fire1944

Home / fire1944

 

Division of Forest Products – Reprint No. 88.

  

[Preprint
from the Journal of the Council for Scientific
and Industrial

Research,
Vol. 18, No. 1. February,
1945.]

  

A Survey of Houses Affected in
the Beaumaris Fire,

January 14, 1944.

  

By G. J. Barrow.*

  

Summary.

To determine the influence of
the type and details of construction on the resistance of houses to
external fire hazards, a survey was made of the damage caused by the
fires which swept the bayside resort of
Beaumaris, Victoria, on January 14, 1944. General observations on the 66 houses
destroyed or damaged and a detailed study of seventeen representative
houses showed that the resistance to fire is determined more bv the details of construction than by the
materials used in the walls. Although the damage was caused primarily
by the external fire, practically all the houses ignited inside, i.e.
in the roof space, in rooms, or under the floors, due to the ingress of
flame, sparks, and embers through openings such as ventilators, eaves,
and windows. Sealing or screening such openings by fine wire mesh
greatly reduced the risk of damage, and the conclusion was drawn that,
from the point of view of resistance to external fires, a house should
be as air-tight as practicable and that any openings which cannot be
eliminated should be screened. A list of recommendations, the adoption
of which should do much to reduce the fire risk, is included.
 

1. Introduction.

On January 14, 1944,
one of the worst conflagrations in the history of the metropolitan area
of Melbourne, swept through the bayside resort of Beaumaris. Beaumaris
is situated on the eastern coast of Port Phillip Bay,
12 miles (19 km) from Melbourne;
some twenty years ago it featured in a closer settlement scheme and an
electric tramway service was extended from Black Rock. The scheme was
abandoned and the tramway system, after operating for a period of five
years from 1926 to 1931, was discontinued and subsequently dismantled.
Since that date there have been no improvements made in the area,
which, although subdivided in detail has been allowed to return to its
natural state. The dwellings in the area are of two main types – 

(i) Small four- or five-room houses of varying
construction, built as seaside dwellings either at the time of the
settlement scheme or since; 
    because of the housing shortage most of them are now
permanently occupied. 

(ii) Large homes constructed either
by the early settlers or, in recent years, by residents who are not
dependent on public transport.

The fire destroyed 58 houses and damaged
eight others, and advantage was taken of the opportunity to obtain
general data on the fire hazard in semi-rural areas and to determine
the influence of the type and details of construction on the resistance
of houses to external fire hazards. 

__________________________________________________________________________________

* An officer of the Division of Forest
Products.

C.12539/44. 

Approximately 100 houses were
inspected during the week following the fire, and from these seventeen
were selected as being representative of the general damage and
conditions. A detailed study of these particular houses was then
made. 

The number of persons interviewed
who were actively involved in the fire (either
owners
, occupiers, or service personnel) was approximately 200,
and from these eyewitnesses data were collected on matters relating to
houses destroyed, damaged, or saved. 

The gathering of the data for the
report was greatly facilitated by the willing co-operation of the
following: – Police officers at Mentone, Black Rock and Cheltenham; the
Town Clerk of the Sandringham City
Council; the superintendent, staff, and linesmen of the
Cheltenham District Depot of the State Electricity Commission; the
district officer of the Melbourne and Metropolitan Fire Brigade;
auxiliary and volunteer firefighters; owners, occupiers, and agents of
the damaged properties. 
 

2. Location of the Fire.

The fire occurred
within an area of some 1,500 acres (600 hectares), bounded on
the north by Bay Road, Cheltenham,.the
west by Reserve-road, the south by the sea, and on- the east by Charman-road; Mentone and Cheltenham, as shown
in Map 1. The area, north of Weatherall-road
is included in Map 1 because, although there is little damage in this
area, it is believed that the main fires through the Beaumaris area originated there. Of the
1,500 acres (600 hectares) included in the map some 700 (280
hectares)
were burnt. 

