Building and Environment 109 (2016) 121e134

Contents lists available at ScienceDirect

Building and Environment

journal homepage: www.elsevier.com/locate/buildenv

Decision support to choose renovation actions in order to reduce

house energy consumption e An applied approach
Franck Taillandier*, Laurent Mora, Denys Breysse
Univ. Bordeaux, I2M, UMR 5295, F-33400 Talence, France

a r t i c l e i n f o a b s t r a c t

Article history: This paper focuses on improving the thermal performances of a set of houses built around 1970 and aims
Received 3 May 2016 to develop a method for decision support regarding the choice of renovation solutions. The approach
Received in revised form goes through two steps based on a multicriteria analysis. The first step allows the owner to choose a
13 September 2016
work package that meets his expectations at best. The second step is to specify the selected works by
Accepted 14 September 2016
providing more detailed technical solutions, like defining the type of insulation (glass wool, sheep wool,
Available online 15 September 2016
etc.). The solution assessment on the various criteria is performed according to models, expert values or
rules set by the experts involved in the project. One of the major challenges of the decision tool is to
Keywords:
Decision support
enable a merging of the project stakeholder points of view, based on a deep interaction between experts
Energy performance and owners. The output is not limited to the prescription of a renovation solution but enriches the house
House renovation owner reflection in order to guide her/him into the whole renovation process. The article presents the
Multicriteria decision-making approach, the method and an application to a house.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction
Construction industry is a key sector for environment: in France
it represents 44% of energy consumption and 21% of total CO2
emissions [6]. In parallel, it is responsible for about half of the
extraction of primary resources, a third of the water consumption
and a third of the waste of the European Union [12]. Since the
annual replacement rate of the building is less than 1% [36],
reducing the energy consumption of the French construction sector
requires the renovation of existing buildings. Several barriers were
identified by a European ERABUILD project [23] for the renovation
of existing building by private owners: the lack of knowledge and
information, the lack of cost effectiveness and of funding. In order
to go beyond these barriers, in addition to current regulation and
State proposals, local initiatives are required. Since 2006, a neigh-
bourhood association (“Des fourmis dans le compteur”, meaning
‘Ants in the meter’, where Ants refers to a well-known LaFontaine
poem in which this animal carefully saves its resources) gathers
house owners in order to share information and provide some help
for the energy renovation of housing in Malartic district at Gradi-
gnan (near Bordeaux, France). The district consists of seven
* Corresponding author.
E-mail address: franck.taillandier@u-bordeaux.fr (F. Taillandier).
hundred houses that have been built at low cost in the 70s and are
thermally deficient. Most of the inhabitants of these houses are also
their owners. This paper focuses on improving the thermal per-
formances of these houses by providing a framework and a method
for supporting decisions regarding the choice of renovation solu-
tions. In the following, a renovation solution is defined as a set of
work actions (e.g. increasing wall insulation, changing the win-
dows, etc.). The paper will expose the work carried out to develop a
methodology for decision support. The issue is not only to propose
a functional tool answering the decision problem but to propose a
global decision support approach allowing the association to
continue its works in autonomy. It requires to be able to identify, for
a given house, the most relevant renovation solution but also to
enclose the tool in a more global approach which can be continued
by association co-operators without the help of experts.
The renovation project selection is a complex decision problem:
there is a large number, even an infinity, of possible renovation
solutions depending on how these solutions are defined, there are
many possible criteria for the selection of a solution, many un-
certainties and constraints, the preferences of the owner can widely
vary, etc. [33] identified 43 decision support tools to select reno-
vation project in design or pre-design phase. These tools compose
an interesting corpus, relevant for many renovation decision sup-
port contexts. However, only few of these tools are really usable for
others case studies than those developed in the papers, because of

http://dx.doi.org/10.1016/j.buildenv.2016.09.019
0360-1323/© 2016 Elsevier Ltd. All rights reserved.
122 F. Taillandier et al. / Building and Environment 109 (2016) 121e134
insufficient availability of these tools or lack of updating of their
databases [33]. Furthermore, few of these tools are adapted to the
specific context of this work, i.e. the renovation of houses in France.
This list of tools can be completed by different existing methods/
approaches which have been proposed in the literature, as expert
system [46], optimization [18,24,55], cost-benefit analysis [15], etc.
These approaches, although interesting, do not answer completely
the decision support issue. Indeed, many of them share the same
drawbacks: it considers (a) only new buildings, without the pos-
sibility to integrate renovation constraints, (b) a single criterion
(building energy, financial criterion, etc.), and it does not take into
account (c) the owner preferences and constraints and (d) un-
certainties. While newly-built houses can be seen as a homoge-
neous asset, it is not the case with a set of 40-years old houses, each
of which having developed some specificities with time (partial
renovation, extension …). The decision support tool must take ac-
count of these differences and propose tailor-made solutions to
each owner.
These methods and tools may be very useful on a technical point
of view, but have a major limitation which decreases their interest
in a real context: they do not integer the owner/inhabitants in a
global process, from the first wish of renovation to the everyday life
after the works. Yet, numerous studies demonstrate that it exists an
important gap between forecast technical assessment on energy
consumption after renovation and real consumption [60] indeed,
the resident behaviour has a strong impact on the energy con-
sumption [61], is highly variable and can evolve after the renova-
tion [62]. The resident has to be placed at the centre of the low-
energy renovation strategy in order to ensure its efficiency. In an
idealistic perspective, the strategy should act simultaneously on
technical system and on owner behaviour, in order to reduce en-
ergy consumption. Different behavioural approaches have been
developed to predict and/or improve resident pro-environmental
behaviour/practices; for instance, Attitude-Behaviour-Context
(ABC) [63], the theory of Interpersonal Behaviour [64], the
Motivation-Opportunity-Ability (MOA) [65], etc. But their real ef-
ficiency for durable behaviour change are discussed: one major
drawback is the focus on individual and the weak consideration of
the context [66,67]. Using a strategy which considers not only in-
dividual, but a more global improvement of the low-energy culture
and an action on broader social structures, appears as more effi-
cient than an individual-focused approach [18,68] suggests to use
community projects, dialogue between owners and different types
of intermediaries and of social learning processes. It is in this
perspective this work takes place, by including the developed tool
in the global renovation process and in direct relation with the local
association, especially as the tool must be usable by the association
in autonomy. The process has to imply directly the house owner
before and after the renovation action, to give her/him the relevant
information [69], to guide her/him during the whole process and to
structure a community sharing a low-energy culture.
In order to reach this objective, we built a tool called ARD-
FOURMI. It was designed as a support for analysis, information
sharing and decision. Structured around a mechanical multicriteria
decision-making tool, the proposed approach includes owners in
order to increase their understanding and their desire for low-
energy renovation actions and to allow them to share knowledge
with others. Furthermore, we assume that this approach could
reinforce their pro-environmental behaviour and improve their
energy use related practices ensuring a better energy gain after the
works. Indeed, the tool is enclosed in a more global process aiming
at developing knowledge, awareness and control over the renova-
tion process and to develop a low-energy culture.
The next parts will detail how the ARD-FOURMI tool works.
2. Method and tool
2.1. Approach
The proposed approach aims at bringing together experts from
different fields and inhabitants of Malartic while considering their
expectations, and constraints as a major item influencing what can
be the “best” solution. Instead of identifying a unique solution
which could be proposed for the whole building asset, the decision
support tool should be able to (a) provide relevant case-specific
renovation solutions, (b) explicitly incorporate the preferences of
the house owners and (c) propose an approach and a result which
can be easily understood by the owner. These three expectations
resulted in an AGILE software development methodology [1]: the
final users, i.e. the Malartic inhabitants, were involved in the tool
development process through an iterative process for the con-
struction of decision strategy which incorporates numerous ex-
changes with the various stakeholders (Fig. 1). Since it is not
economically feasible for each owner to meet all types of experts (in
thermal insulation, in alternative energy, in building material
characterization), our common choice has been to set a team of
experts who could provide a common and multi-view analysis.
The approach uses a knowledge base which has been built from
the literature, the expert team and the Malartic inhabitants.
Knowing that a unique elicitation technique cannot work for all
situations [20], several methods to collect the knowledge were
used depending on the kind of knowledge and its sources (litera-
ture, expert and user):

