NBSIR 79-1922

An Interlaboratory Evaluation of the

ASTM E 84-77a Tunnel Test Modified
By the Consumer Product Safety
Commission for Cellulosic Loose Fill
Insulation

J. Randall

Lawson

Center for Fire Research

National Engineering Laboratory
National Bureau of Standards
Washington, D.C. 20234

November 1979
Final Report

Prepared for:

U.S. Consumer Product Safety Commission

Textile and Mechanical Engineering Group
Bethesda, Maryland 20202

"1;1
I1I1

Ii:

NBSIR 79-1922

AN INTERLABORATORY EVALUATION OF THE

ASTM E 84-77a TUNNEL TEST MODIFIED
BY THE CONSUMER PRODUCT SAFETY
COMMISSION FOR CELLULOSIC LOOSE FILL
INSULATION

J. Randall

Lawson

Center for Fire Research

National Engineering Laboratory
National Bureau of Standards
Washington, D.C. 20234

November 1979
Final Report

Prepared for:

U.S. Consumer Product Safety Commission

Textile and Mechanical Engineering Group
Bethesda, Maryland 20202

U.S. DEPARTMENT OF COMMERCE

Luther H. Hodges, Jr., Under Secretary
Jordan
NATIONAL

J. Baruch, Assistant Secretary for Science and Technology

BUREAU

OF STANDARDS,

Ernest Ambler,

Director

II'

Iii

TABLE

OF CONTENTS

Page
LIST

OF TABLES

iv

LIST

OF FIGURES

iv

Abstract

1.

INTRODUCTION

2.

TEST

3.

PARTICIPANTS

4.

LABORATORY

5.

MATERIALS

6.

EXPERIMENTAL

METHOD

6.1
6.2
6.3

7.

AND

APPARATUS

4
4

SURVEY
AND

SAMPLE

PREPARATION

DESIGN

6.4

Test Procedure
Laboratory
Specimen
Test Density
Order of Testing

TEST

RESULTS

7.1
7.2
7.3
7.4

Statistical
Procedures
Tunnel Test Data
Comparison
of Fire Test
Variability

AND

Preparation

7
7
7
8
8

DISCUSSION

Methods

8
8
10
10

11

8.

SUMMARY

9.

ACKNOWLEDGEMENTS

12

10.

REFERENCES

13

AND

CONCLUSIONS

APPENDIX

A-I

APPENDIX

B-1

APPENDIX

C-1

APPENDIX

D-1

iii

LIST OF TABLES
Page
Table 1.

Characteristics

Table 2.

Tunnel test data, flame spread classifications

Table 3.

E 84 Tunnel Test cell averages ordered

to highest flame spread classification

Table 4.

of the ASTM E 84-77a tunnels

14

16

(FSC)

from lowest
(FSC) .

17

E 84 Tunnel Test results cell standard deviations ordered

lowest to highest flame spread classification
(FSC) ....

Table 5.

Precision

estimates

of flame spread classifications

Table 6.

Comparison

Table 7.

Screen wire comparison using the modified

test procedure

of precisi.on estimates

17

(FSC)

for fire test methods

.
.

18
18

ASTM E 84-77a
18

LIST OF FIGURES
Page
Figure

1.

Example of E 84-77a tunnel used in the interlaboratory

program .........................

19
20

Figure

2.

Details

Figure

3.

Cross section of tunnel at BB of figure 2 .

Figure

4.

Screen wire test frames in place on tunnel ledges

Figure

5.

Screen test frames filled with cellulosic

before testing
...............

Figure

6.

of the E 84-77a tunnel

Examples of char and flaming observed

after test was completed

21
22

insulation
23

by laboratories
.

24

iv

AN INTERLABORATORY EVALUATION OF THE ASTM E 84-77a TUNNEL

TEST MODIFIED BY THE CONSUMER PRODUCT SAFETY COMMISSION
FOR CELLULOSIC LOOSE FILL INSULATION

J. Randall Lawson
Abstract

An interlaboratory
determine

precision

reproducibility
and Materials

evaluation was conducted

estimates

for repeatability

of the American

Society

to
and

for Testing

(ASTM) E 84-77a tunnel test as modified

by the Consumer Product Safety Commission

loose fill insulation.

Six laboratories

for cellulosic
participated

in this study by running tunnel tests on eight cellulosic

loose fill materials.

Each laboratory was surveyed

during the project to examine its conformance

critical details of the test apparatus

with the

and procedure.

The results of the survey showed that none of the

tunnels completely conformed with the specifications
of the modified ASTM E 84-77a standard.

The within-laboratory
for repeatability
cellulosic

coefficients

for the six fire-retardant

insulations

of variation

same materials

between

The between-laboratory

for reproducibility

for the

ranged from 31 to 41 percent with an

average of 35 percent.
that different

treated

ranged from 11 to 23 percent

with an average of 16 percent.

coefficients

of variation

There would be little assurance

laboratories would be able to distinguish

insulations which do and which do not comply with

the CPSC mandatory

level

Key words:

Cellulosic

laboratory
test.

evaluation;

(FSC ~ 25) on a consistent

insulation;
precision;

basis.

flame spread; intertest methods;

tunnel

1.

The interim mandatory

which became effective
insulation

materials

limit according
test procedure

INTRODUCTION

safety standard

on September

8, 1978

meet a prescribed

to a specified
is a Consumer

for cellulosic
[1]1, requires

materials

repeatability2
estimates

a)

[2].

Safety Commission

This report is to provide

and reproducibility3

b)

associated

are required

for several reasons:

To determine

the degree of confidence

placed on numerical

test results

(CPSC) modified
characteristics

precision

version
of

estimates

of

Such

which may be

from a given number

tests from an individual

To determine

the extent of variability

test results

on "identical"

material

This

with this fire test.

of replicate

insulation

(FSC)

fire test for flame spread.

of the current ASTM E 84-77a test for surface burning

building

that cellulosic

flame spread classification

standard

Product

home insulation,

specimens

test laboratory.

assignable

to

of a given

tested in any tunnel conforming

to

the CPSC specifications.