The main fire area south of
Balcombe-road was mainly dense scrub
land covered by a heavv growth of
tea-tree (Leptospermum lævigatum
and L. coriaceum mixed) with
scattered groups of manna gum (Eucalyptus viminalis)
and coast acacia (Acacia sophoræ),
and with groups of white sallow acacia (Acacia floribunda)
further back from the coast; throughout this area were scattered
clearings of from 1 to 5 acres (0.4 to 2 hectares) densely covered with
dry grass and bracken (Pteridium aquilinum) often to a height of 3 feet (900
mm)
; in the lower swamp areas were several acres of swamp tea-tree (Melaleuca squarrosa).
The houses destroyed and damaged extended throughout this
area but were mainly situated within three-quarters of a mile (1.2
km)
from the sea. 

In attempting to retain the
locality in its natural state, the majority of the residents had
allowed their properties to become thickly wooded with native trees and
shrubs, augmenting the natural growth in many cases with exotic
vegetation, thus forming a dense thicket to within 40 to 20 feet (12
to 6 m)
of their dwellings, and in many cases in contact with
them. 

Scrub fires in this area have been
a common occurrence for many years, but are usually localized and are
fairly easily checked before constituting a menace to the populated
sections of the district. The origins and progress of the fire, as
observed by officers of the State Electricity Commission, are shown in
Map 1. It will be noted that there were two distinct fires: the first
swept over the eastern portion of the burnt area, and the second, which occurred half an hour to an hour later,
burnt the western portion, overlapping to some extent the area covered
by the first fire. Both fires burnt down to the water’s edge. 

As will be seen from Map 2, the fire burnt over both scrub land and grass
land, but was very much fiercer and presented a much greater hazard to
houses in the scrub land. In the grassed areas, it was comparatively
easily controlled by breaks such as roads and cultivated areas and by
spraying with garden hoses. 

The fares were fought by personnel
of the Metropolitan Fire Brigade, State Electricity Commission, Police
Department, Civil Defence Organizations,
Army, and local residents and volunteers with fire hoses, garden hoses,
and, where the water supply failed, with beaters. As a result of their
efforts, a number of buildings which were apparently doomed were either
saved or only slightly damaged. 

About 208 houses in the area were
in some danger, but 90 of these, located in cleared areas or in grass
lands, even though in some cases completely surrounded by fire, were
not in immediate danger, as the fire was controllable and was deflected
away from the houses by the fire fighters. The remaining 118 houses
were gravely threatened because of actual contact with flames or
burning debris. 

  

3. Conditions Leading up to the
Fire.

Meteorological data for the day of
the fire, together with the weather conditions for the preceding six
months, as supplied by the Commonwealth Meteorological Services,
Department of Air, Melbourne,
are given in the Appendix

Conditions prior to and during the
fire were ideal for a major conflagration. The six months preceding the
fire were abnormally dry and several perennial species of small flora
(particularly sundews and orchids), for which this and surrounding
districts are noted, bad failed to appear. The day of the fire was hot
with a strong dry northerly wind, the maximum shade temperature being
103.2°F. (39.6°C) and the maximum wind velocity being
54 m.p.h. (86 km/h), the relative humidity falling to 6 per
cent., corresponding to an equilibrium moisture content in wood of
about 2 per cent. 

  

4. Extent of Damage and
Location of Buildings.

The location of the houses in the
affected area, together with information on the type of construction
and the extent of the damage (if any), is given in Map
2

For convenience in locating the
seventeen houses selected for detailed study, Map 2
has been divided into sections A, B, C. Sections A and B have each been
subdivided into three subsections. 

The extent of the damage in
relation to the type of and number of houses seriously threatened by
the fire is given in Table 1. It may appear from this table that the
brick and concrete houses were less liable to danger than timber frame
houses with fibro-cement or ash-cement rendered walls. However,
statistical investigation has shown that there are no significant
differences in the percentages of different types of houses destroyed
or damaged. 

 

Table 1. – Number of Affected Houses in Relation
to Type of Construction and Number Seriously Threatened.

 

Construction

Number

Seriously

Threatened

Number Affected.

Percentage Affected.

Destroyed.

Damaged.

Total.

Destroyed.

Damaged.

Total.

W: W/B

83

41

5

46

50

5

55*

CS/W: CS: ACL/W

15

8

2

10

53

13

66*

B: V: C

20

9

1

10

45

5

50*

Total

118

58

8

66

 

 

 

                     

                   
Notes.- 1. An additional 57 properties
sustained damage to outhouses and fences only. 