for literature: Data and information come from classical reading
of thesis, articles, and project reports.

for expert: To acquire knowledge from experts, a set of semi-
directed interview was done, as recommended by Bernard
(2010) [70] or Davis (2006) [71]. In addition, questionnaires
were used to complete database (as example for insulation
material characteristics). Several experts contributed to the tool/
approach: a thermal design consultant, researchers in building
energy and environment and in building structure, an architect
and a construction economist. These experts were all involved
in the research project FOURMINERGIE which supports the
development of the ARD-FOURMI tool. They were selected ac-
cording to their expertise level, to their interest in the project
and to their motivation and availability to work on the project.

for user (Malartic inhabitants): To acquire operational data/in-
formation, a combination of questionnaires and semi-directed
interviews were used in a first step. Questionnaires are suit-
able to reveal trends by enabling to have results from a large
number of stakeholders. Semi-directed interviews are required
for in-depth analysis of practices. These two approaches are
complementary. Furthermore, a set of real case study experi-
ments was performed in order to improve the tool. The involved
stakeholders were FOURMINERGIE co-operators: the board
members, a set of representative co-operators (chosen from an
initial meeting with the board members), and additional
members of the association whose play the guinea pig role
while testing the tool.
Furthermore, several meetings were organized gathering ex-
perts, association board members and inhabitants in order to allow
a direct communication. In the same way, some of the experts were
involved in the real case study experiments allowing them to un-
derstand the functioning and the interest of the tool as a thinking
support. For instance, the set of criteria is based on many round
trips between stakeholders to confront the opinions, preferences
and objectives of the owners with the ability of experts to provide
 F. Taillandier et al. / Building and Environment 109 (2016) 121e134 123

Fig. 1. Approach to design tool.

relevant data. The process used to obtain the set of criteria is
developed in part 2.3.
The ARD-FOURMI tool combines a spreadsheet interface
(Excel®) which is already known and used by most of the co-
operators, and a calculation engine in Java. This choice was made
according the will to ensure utility and user acceptance [10] of the
tool and of the global approach. The next sections will expose the
functioning of the method and will argue regarding the educational
dimension.
2.2. Defining solutions for MACRO to MICRO levels
The chosen decision strategy uses two successive steps (Fig. 2),
respectively named MACRO and MICRO, both being based on a
multiple-criteria decision analysis taking into account the
uncertainties both on the data and on the models in order to obtain
robust results.
The first step consists in helping the owners to choose the
MACRO solution that best meet their expectations. A MACRO so-
lution is a set of works which remains defined and described at a
generic level (Table 1). For ARD-FOURMI, the use of a consistent set
of works as defining an alternative was privileged better than
comparing directly works as alternatives. Indeed, the use of set of
works as a solution prevents from a combinatorial explosion for the
number of possible alternatives. The strategy consisting in building
all possible solutions would result in a huge number of possible
solutions: a solution could be to isolate just one wall; another, could
differ only by the thickness of the insulating layer, etc. However,
among all these theoretical solutions, many are not relevant,
because impracticable in the real world (e.g. the insulation is often

Fig. 2. Decision support process.

124 F. Taillandier et al. / Building and Environment 109 (2016) 121e134

Table 1 solution recommends insulation, the MICRO solution considers

List of proposed works. various options regarding the insulation material. The insulation
Id Description choice was explicitly demanded by the Malartic inhabitants.
V1 Mechanical ventilation
Indeed, the interviews revealed that many of them were curious
V2 Humidity sensitive mechanical ventilation (HSMV) about the impact of the insulation choice and notably the interest to
V3 Heat recovery ventilation select a natural insulation material. Thus, if the MACRO phase uses
W1 Wall insulation from outside only an aimed U-value for insulation, the MICRO phase enables to
W2 Wall insulation from inside with briquette removing
consider different insulation materials. The current tool database
R1 Roof insulation from outside
R2 Roof insulation from inside considers 16 different types of insulation (Table 3). This list was
R3 Self-supported roof insulation from inside, built considering what is possible according to the geographic area,
D1 Replacement of all the original windows and doors the will of the FOURMI co-operators and the experts' propositions.
D2 Replacement of all the windows and doors
In a further development of the tool, the specification will also
F1 Peripheral floor insulation
F2 floor insulation

Table 3
available according to defined thickness) or inefficient (e.g. List of proposed insulation for MICRO solutions.

