Content uploaded by Lawrence James Cookson

Author content

All content in this area was uploaded by Lawrence James Cookson on Oct 30, 2018

Content may be subject to copyright.

IRG/WP 03-10468

THE INTERNATIONAL RESEARCH GROUP ON WOOD PRESERVATION

Section 1 Biology

A RISK MODEL FOR TERMITE ATTACK IN AUSTRALIA

R.H. Leicester and C-H. Wang

CSIRO Building, Construction and Engineering

PO Box 56, Highett, Victoria 3190, Australia

e-mail: Bob.Leicester@csiro.au

L. J. Cookson

CSIRO Forestry and Forest Products

Bag 10, Clayton South, Victoria 3168, Australia

Paper prepared for the 34th Annual Meeting

Brisbane, Australia

18-23 May, 2003

IRG Secretariat

SE-100 44 Stockholm,

Sweden

A RISK MODEL FOR TERMITE ATTACK IN AUSTRALIA

R.H. Leicester and C-H. Wang

CSIRO Building, Construction and Engineering

PO Box 56, Highett, Victoria 3190, Australia

e-mail: Bob.Leicester@csiro.au

L. J. Cookson

CSIRO Forestry and Forest Products

Bag 10, Clayton South, Victoria 3168, Australia

ABSTRACT

This paper describes a model to predict the risk of termite attack on a house in Australia. It is based on

a survey of expert opinion and data from 5000 houses. The model gives a quantitative estimate of risk,

and as such is useful for the development of risk management systems. An example of the application

of such a system is given.

Keywords termite, risk, hazard, probability, timber.

1. INTRODUCTION

For engineering purposes, it is useful for termite attack to be considered to be a probabilistic event.

This paper describes the development of a model to predict the risk of attack on a house in Australia.

Such a model is useful for assessing (in a quantified manner) the value of various protection strategy

proposals.

First a preliminary model was developed on the basis of opinions elicited from experts on termite

behaviour such as entomologists and pest controllers. These opinions were obtained through the use of

a structured questionnaire. The model was then calibrated through the use of data on 5000 houses.

This data, denoted the “Termite Tally”, was initiated by J. French and reported in detail by Cookson

and Trajstman (2002). It was based on an Australia wide survey undertaken by school children.

The model to be described takes into consideration a great number of factors, such as for example the

age of a house and its surrounding suburbs, the inspection program, the use of physical and chemical

termite barriers, the type of house construction, details of the local environment and climate. Further

details of many aspects of this model can be found in previous papers (Leicester and Wang 2001,

Leicester et al. 2001, Leicester et al. 2002).

2. PROBABILITY MODEL

The probability density function of the time for a house to be attacked by termites is assumed to be of

the type shown in Figure 1. The form of this function was chosen to fit the data found in the Termite

Tally. The equation for the density function is assumed to be

btap

+

= (1)

where a and b are the distribution parameters, and t is the time since time zero, the time at which the

house was constructed. The value of a may be either positive or negative, as shown in Figure 1. The

notation tmax will be used to denote the upper limit of the density function, evaluated from the

assumption that the area under the density graph must be unity, and ta denotes the lower end of the

density distribution. The mean time of attack is given by

2

()(

22 3

max a max a

mean( ) 23

ab

ttt tt

)

3

=

−+ − (2)

An interesting aspect of the risk model is to distinguish between a “true” risk and an “apparent” risk.

The apparent risk is the risk estimate based on the historical memory of the house occupant. This is

obviously less than the true risk as the householder is not aware of all past termite attacks. In the

Termite Tally the average time of occupancy of the householders interviewed was found to be about

11 years. In this risk model the historical memory of the average occupant, denoted by tmem, has been

taken to be 20 years, hence the apparent risk is the probability that the houses have been attacked

during the previous 20 years. Use of the concept of an apparent risk is necessary in interpreting the

data on houses obtained from an occupant survey.

t

max t

a

p

(a) for positive a

p = a + bt

ta = 0

age of house (yrs)

t

max t

p

(b) for negative a

p = a + bt

ta

age of house (yrs)

Figure 1. Probability density functions of the time of a termite attack.