2. Statistical
examination has shown that there is no significant difference between
the percentages marked thus*

LEGEND.- 

W – weatherboard

W/B – weatherboard with brick
foundation 

CS/W – timber frame with
weatherboards to 3-4 feet (0.9-1.2 m) from ground and
fibro-cement sheets above. 

CS – fibro-cement sheet over
timber framework. 

ACL/W – timber frame with
weatherboards to 3-4 feet (0.9-1.2 m) from ground. 

B – brick
and brick stucco covered. 

V – brick veneer 

C – concrete block or slab

It must also be noted that the brick, brick
veneer, and concrete houses were in general located somewhat more favourably than the others, in that they were
generally surrounded by well-kept gardens, which tended to lessen the
severity of the attack. Brick houses which were completely gutted,
leaving the walls only standing, have been considered as being
destroyed in compiling the Table (see Plate 1, Fig. 1). Further details
of the extent of the damage in relation to the type of construction and
location of the houses are given in Table 2. 

 

5. Observations. 

Points of particular interest are illustrated by the
photographs which, with the accompanying captions, are
self-explanatory. 

Detailed records of the construction, surroundings, point of
ignition and extent of the damage to the seventeen representative
houses mentioned previously have been filed and are available to any
one interested, but the following observations should be of general
interest. 

The use of ½-in. (12 mm) mesh wire netting as a
guard against birds, rodents, and possums proved to be of great value
in preventing burning debris from being carried by draughts through
wall ventilators, louvre openings, and
under floor air-vents (see Plate 5, Fig. 1).  

 

Table 2.––House construction and extent of damage
in each section.

  

Section.

Construction.

Destroyed.

Damaged.

Outhouses,

Fences

Damaged.

Undamaged.

A1

Scrub and fence damage only

.   .

.   .

.   .

.   .

A2

W

Corrugated iron roof

6

1

8

1

W

Tile roof

2

.   .

3

.   .

CS/W

Corrugated iron roof

2

.   .

1

.   .

CS

Fibro-cement roof

1

.   .

1

.   .

B

Corrugated iron roof

.   .

.   .

.   .

1

A3

W

Corrugated iron roof

12

.   .

6

7

W/B

Corrugated iron roof

1

.   .

.   .

.   .

W

Tile roof

4

1

.   .

4

CS/W

Corrugated iron roof

.   .

.   .

.   .

3

B:V

Tile roof

2

1

1

2

C

Corrugated iron roof

1

.   .

1

.   .

B1

W

Corrugated iron roof

2

1

1

1

W

Tile roof

.   .

.   .

.   .

1

ACL/W

Corrugated iron roof

.   .

.   .

.   .

1

B2

W

Corrugated iron roof

4

.   .

8

5

W

Tile roof

1

.   .

3

3

CS/W

Corrugated iron roof

.   .

.   .

1

3

ACL/W

Corrugated iron roof

1

.   .

.   .

1

B

Corrugated iron roof

.   .

.   .

.   .

1

B

Tile roof

.   .

.   .

2

5

B3

W

Corrugated iron roof

3

1

8

3

W

Zinc-anneal roof

.   .

.   .

.   .

1

W

Tile roof

3

.   .

3

4

WCS

Tile roof

1

.   .

1

.   .

ACL/W

Tile roof

1

.   .

.   .

.   .

B

Zinc-anneal roof

.   .

.   .

1

.   .

B

Tile roof

5

.   .

4

7

V

Tile roof

.   .

.   .

4

.   .

C1

W

Corrugated iron roof

2

1

5

20

W

Tile roof

1

.   .

4

14

W/B

Tile roof

1

.   .

.   .

.   .

CS/W

Corrugated iron roof

1

1

.   .

4

CS

Fibro-cement roof

1

.   .

1

1

CS

Corrugated iron roof

.   .

1

.   .

2

CS

Tile roof

.   .

.   .

1

.   .

B

Tile roof

.   .

.   .

2

9

B

Corrugated iron roof

.   .

.   .