isolating the floor and not the walls and the roof has only a poor Solution Description
interest). The role of experts was to design first a series of relevant 1 Polystyrene - Panel
and consistent solutions (i.e. of work packages). These solutions 2 Glass wool - Roll
were built following a classical approach where the priority goes to 3 Rockwool - Roll
main weaknesses or easily treatable problems before incrementing 4 Fibreboard - Flexible panel
5 Fibreboard - Rigid panel
the first solution by adding more side work.
6 Cellulose wool - Instilled loose
ARD-FOURMI proposes a list of 27 different MACRO solutions 7 Cellulose wool - Poured loose
(Table 2), which were developed by project experts using consis- 8 Cellulose wool - Panel
tency and efficiency rules and literature analysis [2,34,38]. 9 Cork - Loose
The second step consists in specifying the selected MACRO so- 10 Cork - Panel
11 Hemp wool - Roll
lutions to achieve refined solutions called MICRO. Each element of 12 Hemp wool - Panel
the selected MACRO solution is developed into different alterna- 13 Linen wool - Roll
tives to build a new set of solutions. In the current tool develop- 14 Linen wool - Panel
ment stage, the MICRO solutions are only developed according to 15 Sheep wool - Roll
16 Sheep wool - Panel
the insulation material, i.e. for each surface for which MACRO