An interlaboratory

test program was conducted

to evaluate

the CPSC

modified ASTM E 84-77a test procedure.

The purpose of the test procedure

burning

characteristics

of a material

its surface when exposed

a basis for comparing

lNumbers in brackets
end of this report.

is to determine

the comparative

by evaluating

the flame spread over

test fire.

The test should provide

to a standard

surface burning

characteristics

refer to the literature

of different

references

materials.

listed at the

2Repeatability precision - repeatability or within-laboratory

precision is
defined in terms of the variability between test results obtained in the
same laboratory on the same material [3]
3Reproducibility precision - reproducibility of between-laboratory
precision is defined in terms of variability between test results obtained
in different laboratories on the same material [3].

Ilh'li

Iii

2.

as specified by CPSC requires

The test procedure

E 84 tunnel

(see figures I, 2 and 3).

changes to the test method.

of tolerances

These modifications

on the tunnel.

was to reduce allowable variations

CPSC standard does not require the evaluation

non-substantive

were primarily

A list of these changes is presented

contribution

the use of the ASTM

CPSC made technical

to a number of measurements

these tolerances
equipment.

TEST METHOD AND APPARATUS

the addition

The purpose of

in the procedure

in appendix A.

and

Also, the

of smoke development

or fuel

for these materials.

To briefly describe

the tunnel test for cellulosic

loose fill insulation,

three metal frames 2.53 m (8 ft 3-1/2 in) long and 51 cm (20 in) wide covered
with steel wire screen, wires nominally
in diameter with approximately

1.2 mm

end on the ledges of the test chamber

0.254

0.0254 mm

(0.01

(3/64 in) openings,

(see figure 4).

0.001 in)

are placed end to

In this position

the

test frames with the wire mesh become the interior top surface of the tunnel.
A sample of cellulosic

loose fill insulation

density on the wire screens

specimen

(see figure 5).

is spread at a relatively

uniform

The first 35.6 cm (14 in) of the

(that portion upstream of the burner)

is covered with a piece of

sheet metal.
The specimen length exposed in the tunnel is 7.3 m (24 ft).
The walls and floor of the tunnel are lined with fire brick.
The tunnel has
an inside width of 44.5 cm (17.5 in), a height of 30 cm (12 in) and a length
of 7.6 m
end.

(25 ft).

The tunnel is open at both ends; the "fire" and the "vent"

The specimen,

in a ceiling position with material

exposed through the wire screen, is exposed to diffusion

burners extending
the fire end.

supported by and
flames from two gas

from the tunnel floor and pointed upward 30 cm (12 in) from

A forced draft induced by a blower and controlled

by a damper

system at the vent end of the tunnel pulls air through the opening upstream
of the burners

1.5 m/min

in the fire end.

(240

5 ft/min) is measured

to igniting the exposure

(draft) is maintained
extends

A prescribed

flame.

average air velocity

at the vent end of the chamber prior

After ignition,

and controlled

of 7.3

a constant negative

by the damper system.

pressure

The flame, which

1.36 m (4.5 ft) from the burner, and draft serve to ignite the specimen

and to induce flame spread along the ceiling of the tunnel.

Windows located
on the side of the tunnel allow an observer to record the extent of flame
spread as a function of time over the last 5.9 m (19.5 ft) of the 7.3 m
(24 ft) specimen.
classification

The test duration

is 10 minutes.

The flame spread

(FSC) is based on a scale which has 0 for asbestos-cement

board and 100 for a selected grade of red oak flooring.

for the wire screen is determined
with and without

the screen.

by conducting

factor

tests on the red oak flooring

The FSC for the unscreened

A correction

flooring divided

by the value obtained with the screened flooring gives the screen correction
factor

(SCF).

The FSC calculation

takes into consideration

under the flame spread time-distance

curve.

area under the curve is then multiplied

flame spread classification
are not considered

12].

3.

Five commercial
participated

testing

Each laboratory

possessed

few of the participants

test procedure
anonymity

of one year's experience

insulation.

that exceeded

the laboratory

LABORATORY

laboratory

survey was to determine

each laboratory's

tunnel characteristics

made.

was visited.

dard.

because

of time restraints,

The purpose

of the tunnels

obtained

from the

of tunnel dimensions,

When items were found to deviate

were informed

and

Table I lists the

and information

As shown, the survey included an evaluation

value, the laboratories

of the

so that changes

from the
could be

the survey it was found that none of the tunnels

completely with the CPSC or ASTM E 84-77a specifications.

The
on table I indicate the items that did not conform with the stan-

Most of the tunnels did not conform with the standard because

turbulence

only

Most of the visits were

the characteristics

that were measured,

Upon completing

conformed
asterisks

are assigned

SURVEY

use of the test procedure.

and instrumentation.

specified

the participants'

test data are identified

two visits were made while testing was in progress.

controls

with the modified

the laboratories

made before the testing was under way; however,

operators.

However,

code numbers.

Each participating

to evaluate

with the tunnel.

20 years.

experience

To maintain

to their test results,

4.

laboratory

laboratory

B lists the names of the participants.

had experience

In this report,

by the participants'

and one private

had more than four months'

as it relates

code numbers.

test procedure.

Appendix

for cellulosic

and smoke development

PARTICIPANTS

a minimum

Several of the laboratories

from the

by the SCF to give the reported

laboratories

in this study.

The value developed

Fuel contribution

in this modified

the total area

bricks were not located where specified.

has not been documented,

it is believed

Although

the

it apparently

that the air flow pattern

in the

tunnel is a critical factor in flame spread measurements and is affected by

the placement of the bricks.
A second area where the tunnel was often out
of compliance

was the placement

not use the prescribed

discussion

of the end thermocouple.

wire size for the wire screen support.

of wire mesh study in section

which indicates

One laboratory

the possibility

7.4).