Figure 2 shows plots of the true and apparent risk that a house of age t years will have been attacked.

For the case ta < t < tmax, the true probability that a house has been attacked, denoted by Ptrue, is given

by

()

(

22

true a a

2

b

Patt tt=−+ −

)

(3)

For the case (ta + tmem) < t < tmax, the apparent probability that a house has been attacked, denoted by

Papparent, is given by

(

2

apparent mem mem mem

2

b

Pat tbt

=− +

)

t

(4)

It should be noted that there is a kink in the graph at t = tmem for the apparent rate of attack. This is

because for t < tmem the apparent or observed risk will in fact be the true risk. It is only for t > tmem that

these two risks differ. In addition it should be noted that according to the mathematical model, for the

case t > tmax all houses will have been attacked and the graph for apparent attack should become

horizontal. However, if it is assumed that following remedial action houses may be attacked again,

then to an observer with limited memory there will be some sort of ongoing attack rate. The model for

a second attack has not been studied, but the data from the Termite Tally is in line with the assumption

that the apparent attack rate stays constant for the case t > tmax. This is indicated by the dotted line in

Figure 2.

It is to be noted that two parameters, a and b, are required to define the model. The obvious method

for evaluating these parameters is to fit the model to give the correct mean value and coefficient of

variation of termite attack. However it was found that for practical purposes the coefficient of

3

variation was effectively constant, corresponding to a value of b = 0.0002, and so the model is

effectively defined in terms of a choice of mean value. This mean value is estimated from the

numerous parameters that affect the speed of termite attack, such as for example the type of

construction and location of a house.

t

mem

t

max

t

P

true risk of attack

apparent risk of attack

0

1.0

age of house (yrs)

Probability that attack has occurred

(a) for positive a

t

a+

t

mem

t

max

t

P

true risk of attack

apparent risk of attack

0

1.0

t

a

age of house (yrs)

(b) for negative a

Probability that attack has occurred

Figure 2. Schematic illustration of the cumulative distribution functions of the attack time.

3. ZONATION FOR TERMITE HAZARD

The Termite Tally was used to estimate the termite hazard within a region. This was done by assuming

the hazard to be proportional to the number of gardens that were found to contain signs of termites

attacking wood (e.g. landscape timbers, trees, fences, firewood stacks). By grouping the data

according to mean annual temperature (Leicester et al. 2002) and ago-ecological regions (Cookson

and Trajstman, 2002) two sets of hazard maps have been derived as shown in Figure 3.

N

o Risk

T=25

o

C

T=18

o

C

Zone 1

Zone 2

Zone 4

Zone 3

Negligible

Zone 2

Zone 1

Zone 3

(a) temperature zonation (b) agro-ecological zonation

large sample cluster

small sample cluster

Figure 3. Termite hazard zonation.

Most of the data obtained was from houses clustered around city and town centres near the edges of

Australia as shown in Figure 4, and so it is not possible to choose between the two types of zonation

used solely on the basis of statistical information. Based on subjective estimates of the termite hazard,

it was decided to combine information from the two computed zones and use the zonation shown in

Figure 5 as a first choice for the termite hazard in Australia.

4

110 120 130 140 150

-40

-35

-30

-25

-20

-15

-10

Figure 4 . Location of housing clusters.

(c) Recommended termite hazard zone for Australia

Figure 5. Termite hazard zonation.

Probability that termites are observed in the garden

12 16 20 24 28

Temperature (°C)

0.1

R2 = 0.505

0.2

0.3

0.4

0.5

0.6

0.7

Figure 6 . Effect of temperature on the apparent incidence of termites in the garden.

5

An attempt was also made to assess the effects of rainfall and temperature on the termite hazard. This

was done by first separating the data obtained in the Termite Tally into clusters of data from houses

subjected to the same annual rainfall and temperature. This data was then combined with the estimated

termite hazard to derive the relationships shown in Figures 6 and 7. It is apparent that while there is

some effect of temperature on the hazard level, there is no observable effect due to rainfall.