Table 2
List of proposed MACRO solutions (works set).
 F. Taillandier et al. / Building and Environment 109 (2016) 121e134
cover others domains like the definition of windows or ventilation.
These two steps are based on the same decision method:
ELECTRE III in a stochastic framework [40,47].
2.3. Defining criteria
Defining the set of criteria is a major issue in multicriteria de-
cision problems. The choice of criteria is based on the combination
of literature analysis, expert opinion and inhabitant will. The
starting point is the will to follow a sustainable development logic.
The criteria were thus identified according this objective and
belong to the three pillars (environmental, economic and social) of
sustainable development. Table 4 exposes the whole set of criteria
(finalized set) for the two phases and the corresponding literature
references.
The defining of criteria results from an iterative process as
exposed in part 2.1. More precisely, it goes through the following
steps:
1. Set main objectives, i.e. sustainable perspective.
2. Semi-directed interviews with a set of 15 co-operators proposed
by board members in order to understand their current prob-
lems, their expectations and what seems important for them
regarding the main objectives.
3. Crossing the interview results with literature in order to
establish a first list of criteria.
4. Meeting with experts in order to have their opinion on the
criteria list and their capacity to assess them. There resulted in a
modified list of criteria.
5. Meeting with co-operators to explain the chosen criteria and
use of a questionnaire to obtain their opinion on them (with a
first assessment of preferences between criteria). The list of
criteria was updated.
6. New discussion with experts to assess their capacity to assess
these new criteria, with a final list of criteria.
7. Experimentation: during experimentation, some criteria have
been discussed. One was added (administrative complexity) in
order to consider an aspect which seems important for the
owner and another reformulated; discomfort was split up into
summer and winter comfort in order to be more specific
regarding the owner issue.
The next issue is to assess the different criteria. The assessment
process is exposed in part 2.4.
2.4. Assessment in MACRO phase
2.4.1. Environmental criteria
The aim of the tool is to compare solutions in order to reduce
Table 4
List of criteria.
Domain MACRO phase
Environment Energy efficiency improvement [31]; [35]; [24]; [29]; [7]
Economic Investment cost [16]; [5,59]; [35]
Annual saving on energy [16]; [53,54]; [59]
House value gain
Social Summer discomfort [35]
Winter discomfort [35]
Discomfort during the works period [35]
Contiguity and its constraints
Opportunity [35]
Administrative complexity
125
house energy consumption. All solutions are not equivalent in
terms of potential energy efficiency improvement; the expected
improvement is thus a major criterion. The evaluation of this cri-
terion can be difficult. The thermal performance simulation models
are complex and require a lot of data on houses (nature and con-
dition of materials, air permeability of the envelope, etc.). There-
fore, it was not possible to simulate all Malartic houses to assess the
gain for each of them and for each possible solution. A specificity of
the study must be pointed: when new buildings commonly follow
many common features, it is not the case in existing old housing. In
this case, each house has followed a specific 40-year history, with a
variety of owners and many houses have been modified (exten-
sions, partial renovations …) which makes each case specific. On
one hand the tool must be able to account for this variety (a solu-
tion which would propose to change a component which has been
recently renovated will be difficult to accept) while on the other
hand, it is impossible to have a detailed picture of each house, with
all details about parameters that influence the thermal perfor-
mances of the house. It poses thus a double problem: the avail-
ability of the information and the calculation time. To avoid this
problem, ARD-FOURMI uses a strategy based on a prior typological
analysis of the Malartic houses. The houses of the Malartic area are
divided into three families, with some possible constructive vari-
ants in each family (Fig. 3):
Castille: This type is in L-shape in three versions of respectively
4-, 5- or 6-room. The 4-room version is only on first floor. The two
other versions have an upper floor on a part of the L.
Alicante: It has a classical parallelepiped shape with an upper
floor. It exists in 5 or 6-room versions. Two other versions called
Bermudes and Bale
ares are similar to Alicante but without a
garage-extension on the house side and with different equipment.
Cerac: This type has T-shape. It is slightly more recent than the
two other types. It differs in terms of room arrangement and in
terms of construction materials. It exists into two versions which
differs on the rooms size.
In addition to house type, significant differences exist because
of: (a) house orientation modifying sunlight input, (b) contiguity
with neighbouring houses, (c) works already done, (d) behaviour
and expectations of the owners. A sample of six houses which are
variables regarding these items were selected and have been the
subject of extensive investigation and measurement program; they
are called reference houses. The assessment of the energy perfor-
mance of the solution has been carried out according to the refer-
ence houses and to a typological analysis. Considering the
typological proximity of the different houses of the Malartic
neighbourhood, this sample was considered sufficient in a first
approach. This energy consumption criterion does not consider the
owner behaviour. Indeed, considering the variability of the
behaviour, it is difficult to assess relevantly the future real energy
MICRO phase
Embodied energy [3,53]
Global warming potential [58]; [50]; [32]; [44]
́ Toxicity during the work phase [30]
Long-term toxicity [28]; [45]; [30]
Additional cost [3]; [26,39]
Lost surface [25]; [52,56]
Fire performance [28]; [39,45]; [44]
Durability [39]
126 F. Taillandier et al. / Building and Environment 109 (2016) 121e134
Fig. 3. Three house types of Malartic.
consumption [42]; a typological analysis would require a very
larger sample of owner to be able to make accurate projection. In
this work, the behaviour of the owner is considered separately from
the building performance (the energy consumption criterion con-
siders only this aspect). The behaviour of the owner is considered
through preferences on criteria and social aspects (part 2.4.3).
The thermal simulations were performed with
PLEIADES þ COMFIE [37]. The comparison between simulated
performances of the house in its present condition and simulated
performances of the house with each MACRO solution was quan-
tified in terms of relative gain. This allowed to limit the difference
between reference and other houses regarding the difficulty to fit
the thermal model because of the lack of accurate input data
regarding the actual performance, while an extensive measure-
ment program is not feasible. In order to take into account the
uncertainties according to simulation results and to typological
approach, an uncertainty factor has been introduced. The assess-
ment is obtained by building a range of values from the simulated
value and the uncertainty coefficient.
The simulated energy gains range from 6% (for S2: HSMV) to 70%
(for S26: global renovation integrating external insulation of walls
and roof, double flow ventilation and windows and doors
replacement). Given the characteristics of the houses, it is techni-
cally difficult and economically unrealistic to propose solutions
enabling to achieve the French 4-factor (i.e. the division by four of
greenhouse gas emissions on a period of 40 years). However, the
proposed solutions are upgradable: if an owner chooses a low cost
solution because of his limited budget, he can later upgrade this
solution with additional works for a more efficient renovation, and
the future works will preserve the consistency of the improvement.
The criteria related to the environmental impact of materials, in
a logic of life cycle analysis (embodied energy, potential climate
change, etc.) are not used at this level. Indeed, MACRO solutions do
not specify the materials used and simply refer to aimed U-value for
insulation. These criteria will appear in the MICRO phase (Section
2.4.1).
2.4.2. Economic criteria
Three economic criteria are used: the investment cost, the
annual saving on energy and the house value gain. The investment
cost is the monetary value required to implement the solution. It is
evaluated combining the size parameters of the house and unit
prices. These unit prices were assessed by a construction economist
who has a long-term expertise in house renovation. The calculated
price is an average price corresponding to a rough estimate of the
solution cost. To take into account the uncertainty, the model in-
troduces an uncertainty parameter to build two values for the in-
vestment cost respectively corresponding to lowest and highest
costs.
The annual energy saving calculation is based on two elements:
the amount of saved energy (the same as the one used by envi-
ronmental criteria) and the energy price. The energy price is not a
constant value and is expected to increase in future years. Several
scenarios of energy price evolution based on different assumptions
can be found in literature [11,13]. ARD-FOURMI integrates two
scenarios, which are considered as realistic extreme scenarios
(optimistic and pessimistic). The model also takes into account the
discount rate to assess energy price in constant euros. The annual
gain is calculated over a time horizon (between 1 and 100 years)
defined by the user. The time horizon depends on whether the user
prefers the short or long term. ARD-FOURMI proposes a micro-
economic vision, centred on the owner perspective, and not a
global vision that would consider global gains for the population.
The time horizon is a subjective value depending on the age of the
owner, on his will to stay for a long time in this house, on his house
purchasing credit, etc. The value used in ARD-FOURMI corresponds
to the average of the gain on the time horizon (annualized average).
The economic gain is defined by a range between the value corre-
sponding to the optimistic energy price scenario (minimum value)
and that corresponding to the pessimistic energy price scenario
(maximum value).
The renovation works will allow to increase the value of the
house (higher resale value for a more energy-efficient house). ARD-
FOURMI compares the value of the house without solution on a
given horizon with that of the house with the implemented solu-
tion. A linear depreciation of the house value based on the life-time
of the house is assumed. When performing a work, the value of this
work adds to the house value. If the works consist in element
(insulation, windows, etc.) replacement and if this element has a
positive residual value considering its life-time and its age, its re-
sidual value is subtracted to the house residual value. For instance,
considering the works ‘Replacement of all the windows and doors’
the residual value is calculated by adding the cost of the new
windows and door and by subtracting the residual value of the
former windows and door. The model also takes into account the
loss of living space resulting for example from an inside insulation
 F. Taillandier et al. / Building and Environment 109 (2016) 121e134
which may induce a decrease of the house value. Consequently,
some solutions can lead to decrease the house value. As for other
criteria, an uncertainty coefficient is defined to evaluate the lowest
and highest values for this criterion.
The FOURMINERGIE team, combining arguments and knowl-
edge from experts in thermal simulations, the construction econ-
omist and owners, has decided to not consider the return time as a
criterion whereas it is a common criterion in building energy de-
cision support. Indeed, for the Malartic houses, the return time is
relatively high (10e50 years) and is thus not incentive for owners at
individual scale.
2.4.3. Social criteria
ARD-FOURMI tool considers five social criteria: summer
discomfort, winter discomfort, discomfort during the works period,
contiguity and its constraints, opportunity and administrative
complexity.
The summer discomfort is the number of days during which the
temperature inside the house exceeds a threshold value. This cri-
terion is evaluated in the same way than the energy gains from the
thermal simulation coupled with typological analysis. Winter
comfort is not evaluated quantitatively as summer comfort but
qualitatively: indeed, the thermal simulation is based on a defined
minimum temperature (model hypothesis), assuming that it en-
