wg7 final report 9mai05 - framemethod.net · fire risk evaluation to european cultural heritage...
TRANSCRIPT
FiRE-TECH Fire Risk Evaluation To European Cultural Heritage
Fire Risk Evaluation to European Cultural Heritage
WG7
Define, apply and document a quantitative method to help decision-
making
Final report
May 2005
Convenor: CSTB Project team: CSTB, IBMB Other Participants: RUG, WFR, IST, TNO
Authors Michel Curtat, Dr. Sc. - CSTB Prof. Dr. Ir. Paul Vandevelde, Prof. Ir. Arch. André De Naeyer, Ir. Emmy Streuve - UGent Ir. Leen Twilt, Arnoud Breunese M. Sc. - TNO Prof. Ildefonso Cabrita Neves, Prof. Joaquim Valente, Prof. João Ventura - ST
FIRE-TECH WG7Final report, CSTB, May 2005
2/88
TABLE OF CONTENTS
1. DECISION MAKING IN FIRE-TECH.................................................................. 4 1.1 Introduction ............................................................................................. 4 1.2 Main features of approaches on help to Decision Making .......................... 4
2. LOCATION OF DECISION-MAKING IN THE WHOLE UPGRADING PROCESS ......................................................................................................... 7
• Phase 1. Preliminary steps .................................................................... 7 • Phase 2. Analyse the present level of safety and protection .................. 8 • Phase 3. Possible actions to improve safety and protection .................. 9 • Phase 4. Decision Making (WG7)......................................................... 10
3. HIERARCHY METHODS AND TOOLS........................................................... 12 3.1 Main features of the hierarchy methods ................................................. 12 3.2 Qualitative aspects of the hierarchy network.......................................... 13 3.3 Various naming of the network rows in hierarchy methods..................... 14 3.4 Choosing actions and weights in a hierarchy approach ........................... 15 3.5 Steps of the calculation in a hierarchy approach .................................... 17
4. CSTB ALADIN DECISION MAKING TOOL .................................................... 20 4.1. Bases of the model ALADIN.................................................................... 20 4.2. Quantitative aspects of ALADIN............................................................. 20 4.3 Running ALADIN programme. User’s guide. ............................................ 21 4.4 Examples of runs with ALADIN ............................................................... 21
4.4.1 Simple illustrative example ..................................................................................................................... 22 4.4.2 More realistic example............................................................................................................................. 23
4.5 Using ALADIN for practical applications.................................................. 25 4.6 Example of a study with ALADIN for comparison of 3 alternatives .......... 26
5. THE IST COST/EFFECTIVENESS SPREADSHEET ...................................... 34 Index hierarchy methods.............................................................................. 34 5.1 Introduction ........................................................................................... 35 5.2 Input ...................................................................................................... 35
5.2.1 Policy ....................................................................................................................................................... 35 5.2.2 Objectives ................................................................................................................................................ 36 5.2.3 Strategies ................................................................................................................................................. 36 5.2.4 Fire safety measures ................................................................................................................................ 37
5.3 Scores and weights ................................................................................. 42
FIRE-TECH WG7Final report, CSTB, May 2005
3/88
5.4 Grades of implementation of the fire safety measures ............................ 47 5.5 Costs associated to the implementation of each class of fire safety measures. ..................................................................................................... 49 5.6 Output.................................................................................................... 50
6. OPTIMISATION OF COST /EFFECTIVENESS............................................... 61 6.1 ALADIN APPROACH ................................................................................ 61 6.2 Application made by TNO using the first version of the IST COST/EFFECTIVENESS spreadsheet ............................................................ 63
7. CONCLUSIONS ON HELP TO DECISION-MAKING...................................... 69 ALADIN......................................................................................................... 70 The IST COST/EFFECTIVENESS spreadsheet ............................................... 70 Final comments............................................................................................ 71
8. REFERENCES................................................................................................. 72
ANNEX 1. OTHER DM APPROACH: B. ROY ET AL.............................................. 73
ANNEX 2. AHP (ANALYTIC HIERARCHY PROCESS) .......................................... 74 A2.1 Concepts of AHP................................................................................... 74 A2.2 Theory and Methodology of AHP........................................................... 75
A2.2.1 Weighting influences ............................................................................................................................ 75 A2.2.2 Construction of matrices by pair-wise comparison............................................................................... 76 A2.2.3 Consistency Ratio (C.R.) ...................................................................................................................... 77
A2.3 Other AHP method: Hierarchical Cross-Impact Analysis ....................... 78 A2.4 Previous Applications of AHP to Fire Safety and Fire Protection .......... 79 A2.5 Research under work in AHP Methods .................................................. 80
ANNEX 3. CULTURAL HERITAGE FIRE RISK INDEX METHOD .......................... 81 A3.1. A Fire Risk Index for Cultural Heritage................................................ 82
A3.1.1 Development of a risk index method .................................................................................................... 82 A3.1.2 Generalised procedure .......................................................................................................................... 83
A3.2 Fictive example cultural heritage risk index method. ........................... 84 A3.3 Conclusions.......................................................................................... 88
FIRE-TECH WG7Final report, CSTB, May 2005
4/88
1. Decision making in FIRE-TECH
1.1 Introduction
People have always had to make decisions. A lot of important decisions were made during
the History (war or peace, rising taxes, etc.), and many smaller decisions are made in
everyday life (e.g. buying a new car or a dwelling is a multi-criteria problem with a strong
cost/effectiveness aspect). A first scientific approach of Decision Making was realised in the
19th Century by mathematicians and economists, such as Pareto, Morgenstein, etc. These
first approaches were dealing with decision problems depending on one criterion and one
variable: they had to solve monocriterion decision problems. Multicriteria problems were
addressed in the middle of the 20th Century (Koopsman, Kuhn & Tucker, etc.). The question
was to rank a set of alternatives (each of them based on the same variables having different
values) or to find the best alternative in the set. The first international conference in Multiple
Criteria Decision Making (MCDM) was organised in 1972 and followed by many research
projects. We have now a lot of works done on MCDM or MCDA (Multiple Criteria Decision
Aid), where the decision making procedure is based on different theoretical basis (utility
based [Fishburn], overranking [Roy], etc.). We chose in FIRE-TECH the hierarchy approach that offers the capability of weighting (grading) in simple way alternative solutions of safety/protection actions.
The words: "Decision-Making", "Decision Analysis", "Help to Decison-Making", are more and
more used, mainly in the domains of human activities connected with management,
business, allocation of resources ... The concepts, tools and methods mentioned in many
papers are various, more or less complex and partially validated. Application to Fire Safety
and Fire Protection is recent and still rare.
1.2 Main features of approaches on help to Decision Making
• The selection of the best alternative or a ranking of all the alternatives is aimed.
• Inefficient alternatives must be eliminated.
• The identification of the most promising alternatives can be done solving a mathematical
programming problem.
FIRE-TECH WG7Final report, CSTB, May 2005
5/88
• The solution must be: not too much influenced by the uncertainty of the results of the risk
analysis step, nor by slight variations of the weights given as inputs.
Explaining the general engineering decision making process, Krick (1969) states the
following:
“Although the specifics vary from situation to situation, in almost every instance these four
steps must be taken before an intelligent decision can be reached:
• Objectives, strategies and measures, the “criteria” (presented below) must be selected
and their relative weights determined.
• The performance of alternative solutions must be predicted with respect to these criteria.
• The alternatives must be compared on the basis of these predicted performances;
• And then, a choice must be made.”
Within the FIRE-TECH programme dealing with the optimisation of fire safety and fire
protection in Cultural Heritage, we have developed an approach presenting the following
characteristics:
• "Help to Decision-Making" has in Fire-Tech to be understood very widely, i.e. we think of
bringing to a person in charge of fire protection of a Cultural Heritage any kind of help
FIRE-TECH may deliver. In other words, we are not thinking of considering only the
"decision analysis" task, to be performed when all the technical and economical elements
of knowledge has been gathered, but of all the steps from the very intention of evaluating
or improving the level fire safety to the final actions of renovation, equipment,
organisation, ...
• Methods and tools should not be very complex, for several reasons:
- the final user may not be able to use an expensive approach as she or he will may have a limited amount of money and a too short time available;
- the basis of method and tools have to be understood by the persons who will do the work.
• Method and tools have to have been experimented and validated. Because of the novelty
of the Decision-Making approaches, some of the complex approaches are still in their
development state; These approaches need more research work before practical
application.
FIRE-TECH WG7Final report, CSTB, May 2005
6/88
• The Decision-Making part of the process has to offer the following characteristics:
- Be based on a quantitative approach
- Give a general view of inputs and outputs
- Rank alternate solutions
- Optimise efficiency/cost
- Be open to more complex approaches.
The Decision-Making approach retained in FIRE-TECH is based on the “hierarchy analysis”
that uses a decomposition of the main goal in “sub-goals” at several levels below the goal top
level. At each level, several “boxes” corresponding possible actions at this level are graded
for each box at the upper level (more “general”). These grades, or weights, express the
influence of a row on actions of the upper row. We assume that the influences are linear.
Then matrix or linear combination formulae are used to calculate the dependency of any box
on any other box below it. The § 3 presents the hierarchy analysis and the tools developed in
FIRE-TECH.
The following chapter § 2 present the location of the Decision-Making task in the whole process. The hierarchy approach, retained here, is described in chapter § 3.
The tools developed for FIRE-TECH, CSTB ALADIN programme and IST Cost/Effectiveness spreadsheet, are presented in chapter § 4 and chapter 5 respectively.
Chapter § 6 exposes the way the important question of optimisation of the ratio EFFECTIVENESS/ COST was addressed with ALADIN and with the IST Cost/Effectiveness spreadsheet.
ANNEX 1 gives a few words on the overranking Decision-Making methods.
ANNEX 2 brings details on Analytic Hierarchy Process.
ANNEX 3 presents the “Cultural heritage fire risk index", by Fredrik Nystedt.
NB: This report was written after several case studies had been realised (WG 8).It takes into
account comments made by FIRE-TECH partners. The case studies themselves offer the
opportunity to look at the way the tools: IST spreadsheet or ALADIN, both presented below,
have been applied.
FIRE-TECH WG7Final report, CSTB, May 2005
7/88
2. Location of Decision-Making in the whole upgrading process
WG7, in collaboration with other WGs, proposed a general frame and definitions of the
phases and steps of the general cost-effectiveness study (WG7 draft report,”Presentation of
the fire-tech method to help to decision-making in fire protection of cultural heritage”,
February 2004, Michel Curtat). The global approach was re-formulated after some minor
modifications and largely illustrated by several case studies in the WG8 report (September
2004). The location of the Decision-Making actions in the whole process is clearly presented
in Figure 2: flowchart of the effectiveness / cost study, commented in details in the
WG8 report.
The following lines give comments on the phases of the whole process of improvement.
• Phase 1. Preliminary steps
Phase 1.1. What is awaited at the end?
The goal of the work has to be fixed at the very beginning. It is useful too to know what are
the reasons to perform an upgrading of the level of safety and protection of the Cultural
Heritage, e.g.: is it a general care for the improvement of the protection of the contents, or is
there a new standard or a new regulation, or did a recent fire show a special need of
protection, or does the authority in charge of the Cultural Heritage have a new policy in
safety …?
It is useful too to have exchanges with the persons who are asking for the study (a safety
commission, a Cultural Heritage authority, the Cultural Heritage manager …) and with the
persons making the final decision and the persons in charge of the financing.
Phase 1.2. Definition of time and money available
The awaited duration of the study as well as the money available for the study and for the
improvement (renovation, new equipments …) have of course to be known. Several
alternatives on the improvements funding and delay may be presented, e.g. if several steps
of the improvements can be considered on several years. In this later case, priorities of
actions have to be established with the Decision-Making tools.
FIRE-TECH WG7Final report, CSTB, May 2005
8/88
• Phase 2. Analyse the present level of safety and protection We have distinguished here two approaches, even if the best way is to use both of them:
a “prescriptive and expert approach” and a “performance based approach” based on fire
safety engineering.
Phase 2.1 Prescriptive and expert analysis
• Experience of previous fires
The knowledge of previous fires in a similar C H can be used to suggest fire scenarios
and/or, directly, lead to possible solutions of protection, the efficiency of which having been
demonstrated. Information on previous fires in Cultural Heritage has been provided by
WG2 report.
• Behaviour of ancient materials
Knowledge gained on the characteristics of ancient material can be applied to find frail
elements in the Cultural Heritage, and bring ideas of protection at step 3. The work done in
WG3 brings information on the behaviour of these particular materials.
• Observation and collection of comments from everyday life of the Cultural Heritage
It might be very efficient to gather comments and observations on the behaviour of the
Cultural Heritage, mainly on the occurrence of small incidents or on the discovery of “weak”
aspects. Such a qualitative analysis is an important part of this step.
• Complying to regulations and exchanges with authorities
The persons in charge of Fire Safety of the Cultural Heritage have to know the regulation
concerning the Cultural Heritage. Compliance to existing regulations is very important. Even
if a Performance Based Study is carried out, it has to be executed within a regulatory frame.
WG1 has presented a global view on the European regulations dealing with Fire
Safety/Protection in Cultural Heritage.
The compliance of any new action of improvement the fire safety or protection level has to
be checked.
FIRE-TECH WG7Final report, CSTB, May 2005
9/88
• Expert analysis
Expert judgement, exploitation of results from tests, knowledge from fires in various
conditions …, can bring an important contribution to this phase.
At this step, we could add the possibility of considering advises given in some cases by
associations or authorities.
Phase 2.2 Performance based Quantitative Risk Analysis
Construction of scenarios and fire simulation with computer fire ‘fire, smoke, egress …)
models is very useful as it can give more light or bring a new insight on some difficulties
occurring at certain times. It brings a quantified view of the unwanted consequences, at
which time some unacceptable state is reached, and informs on the speed of fire
aggressions. Fires simulations can show the efficiency of suggested new measures, by
showing the differences in conditions of hazards.
The choice of realistic scenarios and an estimation of their probabilities have to be realised
by an fire expert.
FSE models can then be used again to evaluate the efficiency of possible actions. WG6 has
presented the method and tools of the Risk Analysis step.
• Phase 3. Possible actions to improve safety and protection
After the previous phase, a list of acceptable measures has to be produced, accompanied
with characteristics and cost of each measure. The decision-making tools will help to choose
in this list the most convenient measures. Depending on the Cultural Heritage, on the goal
and nature of the study, a few possible actions can be well defined, or, at the other hand, one
may still have a view on a large variety of them. Two models, presented below, ALADIN and
the IST Cost/Efficiency spreadsheet, are the tools developed in Fire-Tech. ALADIN seems
to be more adapted for the 1st case, the IST Cost/Efficiency spreadsheet for the 2nd case (see
below § 4, § 5, § 6).