Another

of some basic problems

(See the

area of variability
in the control

IIH

II

1111,

I,

,1,,1 .. 11

did

II,dl "I'

of

tunnel operating
participants.

conditions

is the wide heating rate range recorded by the

The standard calls for an approximate

heating rate for the tunnel burner.

other operating

parameters

and tunnel geometry,

such as air velocity,

This represents

which is approximately

by the standards,
room temperature

As shown in table 1, the gas heating

(4805 BTU/min)

a variation

to 5.6 MJ/min

of almost 527 KJ/min

10% of the recommended

could have a significant

One potential

burner design and location,

the flame from the gas burner should extend 1.37 m

rate ranged from a low of 5.1 MJ/min

variation

(5000 BTU/min)

With this flow rate and maintaining

~ 152.5 rom (4.5 ft ~ 6 in) downstream.

BTU/min).

5.3 MJ/min

(5290

(500 BTU/min)

heating rate.

This type of

influence on the tunnel test results.

problem that was observed which is not directly

specified

as it relates to running the test, are variations

in test

and humidity.

Some of the participating

located in cold climates and some are not.

laboratories

While observing

are

tests in progress,

it was noted that a number of tunnel operators were not able or did not
attempt to maintain
room.

uniform temperatures

In some cases, it is doubtful

and humidity were maintained.

velocity

in the tunnel.

and humidity

that consistent

conditions

in the test

air supply temperature

This variation would result in changing

Significant

variations

were caused primarily

air
by the

opening of outside doors to remove samples from the tunnel test room and to
vent smoke from the building.

In one case it was snowing and the building

doors were left open until the tunnel floor temperature

was down to the

prescribed 40.5 ~ 2.8C (105 ~ 5F). At this time the building doors were
closed, the sample placed in the tunnel, and the test was started.
By this
time, the 3.91 m
(96F).

(13 ft) floor thermocouple

It is known that the moisture

rials varies with humidity

reading had dropped

to 35.6C

content of paper and wood base mate-

(and temperature)

level and is a significant

factor

in flame spread measurement.

In one laboratory,

the tunnel air velocity was reset four times during

a single day's operation because of changes in barometric

front approached

the area.

In other laboratories

pressure

these adjustments

as a storm
are not

routinely made.

The effects of some of the factors identified

in the survey may have

only a minor influence on the test results, but a combination

factors may introduce

significant

differences.

of these

5.

MATERIALS

Six loose fill cellulosic

the horne insulation
are typically

treatments

insulation

market were chosen

installed

In addition,

AND SAMPLE PREPARATION

for the program.

materials

were also chosen to provide

Neither

visible

characteristics

of these products

The materials

and chemical

characteristics.

components

without

The processes

were selected

to provide

the cotton product

processes.

but their

treated materials,

were cotton

fire-retardant

the

The primary
fibers and ground

chemicals

to the mate-

for three of the waste paper materials

and chemicals

for the remaining

components

of the untreated

were

waste paper product,

and the ground wood product were applied

The primary

perfor-

treated cellulosic

was ground waste paper.

for adding

chemical

a broad range of physical

Of the fire-retardant

The chemicals

applied by dry mixing,

by blowing machines.

are used as horne insulation,

of four materials

rials also varied.

All of these materials

a base line for comparative

for the other two treated materials

wood chips.

being sold in

fire-retardant

are similar to fire-retardant

insulations.

primary component

currently

in horne attics and wall cavities

two cellulosic

mance.

materials

through wet

cellulosic

products

were

waste paper and ground wood.

Twenty-five
Upon receiving

bags of each material

each product

were obtained

from the manufacturer.

a code letter was assigned

to the material.

code letters A through H were used for identification.

then blended
containing

using the following

procedure.

Twenty-five

enough to hold the quantity

bags were used to protect
of material

required

the specimens

to each of the sample boxes.

bag starting

were distributed

group A had been distributed.

product
by one.

consecutively.
The plastic

shipment.

A bag

and a fraction

an equal distribution

of

of the product

to the bottom.

The samples

order so that each box being filled received

was followed

until all of the samples

To ensure that each laboratory

from

sample would

cross section of the group, when the next bag within

group was sampled the order of the laboratory

Each material

set and placed the last sample in box number

group was prepared

using the above procedure.

'Iu,

II

Iii

,I

sample boxes was indexed

This placed the top sample of the second bag from group A in box

number two of the laboratory

one.

during

at random, weighed,

at the top and progressing

This procedure

an average

boxes

Samples of equal weight were taken from the

in consecutive

an equal portion.

from moisture

which would produce

was

a single bag was not large

for a single tunnel test.

from group A was selected

the weight calculated

represent

corrugated

two large plastic bags each were set out and numbered

Two plastic bags were used per sample box because

material

Each material

The

Iidl

Iii,

After the samples were mixed the boxes were weighed

were sealed.

The samples were then placed into laboratory

numbers were recorded,

contained
tion.

and the boxes were prepared

the equivalent

as obtained

from 11.4 to 18.2 Kg (25 to 40 lb).

Each participating
materials.

three boxes, sufficient

DESIGN

laboratory was shipped three samples of eight different

A letter of instruction was also sent with a copy of
in the August

8, 1978 Federal

pointed out that technical non-substantive

made to the ASTM E 84-77a test procedure.

The letter also requested

of the wire screen to the red oak flooring and explicit

given for the procedure

for developing

See appendix C which describes

6.2
The laboratories

Laboratory

factor

were
(SCF).

used.

Specimen Preparation

were informed that each box, containing

represented

that the

the attachment

instructions

the screen correction

the procedure

Register.

changes had been

strictly comply with part l209.4(c) (9) regarding

bags of insulation,

ranged

Test Procedure

the test procedure which was published

laboratories

bag of insula-

from the manufacturer

providing

EXPERIMENTAL

6.1

The instructions

Each box

tests, for each of the eight materials.