sures the comfort of house inhabitant. However, several owners
have mentioned during interviews winter discomfort which has
many sources (cold walls, temperature heterogeneity, lack of air-
tightness, etc.). This discomfort could be a main reason to
perform works and has to be considered in the decision support
process. To assess it, each owner must fill a questionnaire to analyse
if she/he encounters a winter comfort problem, and if yes, what is
its source. From this inventory, each solution is assessed qualita-
tively by the expert team in terms of gain regarding this issue. For
instance, the windows replacement improves winter comfort if the
source of discomfort is the feeling of cold near the windows.
Therefore, solutions integrating this task will have a positive
assessment regarding this criterion.
The discomfort of the work, contiguity issues and opportunity
criteria are assessed from answers to a questionnaire, drawn up
with an architect and designed to the owner. The contiguity crite-
rion corresponds to problems with some type of works regarding
semi-detached house. It appears mainly with external works
(external insulation for example), which cannot be carried out
independently on two contiguous houses. It depends on the kind of
work, on the relation between neighbours (bad relations can lead to
more problems) and to the works the neighbour has already done
or will do. The opportunity criterion considers the opportunity to
do works in relation with the house condition. As an example, if
electricity is not up to standards, add an interior insulation can
provide the opportunity to renovate the electric network;
conversely, if the interior painting is new, it would not be inter-
esting to remake a renovation. Works can also induce some
discomfort, whose level depends on the type of works, their
duration and the owner behaviour. The evaluation of the discom-
fort of the work involves crossing the objective impact of the works
and the subjective ability of owners to accept this impact. The
possibility of the owner to move to another place during work time
can reduce the discomfort and change the value of this criterion.
Administrative complexity reflects the administrative proced-
ure which has to be done with work. For some work packages, it
may be necessary to make many administrative demands. For
example, making work on the roof requires the permission of the
local authority; it goes through assembling files to justify the me-
chanical strength of the house structure with the new loads and the
aesthetic/urban integration of the chosen roof coating. The file
127
assembling can be complex and/or time consuming. In addition, in
some cases, there is a risk of refusal of the work from local
administration.
2.5. Assessment in MICRO phase
The criteria used in this second phase are different, as the ele-
ments underlying the choice. Refined information about the solu-
tions is available at this stage, namely regarding the materials.
However certain criteria are no longer useful because they do not
allow to compare solutions. For example, opportunity value is
considered to be identical for all kinds of insulation and can no
longer be a decision criterion.
2.5.1. Environmental criteria
The environmental criteria in MICRO phase can be divided into
two families, those related to the material life cycle (the embodied
energy and global warming potential), and those related to the
toxicity of the materials used (toxicity during the work phase and
long-term toxicity during the life of the house). Although all ma-
terials are in conformity with technical approvals, many owners
wanted to clarify the toxicity issue of insulation material. These two
toxicity criteria were then added to the criteria set.
The first two criteria, embodied energy and global warming
potential enter a life cycle logic [14,22]. They are measured from
INIES databases [9] and KBOB [27] indicating the value of these
criteria (normalized to one m3 of material) for each solution type.
ARD-FOURMI computes the global value of these criteria consid-
ering the actual amount of the material used in house (according to
the established house quantity survey). Extreme values for these
two criteria are the minimum and maximum values found in the
databases. Values also depend on the product and the manufac-
turer and can then vary for a same insulation material.
The criteria related to toxicity were evaluated by the experts of
the project, using a qualitative 5-points scale ranging from “no
impact” to “very high risk of toxicity”. The toxicity during the work
phase reflects the risks faced by people in charge of the imple-
mentation of the insulators. It represents a double issue: to owners
wishing to carry on the works by themselves, which can be directly
subject to the risk and for the health of workers which would be in
charge of the works. This criterion is then placed between envi-
ronmental and social dimensions. Toxicity during the life of the
house indicates the long term risks, especially when the material
deteriorates and can be in airborne. The toxicity of insulation is a
controversial issue; no scientific study has, up to our knowledge,
been able to make a final determination on the danger during the
construction phase or during the degradation of materials of
insulation materials available currently on French market (asbestos
is no more authorized in France). The criterion is measured in terms
of risk, i.e. it combines the uncertainty (or presumption) on the
material toxicity and the impact of this presumed toxicity.
2.5.2. Economic criteria
At MICRO phase, the only economic criterion considered is the
additional cost of the different solutions with respect to a reference
solution consisting in selecting glass wool for all surfaces to be
insulated. This additional cost is computed using classical rules
combining unit cost price given by the construction economist and
the house size.
2.5.3. Social criteria
Three social criteria are considered in this phase: the lost sur-
face, the fire protection rating and the durability of the material.
Lost surface corresponds, in case of insulation of the walls from the
inside, to the surface taken by the insulation which varies according
128 F. Taillandier et al. / Building and Environment 109 (2016) 121e134
to the insulation thickness since it is a function of its thermal
resistance. For insulation at the top floor, it corresponds to the loss
of free height.
The fire rating corresponds to the sorting of the material into
Euroclass from the standard EN 13501e1 (2013). Durability reflects
the time during which the material is able to ensure its insulation
function. The criterion considers not only the life of the material but
also its ability to maintain its isolation along time. This criterion is
evaluated qualitatively on a to 5-grade scale (from “very little du-
rable” to “very durable”) by the experts of the FOURMINERGIE
project.
2.6. Decision support
The decision support process of ARD-FOURMI is based on
ELECTRE III method. It allows comparing different solutions and
prioritizing them; then, it enables to retain only the most inter-
esting solutions. To prioritize solutions, ELECTRE III builds out-
ranking relations between them. The tool follows a 4-stage process:
(1) Calculation of concordance and discordance indexes, i.e. indexes
reflecting the preference or the veto of a solution relative to
another, on a given criterion, (2) Calculation of overall concordance
indexes: aggregation of concordance indexes according to the set of
criteria using a weight value for each criterion, (3) Calculation of
credibility degree: integration of veto effects to overall concor-
dance, (4) Establishment of outranking relations: from credibility of
degrees, defining relations (outrank, indifference and incompara-
bility) between solutions and prioritization. From outranking re-
lations, ELECTRE established two rankings: ascendant (from the
worst to the best solution) and descending (from the best to the
worst). These two rankings are then merged into one final ranking.
It must be noted that the owner (i.e. the person who will take the
final decision) can specify her/his preferences through the relative
weights for each criterion and thresholds values, as veto.
An important aspect of decision support methods is the
robustness of the method, i.e. the ability of the method to return a
result insensitive to local variations (uncertainties) of the data or
parameters [41]. Indeed, multi-criteria methods use several pa-
rameters like weights and thresholds which may have a major
impact on the result. In order to ensure the robustness of the re-
sults, the ELECTRE method is coupled with Monte-Carlo simula-
tions; the tool uses the stochastic version of ELECTRE III [47]. In this
version, the ranking with ELECTRE is processed many times
through Monte-Carlo simulations, using for input data a draw ac-
cording to predefined distribution laws. This process enables to
assess the sensitivity of results to uncertainty on input data and
parameters. A uniform probability law on the range (between lower
and upper bounds) of input variables was chosen. For parameters
(such as weights), ARD-FOURMI uses a Gaussian law considering
constraints in parameters values (e.g. veto threshold is always su-
perior to preference threshold). The final results come from the
analysis of the statistical distribution of results.
3. Application
3.1. Case study presentation
To illustrate the process, we present the application of ARD-
FOURMI tool to one of the houses of the Malartic district. The
experimentation consisted in a meeting with the house owner.
During this meeting, firstly, the owner provided information about
her house: type, year of construction, works already carried out, etc.
An Excel interface allows the owner to document the data related to
her/his house. The tool provides default values for most of the
values related to the house (dimension characteristics of the house,
energy consumption …) from its typology. Obviously, if the owner
has more accurate information regarding her/his house, she/he can
update the default values. The experimentation house was of type
Castille T5 (Fig. 4). Dating from 1975, with a living area of 118 m2, it
has an initial consumption of 169 kWh/m2/year. It has been
partially renovated in 2010 when the original windows were
replaced with double-glazed windows on the north side of the
house.
The owner of the house must also answer a questionnaire to
assess social criteria (see section 2.3.3). For this owner, the execu-
tion of works inside the house poses a problem, although she was
not completely opposed to this alternative. The house has not
joined house and the owner has no particular concern with the
neighbourhood. Finally, the owner must fix the weights associated
with each criterion. In our case, the owner wanted to highlight the
environmental aspects, but she also paid attention to the comfort
and quality of use. From this preference elicitation, the tool pro-
poses a weight for each criterion. In parallel, the tool has a set of
pre-filled weights corresponding to typical profiles (“Environ-
mental promotion”, “Financial Risk limitation”, etc.). The owner can
test other weightings and see the influence of weights on the final
result. In this application, two solutions (S24 & S25) were dis-
qualified in reason of the technical impossibility for this house to
perform them at a reasonable cost while they contain floor insu-
lation: these houses were built on a rigid slab, visible on Fig. 3,
without any additional foundation, and while it would be inter-
esting for thermal efficiency to improve the ground isolation, the
corresponding cost is not acceptable.
3.2. Results
From the house data, the answer to questionnaire by the owner
and the expression of her preferences, the software assessed the
solutions and proposed the most relevant solutions for the user. For
the two phases, 10,000 simulations were performed. For each
simulation, the 26 MACRO solutions are ranked (Index 1 corre-
sponds to the best ranked solution). Table 5 provides the average
rank of the 26 MACRO solutions and the standard deviation on
these ranks which constitutes the results for the MACRO phase. The
lower the average rank, the better (i.e. more relevant considering
decision's maker preferences) the solution is. During experimen-
tation, the meaning of this result was directly explained to house
owner. Furthermore, other types of result as the comparison of
solution regarding each criterion, or radar chart were available
allowing the owner to well understand the results. Moreover, in the
proposed approach, the use of the tool is always done with a
referent able to explain all the concepts and to guide its use (this
will be detailed in part 4.3).
Fig. 4. House used in the experimentation.
 F. Taillandier et al. / Building and Environment 109 (2016) 121e134 129