Phase 3.1. Possible technical actions (measures)
Once points where improvements are needed have been identified, it is necessary to know
what the possible technical solutions are, what are their costs, efficiency, reliability, etc. WG
4 wrote a report containing such useful information, and presenting a large list of technical
measures.
FIRE-TECH WG7Final report, CSTB, May 2005
10/88
Phase 3.2. Possible non technical actions (measures)
A number of measures are not merely technical: emergency planning, teaching and training
employees, salvage operation management, organisation of local fire-fighting, conception of
signs, etc. The effectiveness of these actions can be evaluated by expert judgement and/or
by Fires Safety Engineering.
• Phase 4. Decision Making (WG7)
We are then at the point where several safety or protection alternatives are to be evaluated.
Decision-Making tools will then be used to choose one of them, or in order to rank them
either to keep one pf them, or to establish priorities for an improvement during several
years. The present report brings information on the method and on two tools used for
applications to case studies: the ALADIN programme and the IST “cost/effectiveness index”
spreadsheet.
UGentIr. Emmy Streuve ©
PHASE 2: Analysis of present level of safety
Performance BasedAnalysisPrescriptive Analysis
1. Performancecriteria
3. Riskanalysis
event trees
2. Developdesign
fire scenarios
Objectives Acceptancecriteria
Selectionoffire scenarios
Design fire scenarios
Building characteristics
Occupancecharacteristics
Risk level
Fire curves: Quantification of fire development
Analysis of damage
Selectionevents
Selectionfire safety
design
Prescriptivein the
Existingregulations
Informationout of
previousfires
Guidelinesonthe firebehaviourof ancientmaterials
4. Risk evaluationPHASE 3: Possiblefire safetyactions
Prescriptivecritaria for fire
protectionaction
Selected firesafetydesign meets
the performance criteria?
Figure 2.1: Analysis of the present level of safety and protection.
FIRE-TECH WG7Final report, CSTB, May 2005
11/88
Figure 2.2. Flowchart of the cost /effectiveness study.
Cost effectiveness study
Phase 2:
2’. Prescriptiveanalysis
2”. Performancebased analysis
2. Analysis of present level of
safety
3. Possible fire safety actions
1. Preliminary steps
1.1. Definitionof main goal
1.2. Time &money available
4’. Technical optimisation
4”. Technical& financial
optimisation
4. Decision making
Phase 3:
Phase 4:
Approach 1 Approach 2
Approach 1 Approach 2
UGentIr. Emmy Streuve ©
FIRE-TECH WG7Final report, CSTB, May 2005
12/88
3. HIERARCHY METHODS and TOOLS
Hierarchy approaches were retained here as decision-making tools because they seemed
simply consistent with the whole process and because they do not imply heavy calculation.
As other Decision-Making methods, the hierarchy approaches:
- Facilitate the development of logical hierarchical structures for complex decisions,
- quantify the efficiency of assessments and expert’s opinions,
- assign weights (importance of influence) to different possibilities relative to one another:
ranking a set of alternatives in order to retain the most appropriate one(s).
3.1 Main features of the hierarchy methods
A general question is addressed, the goal of the decision making, depending on several
variables, and for which several alternative solutions can be imagined and have to be
retained, ranked, or eliminated.
The decision problem is decomposed in several “possible actions” at different depths of
analysis, from global goals to precise solutions of partial aspects.
At each level of “generality”, a judgement is made on the relative efficiency of each
component of this level on each component of the levels above.
Calculations are executed with linear models of indirect influences.
The outputs are the relative influence of all the components are calculated so that the best
solution(s) (the best alternative) can be identified and weighted, or a ranking be established.
FIRE-TECH WG7Final report, CSTB, May 2005
13/88
3.2 Qualitative aspects of the hierarchy network
Table 3.1. Levels (rows) of the hierarchy network
• Policy = main goal
• Objectives
• Strategies (or: tactics)
• Measures (or: components, or: parameters)
• (Grades, in some approaches)
The main goal of the decision to make is at the top level of the hierarchy. Policy is this
very general goal. We have one policy. Here we can call it: “Reduce fire risks” or
“Optimise fire safety/protection”. These words are written in a single box at the first
(upper) level.
The levels below the top contain “actions” one can think of for the success of the policy.
These actions are described more and more precisely as we go down the levels.
Objectives are the actions directly connected to policy. They are mentioned at the
second (lower) level in a few boxes. They are called, e.g.: “Protect the firemen”, “Protect
the precious contents”, ”Safeguard continuity of activity”, …
At the third level, several boxes contain the Strategies. Strategies are actions necessary
to reach the objectives. E.g.: Reduce the probability of fire start», «Limit fire
development”, “Facilitate egress”…
Measures (or: parameters) are described at the fourth level in a certain number of boxes.
Measures are “practical” solutions to serve the strategies. E.g.: “Means for fire detection”,
“Means for fire suppression”, “Systems of smoke control”, “Emergency and alarm
signs”…
FIRE-TECH WG7Final report, CSTB, May 2005
14/88
A list of measures can be established from expert knowledge. Such a list, proposed in
WG7, is given in the presentation of ALADIN (Chapter 4). The IST cost/effectiveness
index spreadsheet contains another list of 19 measures that was established during the
development of this tool. It is presented below (Chapter 5). The weight a measure can
receive is of course depending on the characteristics of the cultural heritage being
studied and on the fire scenarios.
Grades may be introduced at the lowest level in “index methods”, as guiding values to
help to the choice of a measure. E.g., a list of fire resistant partitions can be given with
values assessing their relative efficiency. Each grade is associated with one measure:
using grades is a way to assess a measure within the global calculation. ALADIN uses no
grades (or we can say that grades = 1 in ALADIN). Further weighting of the efficiency on
alternatives is possibly realised on the outputs of ALADIN
3.3 Various naming of the network rows in hierarchy methods
Close hierarchy methods are using different words for the rows in a hierarchy network. The
following Table 3.2 aims at the presentation of the different terms used.
Table 3.2. Words for Quantitative aspects and outputs
Level In papers by Saaty, Marchant, Shields
Vocabulary of Dublin University
In FIRESEM
(Annex 3)
In Fire Tech
1 (top)
Policy Main goal Policy Policy
2 Objectives Criteria Objectives Objectives
3 Tactics Sub-Criteria Strategies Strategies
4 Components Sub-sub-criteria
Or: decision units
Parameters Measures
5 Sub-components (possibly)
Survey items
Grades for measures
FIRE-TECH WG7Final report, CSTB, May 2005
15/88
In hierarchy approaches:
- Each box receives given values according to the relative influence of all the boxes in the
row above. “Relative” means that the figure is expressed relatively to the other boxes of
the same level.
There are several ways to calculate the weights from levels and boxes: either direct explicit
calculations when it is possible, or matrix calculations.
- The calculations are made by summation of combined influences (see the IST
cost/efficiency index spreadsheet) or by a matrix calculation (see ALADIN).
Details are given in the following paragraphs.
3.4 Choosing actions and weights in a hierarchy approach
Definition of all actions and weights is based on existing knowledge about the various
components that play a role to improve fire safety and fire protection. The choice of the
values of the relative influences of “boxes” at level l+1 to each box of level is of course an
expert’s work, depending on the characteristics of the study being carried out. Among the
criteria for choosing possible actions, we find: performance, suitability, reliability, architectural
acceptability …
Another important aspect is that the following conditions have to be fulfilled (for each given
Cultural Heritage):
- The list of possible actions has to be as complete as possible;
- To avoid bias in the calculated influences, the concepts behind all the actions at the
same hierarchical level have to be clearly different. In other words, the intersection of the
semantic contents of any pair of boxes present on the same line must be empty.
Choice of objectives The user may be interested in one objective, e.g. the protection of the contents. Then the
user will put 1 into the “contents” box, and 0 into the other boxes (in IST Cost Effectiveness
Index spreadsheet), or will not create these useless other boxes (in ALADIN). The user may
e.g. be interested both in “contents” and in “safety for people”, and then give e.g. 0.40 to
“contents” and 0.60 to “safety of persons”.
FIRE-TECH WG7Final report, CSTB, May 2005
16/88
Choice of strategies The user defines the strategies he is thinking of (in ALADIN) and gives values into the
corresponding boxes (in ALADIN and in the IST Cost Effectiveness Index spreadsheet). In
the IST Cost Effectiveness index spreadsheet, a list of strategies is offered. The boxes
corresponding to rejected strategies receive the value 0.
Choice of measures The same can be said for measures: in the IST Cost Effectiveness Index spreadsheet,
values are to be given into pre-existing boxes, in ALADIN one has to create the boxes
corresponding to the retained measures and put numbers into them.
General lists of possible measures are presented with ALADIN and the IST Cost
Effectiveness Index spreadsheet.
The large diversity of fire conditions does not make possible the establishment of
“recommended” measures for a given Cultural Heritage. The risk analysis step is a very
useful step to help the derivation of a list of possible measures as this step underlines the
weak points of the cultural heritage.
The WG8 report presents also an example of grading possibilities for technical measures.
Then, even if the measures and the associated weights have to be fixed for each Cultural
Heritage for each given protection/safety goal, the work done in FIRE-TECH brings helps to
choose them:
- The quantitative risk analysis and/or qualitative analysis phases show weak points for
which strategies and measures can be imagined;
- The knowledge from previous fires, from WG2 report and other sources, gives
information to the efficiency of measures.
- The WG4 report describes many technical measures and gives precious comments
on them.
- The WG8 case studies present examples of possible solutions.
FIRE-TECH WG7Final report, CSTB, May 2005
17/88
3.5 Steps of the calculation in a hierarchy approach
As we said, direct influences are graded by the user who gives values as input data.
The other, “indirect”, influence grades are calculated as presented at Figure 3.1, in which
the presentation uses matrix products.
The results can also be expressed by the linear formulae corresponding to these matrix
products. The only difference from the following Figure is that we use in Fire-Tech the word
“measure” instead of “parameter”.
Figure 3.1. Schematic summary of the hierarchical approach (Watts, 1995).
(See also Figure A2.3)
The following lines give details on the required inputs and the calculated results in output.
The following figures present, on an example of network, the nature of the required inputs
and of the outputs.
The following chapters on ALADIN and on the IST Cost-Effectiveness spreadsheet will
present several examples of quantitative results.
FIRE-TECH WG7Final report, CSTB, May 2005
18/88
• Policy Improve fire safety & fire protection of the CH
• Objectives
e.g. protect persons, protect contents, protect surroundings …
• Strategies e.g. reduce probability of fire start, facilitate fire fighting and salvage, …
• Measures e.g. means for detection, facilitate access …
• Parameters (possibly) grade efficiency of each measure
M
P
M
P
M
P
M
P
M
P
M
P
P
O O O
S S S S
Figure 3.2. Example of a hierarchy network.
Parameters can be used (IST spreadsheet), or not (Aladin), according to the way the user
wants to exploit the results. With no parameter, a relative assessment of the influence of the
considered measures is made. With parameters, one can calculate an “absolute” index of
effectiveness, directly connected to the linear combination of the grades given to
the measures.
Required inputs (figure 3.3) : the user has to give the weights of each mesure on each
strategy, the weights of each strategy on each objective and the weights of objectives on
the policy.
Outputs : they are presented at figure 3.4. we obtain the weights of strategies on policy, the
weight of measures on objectives and the weights of measures on policy.
Pratical examples are given in the following pages.
FIRE-TECH WG7Final report, CSTB, May 2005
19/88
M
P
M
P
M
P
M
P
M
P
M
P
O O O
S S S S
S S S S
P
O O O Objectives on Policy
Strategies on Objectives
Measures on Strategies(Only M(j)/ S(2) represented)
Figure 3.3 Examples of inputs.
O O O M M M MM M
Measures on objectives S
M M M MM M P
Measures on PolicyO S
S S S S P
Strategies on Policy
O
Figure 3.4 Influences obtained as Outputs.
FIRE-TECH WG7Final report, CSTB, May 2005
20/88
4. CSTB ALADIN Decision Making Tool
4.1. Bases of the model ALADIN
The basic concepts are the same as in AHP (cf. Annex 2). In order to produce a tool simple
to use with fast calculation time, we use here fully consistent matrices, i.e. we always have
exactly: aij = 1/aji in matrix A. Then we assume that the expert using the programme gives
directly the weight of each box on level l+1 to each box of level l, relatively to the other boxes
of level l+1. This has to be done for all the boxes.
The definitions of policy, objectives, strategies and measures are fully open in ALADIN.
Chapter 6 presents the definitions used in the IST Cost/Effectiveness spreadsheet that can
of course be used too with ALADIN. But for a simple problem, the number of row and boxes
can be much reduced when using ALADIN (in which one cannot consider more than 6
measures at a time).
4.2. Quantitative aspects of ALADIN
Let us take the example of the box number (l,n) at level l. We consider that this box is
connected to all the boxes at the lower level l+1. Each link is given (by the user) a value. This
value represents the weight of each box number at level l+1 on the box number (l,n) at level
l. If the influence of the action described in a box of level l+1 is negligible relatively to the
other links starting from level l+1, the weight is set to 0. If the influence of the action is strong
relatively to other actions of the same level, the number is higher. The numbers are
normalised so that the sum of all the weights from row l+1 the box number (l,n) be 1 (100 %).
We suggest using the scale proposed by Saaty, given above in Annex 2.
In order to:
• avoid a too heavy work in entering input data,
• lead the user to keep a limited number of really appropriate actions (measures) or group
of actions (of measures), keep enough legibility to the outputs, we fixed in the programme
the following limits:
- the numbers of rows is limited to five,
- the number of boxes in a row was limited to six.
FIRE-TECH WG7Final report, CSTB, May 2005
21/88
4.3 Running ALADIN programme. User’s guide.
ALADIN is a FORTRAN 90 programme developed at CSTB by Michel Curtat and Christine
Jourdain. Running ALADIN is very easy: one has just to answer simple questions. The
computer time is not visible on a modern computer.
Input data
• The programme asks the following input data concerning the structure of the networks:
- How many rows (the levels)? One can use all the set of levels presented above (from
policy to measures), or as well work on a simpler hierarchy corresponding to the
actual problem.