6.

cellulosic

for shipment.

lot

A total of 24 boxes of cellulosic

were shipped to each participant,

for three replicate

groups,

sample weight of one manufacturer's

The original bag weights,

materials

and the plastic bags

one sample.

It was requested

two plastic
that each

sample of cellulosic material be blended by dumping the two bags of material

into an insulation mixer/blower, mixed for a minimum of two minutes, and then
blown.

The material was to be blown into a container

using the procedure

specified

in the standard.

6.3
Since there was a significant
one material
specified.
40 Kg/m3

difference

All specimens except material

(2.4 and 2.6 lb/ft3).

in the physical

It was requested
(1.5 lb/ft3)

(1.4 to 1.6 lb/ft3).

7

were

38 and 42 Kg/m3

H be tested at a

with an allowable

of

to be tested at

range between

that material

properties

test densities

H were specified

density with an allowable

nominal density of 24 Kg/m3

26 Kg/m3

Test Density

as compared to the others, two different

(2.5 lb/ft3)

and then conditioned

range of 22 to

6~4 Order of Testing

Two red oak flooring reference

tests were conducted

select red oak flooring used for calibrating

laboratory.

One of the tests was conducted

the red oak flooring

and one was conducted

initial tunnel calibration

tested.

with the wire screen attached

without

materials

the screen.

factor was calculated

materials

were

one

As shown in appendix

using the screened

to

After the

tests were completed,

screen covered red oak test was run.

screen correction

The standard

the tunnel was supplied by the

with the red oak, the cellulosic

When all of the cellulosic

additional

first.

C, the

and unscreened

red

oak test results.

7.

TEST RESULTS AND DISCUSSION

7.1

Six laboratories
three replicate
participants

Statistical

participated

were compiled

that is a consumer

and compared
The standard

Committee
methods

Each laboratory

The data generated

with the requirements

ran

by the

of the CPSC

states that any cellulosic

insulation

product must have a flame spread rating of 0 to 25 II].

Also, the data were statistically

Recommended

to Determine

in the test program.

tests on each of eight materials.

interim safety standard.

"Tentative

Procedures

Practice

the Precision

E 11 [3].

analyzed

using the methods

for Conducting

of Test Methods"

Appendix

presented

an Interlaboratory

which

Test Program

is being prepared

D is a brief presentation

in the

by ASTM

of the statistical

used in this report.

7.2

The data as received

This table exhibits

Of the materials

chemicals.

tively wide ranges.

statistical

from the laboratories

are presented

in table 2.

the data cells4 which are made up of three replicate

tests on each material.

with fire-retardant

Tunnel Test Data

analysis.

tested, D and F were not treated

These products

show test results with rela-

One data unit located in cell 2 D was not used in the

The laboratory

not fill the length of the tunnel,

reported

that the test specimen

did

and the flames spread to the end of the

specimen.

4cell - each of p laboratories makes measurements on each of q materials.

Each cell consists of n replicate
This gives rise to p x q "cells".
measurements [3].

IIH:il

IIIII

Also shown on table 2 are the results obtained

calibration

tests and the screen correction

lating the flame spread classification

calibration

factors

numbers.

(SCF) used for calcu-

The column of red oak

tests shows results for the unscreened

select red oak tested without

for the select red oak

and screened

the wire screen exhibits

116 with an average of 102 and a standard deviation

covered red oak used for developing

the correction

for calibrating

factor produced

factor has a range from 56

of 6.8.

The screen

The data indicate

that the

the screen correction

uniform results.

the mean values of each test cell with the materials

from the lowest to highest

flame spread classification

table.

of 0.11.

the tunnel and developing

relatively

Table 3 presents
ordered

The screen

factors ranged from a low of 1.32 to a high of 1.63 with an

average of 1.49 and a standard deviation

procedure

The

an FSC range of 93 to

of 8.8.

to 77 with an average of 68 and a standard deviation

correction

samples.

flame spread classification.

for each material

The average

is shown at the bottom of the

To the right of the cell averages are two columns which show the

laboratory

test results compared to the criteria

interim standard.

For a material

to pass, it must have a FSC ~ 25.

be seen, the number of passing materials

one laboratory

cell averages

ranged from 0 to 6.

passed all of the materials

and two laboratories

As can

At the extremes,

except the two untreated

failed all of the materials.

for materials

part of the variation

set forth in the CPSC

products,

As shown in the table, the

F and D have a wide range.

is the result of materials'

It is believed

non-uniformity,

even though

major efforts were made to blend materials

from different

variation

may also be due to environmental

variables

(humidity, temperature)

which would be expected to be particularly

sensitive

for untreated

Table 4 presents

the cell standard deviations

material and laboratory.

bottom of the table.

The precision

bags.

that

Part of the
cellulose.

for each cellulosic

The pooled standard deviations

are shown at the

estimates

for the fire-retardant

cellulosic

flame spread classifications

are given in table 5.

For each material

table gives the average FSC and the coefficients

for repeatability

and reproducibility.

F and D which were not fire-retardant

of variation

The precision
treated,

test was not under control for these materials

standard deviations in table 4.

materials
the

calculated

estimates

for materials

are not included because

the

as can be seen by the cell

Looking

at fire-retardant

treated materials,

significant

degree of variation

coefficient

of variation

ducibility

coefficient

exists.

it is apparent

The range for the repeatability

is 11 to 23 with an average of 16.

of variation

that a

The repro-

ranged from 31 to 41 and exhibits

an

average of 35.

7.3

Comparison

The author compared

study involving

of Fire Test Methods

the precision

of three test methods

the attic floor radiant panel

[4].