400 600 800 1000 1200

Mean annual rainfall (mm)

Probability that termites are observed in the garden

0.15

0.25

0.35

0.45

T = 18 - 20 C

T = 16 - 18 C

regression

Figure 7 . Effect of rainfall on the apparent incidence of termites in the garden.

4. THE EXPERT OPINION MODEL

4.1 The Base Model

The base model has been derived on the basis of expert opinion. It applies to a house surrounded by

50 m of termite-free land. The distance of 50 m was chosen because this is about the limit of the

foraging distance of most termite species. The model used endeavours to estimate four sequential

event times t1 – t4 as defined in Table 1 and illustrated in Figure 8.

Relevant data on these four event times were obtained via a limited survey of expert opinion. Details

of this survey, together with the analysis procedure used to process the data has been described

elsewhere (Leicester and Wang 2001, Leicester et al. 2001).

Stage 4

Destruction

Stage 3 Stage 2

Stage 1

N

est

Figure 8. Illustration of termite progress.

In the survey, a set of parameters affecting each event time was obtained from experts. The set chosen

is listed as P1, P2, …, P14, as tabulated in Table 1. For each parameter, the experts were asked to list

the importance of the parameter with regard to its influence on the relevant event time; this importance

6

was rated on a scale of 1−10, with 10 being the most important; examples of the importance

parameters chosen are also given in Table 1.

Table 1. List of event times and associated parameters

Event time Influencing parameter Importance

factor

t1

the time taken for the establishment

of a mature colony within a distance

of 50 m from the target house

P1: geographical location

P2: age of surrounding suburbs

P3: number of potential nest sites

8

5

9

t

2

the time taken for the termite

foraging galleries to progress to a

house 20 m away from the nest site

P4: geographical location

P5: soil condition

P6: food source

8

6

7

t3

the time taken for termites to

penetrate or bypass a chemical or

mechanical barrier, if any

P7: geographical location

P8: period between inspections

P9: maintenance parameter

4

10

7

t4

the time taken (after penetrating the

barrier) to reach and cause failure of

a timber member

P10: geographical location

P11: ground-contact building element

P12: period between inspections

P13: type of material attacked

P14: timber environment

8

5

9

7

7

For each parameter Pj, there is an associated parameter factor kj. This factor kj is given the value of +1,

0 or −1 depending on whether the parameter has been chosen to correspond to low, medium or high

hazard situations respectively (Leicester et al. 2001).

The mean values of these times are given by

mean (t1) = 16.7(1 + 0.236 k1) (1 + 0.148 k2) (1 + 0.266 k3)

mean (t2) = 4.9 (1 + 0.0771 k4) (1 + 0.0578 k5) (1 + 0.0675 k6)

mean (t3A) = 20.9 (1 + 0. 454k7) (1 + 0.649 k8)

mean (t3B) = 27.4 (1 + 0. 497k7) (1 + 0.710 k8) (5)

mean (t3C) = 37.9 (1 + 0.222 k7) (1 + 0.556 k8) (1 + 0.389 k9)

mean (t3D) = 24.2 (1 + 0.205 k7) (1 + 0.513 k8) (1 + 0.359 k9)

mean (t3E) = 0.0

mean (t4) = 14.5 (1 + 0.455 k10) (1 + 0.284 k11) (1 + 0.512 k12) (1 + 0.398 k13) (1 + 0.398 k14)

The five values of t3 refer to as follows:

t3A: granite-guard,

t3B: termimesh,

t3C: toxicant chemical,

t3D: repellant chemical

t3E: no barrier.

4.2 The Practical Model

For practical application, the base model must be modified so that the target house is closer than 50 m

to the adjoining suburbs. In addition, the model must allow for the possibility that there may be mature

nests nearby at time zero, the year in which the house is constructed. This configuration is illustrated

in Figure 9. A rough approximation to the estimate of the attack time by an expert, denoted by texpert, is

()

()()

expert garden 2 garden suburb 2 suburb 1 2 3 4

10

11

20

d

tPt P P t P ttt

+

=+− +− ++

t

+

(6)

7

where Pgarden is the probability that a mature nest exists in the garden at time zero, and Psuburb, is the

probability that a mature nest exists in the suburb at time zero.