- From the second row (the first row is the single top box, the policy) to the last (lowest)
row, how many boxes?
• The programme asks for the input weights:
- In each row (from the lowest level), and in each box in each row, what is the weight of
any box to each given box of upper level? (See also § A2.2). We use the scale: 0, 1,
3, 5, 7, 9.
The sums of the values for each row are normalised to 1 (100%) after the inputs have
been given.
Output data
• A drawing of the network with numbers given to the boxes.
• For all boxes: all the given and calculated weights.
The computer time is very short.
4.4 Examples of runs with ALADIN
We have chosen to present here examples studied with Christine Jourdain in her Master’s
report. This report (in French) presents ALADIN and gives several other examples. The first
two examples given here are aiming at the presentation of ALADIN’s results in order to help
the understanding of the inputs and outputs and of the way ALADIN calculates the influences
These examples are not directly useful any application to a study of the fire safety of a
cultural heritage. The third example comes from the WG8 case study “Virtual museum”.
FIRE-TECH WG7Final report, CSTB, May 2005
22/88
4.4.1 Simple illustrative example
In the Table 4.1 below, the weight 0.0001 means “no influence”. The number zero is not
used because it leads to problems of “division by zero”. Values of 0.01, 0.001 seem to be
acceptable too. In the present version of the programme, when the user enters 0, the
programme takes 0.001.
Table 4.1 Weights of influence.
FIRE-TECH WG7Final report, CSTB, May 2005
23/88
4.4.2 More realistic example
Figure 4.1. Structure of the network on Fire Safety/Protection imaginary case.
FIRE-TECH WG7Final report, CSTB, May 2005
24/88
Table 4.2 Calculated weights
Figure 4.1 shows the hierarchy structure for an “academic” fire safety/protection study. The
thickness of the arrows is proportional to the given input weights (from 1 to 9).
Table 4.2 gives all the influence weights. The representation is “box to box” and the “steps”
correspond to the levels: step 1 is the top (policy), the objectives box, and the strategies
boxes. The measures boxes (9 to 14) are labelled on the first column only as they have no
lower level.
Comments for this example of calculation: Inert gas, access to firemen and fire protection of
the structure represent 66 % for the protection of the historical content. For protection of
people, access to firemen represent 47%. Separation (creation of compartments) is the
strong factor.
FIRE-TECH WG7Final report, CSTB, May 2005
25/88
Sensibility analysis
The short calculation time of ALADIN authorises the users to perform sensitivity analysis: this
can be very useful in situations where they are not very confident in a given weight.
In the Master’s report by Christine Jourdain, a sensibility study was realised in order to show
how variations of the given weights can influence the conclusions. As the calculation is
based on matrix algebra, the propagation of small numerical variations is linear and does not
lead to dramatic effects on the outputs.
It is not realistic to enter close numbers, differing e.g. by 1 %, even if the programme itself
will show slightly different outputs, because the difference between the corresponding
options is meaningless if compared to the uncertainty of all the other steps of the whole
process: expert judgement, fire modelling, etc. We decided to keep the scale by Saaty given
in Annex 2.
4.5 Using ALADIN for practical applications
- ALADIN can follow the decomposition presented above on: “policy, objectives, strategies,
measures”. ALADIN was not conceived to deal with many measures and/or many
strategies, as the maximum number of boxes in a row is 6. ALADIN can be used for
“zooming” on partial influences, e.g. in using it on a reduced “one given objective,
strategies, measures” network.
- It is often necessary to weigh the global influence of a set of components of a given fire safety system, rather than weighing each individual component separately. ALADIN can be used to calculate e.g. the combined weights on policy and
objectives of given strategies and measures, as well as to compare the combined
weights of several alternatives (established by one hesitating person or by several
experts).
- The results can be presented under the format of a decision matrix that can be exploited
directly by user or can be studied with overranking tools [B. Roy et al.]
- ALADIN has been used in several WG8 case studies (cf. WG8 report).
FIRE-TECH WG7Final report, CSTB, May 2005
26/88
4.6 Example of a study with ALADIN for comparison of 3 alternatives
(WG 8, CSTB case study: “Fire protection of a virtual museum of modern art”)
In this case, the decomposition is very simple. The lower third row is “measures”, the second
row is “objectives” and the first one box row is “Fire Safety”. We have directly connected the
“measures” level to the “objectives” level. This simplification makes the numerical results
more legible. If the Cultural Heritage was more complex (here it is limited to one room and a
corridor connected to a restaurant), or if the expert(s) was (were) not able to retain a few
measures only, ALADIN can be applied on several simpler “subproblems” or one can use the
IST tool where many measures are considered. In the WG8 reports on case studies, other
more complex examples are given. Figure 4.2 presents the simple network of this case
study, with arrows for the example of “protection of contents”.
Figure 4.2: the hierarchy network for the case study of a virtual museum
FIRE-TECH WG7Final report, CSTB, May 2005
27/88
Tables 4.3 and 4.4 present the set of measures on which the 3 experts agreed as being
possible, but to which they did not associate the same weights. This list comes from expert
judgement that may have been helped by a knowledge on previous fires, application of Fire
Safety Engineering. The measures considered are convenient are the following:
- fire detection close to the artworks (“local” detection),
- sprinklers,
- water mist system,
- smoke detection,
- mobile shutter, ”curtain”,
- formation and training of the staff.
At this step, the acceptability and reliability of the measures have not been considered. It will
be assessed later, at the phase of optimisation of cost/efficiency.
Table 4.3. Possible measures.
FIRE-TECH WG7Final report, CSTB, May 2005
28/88
The experts had then to enter weights for each measure, and later for each objective. This is
given at Table 4.4 where values are not yet normalised.
Table 4.4 Input values for measures
Table 4.5. Input values for objectives
Expert Safety/protection {contents} {persons} {building} 1 1 1 1 2 2 2 1 3 2 2 1
The following figures give ALADIN results for all the weights calculated according to the 3
alternatives (3 experts).
Figure 4.3 shows the 3 sets of weights for people safety.
Figure 4.4 gives the weights for the contents.
Figure 4.5 presents the weights given on the protection of the building.
Expert Measure {LocDetect} {Sprinklers} {Water mist} {SmokeDetn} {Curtain} {Staff} 1 contents 7 7 0 5 0 5 persons 7 7 0 3 0 3 building 7 7 0 5 0 0 2 contents 7 0 7 1 7 3 persons 7 0 7 5 7 5 building 7 0 7 3 7 3 3 contents 7 0 7 1 0 3 persons 5 0 3 3 0 5 building 3 3 3 7 0 3
FIRE-TECH WG7Final report, CSTB, May 2005
29/88
Figure 4.6 shows the weights of the objectives on the general policy of safety/protection of
the cultural heritage.
Figure 4.7 shows the weights of the measures on the general policy of safety/protection of
the cultural heritage.
Table 4.6 gives the numerical values of the measures on the general policy of
safety/protection of the cultural heritage. The three columns represent the values associated
to the choices of the three experts. The “criteria” are given on the rows. Table 4.6
summarises the weights of the 3 alternatives on the policy. We can see that some values are
very close (e.g. 0.34 and 0.31, or 0.25 and 0.26, or 0.12 and 0.14). For these situations, the
influences have to be considered as equivalent. In other cases, e.g. for “sprinklers” or “water
mist”, the values are very different, because one measure was excluding the other one.
The cost has been roughly estimated for each alternative; C2 > C3 > C1. The person in
charge of the decision can then exploit this matrix that brings a synthesis of the decisions
elements. The question of optimisation of effectiveness/cost is addressed in the chapter 6.
0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 0,4
{LocDetect} on {People}
{Sprinklers} on {People}
{Water mist} on {People}
{SmokeDetn} on {People}
{Curtain} on {People}
{Staff} on {People}Expert 3Expert 2Expert 1
Figure 4.3: Weights of measures on the Objective “safety of people”.
FIRE-TECH WG7Final report, CSTB, May 2005
30/88
0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 0,4 0,45
{LocDetect} on {Contents}
{Sprinklers} on {Contents}
{Water mist} on {Contents}
{SmokeDetn} on {Contents}
{Curtain} on {Contents}
{Staff} on {Contents} Expert 3Expert 2Expert 1
Figure 4.4: Weights of measures on the Objective “protection of content”.
0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 0,4
{LocDetect} on {Building}
{Sprinklers} on {Building}
{Water mist} on {Building}
{SmokeDetn} on {Building}
{Curtain} on {Building}
{Staff} on {Building} Expert 3Expert 2Expert 1
Figure 4.5: Weights of measures on the Objective “protection of the building”.
FIRE-TECH WG7Final report, CSTB, May 2005
31/88
0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 0,4 0,45
{Contents} on {FS inMuseum}
{People} on {FS in Museum}
{Building} on {FS inMuseum}
Expert 3Expert 2Expert 1
Figure 4.6: Weights of Objectives on the Policy safety/protection of the Cultural Heritage.
0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 0,4
{LocDetect} on {FS inMuseum}
{Sprinklers} on {FS inMuseum}
{Water mist} on {FS inMuseum}
{SmokeDetn} on {FS inMuseum}
{Curtain} on {FS in Museum}
{Staff} on {FS in Museum} Expert 3Expert 2Expert 1
Figure 4.7: Weights of measures on the Policy safety/protection of the Cultural Heritage.
FIRE-TECH WG7Final report, CSTB, May 2005
32/88
Table 4.6: the decision matrix
276367196COST (k€)0,220,140,12{Staff} on {FS in Museum}
0,170,100,21{SmokeDetn} on {FS in Museum}
0,000,250,00{Curtain} on {FS in Museum}
0,260,250,00{Water mist} on {FS in Museum}
0,030,000,34{Sprinklers} on {FS in Museum}
0,310,250,34{LocDetect} on {FS in Museum}
A3A2A1
276367196COST (k€)0,220,140,12{Staff} on {FS in Museum}
0,170,100,21{SmokeDetn} on {FS in Museum}
0,000,250,00{Curtain} on {FS in Museum}
0,260,250,00{Water mist} on {FS in Museum}
0,030,000,34{Sprinklers} on {FS in Museum}
0,310,250,34{LocDetect} on {FS in Museum}
A3A2A1
Acceptability and reliability of the measures
Up to now, these aspects had not been considered. We must mention they could have been
considered in weighting the measures. In this case study, we assume that the characteristics
considered for the measures were only efficiency. Then we can now consider new
characteristics in order to grade the alternatives. Another characteristic that can have a
strong influence on the conclusions of a study is the architectural acceptability. The following
table shows the results obtained after taking into account acceptability and reliability, and
cost (very approximate here), as presented in the report of WG4.
FIRE-TECH WG7Final report, CSTB, May 2005
33/88
Table 4.7 . New matrix with acceptability, reliability and cost.
Conclusion
The “best” alternative is finally alternative 2 because its cost is less than the maximum of 300
k€ and because the associated measures have a high level of reliability and acceptability.
FIRE-TECH WG7Final report, CSTB, May 2005
34/88
5. The IST COST/EFFECTIVENESS spreadsheet
Index hierarchy methods
Several hierarchy methods are calculating an index on the same basic concept of linear
dependency as the hierarchy methods presented above.
Following John M. Watts, JR. in « Fire Protection Performance Evaluation for Historic
Buildings » (Journal of FIRE PROTECTION ENGINEERING, Vol. 11, Nov. 2001):
“Indexing is a form of multi-attribute decision analysis that produces an accumulated score
of positive and negative system attributes contributing to the overall objective of an area of
concern. For example, the wind-chill factor accounts for both wind speed and temperature
to describe how cold it feels. In risk analysis, multi-attribute evaluation is usually referred to
as risk indexing. Fire risk indexing has been used for existing buildings for the last three
decades and has recently been applied to a class of historic buildings.”
The effectiveness index is calculated with the following expression:
Where G (k) is the grade of each measure M(k), M are the measures, ST the strategies, OB
the objectives and PO the policy.
IF the G (k) values are < 1, E is < 1. This the case for the two index methods presented
below. In ALADIN, we assume that the grading of a measure is included in the value given to
the measure before running ALADIN. Then E=1 in ALADIN that considers that the policy is
fully satisfied by the choice made and for which comparisons of alternatives are realised on
the outputs obtained after ALADIN’s runs.
This report describes:
♦ The index method reported in WG6 as a method of risk analysis ("Cultural heritage fire
risk index", report by Fredrik Nystedt, September 2003) but may as well be considered an
evaluation tool for DM. ANNEX 1 presents this method, which has the same basis as the
following spreadsheet.
( ) ( ) ( ) ( ) ( )∑∑ ∑= = =
⋅⋅⋅=6
1
5
1
19
1i j kkGkjMjiSTiOBPOE
FIRE-TECH WG7Final report, CSTB, May 2005
35/88
♦ The IST Cost/Effectiveness Spreadsheet developed in WG7 and applied in WG8 case
studies. The following lines come from IST (The Excel sheet “Cost_Effectiveness.xls”,I.
Cabrita Neves, Joaquim Valente, João Ventura, September 2004). The corresponding
spreadsheets are given in a separate Excel file.
Even if chapter 6 is addressing the question of optimisation of efficiency vs. cost, the way
the IST tool optimises the cost/efficiency ratio is presented here. In chapter 6 an example of
Cost/Effectiveness optimisation using this Excel sheet will be presented (from WG8 TNO
Case Study). Other WG8 Case Studies are also using this tool (see WG8 report).
5.1 Introduction
When studying the improvement of the fire safety conditions in a historic building several
alternatives will normally be considered. These alternatives need to be compared in terms of
the effectiveness of the whole set of fire safety measures in relation to the pre-defined Policy
or in relation to each pre-defined Objective. Different fire safety alternatives have different
costs. The relation between the improvement in effectiveness achieved by each alternative
and the corresponding cost (effectiveness improvement per Euro) is a crucial parameter for
the final decision.
The Excel sheet “Cost_Effectiveness.xls” was developed as a practical tool to perform that
task. A hierarchical approach is used to compare different fire safety alternatives.
5.2 Input
For a given study case, the following should be defined at first:
• Policy
• Objectives
• Strategies
• Fire safety measures
5.2.1 Policy
In general, the policy will be “to reduce the fire risk”.
FIRE-TECH WG7Final report, CSTB, May 2005
36/88
5.2.2 Objectives
The interested party in the study should be clearly identified prior to defining the objectives,
because these depend on who promotes the study.