The comparisons

made using data obtained

from a report by Lee and Huggett

ASTM E 84-70 tunnel test

[5], results

involving

[6], and data obtained

attic floor radiant panel study on thermal

was reasonable
comparison

when compared

to other fire test methods.

one step further,

of the other test methods

the precision

from the

To carry this

of the data obtained

treated materials
(see table 6).

when compared

coefficient

of variation

test method

is 67% higher.

Table 6 indicates

that the CPSC

coefficient

The results when

panel test show that the median

for reproducibility

of the modified

It is also 30% greater

of variation

from this

is compared with the precision

to the other test methods.

compared with the attic floor radiant

coefficient

and Adams

The conclusion

ASTM E 84-77a test has a notably higher reproducibility

of variation

cibility

[4].

the

was that the attic floor radiant panel test precision

study on the fire-retardant

modified

insulation

were

concerning

from a report by Benjamin

the flooring radiant panel test

from this comparison

in a recent

obtained

ASTM E 84-77a

than the median

reprodu-

with the ASTM E 84-70 test with

carpet.
Some of this variability as compared to the carpet materials
be attributed to the known random characteristics of the cellulosic

can

materials.

7.4

Variability

Through test observations and data analysis, sources of variability

that could influence the test results have been identified.
The most
significant
between

variable

influencing

the wire screen specimen

approximately

2.53 m

test results
frames.

appears

to be the two joints

These joints are located at

(8 ft 3-1/2 in) and 5.06 m (16 ft 7 in) as measured

along the tunnel length from the fire end of the test chamber.
laboratories
observed

reported

the flame spread distance

char distance

tests, 57% indicated

the test frame joint.

and presence

occurred within

4 shows a picture

screen wire joints open and sagging.

the test, and the

of flaming after the test.

that these phenomena

Figure

during

Five

Figure

Out of 120

30 cm

(1 ft) of

of the test frames with the

6 shows examples

of Charring

10

,IU,II

Illi

:1,

1.1'1

II,

and flaming observed by laboratories

one case a laboratory

reported that the screen joint appeared

with the fiame travel.

was generally
laboratory

Observation

recognized

In

to interfere

of the joints showed that the wire screen

loose and sagging after conducting

reported bridging

two or three tests.

One

this problem before starting the test program and

the test frame gaps with small pieces of wire screen.

test results from this laboratory

the test frame joints.
throughout

after the tests were, completed.

showed the least variations

Most of the laboratories

the entire program.

The

associated

with

used the same test frames

Another problem also associated

with the wire

screen is its tendency to develop deep and random sags which alter the air
flow patterns

and localized radiation

characteristics

sags in the wire screen were measured

normal plane of the specimen

to be 3.8 cm

in the tunnel.

(1.5 in) deep

Some

(below the

surface) and 25.4 cm (10 in) along the tunnel

lengths.

Laboratory 3 attempted to avoid wire sag by using a stainless steel wire

screen with a wire diameter of 0.508 mm (0.020 in). This screen does not
meet the specifications but reduced the extent of sagging.
A limited study
was conducted to ascertain the effect on the FSC. Two sets of identical
cellulosic

insulation

specimens

production

line were tested.

shipped to the laboratory.

specified

selected consecutively

The bags of insulation were randomized

The laboratory

in the interlaboratory

other tested using galvanized

followed

the test procedure

stainless

steel screen as specified

slightly lower but more uniform

values than that obtained

completion
developed

of testing,
numerous

stainless

wire screen sag increases

steel wire

screen.

while the stainless

At the

screen wire had

steel remained

This limited study supports the conclusion

that

SUMMARY AND CONCLUSIONS

participated

in an interlaboratory

designed

to determine

the repeatability

modified

ASTM E 84-77a test procedure

test program

and reproducibility

for cellulosic

of the CPSC

insulation

as published

in the August 8, 1978 Federal Register.

Eight loose fill cellulosic

were tested.

treated

For the six fire-retardant

coefficients

rela-

test variability.
8.

laboratory

(see

flame spread classification

it was noted that the galvanized

irregularities

Six laboratories

in the standard

from the standard galvanized

tively flat and straight.

as

steel screen wire and the

As shown in the table, the large diameter

screen produced

and

program except that one set of three speci-

mens was tested with the large diameter

table 7).

from a manufacturer's

of variation

insulations

for repeatability

percent with an average of 16 percent,

the within-

ranged from 11 to 23

and the between-laboratory

11

materials

coefficient

of variation

for reproducibility

of 35 percent.

There would be little assurance

would be able to distinguish

between

comply with the CPSC mandatory

of the variability
material,

dispersion

to problems

and irregular
temperature,

joints.

level

to the variability.

laboratories

which do and which do not

(FSC ~ 25) on a consistent

to variability

of the chemical

fire retardant,

and exposure

procedures

basis.

in the cellulose

with the wire screen specimen

Also, laboratory

humidity,

that different

insulations

may be attributed

including

attributed

ranged from 31 to 41 percent with an average

frames,

Part

insulation

and part can be

such as sagging

for maintaining

test room

flame gas flow rates may have contributed

The coefficients

of variation

were significantly

higher

than those for the attic floor radiant panel

9.

Appreciation

is extended

provided material
Product

materials

and arranging

Statistical

Engineering

statistical

consultation

Research,

National

project.

Messrs.

ACKNOWLEDGEMENTS

to the cellulosic

for the test program.

Safety Commission

Mr. Bernard

manufacturers

Schwartz

assisted

in obtaining

to ~BS.

Division,

Bureau of Standards,

National

National

Lee Dunlap,

used in this project.

