Page 328 - Pipeline Risk Management Manual Ideas, Techniques, and Resources

Receptor vulnerabilities 14/305
tainty in the sample data set. A confidence interval bounds pipeline-related fires in Canada each year, compared to 70,000
the uncertainty associated with the point estimate. For example, other fires and 9,000 forest fires. Their conclusion is that gas
a leak rate estimated to a 95% confidence level has a 95% pipelines generally pose little threat to the environment based
probability of including the true leak rate. on the low incident of fires initiated by gas pipelines [95].
When the number of data points available is small, the confi- *eats from more persistent pipeline releases include con-
dence limits are wide, indicating that not enough information is tamination scenarios as discussed in Chapter 7. Case Study C
available to be confident that all future data will be close to the presents a case where damage states for various environmental
small data set already obtained. Data on pipeline failure rates receptors were defined.
are limited. The use of upper limits of statistical confidence
intervals, especially at a high 95% confidence level, would not Population
present meaningful representations of true failure potential. It
will present unrealistically large predictions, strictly as a result Many consequence assessments focus on threats to humans. To
ofthe small number ofdata points available. estimate potential injury and fatality counts, the population in
Such uncertainty-adjusted predictions do not represent best the hazard zone must be characterized. This includes exposure
estimates of failures. It may be theoretically correct to say, for times, which can be estimated by characterizing population
example, that “one can be ninety-five percent confident that densities at any point in time. This includes estimating
there is no more than a one in ten chance of a spill in this area”
as a result of a statistical confidence calculation on limited spill Permanent population
data. However, the best estimate of spill probability might be Transitory/occasional population
only one chance in ten thousand. Special population (restricted mobility).
An alternative to the normal calculation of confidence inter-
vals or bounds about the mean leak frequency is available A thorough analysis would necessarily require estimates of
for instances where the data set is very small. The confidence people density (instead of house density), people’s away-from-
intervals can be calculated using methods that assume a home patterns, nearby road traffic, evacuation opportunities,
Poisson distribution ofthe leak frequency data [86]. time of day, day of week, and a host of other factors. Several
The use of confidence intervals in risk communications is methods can be devised to incorporate at least some of these
discussed in Chapter 15. considerations. An example methodology, from Ref. [67], is
discussed next.
According to Ref [67], average population densities per
IX. Receptor vulnerabilities hectare can be determined for a particular land use by applying
the following formula:
An “estimate of risk expressed in an absolute terms” modeling
approach requires identification of a hazard zone and a charac- Population per hectare = [ 10,000/(area per person)] x (“YO area utilized)
terization of receptors within that zone. A doseresponse type x (%presence)
assessment, as is often seen in medical or epidemiological stud-
ies, may be necessary for certain receptors and certain threats. This reference describes the process of population density esti-
Focusing on possible acute damages to humans, property, and mation as follows (excerpt, but not direct quote):
the environment, some simplifying assumptions can bemade, as Indoor population densities have been based on the number of square
discussed below and as seen in the case studies in this chapter. meters required per person according to the local building code.
As noted in Chapter 7, a robust consequence assessment Residential dwellings are not covered in this building code, but have
sequence might follow these steps: been assigned a value of 100 m2 per person, on the basis of a typical
suburban density of 30 persons per hectare and one-third actual
1. Determine damage states of interest (see discussions this dwelling area. For nonresidential use, available floor space has been
chapter) set at 75% ofthe actual area, to allow for spaces set aside for elevators,
2. Calculate hazard distances associated with damage states of corridors, etc. Based on the above, the indoor populations shown in
interest Table 14.21 have been estimated.
For rural and semirural areas, the outdoors population is generally
3. Estimate hazard areas based on hazard distances and source expected to be greatest on major roads (excluding commercial areas).
(burningpools, vapor cloud centroid etc.) location (seepar- If an appropriate value for vehicular populations can be determined
tide truce element in Table 7.6) then this can be conservatively applied to all outdoor areas. Assuming
4. Characterize receptor vulnerabilities within the hazard that a major rural road is 10 m wide, 1 hectare covers a total length of 1
areas km. For rural areas, an average car speed of 100 km/Iu and an average
rate of I car per minute has been assumed. Based on this and an aver-
This process is rather essential to absolute risk calculations. age of 1.2 persons per car, an outdoor population density of 1 person
per hectare has been determined. Using 60 kmihr and a 30-second
Environmental damages are often very situation dependent average separation, a population density of 4 people per hectare is
given the wide array of possible biota that can be present and applied to semirural areas. For rural commercial outdoor areas and
exposed for varying times under various scenarios. Thermal urbdsuburban outdoor areas, the population values given shown in
radiation levels for non-piloted ignition of wood products can Table 14.22 are suggested.
be used as one measure of an acute damage state. A drawback
might be the uncertainty surrounding the spread of a fire, once Other typical population densities from another source (Ref.
ignition in some portion of the environment has occurred. One [43]) are shown in Table 14.23. (Discussions regarding valua-
Canadian study concludes that there are on average about two tions placed on human life can be found in Chapter 15.)