The following 6 objectives are pre-defined in the spreadsheet:
• OB1 - Protect the occupants
• OB2 - Protect the firemen
• OB3 - Protect the building
• OB4 – Protect the contents
• OB5 - Safeguard continuity of activity
• OB6 - Protect the environment
If necessary, these objectives can be adapted or complemented by additional ones. It is
advisable to write a short description of what is understood by each of the objectives in
relation to each study case. This will help when setting the weights of each objective in
relation to the policy.
5.2.3 Strategies
The following 5 strategies are pre-defined in the spreadsheet:
• ST1 - Reduce the probability of fire start
• ST2 - Limit fire development/ propagation
• ST3 - Facilitate egress
• ST4 - Facilitate fire fighting and rescue operations
• ST5 - Limit the effects of fire products
It is advisable to write a short description of what is understood by each of the strategies and
to have these definitions always in mind when setting the weights of each strategy in relation
to each objective. The following definitions may be used as starting point and be improved,
if necessary.
FIRE-TECH WG7Final report, CSTB, May 2005
37/88
ST1 - Reduce the probability of fire start
This strategy will include every measure that may difficult the beginning of a fire.
ST2 - Limit fire development/ propagation
This strategy will include every measure that may delay the development or limit the extent of
a fire inside the compartment of fire start and also those that may prevent the fire
propagation to other compartments in the same floor, to other floors and to other buildings.
ST3 - Facilitate egress
This strategy will include every measure that may contribute to the quick and safe movement
of the occupants towards a safe place, normally outside the building.
ST4 - Facilitate fire fighting and rescue operations
This strategy will include every measure that may facilitate fire fighting at an early stage by
the occupants, by the staff or by an existing on site fire brigade and also, at a second stage,
the fire fighting and rescue operations by the firemen.
ST5 - Limit the effects of fire products
This strategy will include every measure that may limit the effects of smoke on people,
building and contents, and the effects of contaminated water on the environment.
5.2.4 Fire safety measures
The following 19 classes of fire safety measures are pre-defined in the spreadsheet:
• M1 - Reaction to fire
• M2 - Fire resistance of the structure
• M3 - Fire resistance of partitions
• M4 - Size of fire compartments
• M5 - Characteristics and location of the openings on the facades
• M6 - Distance between buildings
• M7 - Geometry of egress paths
• M8 - Access for the firemen
FIRE-TECH WG7Final report, CSTB, May 2005
38/88
• M9 - Means for fire detection
• M10 - Means for fire suppression
• M11 - Smoke control
• M12 - Emergency and alarm signs
• M13 - On site firemen
• M14 - Fire brigade
• M15 - Maintenance of fire safety systems
• M16 - Education for fire safety
• M17 - Emergency planning + training
• M18 - Salvage operation management
• M19 - Periodic inspection of the building
The measures should be adapted to each. It is advisable to write a short description of what
is understood by each class of fire safety measures and to have these descriptions always in
mind when setting the weights of each class of fire safety measures in relation to each
strategy. The following descriptions may be used as starting point and be improved/ adapted,
if necessary.
M1 - Reaction to fire
This class of fire safety measures refers to the control of the reaction to fire characteristics of
the construction materials used in the building, including structural materials, coating
materials and finishing materials, applied on interior as well as on exterior surfaces.
M2 - Fire resistance of the structure
This class of fire safety measures refers to the fire resistance of the load bearing elements.
FIRE-TECH WG7Final report, CSTB, May 2005
39/88
M3 - Fire resistance of partitions
This class of fire safety measures refers to the fire resistance of all those elements from
which it is expected an adequate performance on preventing the fire spread between distinct
fire compartments. It includes the integrity and the insulation criteria, and also, if adequate,
the load bearing criterion.
M4 - Size of fire compartments
This class of fire safety measures aims at limiting the consequences of a fire and at
facilitating the fire fighting operations. In new buildings, different floors are usually distinct fire
compartments. This is often not so in old buildings, where the whole building builds up
sometimes one single fire compartment. Correcting this in historic buildings is frequently
difficult. Compensation by other fire safety measures is then the solution. Nevertheless, in
some cases, it may be possible to improve the fire resistance of existing elements in order to
reduce the size of fire compartments.
M5 - Characteristics and location of the openings on the facades
This class of fire safety measures aims at limiting the fire spread between storeys. It includes
the distance between windows in the same vertical, the fire resistance of the window pane,
and the possible existence of balconies.
M6 - Distance between buildings
This class of fire safety measures is intended to limit the fire propagation between facing
buildings. In old urban areas this is frequently a critical factor.
M7 - Geometry of egress paths
This class of fire safety measures is intended to facilitate the quick egress of the building. It
includes the width of corridors and stairs, the length of a path from any point within the
building to a protected egress path and the number and distribution of alternative egress
paths.
M8 - Access for the firemen
This class of fire safety measures is intended to facilitate the fire fighting and rescue
operations by the firemen. It refers to the geometry of the streets leading to the entrances
and to the penetration points of the building and the eventual existence of obstacles to the
fire brigade operation.
FIRE-TECH WG7Final report, CSTB, May 2005
40/88
M9 - Means for fire detection
This class of fire safety measures has the purpose of detecting the fire as early as possible in
order to alarm the occupants, transmit the alert to the fire brigade and begin the fire fighting
operations. Examples are: detection by the occupants during day time; detection by a
watchman during the night; automatic smoke detection; automatic heat detection; automatic
flame detection; CCTV system connected to a security room (this may also be a means of
detecting suspicious behaviour of potential arsonists).
Automatic intrusion detection may also be included in this class, in the sense that intrusion
can lead to arson.
M10 - Means for fire suppression
This class of fire safety measures has the purpose of extinguishing the fire within the shortest
period of time. Examples: hose reels; automatic water sprinkler systems; portable
extinguishers; water mist systems; gaseous suppression systems.
M11 - Smoke control
This class of fire safety measures has the purpose of keeping the egress paths free from
smoke, and of limiting the effects of smoke on people, contents and building. It also aims at
facilitating the action of firemen.
M12 - Emergency and alarm signs
This class of fire safety measures has the purpose of warning the occupants about an
existing fire, to give them indications about the evacuation process and to guarantee enough
visibility during the evacuation. It includes such components as emergency lighting, voice
alarm systems, and bell and siren systems.
M13 - On site firemen
This class of fire safety measures has the purpose of initiating the fire fighting as early as
possible. On site fire brigades possess a good knowledge of the building, receive training
and are on site, which increases their effectiveness.
M14 - Fire brigade
This class of fire safety measures has the purpose of fighting the fire once it has been
detected and the alert transmitted. The time the fire brigade needs to arrive and their
equipment are important factors to consider.
FIRE-TECH WG7Final report, CSTB, May 2005
41/88
M15 - Maintenance of fire safety systems
This class of fire safety measures aims at reducing the probability of failure of the fire safety
systems. Inspection and testing of the fire safety systems should be done on a regular basis.
M16 - Education for fire safety
This class of fire safety measures is intended to increase the level of knowledge of the
occupants about the fire phenomena, making them aware of the correct behaviour in order to
prevent fire initiation and to limit their consequences.
M17 - Emergency planning + training
This class of fire safety measures is intended to anticipate every possible risk, to establish
the right procedure for every situation and to test these procedures.
M18 - Salvage operation management
This class of fire safety measures consists in defining and testing in advance the set of
adequate procedures to minimize the damage to the contents of the building. Examples are:
identification, ranking and labelling of the most valuable items; transport of valuable items to
a previously identified safe place; definition of the adequate storing conditions for each item.
M19 - Periodic inspection of the building
This class of fire safety measures aims at identifying possible sources of fire start. Examples
of fire safety measures from this class are: identification of malfunctioning devices;
identification of undue storage of objects in the egress paths; identification of unnecessary
accumulation of easily combustible materials.
For examples of general fire safety measures and of those which are more suitable for the
protection of Cultural Heritage, reference is made to the FiRE-TECH WG4 Report.
FIRE-TECH WG7Final report, CSTB, May 2005
42/88
5.3 Scores and weights
The second and very important task will be the definition of the weights (importance) of the
parameters in each level on the parameters in the level above. This can be more
conveniently achieved by attributing scores to each parameter representing their importance
in relation to each of the parameters in the level above. A scale like the one shown in table
5.1 could be used. The spreadsheet then automatically transforms these scores into
normalized weights.
Preference / importance Scores
None 0
Very little 1
Weak 3
Medium 5
Strong 7
Absolute 9
Table 5.1 - Scale for setting the scores
(See also A2.2)
The scores to be defined are:
• OB(i) - representing the importance of each objective i on the policy
• ST(ji) - representing the importance of each strategy j on each objective i
• M(kj) - representing the importance of each fire safety measure class k on each strategy j
FIRE-TECH WG7Final report, CSTB, May 2005
43/88
a) Importance of each objective i on the policy, OB(i)
Even if we keep in mind that our main concern within the scope of the FiRE-TECH project is
the protection of Cultural Heritage, an integrated approach to fire protection cannot disregard
objectives like the protection of occupants and the protection of firemen. The definition of the
scores representing the importance of each objective on the pre-defined policy is clearly a
political task, and the values depend on the study case and on whom is the party interested
in the study.
The spreadsheet transforms these scores into normalized weights OB (i) such that
( ) 16
1=∑
=iiOB (1)
OB1 - Protect the occupants
OB2 - Protect
the firemen
OB3 - Protect
the building
OB4 - Protect
contents
OB5 - Safeguard
continuity of activity
OB6 - Protect the
environment PO -
Reduce fire risk
Scores OB(i)
Normalized OB(i)
Table 5.2 – Table of scores and normalized weights OB(i)
FIRE-TECH WG7Final report, CSTB, May 2005
44/88
b) Importance of each strategy j on each objective i, ST(ji)
The choice of these scores has to be based on technical knowledge and may depend on the
study case.
ST1 -
Reduce the
probability of
fire start
ST2 - Limit fire
development/
propagation
ST3 -
Facilitate
egress
ST4 -
Facilitate
fire fighting
and rescue
ST5 - Limit
the effects
of fire
products
ST(j1) OB1 - Protect
the occupants Normalized
ST(j1)
ST(j2) OB2 - Protect
the firemen Normalized
ST(j2)
ST(j3) OB3 - Protect
the building Normalized
ST(j3)
ST(j4) OB4 - Protect
contents Normalized
ST(j4)
ST(j5) OB5 -
Safeguard
continuity of
activity
Normalized
ST(j5)
ST(j6) OB6 - Protect
the environment Normalized
ST(j6)
Table 5.3 - Table of scores and normalized weights ST (ji)
The spreadsheet normalizes these scores, giving rise to weights ST(ji) such that
( ) ( )6,115
1==∑
=
ijiSTj
(2)
FIRE-TECH WG7Final report, CSTB, May 2005
45/88
c) Importance of each fire safety measure class k on each strategy j, M(kj)
The choice of these scores has to be based on technical knowledge and may depend on the
study case.
The spreadsheet normalizes these scores, giving rise to weights M(kj) such that
( ) ( )5,1119
1==∑
=
jkjMk
(3)
The breaking of Table 5.4 (as well as Tables 5 and 6) is only due to formatting reasons.
M1 - Reaction to fire
M2 - Fire resistance of
structure
M3 - Fire resistance of
partitions
M4 - Size of fire compartments
M(k1) ST1 - Reduce the
probability of fire start Normalized
M(k1)
M(k2) ST2 - Limit fire
development/ propagation Normalized
M(k2)
M(k3) ST3 - Facilitate
egress Normalized M(k3)
M(k4) ST4 - Facilitate
fire fighting and rescue Normalized
M(k4)
M(k5) ST5 - Limit the
effects of fire products Normalized
M(k5)
Table 5.4 a) - Table of scores and normalized weights M(kj)
FIRE-TECH WG7Final report, CSTB, May 2005
46/88
M5 - Characteristics and location of
openings on the facades
M6 - Distance between buildings
M7 - Geometry of egress paths
M8 - Access for the firemen
M9 - Means for fire
detection
M(k1) ST1 - Reduce the
probability of fire start Normalized M(k1)
M(k2) ST2 - Limit fire
development/ propagation Normalized M(k2)
M(k3)
ST3 - Facilitate egress Normalized M(k3)
M(k4) ST4 - Facilitate fire
fighting and rescue Normalized M(k4)
M(k5) ST5 - Limit the effects
of fire products Normalized M(k5)
Table 5.4 b) - Table of scores and normalized weights M(kj)
M10 - Means for fire
suppression
M11 - Smoke control
M12 - Emergency and
alarm signs
M13 - On site firemen
M14 - Fire brigade
M(k1) ST1 - Reduce the
probability of fire start Normalized M(k1)
M(k2) ST2 - Limit fire
development/ propagation Normalized M(k2)
M(k3)
ST3 - Facilitate egress Normalized M(k3)
M(k4) ST4 - Facilitate fire
fighting and rescue Normalized M(k4)
M(k5) ST5 - Limit the effects
of fire products Normalized M(k5)
Table 5.4 c) - Table of scores and normalized weights M(kj)
FIRE-TECH WG7Final report, CSTB, May 2005
47/88
M15 - Maintenance of
fire safety systems
M16 - Education
for fire safety
M17 - Emergency planning +
training
M18 - Salvage
operation management
M19 - Periodic inspection of the building
M(k1) ST1 - Reduce the
probability of fire start Normalized M(k1)
M(k2) ST2 - Limit fire
development/ propagation Normalized M(k2)
M(k3)
ST3 - Facilitate egress Normalized M(k3)
M(k4) ST4 - Facilitate fire
fighting and rescue Normalized M(k4)
M(k5) ST5 - Limit the effects
of fire products Normalized M(k5)
Table 5.4 d) - Table of scores and normalized weights M(kj)
5.4 Grades of implementation of the fire safety measures
In a given study case, the existing situation will be characterized by the fact that each class of fire safety measures is already implemented to a certain degree. This implementation grade ranges from zero, when the class of fire safety measures is totally absent, to the value of one, when the class is fully and satisfactorily implemented.
In old buildings, the task of choosing some of these implementation grades is more difficult as a consequence of the difficulty in making a proper characterization of relevant properties, such as the reaction to fire of ancient materials, or the fire resistance of building components and assemblies, for instance. Nevertheless, having components with an assumed unsatisfactory fire performance is not the same as not having them at all. This has to be translated into an implementation grade, no matter how difficult it may appear. Expert judgement is very useful and the Delphi method might be a way to achieve this purpose.
Sometimes, due to the constraints in historic buildings it may be only possible to upgrade some of the components of a given system. If this is the case, an evaluation of the effect of that improvement on the resulting implementation grade has to be made on a technical basis.