Table 5 user preferences. It is interesting to note, however, that the three

MACRO phase results. proposed solutions are very different in terms of financial cost: (a)
Solution Average rank Standard deviation on rank V 43 600, (b) V 28 039 and (c) V 16 716. They can be seen as three
S0 23.15 1.2
levels of completeness of the same strategy: solution (c) provides a
S1 10.13 1.58 first level of intervention with ventilation and wall insulation; so-
S2 11.89 3.25 lution (b) proposes to add the change of windows and door and
S3 2.56 1.4 solution (a) adds roof insulation to solution (b). During the exper-
S4 16.64 3.48
imentation, the owners selected solution (a) i.e. S26 which corre-
S5 5.27 1.69
S6 15.67 1.68 sponds to the extensive renovation. This choice was argued by the
S7 14.43 2.08 ranking of this solution (the best among all the solutions) and on
S8 19.4 1.43 the desire of the owner to make all the necessary works at once in
S9 14.96 3.22
order to reach a high level of thermal performance. This choice
S10 18.55 1.86
S11 3.38 1.62
corresponds to the FOURMINERGIE project philosophy which
S12 22.48 1.79 promotes global renovation.
S13 5.48 2.58 The standard deviation (Table 4) and the boxplot graph (Fig. 5)
S14 15.59 2.24 give information about result uncertainties. The standard de-
S15 7.92 3.14
viations are between 0.99 and 4 (on a [1e25] scale), showing that
S16 20.12 1.29
S17 20.35 3.5 the ranking is highly dependent on the input data (values on
S18 9.21 1.66 criteria) and parameters (weights, thresholds). These variations are
S19 6.95 1.41 particularly visible from the extreme values (minimum and
S20 22.82 0.99 maximum) from each solution: 23 solutions out of the 26 have a
S21 21.86 2.49
S22 11.12 4
maximum rank higher than 20 and 22 solutions have a minimum
S23 3.88 2.15 value lower than 7. Considering the extreme values, it is thus
S26 2.12 2.08 difficult to establish a ranking. It is more relevant to analyse the
S27 10.13 3.63 statistical distribution of these solutions. It appears that the two
most relevant solutions, i.e. S3 and S26 have 75% of their ranks
lower or equal to rank 3. This means a robust integration of these
The three MACRO solutions recommended by the tool are (i.e. two solutions in the ‘top 3 solutions’. The third solution, S11 is
those which have the lower average rank): (a) S26 - Roof insulation challenged by S23: if S11 has a better average and median rank,
from outside, Replacement of the windows and doors, Wall insu- considering the uncertainties, the two solutions are rather close.
lation from outside and Humidity sensitive mechanical ventilation, The house owner selected S26, but considering uncertainties, S3
(b) S11 - Wall insulation from outside, Replacement of all the and S26 have their own advantages: S26 has the best median and
windows and doors and Humidity sensitive mechanical ventilation, average ranking, but S3 is very good considering the maximum
and (c) S3 - Wall insulation from outside and Humidity sensitive value. This solution could appear as less risky than solution S26 but
mechanical ventilation. it is also less efficient, explaining the decision-maker choice. A
From these results, the owner decided to choose the solution 26 sensibility analysis could be interesting in order to identify the
with roof and wall insulation. From this solution, ARD-FOURMI parameters or the data with the highest impact on the variability of
proposed a set of MICRO solutions corresponding to the choice of the results; it may enable to acquire additional information on
insulation type for roof and walls; the insulation can be different them and perhaps to reduce uncertainties.
regarding roof and walls. Considering the 16 insulation types of the The MICRO phase led to the recommendation of cork for walls
database and that some insulation types are not usable for outside and roof. The material is the same but the laying is different be-
walls or roof insulation, 36 MICRO solutions are proposed. As for tween walls (panel) and roof (loose) which is explained by the
MACRO phase, 10,000 simulations were performed giving 10,000 match between the laying and the type of surface. The choice of this
ranking (from 1 to 36). Table 6 exposes the MICRO solution average material is consistent with the preference of the owner who priv-
ranking for this phase; the three solutions proposed to the owner, ileged the environmental dimension and weighted it against the
i.e. those having the best ranking are (a) 24 e Cork for walls and cost aspect. The choice between the different natural insulation
roof, (b) 12 - Cork for walls and fibreboard for roof and (c) 16 e Cork materials was done on several aspects: firstly, the possibility of
for walls and cellulose wool for roof. The owner has chosen in a first using them in isolation from the outside and secondly the dura-
approach to retain Solution 24 using cork for walls and roof but she bility of these materials compared to some other natural insulation.
would like to have more precise information on cork and wood By analysing the uncertainties through the boxplot (Fig. 6), it
fibreboard panels (used in the second ranking solution) to confirm appears that the uncertainties on the results are less important
or to change her first decision. than for MACRO solution ranking. The uncertainties are more
important for average solution than for best and worst solution.
4. Discussion Then, solutions S17 to S20 are the worst in any case; every of these
solutions contain ‘Cellulose wool - Poured loose’ which seems to be
4.1. Discussion about experimentation results not relevant considering this house and owner preferences. In
parallel, the three best solutions, even considering uncertainties,
The MACRO phase results in three solutions (S26, S11 and S3) outrank all other solutions confirming the robustness of the results
which were relatively close in terms of average ranking (a) 2.12 (b) (weak sensibility to parameter variation). These three solutions
2.56 and (c) 3.38. The three resulting solutions, which however had contain ‘Cork e Panel’ which appears as the best solution regarding
the best rankings among the 26 proposed solutions, are close to the wall insulation, but differs regarding roof insulation. On this last
best theoretical classification that would be 1, but none of them has point, if ARD-FOURMI gives a real preference to ‘Cork - Loose’
an average ranking between 1 and 2 which would ensure its strict considering its average rank of 1 which is the best theoretical value
domination on all others. Then, if none of the solutions strictly and its standard deviation of only 0.02. ARD-FOURMI has not the
dominates the others, several offer equivalent trade-off regarding role to take decision, but just to help the owner to take a carefully
130 F. Taillandier et al. / Building and Environment 109 (2016) 121e134