1

7

             

            The
fire burnt the fences of fourteen houses but was stopped before it
could constitute a menace to the houses, which were situated
either 
            in
cleared areas or in grassland. 

Flywire proved in a number of cases to be an
excellent spark arrester, House B3-6 being a case in point. This
dwelling had large openings across the front covered by fly-wire;
immediately behind the wire were large canvas blinds which were down
during the whole period of the surrounding fire. A. general profusion
of sparks and burning material was swept against this portion of the
house, but an examination of the framing ledge between the fly-wire and
the blinds showed it to be free of any charred material or ash. 

Both hessian
and canvas showed little resistance to sparks, hessian
bags lying in yards and canvas blinds up to a distance of a mile (1.6
km)
from the fires being burnt. 

Wall ventilators and air vents
below floor level, where unprotected by some form of screening. proved disastrous in a number of cases. In some
cases where the fire appeared to be controlled a sudden draught swept
sparks up to the dwelling and in through these vents. This point was
well illustrated in the fighting of the fire at House B2-1. While a
fire was being fought in the eaves and ceiling at the rear of the
house, the draught through the underfloor
vents near, the ground at the back of the house swept burning material
under the house to such an extent that a fire under the floor became
uncontrollable. A similar type of attack occurred at Houses B3-9, C1-1,
and C1-3. 

Houses with eaves either boxed with
fibro-cement sheets or completely boarded had a greater degree of
safety than those with eaves left open for ventilation. It was reported
that the majority of fires started in the roofs. 

Fibro-cement sheets were found to
be fire retardant but not fire proof; when subjected to intense heat
they cracked, flaked, and collapsed. Where water could be poured or
sprayed to keep them cool they proved a most effective check to the
spread of fire. This point is shown in Plate 2, Fig. 2, the
fibro-cement sheets are still intact above the steel door of the
garage. The fire within the garage was so fierce that the front of the
garage was sprayed with a fire hose for more than fifteen minutes
before the fire behind the door burnt out. 

Several fires became uncontrollable
because shingles on the gable ends of the buildings ignited and the
draught swept the flames into the roof through large roof ventilators.
It would be preferable to replace these by a number of small
ventilators, as shown in Plate 6, Fig. 1. 

Badly fitting Marseilles
pattern tile roofs were a source of danger; houses which either had a
reasonably good chance of escaping the fire or were considered to be
saved as the fire had passed, were either damaged or destroyed through
the access of sparks and small burning debris under the tiles. 

There were several cases of roof
fires in houses roofed with corrugated iron; generally these fires
entered through large roof ventilators in gable ends or ignited the
protruding eaves and burnt through to the rafters and ceiling joists.
The position of the guttering prevented the updraught
from carrying the sparks up under the corrugations into the roof, however it is a possibility that in some
cases an accumulation of burning debris, lodged in the guttering, may
have been swept under corrugations by an eddy of wind.

Compared with that of sealing a tile roof
with its multiple openings there is little difficulty in sealing a
corrugated iron or corrugated fibro-cement roof. This can be done by
scalloping the fascia board to fit the corrugations, by nailing a
scalloped sheet of iron to the fascia, or filling the spaces under the
corrugations with fibro-cement or mortar. 

Throughout the fire several houses were seen to ignite
although there was no apparent contact with the flame, sparks, or
burning debris; this ignition took place both in the roofs and in the
walls. 

In a number of cases the proximity of high trees, preceded by
very low scrub or grass, proved an effective fire break. This applied
only in cases where branches of the trees were not located under the
eaves. Due to the hurdling effect of the high trees the draught caused
by the fires swept the burning material over the house. In the majority
of cases the trees either failed to ignite or burnt slowly and so
proved controllable (Fig. 1 and Plate 6, Fig. 2). It must be emphasized
that high trees gave protection only when surrounded by
grasslands or by low scattered scrub. 

In the case of unattended houses (where fire fighting was
impossible) the normal time from ignition to destruction was seven to
fifteen minutes for brick or brick veneer and ten to 25 minutes for
timber frame houses. 