12

assistance

prepared

provided

Flammability
throughout

and Tom Prather

Bureau of Standards,

the

~1rs. Mary G. Natrella,

and Mr. Sanford Davis, Furnishings

provided

who

of the Consumer

for their delivery

Dennis Sullivan,

insulation

and the field offices

Bureau of Standards,

for Fire Research,

the

of the Center

the test specimens

y0.020*
yy
y0.0591t
231
2SB
56needed
5B
1USB
27.9
34.7
22.5
234GS
GS
SS
12
SS
GS
SB
0.059lt
0.011
0.01
0.009
28.9
39.7
As
0.061St
O.0591t
0.011
0.0534t
0.059lt
0.009
of
tunnel (ft )
(oo./year)
Sm::>ke
ern
Type
ofmeter
metalfran LaboratQriea

16
to
y 40
Stardard
'" steel
0.0469
0.01
:t 0.001 6 SB

*These characteristics
do oot meet the standards
tMeasurements made by microsoopic means.
mI = Bottled methane

HW= Hot wire anenareter

Table 1 (Qlnt1nued1

specified

in the J\S'lK E 877. standard.

CA = Chrane1-alume1
ex; = City qas
FW = Rotating vane aJlE!IIQIlBter
SS Stainlesa

15

GS = Galvanized steel
steel

SB = Brooke I:xrID


H
D*
F*
Factor
Oak
Screen
35.3
162.0
95.8
12.9
10.7
60.1
35.0
20.6
44
111.
27.3
0
32.8
29.2
87.3
29.3
32.7
85.0
G
24.4
52
12.3
13.3
12.2
56.7
16.2
53.9
8.1
8.0
1. 103.0
566
181
210
26
13
23
18
96
3
1
36
26
63
Correction
72.00
20.6
148.1
87.4
81.
25.1
41.6
41.
32.6
28.6
28.5
97.0
37.6
33.2
7
6(s)
1.475
116.26
17
30.6
144.4
116.6
28
70
27.2
75
31.
34.0
38.9
2
94
68.7
16.6
149.9
19.9
36
25
32
67
31.1
15.0
87
21
35.5
29
34.0
35.5
97.5
1. 32
28.5
63.1
36.8
40.9
50.8
24.5
36.7
77.4
105.55
46.0
37.3
Table
2. data, flameMaterial
77.2
59.3
68.5
50.0
61.4
12.7
8.2
8 108
408
271
31
23
92.8
71.36
29.0
87.0
29.1
84.8
37.5
97.7
67.1
35.1
7lt
95.5
56.3
77.3
75.36
(s)
57.6
74
.
~
69
89
.3
!!
It was reported(FSC)
that the material
Tunnel
test
spread classifications
(s) the flames spread to the end of the specimen.
and
Select
(s)
Red oak
flooring
screen.
t This
value
was nottested
used with
in thewire
analysis.
fire-retardant
chemicals.

did not

Red
of the tunnel
with

16

Iii

i,l!

~, II

'!

'"

'II'

!!

B
7E20.5
27.3
F
G
E
C
*
Fail
4
30.5
1
22.2
8
19.7
29.0
23.7
33.0
6
2
6
284
109.2
22.1
29.4
32.7
Pass
D*
12.0
238.0
14.7
24.0
28.3
4
29.5
43.1
41.6
30.7
87.3
78.5
0
11.
1
2
13.4
0.8
0 Tunnel
28.5
25.8
31.3
33.7
81.0
69.0
34.1
111.4
Table Test
3. cell averages
25.0
127.9
38.8
44.2
34.9
29.1
32.4
28.5
34.0
88.4
12.2
14.9
93.4
54.7
32.0
Material
186.7
110.8
107.8
35.4
ordered
Average
to highest flame spread classification
(FSC)
treated

with

fire-retardant

from lowest

chemicals.

--

H
C
E
F*
2.22
5.06
6.30
2.31
1.
73
2.38
2.34
2.14
2.46
0.58
4.12
Table
4.
D*2.17
33.51
58.48
1.46
6.40
5.09
6.04
2.31
8.48
32.66
2.37
6.56
2.89
2.43
6.36
0.10
0.23
6.12
5.29
5.77
81.65
2.88
4.62
7.63
3.74Tunnel
16.39
0.81
8.23
10.66
The wide
1.36
4.04
0.00
6.25
2.51
157.87
2.52
3.43
Material
E
84
Test results cell
standard deviations ordered
!!
.
Standard
Deviation
lowest
to which
highestmakes
flameit spread
classification
(FSC)
t the test was out of control materials
with these
inappropriate
to include
pooled standard deviations.
A

were not treated

with fire-retardant

chemicals.

17

20.5
29.4
22.1
31.
3Table
27.3
35.88
3.74
2.88
15.71
13.03
32.90
35.99
41.
00
Deviation
Standard
30.57
32.7
7.63
3.43
10.96
23.33
11.
34.64
74
18.24
8.41
Deviation
4.62
9.67
7.95
Coefficient
Coefficient
Standard
4.12
15.09
8.34
10.83
of
Variation
5.
Laboratory
Repeatability
Reproducibility
Precision
Repeatability
Reproducibility
estimates
of flame 35.16
spread
Average

Table

6.

Comparison

16.06
(FSC)
classifications

of precision

estimates

for fire test methods

11
10
~1edian
Median
12
21
7to
16
27
35
87
15
13
30
31
to 43
41
Repeatability
Range

Coefficient of Variation
Reproducibility

Test
Coefficient
Range of Variation
Method
Material
8 11
to to
15 23
8 to 27
19
4
CPSC Modified
Fire-Retardant
Treated
ASTM E 84-77a
Loose Fill Cellulose
ASTM E 84-70

Carpet

Attic Floor
Radiant Panel

Loose

Flooring
Radiant Panel

Carpet

Fill Cellulose

Table

7.

Screen wire comparison using the modified

ASTM E 84-77a test procedure

0.254 rom (0.01 in)

Galvanized Screen
Wire
(FSC)

0.508 rom (0.020 in)

Stainless Steel Screen
Wire
(FSC)

16.06
21.14
17.00
Average

16.11
15.89
16.11

18.07 + 2.70

Average

16.04 ~ 0.13

18

II

IIII

, '!

nj
,..,

lJl

o
,..,

0..