FIRE-TECH WG7Final report, CSTB, May 2005
48/88
It is often necessary to weigh the global influence of a set of components of a given fire safety system, rather than weighing each individual component separately. The final grade is naturally influenced by the assumed contribution of each component. This is particularly true when evaluating the situation in old buildings where the fire safety conditions are unsatisfactory. Nevertheless, the partial contribution of each component has to be evaluated and taken into account. The fire safety level is the result of the summation of these partial contributions.
There are often classes of fire safety measures that cannot be applied or improved. Others give only small improvements. But the joint contribution of several of them can result in a lower fire risk.
The choice of the implementation grades should be adapted to each study case and be based on the existing knowledge about the suitability, performance and reliability of the various components that play a role on a given fire safety measure class. In this task, the report of FiRE-TECH Working Group 4 may be a useful tool.
The spreadsheet is prepared to receive implementation grades for an initial situation and for two additional fire safety alternatives.
Fire safety measures M1 -
Reaction to fire
M2 - Fire resistance of
structure
M3 - Fire resistance of
partitions
M4 - Size of fire compartments
M5 - Characteristics and location of
openings on the facades
Grade of implementation G(k)
Initial situation Grade of
implementation G(k) Alternative 1
Grade of implementation G(k)
Alternative 2
Table 5.5 a) - Table of implementation grades G(k)
Fire safety measures M6 - Distance
between buildings
M7 - Geometry of egress paths
M8 - Access for the firemen
M9 - Means for fire
detection
M10 - Means for fire suppression
Grade of implementation G(k)
Initial situation Grade of
implementation G(k) Alternative 1
Grade of implementation G(k)
Alternative 2
Table 5.5 b) - Table of implementation grades G(k)
FIRE-TECH WG7Final report, CSTB, May 2005
49/88
Fire safety measures M11 -
Smoke control
M12 - Emergency and
alarm signs
M13 - On site firemen
M14 - Fire brigade
M15 - Maintenance of fire safety systems
Grade of implementation G(k)
Initial situation Grade of
implementation G(k) Alternative 1
Grade of implementation G(k)
Alternative 2
Table 5.5 c) - Table of implementation grades G(k)
Fire safety measures
M16 - Education for fire safety
M17 - Emergency planning +
training
M18 - Salvage operation
management
M19 - Periodic inspection of the
building
Grade of implementation
G(k) Initial situation
Grade of implementation
G(k) Alternative 1
Grade of implementation
G(k) Alternative 2
Table 5.5 d) - Table of implementation grades G(k)
5.5 Costs associated to the implementation of each class of fire safety measures.
Once defined the fire safety measures to be implemented in each fire safety alternative, the
corresponding associated costs have to be evaluated and put in the spreadsheet.
Fire safety measures M1 - Reaction to fire
M2 - Fire resistance of
structure
M3 - Fire resistance of
partitions
M4 - Size of fire compartments
M5 - Characteristics and location of
openings on the facades
Costs Ck of implementing alternative 1 Costs Ck of
implementing alternative 2
Table 5.6 a) – Costs of implementing alternatives 1 and 2
FIRE-TECH WG7Final report, CSTB, May 2005
50/88
Fire safety measures M6 - Distance
between buildings
M7 - Geometry of egress paths
M8 - Access for the firemen
M9 - Means for fire detection
M10 - Means for fire suppression
Costs Ck of implementing alternative 1 Costs Ck of
implementing alternative 2
Table 5.6 b) – Costs of implementing alternatives 1 and 2
Fire safety measures M11 -
Smoke control
M12 - Emergency and
alarm signs
M13 - On site firemen M14 - Fire brigade
M15 - Maintenance of
fire safety systems
Costs Ck of implementing alternative 1 Costs Ck of
implementing alternative 2
Table 5.6 c) – Costs of implementing alternatives 1 and 2
Fire safety measures M16 - Education for fire safety
M17 - Emergency planning +
training
M18 - Salvage operation management
M19 - Periodic inspection of the
building
Costs Ck of implementing alternative 1 Costs Ck of
implementing alternative 2
Table 5.6 d) – Costs of implementing alternatives 1 and 2
5.6 Output
For a global view, the output sheet shows the input grades of implementation of each class
of fire safety measures for the initial situation and for the two additional fire safety
alternatives, and the corresponding costs.
FIRE-TECH WG7Final report, CSTB, May 2005
51/88
The output sheet also contains the following results:
1) the influence of each measure on the policy and on each objective, assuming the full implementation of all measures (implementation grades equal to one)
Influence of each measure on
MEASURES PO
Reduce fire risk
OB1 Protect
the occupants
OB2 - Protect the
firemen
OB3 - Protect
the building
OB4 - Protect
contents
OB5 - Safeguard
continuity of activity
OB6 - Protect the
environment
M1 - Reaction to fire
M2 - Fire resistance of
structure
M3 - Fire resistance of
partitions
M4 - Size of fire compartments
M5 - Characteristics and
location of openings on the
facades
M6 - Distance between buildings
M7 - Geometry of egress paths
M8 - Access for the firemen
M9 - Means for fire detection
M10 - Means for fire suppression
M11 - Smoke control
M12 - Emergency and alarm signs
M13 - On site firemen
M14 - Fire brigade
M15 - Maintenance of fire safety
systems
M16 - Education for fire safety
M17 - Emergency planning + training
M18 - Salvage operation
management
M19 - Periodic inspection of the
building
Table 5.7
FIRE-TECH WG7Final report, CSTB, May 2005
52/88
2) the weighted influence (effectiveness) of each measure on the policy and on each objective for the initial situation.
Initial situation – Weighed influence (effectiveness) of each measure on
MEASURES PO –
Reduce fire risk
OB1 – Protect
the occupan
ts
OB2 – Protect the
firemen
OB3 – Protect
the building
OB4 – Protect
contents
OB5 – Safeguard
continuity of activity
OB6 – Protect the
environment
M1 – Reaction to fire
M2 – Fire resistance of
structure M3 – Fire
resistance of partitions
M4 – Size of fire compartments
M5 – Characteristics and
location of openings on the
facades M6 – Distance
between buildings M7 – Geometry of
egress paths M8 – Access for
the firemen M9 – Means for fire
detection
M10 – Means for fire suppression M11 – Smoke
control M12 – Emergency and alarm signs
M13 – On site firemen
M14 – Fire brigade
M15 – Maintenance of fire safety
systems M16 – Education
for fire safety M17 – Emergency planning + training
M18 – Salvage operation
management M19 – Periodic
inspection of the building
Table 5.8
FIRE-TECH WG7Final report, CSTB, May 2005
53/88
3) the weighted influence (effectiveness) of each measure on the policy and on each objective for the fire safety alternative 1
Alternative 1 - Weighed influence (effectiveness) of each measure on
MEASURES PO -
Reduce fire risk
OB1 - Protect the
occupants
OB2 - Protect the
firemen
OB3 - Protect
the building
OB4 - Protect
contents
OB5 - Safeguard
continuity of activity
OB6 - Protect the
environment
M1 - Reaction to fire
M2 - Fire resistance of structure
M3 - Fire resistance of partitions
M4 - Size of fire compartments
M5 - Characteristics and location of
openings on the facades
M6 - Distance between buildings
M7 - Geometry of egress paths
M8 - Access for the firemen
M9 - Means for fire detection
M10 - Means for fire suppression
M11 - Smoke control
M12 - Emergency and alarm signs
M13 - On site firemen
M14 - Fire brigade
M15 - Maintenance of fire safety systems
M16 - Education for
fire safety M17 - Emergency planning + training
M18 - Salvage operation
management M19 – Periodic
inspection of the building
Table 5.9
FIRE-TECH WG7Final report, CSTB, May 2005
54/88
4) the weighted influence (effectiveness) of each measure on the policy and on each objective for the fire safety alternative 2
Alternative 2 - Weighed influence (effectiveness) of each measure on
MEASURES PO –
Reduce fire risk
OB1 – Protect
the occupants
OB2 – Protect the
firemen
OB3 – Protect
the building
OB4 – Protect
contents
OB5 – Safeguard
continuity of activity
OB6 – Protect the
environment
M1 – Reaction to fire
M2 – Fire resistance of structure
M3 – Fire resistance of partitions
M4 – Size of fire compartments
M5 – Characteristics and location of
openings on the facades
M6 – Distance between buildings M7 – Geometry of
egress paths M8 – Access for the
firemen M9 – Means for fire
detection
M10 – Means for fire suppression
M11 – Smoke control
M12 – Emergency and alarm signs
M13 – On site firemen
M14 – Fire brigade
M15 – Maintenance of fire safety
systems M16 – Education for
fire safety M17 – Emergency planning + training
M18 – Salvage operation
management M19 – Periodic
inspection of the building
Table 5.10
FIRE-TECH WG7Final report, CSTB, May 2005
55/88
5) the improvement in effectiveness (%) due to the implementation of the fire safety alternative 1
Improvement in effectiveness (%) due to implementation of Alternative 1
MEASURES PO –
Reduce fire risk
OB1 – Protect
the occupants
OB2 – Protect the
firemen
OB3 – Protect
the building
OB4 – Protect
contents
OB5 – Safeguard
continuity of activity
OB6 – Protect the
environment
M1 – Reaction to fire
M2 – Fire resistance of structure
M3 – Fire resistance of partitions
M4 – Size of fire compartments
M5 – Characteristics and location of
openings on the facades
M6 – Distance between buildings M7 – Geometry of
egress paths M8 – Access for the
firemen M9 – Means for fire
detection
M10 – Means for fire suppression
M11 – Smoke control
M12 – Emergency and alarm signs
M13 – On site firemen
M14 – Fire brigade
M15 – Maintenance of fire safety
systems M16 – Education for
fire safety M17 – Emergency planning + training
M18 – Salvage operation
management M19 – Periodic
inspection of the building
Table 5.11
FIRE-TECH WG7Final report, CSTB, May 2005
56/88
6) the improvement in effectiveness (%) due to the implementation of the fire safety alternative 2
Improvement in effectiveness (%) due to implementation of Alternative 2
MEASURES PO -
Reduce fire risk
OB1 - Protect
the occupants
OB2 - Protect the
firemen
OB3 - Protect
the building
OB4 - Protect
contents
OB5 - Safeguard
continuity of activity
OB6 - Protect the
environment
M1 - Reaction to fire
M2 - Fire resistance of structure
M3 - Fire resistance of partitions
M4 - Size of fire compartments
M5 - Characteristics and location of
openings on the facades
M6 - Distance between buildings M7 - Geometry of
egress paths M8 - Access for the
firemen M9 - Means for fire
detection
M10 - Means for fire suppression
M11 - Smoke control
M12 - Emergency and alarm signs
M13 - On site firemen
M14 - Fire brigade
M15 - Maintenance of fire safety
systems M16 - Education for
fire safety M17 - Emergency planning + training
M18 - Salvage operation
management M19 - Periodic
inspection of the building
Table 5.12
FIRE-TECH WG7Final report, CSTB, May 2005
57/88
7) the increase in effectiveness (%) per kEuro due to the implementation of the fire safety alternative 1
Increase in Effectiveness (%) per kEuro – Alternative 1
MEASURES PO –
Reduce fire risk
OB1 – Protect
the occupants
OB2 – Protect the
firemen
OB3 – Protect
the building
OB4 – Protect
contents
OB5 – Safeguard
continuity of activity
OB6 – Protect the
environment
M1 – Reaction to fire
M2 – Fire resistance of structure
M3 – Fire resistance of partitions
M4 – Size of fire compartments
M5 – Characteristics and location of
openings on the facades
M6 – Distance between buildings M7 – Geometry of
egress paths M8 – Access for the
firemen M9 – Means for fire
detection
M10 – Means for fire suppression
M11 – Smoke control
M12 – Emergency and alarm signs
M13 – On site firemen
M14 – Fire brigade
M15 – Maintenance of fire safety
systems M16 – Education for
fire safety M17 – Emergency planning + training
M18 – Salvage operation
management M19 – Periodic
inspection of the building
Table 5.13
FIRE-TECH WG7Final report, CSTB, May 2005
58/88
8) the increase in effectiveness (%) per kEuro due to the implementation of the fire safety alternative 2
Increase in Effectiveness (%) per kEuro – Alternative 2
MEASURES PO –
Reduce fire risk
OB1 – Protect
the occupants
OB2 – Protect the
firemen
OB3 – Protect
the building
OB4 – Protect
contents
OB5 – Safeguard
continuity of activity
OB6 – Protect the
environment
M1 – Reaction to fire
M2 – Fire resistance of structure
M3 – Fire resistance of partitions
M4 – Size of fire compartments
M5 – Characteristics and location of
openings on the facades
M6 – Distance between buildings M7 – Geometry of
egress paths M8 – Access for the
firemen M9 – Means for fire
detection
M10 – Means for fire suppression
M11 – Smoke control
M12 – Emergency and alarm signs
M13 – On site firemen
M14 – Fire brigade
M15 – Maintenance of fire safety
systems M16 – Education for
fire safety M17 – Emergency planning + training
M18 – Salvage operation
management M19 – Periodic
inspection of the building
Table 5.14
FIRE-TECH WG7Final report, CSTB, May 2005
59/88
9) for the initial situation, for the fire safety alternative 1 and for the fire safety alternative
2, the influence of the implemented fire safety measures on the policy (Effectiveness
index E(PO)), and the influence of the implemented fire safety measures on each
objective (Effectiveness index E(OBi)), taking into account the relative contribution of
that objective to the policy.
( ) ( ) ( ) ( ) ( )∑∑∑= = =
⋅⋅⋅=6
1
5
1
19
1i j kkGkjMjiSTiOBPOE (4)
( ) ( ) ( ) ( )∑∑= =
⋅⋅⋅=5
1
19
1)(
j k
kGkjMjiSTiOBOBiE (5)
MEASURES PO –
Reduce fire risk
OB1 – Protect the occupants
OB2 – Protect the
firemen
OB3 – Protect the
building
OB4 – Protect
contents
OB5 – Safeguard
continuity of activity
OB6 – Protect the
environment
E(PO), E(OBi)
Initial situation
E(PO), E(OBi)
Alternative 1
E(PO), E(OBi)
Alternative 2
Table 5.15
10) for the fire safety alternatives 1 and 2, the global ratio Effectiveness/ Cost in relation to
the policy (increase in effectiveness obtained by each fire safety alternative in relation
to the policy per kEuro spent)
1
19
1
01 )()(
−
∑=k
kC
POEPOE (6)
2
19
1
02 )()(
−
∑=k
kC
POEPOE (7)
FIRE-TECH WG7Final report, CSTB, May 2005
60/88
11) for the fire safety alternatives 1 and 2, the ratio Effectiveness/ Cost in relation to each
objective (increase in effectiveness obtained by each fire safety alternative in relation to
each objective per kEuro spent).