A suitable equation for estimating Pgarden and Psuburb is

suburb suburb

garden suburb

suburb

100, 100 years;

1 100 years.

tt

PP t

≤

==

>

(7)

where tsuburb denotes the age of the suburb at time zero, the year in which the target house was built.

b

uildings of older

built-up suburbs

possible nest site

existing at time

zero

target house

possible nest site

existing at time zero

d

d = minimum

distance from

older built up

suburbs

Figure 9. House and land surrounded by existing buildings and nest sites at time zero.

5. MODEL CALIBRATION

As discussed in a previous paper (Leicester et al. 2002), data from the Termite Tally indicates that for

the probability distribution function of attack times, suitable calibration choices are for a parameter b =

0.0002, and the mean time to attack mean(t) = 1.5 mean(texpert), where mean(texpert) denotes the meant

time estimate based on expert opinion. Figure 4 shows how predictions on apparent risk, based on

these assumptions, match the Termite Tally data for average hazard zone conditions. Taking into

account the scatter of the Termite Tally data, the model appears to give as good a fit as can be

expected for average conditions. Reasons for the kink and the dotted extension of the predicted graph

of apparent attack have been explained in section 2.

0

0.2

0.4

0.6

0 20406080100

Apparent probability that

house has been attacked

Age of house (years)

Model

mean(t) = 44 yrs

Termite Tally:

Temperature (zone 2)

Agro-ecological (zone 2

& 3)

Temperature zonation (zone 2)

Agro-ecological zonation

(zone 2 and 3)

Figure 10. Comparison of apparent risks derived from the model and the Termite Tally.

8

6. APPLICATION OF THE PROPOSED MODEL

An example of an application of the termite attack model is illustrated in Tables 2 and 3. Here the aim

is to specify the minimum requirements for inspection and termite barriers so that for all new houses

the risk of attack is less than 20 per cent in the next 50 years, a risk less than that which currently

exists for the average house in Australia. The procedure is to evaluate a total hazard score according to

Table 2, and then to use Table 3 to derive the level of inspection and termite barrier required. The

hazard scores in Table 2 have been chosen so as to have a value of zero for the low hazard situation.

The range of the scores have been chosen essentially by trial and error to produce the correct target

risk, using the importance values given in Table 1 as a starting point.

Table 2. Hazard scores for termite attack

Location Zone(1) Hazard score

B 0

C 2

D 4

(1) This risk for zone A is considered to be negligble.

Age of suburb(2) Hazard score

<10 yrs 0

10-70 yrs 2

>70 yrs 4

(2)A suburb refers to an area in which atleast 20% of the land

is covered by buildings.

Distance to nearest boundary Hazard score

>8 m 0

2—8 m 0.5

<2 m 1.0

Quantity of wood in garden and

under house(3)

Hazard score

low 0

medium 0.5

high 1.0

(3) See Appendix A1.

Hazard related to ground contact(4) Hazard score

low 0

medium 1

high 2

(4) See Appendix A2.

Hazard related to type of

construction materials(5)

Hazard score

low 0

medium 1

high 2

9

(5) See Appendix A3.

Hazard related to exposure of

material(6)

Hazard score

low 0

medium 1

high 2

(6) See Appendix A4.