Table 6
MICRO phase results.

Solution Walls Roof Average rank Standard deviation

1 Polystyrene - Panel Glass wool - Roll 13.89 3.13

2 Rockwool - Roll Glass wool - Roll 16.93 3.33
3 Wood fibreboard - Rigid panel Glass wool - Roll 14.6 1.86
4 Cork - Panel Glass wool - Roll 10.94 1.36
5 Polystyrene - Panel Rockwool - Roll 28.89 0.97
6 Rockwool - Roll Rockwool - Roll 28.85 1
7 Wood fibreboard - Rigid panel Rockwool - Roll 29.07 0.97
8 Cork - Panel Rockwool - Roll 23.61 1.62
9 Polystyrene - Panel Wood fibreboard - Flexible panel 20.64 2.34
10 Rockwool - Roll Wood fibreboard - Flexible panel 22.27 2.99
11 Wood fibreboard - Rigid panel Wood fibreboard - Flexible panel 7.74 1.9
12 Cork - Panel Wood fibreboard - Flexible panel 1.87 0.34
13 Polystyrene - Panel Cellulose wool - Instilled loose 9.8 2.72
14 Rockwool - Roll Cellulose wool - Instilled loose 12.88 3.64
15 Wood fibreboard - Rigid panel Cellulose wool - Instilled loose 6.68 0.86
16 Cork - Panel Cellulose wool - Instilled loose 2.85 0.75
17 Polystyrene - Panel Cellulose wool - Poured loose 34 0.01
18 Rockwool - Roll Cellulose wool - Poured loose 35.71 0.7
19 Wood fibreboard - Rigid panel Cellulose wool - Poured loose 34 0.01
20 Cork - Panel Cellulose wool - Poured loose 33 0
21 Polystyrene - Panel Cork - Loose 16.42 2.84
22 Rockwool - Roll Cork - Loose 15.98 2.24
23 Wood fibreboard - Rigid panel Cork - Loose 6.47 1.86
24 Cork - Panel Cork - Loose 1 0.02
25 Polystyrene - Panel Hemp wool - Roll 22.92 2.97
26 Rockwool - Roll Hemp wool - Roll 27.42 1.25
27 Wood fibreboard - Rigid panel Hemp wool - Roll 13.51 3.74
28 Cork - Panel Hemp wool - Roll 4.61 0.67
29 Polystyrene - Panel Linen wool - Roll 22.11 2.53
30 Rockwool - Roll Linen wool - Roll 24.34 3.76
31 Wood fibreboard - Rigid panel Linen wool - Roll 11.48 1.61
32 Cork - Panel Linen wool - Roll 3.69 0.73
33 Polystyrene - Panel Sheep wool - Roll 17.58 4.25
34 Rockwool - Roll Sheep wool - Roll 27.68 1.78
35 Wood fibreboard - Rigid panel Sheep wool - Roll 25.39 1.8
36 Cork - Panel Sheep wool - Roll 21.69 2.41

