbwhen H > 0.13 Lb. Cantilever fire walls should also have a lateral strength sufficient to resist thermal stress induced by the fire, volumetric expansion and buoyancy pressures generated by fire gases (see Chapter 4), and the pull of flashing attached to the roof cover. The design basis lateral load specified in FM Data Sheet 1-22 (1985) for both cantilever walls and tied walls is 24 kg/m²(5 lb/ft²). This strength for cantilever walls usually entails use of reinforcing bars (with sizes and spacing determined

d_w

d_w d_s d_b

L_b

T_b Heated beam

Fire

Fire

d_b= asL_b(T_b− T0)

d_w= H

H T₁

T₂ T₁ T₂

d

a_cH²(T₂− T1) 2d

Figure 3.8. Deflections relevant to cantilevered fire wall

from concrete design manuals) extending from the concrete footing up to more than half the wall height as indicated in Figure 3.6. Pilasters with reinforcing bars or encased steel columns are also used to achieve lateral stability.

A double fire wall consists of two one-way fire walls situated back-to-back with a minimum spacing as specified above for cantilevered fire walls. The only connections between the walls

should be the roof flashing which should consist of separate pieces with only frictional resistance holding them together. When constructed to these specifications, double fire walls are considered very reliable. They are often used when an addition to a plant requires a fire wall separation from the existing structure.

3.4.3 FIRE WALL LOSS EXPERIENCE

There have been numerous examples of fire walls providing effective isolation during severe fires. The General Motors Livonia fire described in Appendix B is one example where one fire wall isolated a relatively small part of the facility, and several other fire walls were needed. The Ford Cologne, Germany warehouse fire (also described in Appendix B) is a good example of a double brick fire wall confining a severe fire with the help of manual firefighting so that a large portion of the warehouse remained almost undamaged.

There are several examples of ineffective fire walls. One of the classic examples of fire wall failure is the K Mart fire also described in Appendix B. Failure of the K Mart fire walls has been attributed (Best, 1983) to the use of a tied wall with unprotected adjacent steel columns and framework, and the lack of sufficient reinforcement to resist the lateral forces developed by the collapsing steel (FM Data Sheet 1-22, 1985). Another important weakness was the use of water deluge curtains instead of fire doors at several openings in the fire wall.

The 1987 Sherwin-Williams warehouse fire also penetrated a fire wall, which in this case was a 16.5 cm (6.5 in) thick concrete cantilevered wall. Wall failure was attributed (Isner, 1988) to flammable liquid fire spread under the fire doors such that the wall was exposed to fire on both sides. This two-sided fire exposure caused massive spalling of the concrete wall panels and collapse of some panels. Both the Sherwin-Williams fire and the K Mart fire demonstrate the importance of providing effective fire doors to maintain fire wall integrity.