Evaluation of hazard score total

Hazard factor Hazard score

Location zone

Age of suburb

Distance to boundary

Wood in garden

Ground contact

Construction material

Timber exposure

Hazard Score Total:

10

Table 3. Specification of termite management requirements

Barrier type(1)

Period between

inspections

(yrs)(2)

Period between

treatments

(yrs)(3)

Maximum

acceptable

hazard score

total

Graded crushed

stone

<1

1–5

>5

–

–

–

9.5

7.5

3.5

Stainless steel

mesh

<1

1–5

>5

–

–

–

10.0

8.0

4.0

<1

Tm

2Tm

>8Tm

no limit

no limit

10.5

1–5

Tm

2Tm

>8Tm

13.5

10.5

7.5

Toxic chemical

>5

Tm

2Tm

>8Tm

6.5

5.0

4.0

<1

Tm

2Tm

>8Tm

14.0

11.0

8.5

1–5

Tm

2Tm

>8Tm

9.5

8.0

6.5

Repellant chemical

>5

Tm

2Tm

>8Tm

5.0

4.0

3.5

No barrier(4) <1

1–5

>5

–

–

–

5.5

4.0

2.5

Notes:

(1) For barriers placed and maintained according to AS 3660.1

(2) For inspections carried out in according to AS 3660.2

(3) Tm denotes the period between retreatments as recommended by the chemical manufacturer

(4) The term ‘no barrier’ denotes the absence of a house perimeter barrier, such as that provided by graded

crushed stone, stainless mesh or chemicals

6. CONCLUDING COMMENTS

Figure 11 shows the risk for the three cases of hazards corresponding to all the hazard parameters

having the values of kj = 1, kj = 0 and kj = –1. The envelope represents the extremes of risk predictions

by the model. For houses of a given age, there is such a wide range of possible risk that it is obviously

not economical to treat all houses similarly. Use of a risk prediction model represents a practical tool

for cost optimised decisions to mitigate termite attacks.

There are two major improvements that can be made to the model. The first is to obtain information in

order to effectively interpolate the hazard zones between the major city clusters covered by the data of

11

the Termite Tally. The second is to obtain more accurate quantitative information on the performance

of termite barriers.

The model is not perfect. However it does provide a framework into which knowledge or opinions can

be placed as this becomes available. The model provides a tool for making a quantitative assessment

of the monetary value of obtaining new knowledge, using new barrier systems and applying new

attack mitigation strategies.

0

0.5

1

1.5

0 50 100 150 200 250

age of house (years)

Apparent probability that a

house has been attacked

mean

(

tm

ode

l

)

= 15

y

rs

45 yrs

180 yrs

(i) apparent risk of termite attack

0

0.5

1

1.5

0 50 100 150 200 250

age of house (yrs)

true probability that a house has

been attacked

mean(tmodel) =

15 yrs

45 yrs

180 yrs

(ii) true risk

Figure 11. Possible range of apparent and true risks of termite attack.

7. ACKNOWLEDGEMENTS

This project was undertaken in collaboration with, and funded by the Forestry and Wood

Products Research and Development Corporation, Australia. The authors are indebted for

valuable discussions and advice from Berhan Ahmed, John French, Jim Creffield and Doug

Howick.

8. REFERENCES

Cookson, L.J. and Trajstman, A. (2002). Termite survey and hazard mapping. CSIRO FFP Report No.

137. http:/www.ffp.csiro.au/wft/wpc/termmap/index.htm.

Leicester, R.H., Wang, C-H., Cookson, L.J. and Creffield, J. (2002). A model for termite hazard in

Australia. Proceedings of 9th International Conference on Durability of Building Materials and

Components, Brisbane, Australia.

12

Leicester, R.H. and Wang, C-H. (2001). A probabilistic model of termite attack. Proceedings of

International Conference on Structural Safety and Reliability, Newport Beach, California, June.

Leicester, R.H., Wang, C-H., Cookson, L.J., Wang, X. and Foliente, G.C. (2001). Durability of timber

construction – Termites: Design package and calibration of software. BCE Doc 01/247, CSIRO

Building, Construction and Engineering, Melbourne, Australia.

Standards Australia. AS 1604.1. (2002). Specification for preservative treatment. Part 1: Sawn and

round timber. Sydney.

Standards Australia. AS 3660.1. (2000). Termite management. Part 1: New building work. Sydney.

Standards Australia. AS 3660.2. (2000). Termite management. Part 2: In and around existing buildings

and structures – Guidelines. Sydney.

