PBEE Decision Variable
Considerations
DISCUSSION
PEER TESTBED MEETING
5/23/02
Overall Conclusions
 Multiple considerations involved
 Relevant considerations will differ among decision
makers and decision situations
 Consideration of life-safety, functionality (downtime),
and repair costs are relevant
 Level of information will also vary
 By stage in education in decision process
 moving from probability of loss of life to more refined
probabilistic statements
 Moving from scenario based results to probabilistic results
 Important to be able to communicate probabilistic
aspects and uncertainties
Overall Conclusions
 Useful to consider new ways of conveying
information
 Relative risk considerations as done with medical
procedures
 Comparative information (relative to events within
experience base of client, types of structures)
 Specifics of different metrics – getting closer!
 Life safety: several different potential measures all
relating to probability of loss of life
 Functionality: hours/days (time) to restore full
functionality, which can then be translated into
downtime costs
 Repair/damage: cost of repair
Overall Conclusions
 Considering probabilistic information
 Will depend on decision-maker – many want initially
scenario /deterministic results
 Multiple ways of communicating probabilistic aspects
 Considering uncertainty
 Important to consider
 Split between notions of level of confidence (e.g.,
90% confident), and expressing results as confidence
interval (e.g., expected damage within range of
$400,000 and $900,000).
 Need to consider relevant time horizons for thinking
about events and decisions (e.g., for confidence
intervals) to the extent these are relevant
Small Group Results
 14 individuals participating in PEER
breakout
 10 structural engineering (4 practicing, rest
academic), 4 social science/ management/
architecture/ risk analysis
 Filled out short questionnaire with results
that follow
Decision Considerations
1. Potential Loss of Life – for a given event or combination of events
(Check one)
36% Expected number of lives that will be lost.
36% The probability of any loss of life.
29& The probability that more than X number of lives will be lost. [where
decision-makers can specify the threshold level X in advance]
0% The average number of lives expected to be lost per year.
RELEVANCE RATING mean 9.6 median 10
Decision Considerations
2. Potential Damage to Facilities—Downtime and Functionality
(Check one)
14% Percentage of the building incurring substantial damage [where substantial
damage is defined in advance by relevant decision-makers].
50% Number of hours or days before full functions can be resumed in the facility.
29% Dollar value of lost business and other costs associated with “downtime”.
7% The average annual economic loses per year, expected to occur as a result
of earthquake.
RELEVANCE RATING mean 8.4 median 9
Decision Considerations
3. Potential Repair Costs [DIRECT DAMAGE TO FACILITY]
(Check one)
64% Costs of repairing the facility to bring it back to full functions.
21% Percentage of replacement costs that repair costs will constitute.
0% “Risk of ruin” – The likelihood that the costs of repair (and other
earthquake costs) will lead to financial insolvency.
14% The average annual expected cost of repair.
RELEVANCE RATING mean 7.6 median 7.5
Decision Considerations
4.
Potential Externalities – Costs imposed on others because of damage to a
given facility (from the perspective of the damaged facility decision-maker)
(Check one)
64% Costs of damage to other facilities/operations caused by damage to
this facility.
21% Percentage of total damage that occurs to the facility itself versus
as external harm to other facilities (ratio of direct and external
costs).
14% The average annual expected external costs imposed by damage to
a given facility.
RELEVANCE RATING mean 5.2 median 5
Communicating Uncertainty
1. Consider the following ways of presenting probabilistic information about
potential loss of life for a hypothetical structure when fully occupied. Which
do you think decision-makers find most helpful? (Check one)
14% The expected number of fatalities in any given year is 1.3.
21% The probability of one or more fatalities in any give year within this
structure is 15 percent.
21% The probability of 10 or more fatalities in the next 20 years
associated with this structure is 25 percent. [where the number of
fatalities is selected by decision-makers in advance, and the
analysis shows the likelihood of exceeding that amount]
36% If a magnitude 7 earthquake, occurs, the expected number of
fatalities for this structure is 20.
7% Fifty lives are expected to be lost in this structure from the most
severe earthquake that is likely to occur in the next 100 years.
Communicating Uncertainty
2. Consider the following ways of presenting probabilistic information about
potential earthquake losses (i.e. dollar value of lost business, repair
costs, employee costs). Which do you think decision-makers find to be
the most helpful? (Check one)
21% The annualized, expected loss of a damaging earthquake is
$400,000.
43% The probability of earthquake losses of $400,000 or greater in the
next 20 years is 33 percent. [where the dollar level is selected by
decision-makers in advance, and the analysis shows the likelihood
of exceeding that amount]
36% The probable maximum loss associated with a major earthquake
(expected one time every 500 years) is $600,000.
Communicating Uncertainty
3. Consider the following ways of communicating uncertainties about the
potential value of non-life related earthquake losses. Which do you find to
be the most helpful? (Check one)
57% We are 90 percent confident that losses from an earthquake for this
structure will not exceed $800,000.
7% We are very confident that losses from an earthquake associated
with this structure will not exceed $800,000.
36% The dollar value of potential losses for this structure from an
earthquake were it to occur in the next 20 years are expected to be
between $400,000 and $900,000.
Communicating Uncertainty
4. Rate the importance, on a scale of 1 (not important) to 10 (very important),
of the importance of communicating to relevant decision makers:

Impact consequences among potential events in probabilistic terms (versus
strictly as a deterministic one-event scenario)
mean 7.8 median 8

Uncertainties associated with predictions of performance mean 7.4 median 8
Other Considerations
 Individual comments re expressing impacts
 Better to talk about exceedance probabilities on an annual basis for
particular kinds of impacts – providing distributions across different
facilities.
 Often asked to give an overall rating of “life safe” to a building (with
uncertainties embedded into this judgment)
 Externalities almost never considered
 Individual comments re expressing probabilities
 Better to talk about probability of given level of loss on an annual basis
 Requires sophisticated client to present probabilistic results
 Individual comments re expressing uncertainties
 Engineers more comfortable expressing range of potential losses
(confidence interval) rather than a single confidence value.
 Requires sophisticated client to present probabilistic results