Evidence obtained during the survey indicates that there is no
possibility of predicting the behaviour of
a scrub fire. In a number of cases, scrub and trees which
appeared to be burnt out, re-ignited, accompanied by a crackling and
hissing noise suggestive of gas igniting; these flashes of fire would
then jump distances up to several hundred feet hundred feet (100 or
more metres)
igniting any scrub
through which they passed. The rush of flame and the intense heat of
the burning tea-tree caused many unexpected draughts and caused runs of
fire to burn back against the wind for several hundred yards (metres). 

This process of the fire chopping about in several directions
was instrumental in saving some houses by first burning breaks either
near or partially around them. This occurred in the case of House B3-1
where a fire burnt off the scrub to the front, the east and a section
of the trees to the south (rear) of the house, thus leaving only one
side and a corner to be protected when the second fire came through
from the north-west. Inspection of this dwelling after the fire gives
the impression that it was completely surrounded by fire. This was
quite correct, but the fire was not burning on all sides at the same
time. 

 

6. Conclusions.

The survey showed that, in a fire of the type that swept Beaumaris, the chances of a house surviving are
determined more by the nature of the surroundings and the details of
construction than by the materials used in the walls. With two
exceptions, all the really destructive fires started inside the houses,
i.e., in the roof space, in rooms, or under the floors, the immediate
cause of ignition in such cases being the entrance of flame, sparks,
and burning-debris through openings such as ventilators, eaves, and
windows. An air-tight house would appear to have much greater
resistance to fire than one in which free circulation of air is
encouraged. Of course, an air-tight house, even if it could be built,
would be most undesirable for other reasons, but it is necessary in
areas such as Beaumaris, where a serious
fire hazard exists, to make a compromise between fire resistance and
ventilation, by completely boxing the eaves, sealing roofs,
&c. 

In the two exceptions referred to previously, the houses
appeared to ignite externally due to the high temperature of the
surrounding scrub fires. However, these observations may not be
correct, as under conditions such as existed at the time, observers’
reports are often unreliable, and it is quite possible that these
houses also caught alight because of the entrance of flame and
sparks. 

Experience at Beaumaris showed
that the ill effects of openings can be considerably mitigated by
covering them with ½-in. mesh (12 mm) wire netting or
preferably with fly-wire. The wire netting will prevent the entrance of
burning debris, and fly-wire is effective in stopping sparks These
coverings, in a number of cases, saved houses from destruction.
Similarly, fly-wire window-screens and doors are of great assistance in
preventing the entrance of sparks in the event of windows or doors
being accidentally left open during the fire. 

Badly fitting Marseilles
pattern tile roofs appear to be a menace as they provide innumerable
openings through which sparks may enter. Some form of sheet roofing,
e.g. corrugated iron, fibro-cement, or close fitting composition
roofing, appears to be much more satisfactory. Corrugated roofing
should be sealed at the eaves, ridge, hips, and valley by some such
composition as mortar or fibro-cement. 

These simple precautions greatly increase the resistance of a
house to fire, with little, if any increase in cost, and they are
likely to be far more effective than such methods of protection as the
use of non-combustible materials for walls. The impression gained from
a glance at a completely gutted brick house in which the walls are
still standing (see Plate 1, Figs. 1 and 2), is that the damage is much
less serious han in the case of a timber
frame house which is usually burnt to the ground. However a little
consideration will show that the cost of the walls of a timber frame
house is a comparatively small proportion of the total cost, and the
cost of reconditioning a burnt-out brick house is likely to exceed the
cost of rebuilding a timber frame house. 

  

7. Recommendations.

The adoption of the following recommendations in areas where
there is a fire hazard, would do much to reduce the risk: – 

1.. In timber houses, the walls
below floor level should be close boarded, ventilation being provided
by woven wire vents. 

2. All vents should be of the woven wire type or else covered
by a fine mesh. 

3. Large ventilators in gable ends should be eliminated and
replaced by a number of scattered small ventilators with fine mesh
openings. 