APPENDIX

A.1 Procedure for assessing the hazard level due to the quantity of wood occurring in a

garden and under a house

Table A.1 shows in quantitative terms some typical distributions of wood corresponding to low,

medium and high hazard levels of termite attack. For other distributions of wood, suitable estimates

may be made through interpolation of these values.

Table A.1 Definition of hazard level assessment due to occurrence of wood in the

garden and under the house

Hazard

level

Number of potential nesting sites(1) Typical distance between substantial

food source (m)(2)

Low <2 >20

Medium 2–5 5–20

High >5 <5

(1). Examples of potential nesting sites

The following refers to potential nest sites for harbouring mature colonies which are not more than 50 m from

the building.

• Tree (diameter larger than 300 mm)

• Tree stump or untreated pole (diameter larger than 200 mm)

• Untreated landscape timber (e.g., sleepers, retaining walls, length > 1.0 m, height > 0.5 m)

• Woodheap (height >0.5 m, ground contact area 0.5 x 0.5 m, length of periods that bottom layer woodheap

is untouched > 1 year)

• Compost heap

• Wood “stepping stones”

• Subfloor storage (height >0.5 m, ground contact area >0.5 x 0.5 m, length of period which it is untouched

>1 year).

• Solid infill under a varandah

• Any part of a building with water leaking under it

(2). Example of a substantial food source

A typical example of a substantial food source would be a piece of timber equal to or greater than 200 x 50 mm

lying in ground contact.

A.2. Definition of ground contact

Table A.2 gives examples of building construction that leads to high, medium and low hazard

of termite attack related to ground contact characteristics

13

Table A.2 Examples of hazard level assessment due to the nature of the

ground contact of a house

Hazard level

Ground contact elements

Low

• House supported by exposed concrete piers or

steel stumps more than 2 m high

Medium

• Intact concrete slab on ground;

• House on stumps less than 600 mm high with ant

caps and made of concrete or treated timber(1) or

heartwood of durable timber(2)

High

• Construction does not comply with AS 3660.1

• Building not inspectable according to AS 3660.2

• Concealed entry zones of any type

• Floor connected to ground by stair cases of

untreated softwood, untreated non-durable

timber(3), untreated sapwood of durable timber;

• Attached patio with solid infill

• Concrete slab-on-ground with large cracks

and/or unprotected pipe penetrations

• Floors connected to ground by elements

containing hidden cavities (e.g. masonry

construction, deeply grooved elements, members

in imperfect contact).

• Brick veneer house

• Leakage of moisture to ground

• Timber floor less than 600 mm off the ground

(1) treated timber refers to timber treated according to AS 1604.1–2000.

(2) for a listing of timber durable classes 1 and 2, see AS 1604.1–2000.

(3) for a listing of non-durable timber of classes 3 and 4 see AS 1604.1–2000.

A.3. Hazard level related to type of construction material

Table A.3 gives examples of high, medium and low hazard of termite attack related to the type of

material used for construction.

14

Table A.3 Examples of hazard level assessment related to the type of

construction material used

Hazard level Type of construction material attacked

Low

• Treated timber(1)

• Untreated heartwood of durability class 1(2)

Medium

• Untreated heartwood of durability class 2(2)

High

• Untreated hardwoods of durability classes 3 and 4(2)

• Untreated sapwood of all species

• Composite wood boards

(1) treated timber refers to timber treated in accordance AS 1604–2000.

(2) for naturally durable timber classes, see AS 1604–2000.

A.4 Hazard related to exposure of timber

Table A.4 gives a method for assessing the hazard level due to the nature of exposure of timber.

Table A.4. Examples of hazard level related to exposure of timber

k10 Exposure of timber

+1

• High human activity

• High up a building

• Humidity <30%

0 • Exposed to rain

−1 • No disturbance and dark (e.g. wall stud, double leaf

masonry wall, roof member.)

• Exposed to sources of moisture so as to be periodically

wet

• Abandoned houses or mostly vacant holiday houses

• Humidity >90%

15