~
,..,

o
.j..I

nj
,..,

o
..Q

nj

r-l
,..,

Q)
.j..I
s::

-.-I
Q)
..c=

.j..I
s::

-.-I

'tj
Q)
Ul

;j
r-l
Q)
s::
s::

;j
.j..I

f'nj
f'

"'"

co

19

SHEET
VENT

METAL
REMOVABLE

PIPE

TOP

ADJUSTABLE

PANEL

7.6

125')

- LENGTH

OF TEST

SAMPLE

SHUTTER

./

rB

otv

o
__

THERMOCOUPLES
GAS AN-At YSES
VENTS
FOR

PHOTO-ELECTRIC

19

WINDOWS

150 mm x 150 mm

16"

6")

WINDOW

- 70

mm x 300

12*','

mm

Section

FIRE

11*")

- DOOR

AUTOMATIC

LA
u
__ ,

"'"

OPENING

MANOMETER
TUBE.
DIFFERENTIAL
CELL

._;\_
LB

450 mm I17Yz")

DAMPER

Ie

-t

~C:":.]jIoo

mm

Section

C-C

Figure

2.

Details

of the E 84-77a

A-A

tunnel

{WI

END

A1R-SU-PPL

FOR
Y

REMOVABLE

TOP

PANEL

6 '~lI1m
ASBESTOS

50

mm

(2")

CALCtUM

SILICATE

TEMP

980C/1800F)

(MAX

L~")

CEMENT

BOARD

INSULATION

i:~/~:~:;:':;~./\~l
J!__
,

TEST
508

rTI~II,;w~(
228

mm

114

(9"

4~"

.-.....

FIRE

mm

mm

63

(MAX

TEMP

(20")

"~:.;<.~'.:".~:~~:"':.~".<." -,:-;.:

... '.

2~")

BRICK

mm

(4")

tJ~~
}B:~)~~~

50
CASTABLE
(MAX

3.

~~.

GAS

PORTS

f~i~;~:~~t~E

:~~~{!~~~

Figure

SEAL

;;!t:ti,;:

mm

1400C/2600F)

100

""

,~_Wi{,;l\':~'~,~\~
II:"W ATE R

WIDE

'pgi;1i
APPROX

".

SAMPLE

t>.J

....

TEMP

Cross section of tunnel at BB of figure

mm

(2")

REFRACTORY
1370C/2500Fl

VENT END

FIRE END

IV
ol::>

JOINTS

Note: Sk.tchn show screen wire test Ir s II .

loelteel end tI cover the tunnellenlth.

Figure

6.

Examples

of char and flaming observed

by

In

laboratories

the tlllnei. Thr" frames

after

test was

completed

APPENDIX A
Technical Non-substantive Changes Made to the
ASTM E 84-77a Standard by CPSC

The Commission

made these changes in order to:

(A)

Ease some of the restrictive

and test procedure,

specifications

(B)

Simplify

(C)

Eliminate provisions in the requirements

cellulosic insulation, and

(D)

Increase the likelihood

obtained.

for the test apparatus

the test procedures,

that consistent

that are uhrelated

to

test results will be

Changes
1.

Added the following tolerances:

+ 1/16 inch to describe thickness of asbestos fabric gasket tape.

+ 3 inches to describe the fire test chamber length.
3 + 1 inch to describe

Replaced a nominal 2 inches with a tolerance

the insulation for the top of the tunnel.

+ 1 inch to describe the spacing of burners.

+ 1/4 inch to describe the diameter of the vent end flue pipe.
+ 1/16 inch to describe the depth of the embedded thermocouple.
- 0
+ 1/32 inch to describe the asbestos cement board thickness.
+ 1 inch to describe
the air velocity.

the distance

from burner ports for recording

+ 6 inches to describe the distance the test flame must extend

downstream.

+ 6 inches to describe
Tn calibration tests.

the distance

intervals

+ 1/4 inch and + 1/16 inch to describe

the specimen holder.

for making observations

the test frame supporting

+ 2 seconds to describe the duration of the test.

2.

Added

(or deleted) the following

statements

for clarity:

Provision that allows operators to use draft gage tap inserted

through top of tunnel fire end.
Added

"nominal" to dimensions

for steel test frame steel angles.

Added word "nominal" to describe pipe elbow at burner outlet.

A-I

APPENDIX

An alternate air velocity

top of tunnel fire end.
Corrected

reading

two typographical
changed

A (Continued)

range used for draft gage taps in

errors in the ASTM standard.

(a)

Air velocity

from "7.32" to "73.2" m/min.

(b)

Red oak flooring thickness

"25/32 inch."

changed

from "23/32 inch" to

Changed description of screen wire from "galvanized

"steel" to allow stainless steel screen.

steel" to

Loose fill insulation shall be placed on steel screening with wires

normally 0.01 + 0.001 in (0.254 mm + 0.0254 mm) in diameter with
approximate 3/64 in (1.2 mm) openings.
Steel screen correction

factor

(SCF).

A provision to allow the Commission to determine the density of

cellulose insulation for test if manufacturers do not specify.
Deleted terms "delamination"
characteristics.
Deleted

paragraph

and "shrinkage"

8.1 providing

for analysis

as observed

burning

of products

of combustion.

A-2

,Iii

Iii,,,,,,

Iii
I

APPENDIX

Participants
Factory Mutual Research Corporation
Norwood, Massachusetts
Hardwood Plywood Manufacturers
Arlington, Virginia

Associatiori

Southwest Research Institute

San Antonio, Texas
Underwriters' Laboratories,
Northbrook, Illinois

Inc.

Underwriters' Laboratories,
Santa Clara, California

Inc.

Warnock Hersey Professional

Vancouver B.C., Canada

Services Ltd.

B-1

III

,II

, III
,,'

'II

APPENDIX C
Calculation

1.

Determine
the method

2.