4. Eaves should preferably be completely boxed, but if left
open should be covered by fine mesh wire netting. 

5. Badly fitting Marseilles
pattern tiles are a source of danger. 

6. The space under the corrugations of corrugated roofing
should be closed at the eaves, ridges, hips, and valleys. 

7. Fly-wire window-screi6na and doors are beneficial. 

8. Trees and shrubs should be kept clear of the walls. 

9. Stacks of fuel should be well clear of the walls or stored
in properly constructed sheds. 

 

APPENDIX

Weather conditions in Melbourne

(Compiled from data supplied by the
Commonwealth Meteorological Services, Department of Air,
Melbourne.)

Note: The S.I. Units of measurement shown
were derived from the Imperial values in the original version of the
Report.

 

(1) CONDITIONS FOR SIX MONTHS ENDING JANUARY
31, 1944

 

Month.

Mean Temperature (°C)

Rainfall (mm)

Temperature.

Normal.

Rainfall.

Normal.

August

8.8

10.6

24.1

48.0

September

11.7

12.3

50.5

57.9

October

13.8

14.3

21.3

66.5

November

15.7

16.3

79.2

57.7

December

17.6

18.3

17.3

58.2

January

20.3

19.7

18.8

49.0

Total

 

 

211.2

337.3

  

 

(2) CONDITIONS FROM JANUARY 1 TO JANUARY
14, 1944
 

 

Date, 1944

Maximum 

Temperature 

(°C)

Maximum 

Wind 

Velocity 

(km/h)

Wind

Direction

(Prevailing)

Rainfall

(mm)

January 1

38.1

40

NNE

nil

2

26.6

37

WSW

1.3

3

23.6

37

NNW

nil

4

20.4

42

SSW

0.5

5

20.7

37

WSW

4.1

6

22.7

23

SSW

0.3

7

28.9

26

SSE

nil

8

38.9

41

NNE

nil

9

23.6

29

S

nil

10

26.9

40

NSW

nil

11

21.1

26

W

nil

12

21.0

29

ESE

nil

13

33.1

18

WNW

nil

14

39.6

54

N

nil

 

 

(3) CONDITIONS ON JANUARY
14, 1944

 

Times.

Condition.

Local Mean.

Australian

Eastern

Daylight

Saving.

Wind

Velocity

(Average).

[km/h]

Wind

Direction.

Temperature

[°C]

Relative

Humidity.

[%]

Equilibrium

Moisture

Content.

[%]

7

8:20

35

N

26.9

25

5.0

8

9:20

35

NNE

29.5

24

5.0

9

10:20

42

N

31.2

17

3.5

10

11:20

45

N

32.8

13

3.0

11

12:20

42

N

34.2

12

3.0

12

1:20

39

N

36.2

10

2.5

13

2:20

35

N

37.6

7

2.5

14

3:20

37

NNW

38.4

9

2.5

15

4:20

32

N

39.6

9

2.5

16

5:20

26

N

38.4

8

2.5

17

6:20

26

NNW

38.4

6

2.0

 

(a)  Maximum
temperature at Melbourne
(19 km NNW of Beaumaris) on January 14, 1944, was
39.6°C occurring at 1455 L.M.T. 
(b)  The maximum gust of wind at Melbourne
on January 14, 1944,
87 km/h occurring at 1007 L.M.T. 
(c)  The following is a description of the weather at Melbourne
on January 14, 1944:
– 

“Clear at first,
overcast with cirrus cloud at 0740 L.M.T. Almost cloudless at 1040.
Dust and smoke haze greatly reduced visibility.  
At 1640 L.M.T. dense dust haze all around. At 1940 L.M.T. there was a
dense dust haze. Visibility very poor.” 

 

_________________________

H.
E. Daw, Government Printer, Melbourne

C12539/44
– 2 

A copy of this report with the original
colour version of the map below, as a
1.28 MB PDF file, can be emailed to you free by emailing ggd@netspace.net.au with a
request. That map file has better resolution than the 140 KB monochrome
version below, which is only used because of Web site limitations.
Alternatively you can receive it by ordering the BCS Inc. Archive Files
CD-ROM at a cost of $5, postage included. That disc also contains other
BCS Inc. archive material.

 

Order by sending
your name, address and cheque to the
Webmaster, BCS Inc,
PO Box
7016
, BEAUMARIS VIC 3193.

 

 

www.vicnet.net.au/~bcsinc