1
19
1
01 )()(
−
∑=k
k
ii
C
OBEOBE (8)
2
19
1
02 )()(
−
∑=k
k
ii
C
OBEOBE (9)
MEASURES PO –
Reduce fire risk
OB1 – Protect the occupants
OB2 – Protect the
firemen
OB3 – Protect the
building
OB4 – Protect
contents
OB5 – Safeguard
continuity of activity
OB6 – Protect the
environment
Ratio Effectiveness/ Cost
Alternative 1
Ratio Effectiveness/ Cost
Alternative 2
Table 5.16 – Ratios Effectiveness/ Cost
For a better visualization of the results, a graphical representation of the output data can be
found in separate sheets.
The worksheets are write-protected without password, so that the user can always adapt
them to his personal needs.
FIRE-TECH WG7Final report, CSTB, May 2005
61/88
6. OPTIMISATION of COST /EFFECTIVENESS
6.1 ALADIN APPROACH
The example of CSTB WG8 case study (see § 4) will still be considered here.
We will here continue the assessment of the measures by an evaluation of the reliability of
the systems (from the data of the WG 4 report).
The following Table 6.1 presents the acceptability (already considered in the choice of the
measures), the reliability (considered now) and the cost (very approximate here), that must
be less than 300 k€.
Table 6.1
FIRE-TECH WG7Final report, CSTB, May 2005
62/88
The following Table 6.2 presents the weights of the alternatives on the safety/protection of the virtual museum.
Alternative 1 Alternative 2 Alternative 3
{LocDetect} on {FS in Museum} 0,34 0,25 0,31
{Sprinklers} on {FS in Museum} 0,34 0,00 0,03
{Water mist} on {FS in Museum} 0,00 0,25 0,26
{SmokeDetn} on {FS in Museum} 0,21 0,10 0,17
{Curtain} on {FS in Museum} 0,00 0,25 0,00
{Staff} on {FS in Museum} 0,12 0,14 0,22
1,00 1,00 1,00
Table 6.2
Table 6.3 presents the new weight after reliability has been considered. The global reliability is: 0.89, 0.86, 0.82, and does thus not vary much. The cost has to be lower than 300 k€. Then alternative 2 is not acceptable. Alternative 1 is much cheaper than alternative 3, with a slightly higher reliability. It has then to be chosen.
Reliability Alternative 1 Alternative 2 Alternative 3
{LocDetect} on {FS in Museum} 1,00 0,34 0,25 0,31
{Sprinklers} on {FS in Museum} 0,80 0,27 0,00 0,03
{Water mist} on {FS in Museum} 0,60 0,00 0,15 0,16
{SmokeDetn} on {FS in Museum} 0,90 0,19 0,09 0,15
{Curtain} on {FS in Museum} 1,00 0,00 0,25 0,00
{Staff} on {FS in Museum} 0,80 0,10 0,11 0,18
0,89 0,86 0,82 [Costmax - Cost (i)] / Costmax 0,54 0,00 0,25 Cost(i) 169,15 366,65 276,15
Costmax 366,65 366,65 366,65
Table 6.3
This example shows how: - the cost,
- a further (after ALADIN) assessment of the alternative (here reliability), can permit to choose the best ratio efficiency/cost.
More complex tools can also be used to compare alternatives, based on over-ranking methods (Bernard Roy et al., cf. ANNEX 1).
FIRE-TECH WG7Final report, CSTB, May 2005
63/88
6.2 Application made by TNO using the first version of the IST COST/EFFECTIVENESS spreadsheet
(The latest version of the IST spreadsheet incorporates Effectiveness / Cost ratios and graphs, as a consequence of the applicaton made by TNO)
Different measures can be taken to obtain a fire safe situation in the Nieuwe Kerk. As was shown in the WG8 report by TNO (par. 6.4), these measures can be ranked in order of importance for the overall fire safety. Using the modified IST excel sheet, the importance of each measure is given in Figure 6.4
Weights of the measures on PO (Reduce fire risk)
0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100
P1 - Sprinklers
P2 - First aid fire fighting equipment
P3 - Automatic fire detection
P4 - Alaram systems
P5 -Visual signals and evacuation plans
P6 - Smoke control
P7 - Fire resistance of glazing
P8 - Inert insulating materials
P9 - Intumescent materials
P10 - CCTV
P11 - Training of personnel
P12 - Procedures for evacuation of people
P13 - Fire guards during large events
P14 - Guides accompanying vistors to tower
P15 - Control of installations
P16 - Guidelines during renovation
P17 - Burglary alarm
P18 - Limit unnecessary flammable items
P19 -Contact with fire services
Weights
full level of implementation
Figure 6.4 effectiveness of each measure if fully implemented
In the current situation only some measures have been (partially) implemented. Each measure gives a certain contribution to the overall effectiveness index (EI) with its current grade of implementation. The current contributions to the overall EI are compared to the maximum values in figure 6.5. The green values again represent full implementation of a measure, the red value the current level of implementation (G).
FIRE-TECH WG7Final report, CSTB, May 2005
64/88
Weights of the measures on PO (Reduce fire risk)
0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100
P1 - Sprinklers
P2 - First aid fire fighting equipment
P3 - Automatic fire detection
P4 - Alaram systems
P5 -Visual signals and evacuation plans
P6 - Smoke control
P7 - Fire resistance of glazing
P8 - Inert insulating materials
P9 - Intumescent materials
P10 - CCTV
P11 - Training of personnel
P12 - Procedures for evacuation of people
P13 - Fire guards during large events
P14 - Guides accompanying vistors to tower
P15 - Control of installations
P16 - Guidelines during renovation
P17 - Burglary alarm
P18 - Limit unnecessary flammable items
P19 -Contact with fire services
Weights
current level of implementationfull level of implementation
Figure 6.5: Contributions of all measures to EI, for current situation and situation with full implementation of all measures
However, to increase the grade of implementation of a certain measure does not necessarily lead to G=1 for that measure; possibly a measure can relatively easily be improved to e.g. G=0.9, while obtaining G=1.0 would require large investments. Therefore, for each measure an estimation has been made of a feasible grade of implementation. (This estimate only serves as an example, and has not been verified!) In figure 6.6 this proposed new grade of implementation is shown in yellow and compared to current and full grades of implementation.
FIRE-TECH WG7Final report, CSTB, May 2005
65/88
Weights of the measures on PO (Reduce fire risk)
0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100
P1 - Sprinklers
P2 - First aid fire fighting equipment
P3 - Automatic fire detection
P4 - Alaram systems
P5 -Visual signals and evacuation plans
P6 - Smoke control
P7 - Fire resistance of glazing
P8 - Inert insulating materials
P9 - Intumescent materials
P10 - CCTV
P11 - Training of personnel
P12 - Procedures for evacuation of people
P13 - Fire guards during large events
P14 - Guides accompanying vistors to tower
P15 - Control of installations
P16 - Guidelines during renovation
P17 - Burglary alarm
P18 - Limit unnecessary flammable items
P19 -Contact with fire services
Weights
current level of implementationproposed new level of implementationfull level of implementation
Figure 6.6 current, proposed new and maximum grades of implementation of all measures
Weights of the measures on PO (Reduce fire risk)
0.000 0.010 0.020 0.030 0.040 0.050 0.060
P1 - Sprinklers
P2 - First aid fire fighting equipment
P3 - Automatic fire detection
P4 - Alarm systems
P5 -Visual signals and evacuation plans
P6 - Smoke control
P7 - Fire resistance of glazing
P8 - Inert insulating materials
P9 - Intumescent materials
P10 - CCTV
P11 - Training of personnel
P12 - Procedures for evacuation of people
P13 - Fire guards during large events
P14 - Guides accompanying vistors to tower
P15 - Control of installations
P16 - Guidelines during renovation
P17 - Burglary alarm
P18 - Limit unnecessary flammable items
P19 -Contact with fire services
Weights
possible increase of EI if measureis implemented as proposed
Figure 6.7: possible increase in EI due to full or partial implementation of each measure
FIRE-TECH WG7Final report, CSTB, May 2005
66/88
Assume that there is a certain (given) budget available to improve the fire safety of the Nieuwe Kerk. Objective is to increase the effectiveness index as much as possible with this money.
All costs have to be indexed to net market values for the time of investment. For each measure it is estimated what the costs of an improvement would be, and what the new grade of implementation (G) would be after the improvement. The estimated costs are given in Table 6.3. (This estimate only serves as an example, and has not been verified!)
Table 6.3: estimated costs of improvements
measure Current grade of
implementation
Proposed new grade of
implementation
cost of improvement
[k€]
P1 - Sprinklers 0.8 1.0 150
P2 - First aid fire fighting equipment 0.7 1.0 25
P3 - Automatic fire detection 0.5 0.7 25
P4 - Alaram systems 0.8 1.0 25
P5 -Visual signals and evacuation plans 0.1 0.8 50
P6 - Smoke control 0.8 0.8 -
P7 - Fire resistance of glazing 0.7 0.7 -
P8 - Inert insulating materials 0.3 0.6 75
P9 - Intumescent materials 0.4 0.5 25
P10 - CCTV 0.5 1.0 100
P11 - Training of personnel 0.9 0.9 -
P12 - Procedures for evacuation of people 1.0 1.0 -
P13 - Fire guards during large events 0.0 1.0 50
P14 - Guides accompanying vistors to tower
0.2 1.0 200
P15 - Control of installations 0.3 1.0 50
P16 - Guidelines during renovation 1.0 1.0 -
P17 - Burglary alarm 0.8 0.8 -
P18 - Limit unnecessary flammable items 1.0 1.0 -
P19 -Contact with fire services 1.0 1.0 -
The model calculates the effect that the increase of G has on the total EI; therefore it is possible to relate the costs of each measure to the obtained overall effectiveness index.
FIRE-TECH WG7Final report, CSTB, May 2005
67/88
This gives a ranking as follows, see also figure 6.8:
- P13 - Fire guards during large events 0.11 %/k€ (this means that if money is invested in P13, for each 1000 € the EI will increase 0.0011).
- P15 - Control of installations 0.089 %/k€ - P2 - First aid fire fighting equipment 0.062 %/k€ - P3 - Automatic fire detection - P5 -Visual signals and evacuation plans - P4 - Alarm systems - P14 - Guides accompanying visitors to tower - P8 - Inert insulating materials - P9 - Intumescent materials - P10 - CCTV - P1 - Sprinklers
The other measures are either already fully implemented or impossible to improve.
Costs effectiveness per measure (improvement of EI per invested kEuro)[%/kEuro]
0.0000 0.0200 0.0400 0.0600 0.0800 0.1000 0.1200
P1 - Sprinklers
P2 - First aid fire fighting equipment
P3 - Automatic fire detection
P4 - Alaram systems
P5 -Visual signals and evacuation plans
P6 - Smoke control
P7 - Fire resistance of glazing
P8 - Inert insulating materials
P9 - Intumescent materials
P10 - CCTV
P11 - Training of personnel
P12 - Procedures for evacuation of people
P13 - Fire guards during large events
P14 - Guides accompanying vistors to tower
P15 - Control of installations
P16 - Guidelines during renovation
P17 - Burglary alarm
P18 - Limit unnecessary flammable items
P19 -Contact with fire services
Cost effectiveness [%EI / kEuro]
Figure 6.8: cost effectiveness of each measure
Figure 6.8 can be compared to figure 6.7. This shows that the cost effectiveness calculation changes the ranking of measures, compared to a ranking purely on possible improvement per measure.
If the available money is invested according to the above ranking, the increase of the effectiveness index is always maximum that can be achieved for a certain budget. In figure 6.9 is shown, which EI can be achieved if the budget is invested according to the ranking.
FIRE-TECH WG7Final report, CSTB, May 2005
68/88
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
0 100 200 300 400 500 600 700 800
costs (kEUR)
Effe
ctiven
ess Inde
x
none
P13
P13 + P15P13 + P15 + P2
P13 + P15 + P2 + P3
P13 + P15 + P2 + P3 + P5
… + P4 … + P14
… + P8… + P9 … + P10
… + P1
Figure 6.9 optimum EI dependent in available budget
E.g. if a budget of 200.000 € is available, the most cost-effective option is to install the measures P13 - Fire guards during large events, P15 - Control of installations, P2 -First aid fire fighting equipment, P3 - Automatic fire detection and P5 -Visual signals and evacuation plans. The EI will then be increased from 0.666 to 0.805.
It is interesting to see what benefits can be gained from this method. If it is assumed that costs are not taken into account, then the priorities for investing in fire safety measures would probably be based on the possible increase of the overall EI due to each measure. This means that the order of investing is according to figure 6.7 instead of figure 6.8. In that case, the relation between invested budget and obtained EI would be as shown in figure 6.10.
A close comparison of figures 6.9 and 6.10 shows that there are substantial benefits to be gained by a cost effectiveness assessment. If the aim is an EI of 0.8 the required investment can be lowered from 300 k€ to 200 k€. If the aim is an EI of 0.85 the amount goes down from 575 k€ to 425 k€.
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
0 100 200 300 400 500 600 700 800
costs (kEUR)
Effe
ctiven
ess In
dex
Figure 6.10: EI dependent on available budget without cost effectiveness assessment
FIRE-TECH WG7Final report, CSTB, May 2005
69/88
7. Conclusions on Help to Decision-Making
Optimisation of fire safety and fire protection in Cultural Heritage implies the examination of possible alternative technical and organisational actions and, of course, an evaluation of associated costs, in order to retain acceptable solutions. The method of Help to Decision-Making used must be able to produce a selection of the best alternative(s) or a ranking of all the alternatives.
WG7 has developed and applied an approach presenting the following characteristics:
• "Help to Decision-Making" can address any aspect of safety or protection the expert is thinking of.
• The decision problem is decomposed in several “possible actions” at different depths of analysis, from global goals to precise actions.
• The method gives a general network view of inputs and outputs. It quantifies the effectiveness of assessments and expert’s opinion. It assigns weights (importance of influence) to different possibilities relative to one another.
• Identification of the most promising alternatives can be done solving a mathematical programming problem.
• The developed tools are simple to use and not time consuming.