Determine

of Screen Correction

Factor

the FSC of the select red oak without

the screen using

specified in ASTM E 84-77a sections 5, 6 and 7.

the FSC of the select red oak with the screen using the

same method.

Two

before insulation

(2) tests with the screen were conducted,

is tested and one

(1) after.

3.

Take the average of the FSC with the screen.

4.

Divide the FSC of the red oak without

the screen

average of the values with the screen

(step 3) .

Example:

value without

95 = 1 46 SCF

the screen

65

Note:
FSC - Flame spread classification
SCF - Screen correction

factor

C-l

one

(step 1) by the

(1)

~'"I'I

ill

I,III

II

,i

'I

,,,' r

APPENDIX D

STATISTICAL

METHODS

Nomenclature

(CV )

Coefficient

(CVR)

Coefficient of variation
laboratories)

d ..

Cell deviations

n ..

Number of replicates

Total number of laboratories

s ~J..

Cell standard deviation

~J
~J

of variation

Component

(sr)

Pooled standard deviation

of variance between

(sR)j

Standard deviation

(s )

Intermediate

x ..

~J

x.

for reproducibility

laboratories

for repeatability

for reproducibility

variance

quantity

Average for cell (i,j) where

and j the material
Average

(between-

per cell

(within-laboratory)

from average

(sL)

for repeatability

for jth material

D-1

represents

the laboratory

for all laboratories

Pooled Standard Deviation

for Repeatability

for the jth material:

E.

Equation

(1) is applicable

each laboratory

for a given material.

Where there are missing

use equation

(Sr)

is the same for

replicates

(la).

(n .. -1) s~ .

E.

of Variation

(1)

~J

only when the number of replicates

in one or more laboratories

Coefficient

s 2..

1 1) 1)
L.(n .. -l)
~ ~J

(la)

for Repeatability:

(s6
-

(CV ) = 100

(2)

x.
J

Standard Deviation
First calculate

for Reproducibility:

the "deviations

from average"

d .. = x .. ~J

Then calculate

the intermediate

~J

for each cell

X.

(3)

standard deviation

E.

(i,j):

quantity

where:

(d .. )

(4)

~ ...:..2:1

p-l

Using

x- J.

(s)

where:

and

(s ) J. calculate

the "component

of variance"

between

laboratories,

(s6
n
The variance
material

of the total variability

including

both within

(5)

of a single test result on the jth

and between

laboratory

is given by:

(s ).J + (sL)' J
D-2

variability

(6)

Coefficient

of Variation

for Reproducibility

for the jth material:

(7)

0-3

U.S. DEPT. OF COMM.

BIBLIOGRAPHIC

1. PUBLICATION

DATA

NBSIR

SHEET
4. TITLE

OR REPOR1 NO.

79-1922

AND SUBTITLE

5. Publication

AN INTERLABORATORY
EVALUATION OF THE ASTM E 84-77a
TUNNEL TEST MODIFIED BY THE CONSUMER PRODUCT SAFETY
COMMISSION FOR CELLULOSIC LOOSE FILL INSULATION

Date

7. AUTHOR(S)

J. Randall

Lawson

9. PERFORMING ORGANIZATION

NAME AND ADDRESS

NATIONAL
BUREAU OF STANDARDS
DEPARTMENT
OF COMMERCE
WASHINGTON,

12.

Consumer

Textile

15. SUPPLEMENTARY

NAME AND COMPLETE

Product

Safety

and Mechanical

Bethesda,

No.

DC 2023.4

SPONSORING ORGANIZATION

U.S.

11. Contract/Grant

Maryland

ADDRESS (Street,

City, State.

ZIP)

13. Type of Report & Period Covered

Commission

Engineering

Final

Group

20202

NOTES

Document describes

a computer program; SF-185, FIPS' Software Summary, is attached.

16. ABSTRACT (A 20o-word Or less factual summary of mo.t significant

literature tJurvey, mention it here.)

information. If document includes a significant bibliography

or

An interlaboratory
evaluation was conducted to determine precision
estimates for repeatability
and reproducibility
of the American Society
for Testing and Materials
(ASTM) E 84-77a tunnel test as modified by the
Consumer Product Safety Commission for cellulosic loose fill insulation.
Six laboratories participated
in this study by running tunnel tests on
eight cellulosic loose fill materials.
Each laboratory was surveyed during
the projeet to examine its conformance with the critical details of the
test apparatus and procedure.
The results of the survey showed that none
of the tunnels completely conformed with the specifications
of the modified
ASTM E 84-77a standard.
The within-laboratory
coefficients of variation for repeatability
for
the six fire-retardant
treated cellulosic insulations ranged from 11 to 23
percent with an average of 16 percent.
The between-laboratory
coefficients
of variation for reproducibility
for the same materials ranged from 31 to 41
percent with an average of 35 percent.
There would be little assurance that
different laboratories would be able to distinguish between insulations which
do and which do not comply with the CPSC mandatory level (FSC < 25) on a
consistent basis.
17. KEY WORDS (alx to twelve entries; alphabetical
separated by aemicolona)

order; capitalize only the fir.t letter of the firat key word unlea. a proper name;

Cellulosic insulation; flame

test methods; tunnel test.

For Official

Distribution.

interlaboratory

evaluation;

precision;

IX] Unlimited

19. SECURITY CLASS

(THIS REPORT)

21. NO. OF
PRINTED PAGES

Do Not Release to NTIS

UNCLASSIFIED

34

18. AVAILABILITY

o
o

spread;

Order From Sup. of Doc., U.S. Govllfnment Printing

20402, SO Stock No. SN003.Q03Order From National Technical

Information

Office,

20. SECURITY CLASS

(THIS PAGE)

Washington, DC

22. Price

$4.00

Service (NTIS), Springfield,

UNCLASSIFIED

VA. 22161

USCOMM-DC

il

I"

'Idl

,'I Ii

'II