The DM approach retained in FIRE-TECH is based on the “Hierarchy Analysis” that uses a decomposition of the main goal in “sub goals” at several levels “below“ the general goal top level. At each level of the “generality scale”, a judgement is made on the relative efficiency of each component of this level on each component of the levels above; several “boxes” corresponding to possible actions are weighted. From very general to less general (policy, objectives, strategies and measures), the weights given express the influence of a row on the actions of each box of the upper row. We assume that the influences are linear. Then matrix or linear combination formulae are used to calculate the dependency between boxes. The outputs are the relative influence of all the components are calculated so that the best solutions (the best alternatives) can be identified and weighted or a ranking be established.
The DM approach is one step in the global study of a Cultural Heritage and is strongly connected to others tasks. This step is located in phase 4, if:
• Phase 1 is the preliminary phase (qualitative definition of final goal awaited, evaluation of time and money available …),
• Phase 2 correspond to the analyse of the present level of safety and protection of the cultural heritage (Prescriptive and expert analysis, Performance based Quantitative Risk Analysis),
FIRE-TECH WG7Final report, CSTB, May 2005
70/88
• Phase 3 is the collection of possible actions to improve safety and protection (technical actions, non-technical actions…). Phase 3 gives the alternatives to be assessed and compared.
The following two-DM tools have been developed and applied to several cases (Cf. the WG8 report). The applications made have permitted to test the tools and to develop ways to take cost into account as well as the effectiveness aspects of the alternatives.
ALADIN
ALADIN is a FORTRAN programme based on the basic principles of Hierarchy Analysis. It leads to a ranking of alternatives. The use is very simple because the programme is asking simple questions about the network to build and the values to give. The results are the calculated influences. ALADIN can be used to calculate e.g. the combined weights on policy and objectives of given strategies and measures, as well as to compare the combined weights of several alternatives (established by one hesitating person or by a group of several experts).
ALADIN follows the decomposition of levels presented on: “policy, objectives, strategies, measures”. ALADIN was not conceived to deal with many measures and/or many strategies, as the maximum number of boxes in a row is 6 and the numbers of rows is limited to five.
ALADIN can address e.g. the following situations:
• Discussion and negotiation between experts have already limited the set of possible alternatives. The final choice to be made concerns then a few “boxes”.
• “Zooming” on partial influences, e.g. in using it on a small network, e.g.: one given objective, some strategies, and a few measures.
• A decision matrix is wished. The results will be presented under the format of such a matrix. Table 4.5 gives an example of a decision matrix with 3 alternatives in columns. The criteria (e.g. objectives or measures) are in rows. Additional rows can be added to consider new “criteria” such as cost (in order to consider cost/effectiveness), or reliability (if these criteria have not been considered already in the previous weighting). This matrix can be exploited directly or with the help of overranking methods.
The IST COST/EFFECTIVENESS spreadsheet
Several hierarchy methods are calculating an index on the same basic concepts of linear dependency as in ALADIN. Several specific features of the IST Cost/Effectiveness spreadsheet have to be mentioned:
FIRE-TECH WG7Final report, CSTB, May 2005
71/88
• In the IST spreadsheet a global risk index is calculated for each alternative considered because each measure is given a grade of implementation. Another index method is presented in Annex 3.
• The IST tool is based on EXCEL. Many actions are proposed at different levels. Each action not retained receives a weight zero. Retained actions are given relative weights. The utilisation is very simple and histograms give a graphic view of the results.
• The initial state (before any improvement of the protection and safety to the Cultural Heritage) is evaluated and two alternatives can be evaluated simultaneously and then compared to the initial state and compared to each other.
• The relation between the improvement in effectiveness achieved by each alternative and the corresponding cost (effectiveness improvement per Euro) is taken into account. The spreadsheet calculates the cost of each combination of actions and authorises the user to choose the best ratio effectiveness/cost for 2 alternatives or to look for a new alternative leading to a better ratio.
Final comments
The concepts developed here for fire protection and fire safety could be applied to other questions concerning Cultural Heritage. We are nevertheless missing experience since the use of such methods and tools is still very recent. We hope that the two tools presented above will be useful for the success of practical actions and that the experience gained will bring fresh knowledge to enrich future approaches.
FIRE-TECH WG7Final report, CSTB, May 2005
72/88
8. REFERENCES
• Buchanan, John and Sheppard, Phil, Ranking Projects Using the ELECTRE Method, 23 June 1998.
• Bukowski, Richard W.; Nuzzolese, Vincenzo; Bindo, Mirella, « Performance-based Fire Protection of Historical Structures » 20 October 2003.
• FIRE-TECH - WG7 draft report, Presentation of the fire-tech method to help to decision-making in fire protection of cultural heritage, Michel Curtat, February 2004).
• Karlsson, Björn and Larsson, Daniel, “Using a Delphi panel for developing a Fire Risk Index Method for Multi-Storey Apartment Buildings”, Report Lund University, 2000.
• Krick, E.V., “An Introduction to Engineering and Engineering Design”, second edition, John Wiley & Sons, Inc., New York, 1969.
• Magnusson, Sven Erik and Rantatalo, Tomas, Lund University Internrapport 7004, 1998: "Risk Assessment of Timberframe Multi-storey Apartment Buildings. Proposal for a Comprehensive Fire Safety Evaluation Procedure"
• Roy, Bernard, Méthodologie Multicritère d’aide à la Décision, Ed. Economica, 1985.
• Saaty, Thomas Lorie, “The Analytic Hierarchy Process – Planning, Priority setting, Resource allocation”, 1980.
• Shields, J., and Silcock, G., “An Application of the Hierarchical Approach to Fire Safety”, Fire Safety Journal, n°11, 1986.
• Shields, J., Silcock, G. and Y. Bell, “Fire Safety Evaluation of Dwellings”, Fire Safety Journal, n°10, 1986.
• Watts John M., JR. “Fire Protection Performance Evaluation for Historic Buildings”, Journal of FIRE PROTECTION ENGINEERING, Vol. 11, Nov. 2001.
• Watts, John M., Jr., “Fire Risk Indexing,” Section 5, Chapter 10, SFPE Handbook of Fire Protection Engineering, National Fire Protection Association, 3rd ed., Quincy, MA, 2001
• Zhao, C.M., Loa, S.M., Lu, J.A. , and Fang, Z., “A simulation approach for ranking of fire safety attributes of existing buildings”, Fire Safety Journal 39 (2004) 557–579.
FIRE-TECH WG7Final report, CSTB, May 2005
73/88
ANNEX 1. Other DM approach: B. Roy et al. Several methods and tools (ELECTRE, PROMETHEE, GAIA, …) have been developed by
Bernard Roy et al. at the University of Paris and applied and enriched by a ” French school of
DM”. Matrices again are considered: alternatives are on the rows, criteria on the columns. A
basic concept is outranking. Outranking is less concerned with a method for applying
weights to attributes, and more with a holistic comparison of Alternative A to Alternative B.
Roy's utilizes both concordance and discordance measures for accomplishing this. The
concordance measure is a ratio computed by summing the weights for those attributes for
alternative A which are superior to the attributes for Alternative B divided by the weights for
Alternative A as a whole. The closer this ratio is to 1.0, the more superior Alternative A is to
Alternative B. The discordance measure looks the largest difference for the attribute sets of A
over B compared to the largest difference over all alternatives. Multiple Criteria Decision
Aiding with the invention of the family of ELECTRE methods and methodological
contributions to decision aiding that lead to the creation of DM teams.
PROMETHEE is an example of method produced. In the PROMETHEE method, the user
has not to perform a large number of comparisons, as he can directly use the data of the
problem in a simple multicriteria table. The decision-maker has to define his own scales of
measure (without limitation), to indicate his priorities and his preferences for every criterion.
PROMETHEE allows conducting easily and quickly analyses of sensibility. PROMETHEE
calculates too the robustness of the current classification for each criterion.
It is very difficult to compare the awaited advantages and drawbacks of the two methods
(AHP and PROMETHEE). On one side, programming AHP calculations seems lighter, on the
other side PROMETHEE and other closed approaches look more appropriate when the
number of alternatives is higher than a few ones.
These overranking methods can anyway be used for studies as the one presented in § 6.1
by application of the method to an assessment of “distances” between alternatives and
associated costs.
FIRE-TECH WG7Final report, CSTB, May 2005
74/88
ANNEX 2. AHP (Analytic Hierarchy Process)
In the following, we will present AHP (hierarchy method founded by Saaty) and other close
hierarchy methods.
A2.1 Concepts of AHP
Saaty’s Analytic Hierarchy Process is a practical technique for modelling and solving multi-
criteria decision problems. AHP is a logical and structural framework, which allows improving
the performance of complex decisions by decomposing the problem in a particular
hierarchical structure. The incorporation of all the decision criteria retained, and their pairwise
comparison allows the decision maker to determine the trade-offs among objectives.
The application of the AHP method exploits two kinds of knowledge in the process of priority
setting to conceive the “boxes” of the network as well as to fix the weights:
• information gained from the expertise of the participants (subjective judgements, the regulations texts, knowledge from previous fires),
• measured (from tests when available) or calculated information (modelling).
AHP was first developed by Professor Thomas L. Saaty in the 1970’s and since that time has
received wide application in a variety of areas.
The first famous document on AHP is the book by Saaty (1980), “The Analytical Hierarchy
Process”. This book gives:
• a detailed presentation of the concepts of hierarchy, priority and judgement,
• the theoretical and mathematical foundations of the method,
• several examples of applications (e.g. to resource allocation, planning, conflict resolution …).
AHP was initially designed as a decision analysis tool to:
• Facilitate the development of logical hierarchical structures for complex decisions
• Quantify the opinions of subject area experts
• Assign weights to different functional areas relative to one another
• Assign weights to different technical capabilities relative to one another.
FIRE-TECH WG7Final report, CSTB, May 2005
75/88
In the classical Saaty’s AHP, decision makers supply pair-wise comparisons of various alternative “factors” on a ratio scale. These pair-wise comparisons capture decision maker’s subjective judgements and help in ranking a set of alternatives.
The AHP approach is based on three principles:
• Decomposition of the decision problem,
• Comparative judgement of the elements, and
• Synthesis of the priorities.
The pairwise comparison judgement is incorporated in forced reciprocal matrices. The
technique is based on matrix algebra, eigenvalues and eigenvectors.
A2.2 Theory and Methodology of AHP
A2.2.1 Weighting influences
Experts of the domain are asked to give measured or calculated weights, if it is meaningful
possible! The method compares, in a pairwise way, elements on the same level of the
hierarchy - for their influence on a box in a higher level -, indicating which element in the pair
is preferred and by what degree. Comparison of elements in pairs with respect to each
element of the next higher level is made on the basis of the scale of the following table A2.1.
Table A2.1. Scale for AHP weights
Numerical values
Equal preference 1
Weak preference 3
Preference 5
Very strong preference 7
Absolute preference 9
Intermediate values between the stated preferences
2, 4, 6, 8
FIRE-TECH WG7Final report, CSTB, May 2005
76/88
We need to ensure that the numbers given in the boxes are all on the same scale; otherwise
the weighted importance of the criteria won’t matter. To ensure the same scale is considered,
we normalize the numbers.
By combining all above information, we can finally get composite priorities for each of the
decision alternatives.
A2.2.2 Construction of matrices by pair-wise comparison
A simple example of hierarchy with three rows and a few boxes is represented at figure A2.2.
We assume that the experts have to estimate the weights with their own judgement (with no
measurement or calculation). E.g., at level 3, the most « precise » level in this example, the
weights of each strategy (Str 1, Str 2 and Str 3) has to be given for each objective (OB 1 and
OB 2). Let us consider the objective OB1. The following matrix is constructed on the basis of
ratios based on experts’ opinions, where all the possible pairs at row 3 are considered:
Figure A2.2. Example of a simple AHP structure.
(The thicker the arrow, the stronger the influence.)
STR 1 STR 2 STR 3
OB 1 OB 2
Policy
FIRE-TECH WG7Final report, CSTB, May 2005
77/88
The following matrix is constructed on the basis of ratios based on experts’ opinions, where
all the possible pairs at row 3 are considered:
Awaited properties:
aii = 1 ; i = 1, 3
aij = 1/aji ; i = 1, 3
A.r = 3×r, where r is a vector (r1, r2, r3)
In real world, some inconsistency may appear, e.g. aij can be more or less different of 1/aji
because these reports come from experts’ evaluation that may be not fully consistent. Saaty
introduced a “consistency vector”, a “consistency ratio, and a “consistency index” in order to
quantify these deviations and to be able to decide if they are acceptable or not. Saaty’s book
and other papers are presenting these concepts and the associated calculations. To give
here a brief presentation on the mathematical aspects off these concepts, we have taken the
following lines from the web pages of Dublin University on AHP where one can find a
numerical example (several examples are given too in Saaty’s book):
A2.2.3 Consistency Ratio (C.R.)
A consistency vector can be calculated by: calculating the row averages for the normalised
matrix to get weights multiplying the original matrix by the weights to get the product dividing
the product by the weights to get the consistency vector.
SAATY defines the consistency index (C.I.) as:
Where λ is the average value of the consistency vector, and n is the number of potential
decisions being compared.
3
3
2
3
1
3
3
2
2
2
1
2
3
1
2
1
1
1
rr
rr
rr
rr
rr
rr
rr
rr
rr
A =
1..
−−=nnIC λ
FIRE-TECH WG7Final report, CSTB, May 2005
78/88
For each matrix of size n, random matrices were generated and their mean C.I. value, called
the random index (R.I.), was computed. Using these values, the consistency ratio (C.R.) is
defined as the ratio of the C.I. to the R.I.; this C.R. is the measure of how a given matrix
compares to a purely random matrix in terms of the C.I.’s.
Here are some R.I. values for a given matrix dimension n:
n 2 3 4 5 6 7 8 9 10
R.I. 0 0.58 0.90 1.12 1.24 1.32 1.41 1.45 1.51
A value of the C.R. < 0.1 is typically considered acceptable; larger values require the
decision-maker to reduce the inconsistencies by revising judgements.
The following gives a short description of another method: “Hierarchical Cross-Impact
Analysis”, from [DONNEGAN 2002].
A2.3 Other AHP method: Hierarchical Cross-Impact Analysis
HCIA methodology
“Hierarchical Cross-Impact Analysis (HCIA) [DONEGAN et al.] is a multiattribute weighting
strategy developed at the University of Ulster using a similar philosophy to the Edinburgh
approach. According to DONEGAN, the characterisation of HCIA addresses the Cross-
Impact approach at two levels: a “fundamental” level and a “pragmatic” level.
In the fundamental level, a pseudo axiomatic approach is taken to:
- Define a hierarchy (in Fire safety),
- Explain the meaning of interactive importance,
- Propose the notion of partial impact,
- Define a global impact,
- Introduce sequential perturbations.
FIRE-TECH WG7Final report, CSTB, May 2005
79/88
In the pragmatic level a modelling approach is taken to:
- Maintain the assessment interval through each stage of the quantification
- Stretch the component ranking intervals to enhance the important components and decrease the psychological significance of the less-important components
- Perturb consensus weighting with interaction noise.
The partial impact of tactic Tj on component Ci relative to the rth objective Or is written as:
( )rOji TCI /∂ = ( )( )OrTTC iji //
The total impact of the collection of tactics on Ci relative to Or is defined as:
( ) ∑∑==
=∂nj
jrijnj
Oji tTCIr
,1,1/ σ ”
which is a matrix product. The components-to-objectives are defined by:
∑=
=nj
jrijri tn
OC,1
1/ σ
A drawback in HCIA is the absence of consistency check.
A2.4 Previous Applications of AHP to Fire Safety and Fire Protection
J. SHIELDS and G. SILCOCK presented in 1986 the AHP the applicability potential and
practicability of the AHP for use in safety planning and for determining fire safety priorities.
AHP was compared to an “Evaluation Points Scheme” for hospitals using a Delphi technique
(MARCHANT1984). The authors emphasised the fact that AHP takes into account all the
influences (on a given row and between rows) while the Evaluation Points Scheme
interactions and comparisons between elements on the same level are ignored. The practical
application the authors were thinking of is Fire Safety in dwellings and public assembly
buildings.
FIRE-TECH WG7Final report, CSTB, May 2005
80/88
They defined the following levels and “boxes”:
• Policy: Fire Safety.
• Objectives: Life Safety, Property Protection.
• Tactics: Ignition Prevention, Fire Control, Safe Egress, Rescue.
• Components: Occupants, Doors, Communications, Internal Planning, Travel Distance, Flues Ducts.
Further applications of AHP concepts and methods seem to have been carried out under the
names of “Fire Risk Evaluation” and “Fire Risk Index”.
A2.5 Research under work in AHP Methods
Research on the definition of better scale than 1,…, 9: Saaty’s scale may bring problems
because of the lack of steps, for example, between 1 and 2. It has been shown that, for
example, the use of the 99/1 scale (Ma and Zheng, 1991) or the balanced scale (Salo and
Hämäläinen, 1997) can give better results (Pöyhönen et al., 1997a).
Fuzzy logic can be used to deal with the Imprecision in Specification of Pair-wise
Comparisons on Ranking of Alternatives using Fuzzy Analytic Hierarchy Process (B.
Narasimha, Web Site of City University Hong Kong).
Genetic algorithms and other recent concepts are applied to hierarchy approach.
More generally, the concepts and methods in Decision-Making are having now such an
important place in management sciences that several kinds of applied research are trying to
bring together in newer tools the advantages of old or newer concepts.
FIRE-TECH WG7Final report, CSTB, May 2005
81/88
ANNEX 3. Cultural Heritage Fire risk index method
("Cultural heritage fire risk index", report by Fredrik Nystedt, September 2003)
The method is described in several reports and papers mainly from Nordic countries and UK, e.g. in : “Fire Risk Index Method – Multi-storey Apartment Buildings”, 2000, by Björn Karlsson and the report “Using a Delphi panel for developing a Fire Risk Index Method for MultiStorey Apartment Buildings” by Björn Karlsson and Daniel Larsson.
The paper by Sven Erik Magnusson and Tomas Rantatalo, Internrapport 7004, 1998: "Risk Assessment of Timberframe Multistorey Apartment Buildings. Proposal for a Comprehensive Fire Safety Evaluation Procedure", presents a method to verify the overall or general safety, taking into account structural fire endurance, design of fire protection systems, fire brigade capabilities as well as organisational and management factors. This is being achieved by means of an index system. The following lines come from this report.
The system is based on a hierarchy of fire safety decision-making levels (policy, objectives, strategies, parameters, survey items) with each level, e.g. objectives, being described in terms of the quantities defining the immediate lower level. The final result is a risk index describing overall fire safety policy as a function of 13 fire safety parameters. To prove the feasibility of the approach the methodology has been applied to fire risk ranking of multi-storey apartment buildings with wood frame and a frame of non-combustible material respectively. At least at this preliminary stage, the method looks very promising. One important aspect is that in practical use, the risk index calculation is extremely simple. The risk index assessment method is called FiRESEM.
The method used to produce the global or overall fire risk index mentioned above is generally called "multi-attribute evaluation". It is one of the most common and most powerful heuristic decision-making techniques, and an approach supported by a large body of knowledge described in the literature of decision analysis and management science. Multi-attribute evaluation is used to develop simplified but robust models of complex systems. Values are assigned to important attributes of the problem based on professional judgement and experience. These values are then operated on by some combination of arithmetic functions to arrive at a single score or index. The result is compared with other assessments or to a standard.
The second step of the safety evaluation scheme is the verification with respect to three individual fire safety and fire protection objectives:
provide life safety,
prevent fire spread out of compartment of origin,
prevent fire spread to next building.
FIRE-TECH WG7Final report, CSTB, May 2005
82/88
An obvious and appealing alternative for the evaluation of the three individual fire safety objectives is to employ successively the relevant part of the overall risk ranking system described above. The project team took the standpoint that a totally independent system of evaluation procedures was to be preferred. A number of well-developed and recognised risk assessment systems have therefore been reviewed with respect to their practical application to the individual fire safety objective. These systems include the FIRECAM method, the analytical FOSM (first order second moment) method developed at LTH, the L-curve method, the GRETENER method; the CRISPII-method developed at FRS, etc. To the extent that it has been natural and appropriate, more general engineering methods such as deterministic hazard calculation, application of the t-equivalent concept for structural fire endurance, the use of checklist procedures, etc have been reviewed with respect to practical usability.
A3.1. A Fire Risk Index for Cultural Heritage
An important contribution for FIRE-TECH is the recent report by Fredrik Nystedt (September 2003): "Cultural heritage fire risk index". A large part of this report is included in the present paragraph.
A3.1.1 Development of a risk index method
The first step is to identify the hierarchical levels. Table A3.1 shows a five-level hierarchy that has been used in many different applications (Watts, 1995).
Level Name Description 1 Policy Course or general plan of action adopted by an organisation to
achieve safety against fire and its effects 2 Objectives Specific fire safety goals to be achieved 3 Strategies Independent fire safety alternatives, each of which contributes
wholly or partly to the fulfilment of fire safety objectives 4 Parameters Components of fire risk that are determinable by direct or
indirect measure or estimate 5 Survey
items Measurable feature that serves as a constituent part of a fire safety parameter
Table A3.1. Decision making levels
The policy for fire safety in cultural heritage buildings could be that the level of fire safety should meet the requirements in the building regulations or that the safety level should show equivalence with other buildings in the same category. One objective could by “life safety”, with a strategy “to provide safe egress” with a parameter “detector system” and its survey item “type of detectors”.
FIRE-TECH WG7Final report, CSTB, May 2005
83/88
A3.1.2 Generalised procedure
A generalised procedure for ranking fire safety parameters to determine their relative importance is summarised in the following four steps (Watts, 1995).
Identify hierarchical levels of fire safety specification
Specify items compromising each level
Construct and assign values to matrices of each subsequent pair of levels
Combine matrices to yield importance ranking of items.
An example of step 1 is presented in Table A3.1. Step 2 requires lists of objectives, strategies parameters and survey items to be developed. The list of fire safety objectives might include statements on life safety, property protection, environmental protection, continuity of operations and heritage preservation. Strategies used to comply with the objectives are derived. Such strategies could involve ignition prevention, compartmentation, fire detection, fire suppression, etc. It is now possible to construct a matrix of objectives versus strategies, as shown in Table A3.2
Table A3.2. Matrix of objectives vs. strategies (example)
How important is each strategy to achieve each objective? In order to allow for mathematical computations, the values in the matrix can be normalised. If the policy vector is multiplied with the objectives/strategies vector, a new vector that shows the overall importance of each strategy to the fire safety policy. As the procedure continues the next step is to derive parameters for fire safety. Parameters as construction, height, compartmentation, building services, furnishings, equipment, special hazards, detection, alarm, smoke control, fixed suppression, fire department, egress system, personnel and management should be considered. A matrix of strategies versus fire safety parameters are constructed and evaluated. It is now possible to derive a vector illustrating the overall importance of each parameter to the fire safety policy.
Prevent ignition
Compartmentation Fire suppression
Life safety 2 1 2
Property protection
2 3 4
FIRE-TECH WG7Final report, CSTB, May 2005
84/88
Delphi work panel
As the risk index method allows the user to have moderate knowledge on the fundamentals of fire development and risk analysis, the development of the method must be done with state-of-the-art techniques. Many risk index methods were developed by the use of a Delphi work panel, as described by Linstone & Turoff (1975). They defined the Delphi technique as “a method for structuring a group communication process so that the process is effective in allowing a group of individuals, as a whole, to deal with a complex problem". In short, the technique is applied when a small monitor team designs a questionnaire that is sent to a larger respondent group. When the questionnaire is returned the monitor team summarises the results and, based upon the results, develops a new questionnaire for the respondent group. The respondent group is usually given at least one opportunity to re-evaluate its original answers based upon examination of the group response.
The task of the Delphi work panel is to assign weights to all objectives, parameters and strategies. As the process moves along, the answers from the Delphi panel could be used to gradually develop and improve the structure of the index method. Through matrix multiplication of the weights, a relative weight for each Parameter was derived, which allows a single index value for each building to be arrived at. This single index value can then be used for comparison with a similar building.
A3.2 Fictive example cultural heritage risk index method.
Policy
The policy is ensuring an acceptable level of property protection in order to preserve cultural heritage.
Evaluation of life safety of occupants is not included in this method, as it should be covered by adopting the national building regulations.
Objectives
There is one objective in the suggested method, i.e., to provide property protection in the compartment of origin, rest of the building and adjacent buildings.
Strategies
The objective of property protection is fulfilled by the following strategies:
Control fire growth Def. Controlling fire growth by using suppression systems, smoke control and the fire service (including staff response).
Confine fire by construction Def. Provide structural stability, control the movement of fire through containment, use of fire safe materials (passive fire protection).
FIRE-TECH WG7Final report, CSTB, May 2005
85/88
Establish safe rescue Def. Ensure safety of staff and fire service when they carry out rescue operations (fighting the fire, removing valuable objects, etc.)
It is necessary to value which importance a specific strategy in order to assign weights. This is done in a objectives vs. strategy matrix, shown below.
Strategy Objective:Property protection Normalised
Control fire growth 4 4/(4+2+2) = 0.5
Confine fire by construction 2 0.25
Establish safe rescue 2 0.25
Parameters
The following parameters are identified to play the most important role in fulfilling strategies.
Suppression system
Detection system
Smoke management
Compartmentation
Staff / fire service
A matrix on strategy vs. parameters is constructed and shown below.
Parameters Control fire growth
Confine fire by construction
Establish safe rescue
Suppression system 5 (0.36) 5 (0.3.1) 4 (0.27)
Detection system 3 (0.21) 3 (0.19) 3 (0.20)
Smoke management 2 (0.15) 3 (0.19) 4 (0.27)
Compartmentation 1 (0.07) 2 (0.12) 3 (0.20)
Staff / fire service 3 (0.21) 3 (0.19) 1 (0.06)
AUTOMATIC SPRINKLER SYSTEM
Type of sprinkler (N = no sprinkler, R = residential sprinkler, O = ordinary sprinkler) and Location of sprinkler (A = in apartment, E = in escape route, B = both in apartement and escape route)
FIRE-TECH WG7Final report, CSTB, May 2005
86/88
SURVEY ITEMS DECISION RULES
type of sprinkler N R R R O O O
location of sprinkler - A E B A E B
Grade N M L H M L H
(N = no grade, L = low grade, M = medium grade and H = high grade)
PORTABLE EQUIPMENT
N None
F Extinguishing equipement on every floor
A Extinguishing equipement in every apartment
PARAMETER GRADE
SUB-PARAMETERS DECISION RULES
Automatic sprinklers
system
N N N L L L M M M H H H
Portable equipment N F A N F A N F A N F A
GRADE 0 0 1 1 1 2 4 4 4 5 5 5
(Minimum grade = 0 and maximum grade = 5)
FIRE-TECH WG7Final report, CSTB, May 2005
87/88
Sub-parameters
Each parameter is divided into sub-parameters in order to make possible to assign a specific grade (0-5) to the parameter. Here is an example with the parameter “suppression system”.
Calculating weights
The information in the matrices is used to derive a new matrix where on policy vs. parameters. The complete process is described in the WG6 report.
Parameter Weights to fulfil the policy
Suppression system 0.5x0.36 + 0.25x0.3.1 + 0.25x0.27 = 0.325
Detection system 0.5x0.21 + 0.25x0.19 + 0.25x0.20 = 0.203
Smoke management 0.5x0.15 + 0.25x0.19 + 0.25x0.27 = 0.190
Compartmentation 0.5x0.07 + 0.25x0.12 + 0.25x0.20 = 0.115
Staff / fire service 0.5x0.21 + 0.25x0.19 + 0.25x0.06 = 0.167
Figure A3.3. Example of a complete process
Calculating risk index
The risk index is calculated with the following expression:
1
n
i ii
R w g=
=∑
where iw is the weight for a specific parameter and ig is its grade.
FIRE-TECH WG7Final report, CSTB, May 2005
88/88
A3.3 Conclusions
It would be useful to have a cultural heritage fire risk index method available. The index method could be used to give a quick and rough estimate of the fire performance of a large quantity of buildings. The National Property Board could use the risk index to rank which buildings that have the best and the poorest fire safety. However, the method cannot be used to show the exact level of risk, as it very difficult to establish which level of the risk index that is considered acceptable. The risk index will only tell how safe the building is compared to others, i.e., a relative measure of safety. If it is of interest to have a quantified measure of the fire safety, a quantitative risk analysis is preferred. The QRA should also be used when different fire safety alternatives are evaluated. The risk index method identifies the cultural buildings where a more extensive analysis of their fire performance is needed. It is, however, possible to use the risk index to evaluate to which parameter a change is needed to improve the overall fire safety performance.