thought out decision. Consequently, the wish of the owner to requirement. A less compensative method seems more relevant.
obtain more information on Cork and Fibreboard corresponds to Each method has its drawbacks and no method can be relevantly
the project philosophy, i.e. giving knowledge to allow owner to take applied to all problems. Up to now, it exists no process or method to
relevantly their own decision. select the best multi-criteria decision method/tool [21], but [19]
suggest different points to be studied in order to choose a rele-
vant decision-aiding method: the number of decision-makers, the
4.2. Methodological discussion
preferences of decision-makers between pairwise comparisons and
trade-offs, the quantity and the quality of the available inputs, the
The choice of the two-phases approach and the use of multi-
compensation degree of the method, the respect of the funda-
criteria evaluation were partly influenced by the educational
mental hypothesis of the method and the decision support system.
dimension. This approach secures improvements in successive re-
In the ARD-FOURMI decision context, the natural process used by
finements of the solution while the owner can understand the
most of the owners consists in pairwise comparing of the different
various issues which are made explicit. To move in this direction,
solutions. The notion of veto is also often used by the decision-
the tool doesn't display only the most relevant solutions but also
maker. For instance, in one of the experimentation in which the
allows the user to test all the alternatives and compare them with
solution cost was very important for the owner (because of his
those promoted by the tool. This allows owners to be aware of the
limited budget), a combination of a strong weight for the criterion
strengths and weaknesses of each solution.
“Investment cost” and a restrictive veto threshold (as enabled in
The choice of a particular method among the large number of
ELECTRE III) penalizing solution with a cost exceeding budget.
multi-criteria decision-aiding method can have an important
Furthermore, the higher complexity of ELECTRE compared to AHP
impact on the result [57]. In this work, the ranking is made using
and its higher number of parameter is not a major drawback in the
ELECTRE III method, which was successfully used to solve
considered context. Indeed, the high number of parameters is well
numerous problems in different domains [17]. The choice of this
adapted for describing the opinion of the house owner (as for the
method rather than AHP [43] can be discussed. Yet, ELECTRE pre-
threshold) while the stochastic framework [47,48] by considering
sents several drawbacks as its difficulty to use, its low learning
uncertainties on data and parameters enables to qualify the
dimension or its potential rank reversing [8]. Furthermore, AHP is
robustness of choices. Consequently, ELECTRE III method is very
the most used method in the building renovation context [33].
suitable to this decision context.
However several authors, e.g. Refs. [4,8], have shown its limitation
in a sustainable assessment context, notably due to its high
compensation degree among sustainability criteria; i.e. a very low 4.3. Discussion about the use of ARD-FOURMI in a real context
score on a criterion can be compensated by a high score on another
criterion which does not meet the sustainable perspective The output of the tool is not limited to the prescription of a
 F. Taillandier et al. / Building and Environment 109 (2016) 121e134
Fig. 5. Boxplot of MACRO phase solutions.
solution or a given insulation material. It has also an educational
objective: it brings the owner new questions that she/he would not
necessarily have in mind at first. The tool enriches her/his reflec-
tion, as we have experienced during the many workshops with the
association co-operators. According to this, the tool could be really
considered as a real support for decision. It does not aim to provide
direct answers to the owner, but to help her/him to build and to
mature her/his renovation project. In this perspective, the limita-
tion on owner input was not perceived as a drawback by the
owners during the different applications. The possible answers are
firstly a base for discussion. The necessity to choose a defined
answer among a limited set requires reflection for the owner.
Furthermore, it facilitates the performing of the decision support
calculation.
The skill and knowledge of the co-operators and more generally
of the Malartic house owners are strongly heterogeneous. Several
co-operators are really active in the fields of environmental
activism and played a major role in the tool structuration; they are
transmitters of the association values and attach importance to
share skill and knowledge. Otherwise, some co-operators have
already a good knowledge on energy use and renovation actions
and have already maturate their renovation project; they do not
need a decision support, but could find in the tool a confirmation of
their choice. But most of the new co-operators have a low
131
knowledge about renovation action and have lots of questions
about it. The tool is a first step allowing them to begin their
thinking. The use of the tool takes the time of a meeting (about two
hours): we present the principle of the tool, we fill with the owner
the different items, and we calculate directly the results (about
5 min). Results are analysed on the fly. A large part of the meeting is
dedicated to the discussion. Normally, the use of the tool should
lead to deepen the discussion and ask new questions. A time for
maturation is then generally necessary. The next steps (research of
information, discussion with other co-operators, etc.) are beyond
the scope of the tool. The tool belongs to a more global process,
from the wish of the owner to make work or the expression of a
curiosity about it, to the end of the works. This whole process could
be rather long; it took about 2 years for the first co-operators who
has completed the process. The whole process implies other
stakeholders as an architect who participated at this tool and who
can ensure the role of project manager for renovation works (if the
owner want to call upon a project manager). The association has
also a list of recommended companies/craftsmen which have
already worked with co-operators. These companies were not
implied in the tool elaboration in order to avoid conflict of interest:
indeed, some of these companies could favour particular solutions
(for instance, several companies do not propose natural insulation).
They only take action in the last steps of the global process, but they
132 F. Taillandier et al. / Building and Environment 109 (2016) 121e134
Fig. 6. Boxplot of MICRO phase solutions.
know the tool from the co-operators with who they worked.
One major issue is to ensure the continuity of the use of the
ARD-FOURMI tool by the co-operators after the end of the project.
This requires that the tool should be usable without the help of the
research team. Two elements are important for this: the ergo-
nomics of the tool and the training. Regarding the first point, the
different tests and workshops were primordial. As exposed in Fig. 1,
the software development was based on many discussions with the
owners; following AGILE methodology, two final users (house
owners) were regularly involved during the ARD-FOURMI tool
development. They made comments and suggestions on the tool
functioning and on the software interface. This ensures the capacity
of the tool to really answer the user needs and requirements.
Furthermore, the ergonomics was continually improved. Finally, to
ensure the ARD-FOURMI durability, a ‘percolation’ process was
used. Two co-operators (called referents) were trained to use and
explain the tool. They are able to use the tool with other owners on
other houses. In a second time they will be able to train other co-
operators to use themselves the tool. The new trained co-
operators could then train others and so forth. The success of
such operation depends on the tool acceptance by the co-operators
which is determined by its perceived usefulness and its perceived
ease of use [51]; the expressed feeling and desires of some co-
operators (the most active core of the association), and the re-
sults of the applications are promising regarding the tool accep-
tance, but an in-depth analysis should be necessary to assess this
characteristic [49].
An important question is the added value of the tool compared
to a classical discussion/meeting approach. The tool has several
interests: to structure the meeting, to ask reflection to the user
whether it be on the questions or on the answers, to make explicit
decision criteria and user preferences, to assess potential solutions
regarding a set of criteria and to propose a relevant renovation
solution; these two last points are very interesting to enrich dis-
cussion and to be the support of project maturation. Moreover, the
tool contains knowledge which is by its functioning easily trans-
ferable. Finally, the tool has a promotion role: the board members
of the association use it to appealed potential new co-operators as a
first step to join the association. If we have confidence in the pos-
itive impact of the tool, we admit that the success of the renovation
process and the low-energy culture development in the neigh-
bourhood depend strongly on the capacity of the association
members/board to federate, share (knowledge, mean, etc.) and be
active, beyond the use of the tool.
 F. Taillandier et al. / Building and Environment 109 (2016) 121e134
5. Conclusion and perspectives
[3]
We present in this paper a method allowing a house owner to
choose a renovation solution in order to improve the thermal
performances of her/his house. The scientific developments were [4]
carried on with a constant attention paid on the end-user needs [5]
and the specificities of the application field, which make the
[6]
problem highly complex. A tool based on this method was devel-
oped and was applied to different real cases during an experi- [7]
mentation phase. If the proposed tool is already usable, several
opportunities are available to improve it. On the one hand, it would
[8]
be interesting to enrich the solutions, either in the MICRO and
MACRO phases, integrating new solutions in the database. In
MACRO phase, solutions could contain actions according to heating [9]
production (installation of a wood stove for example) while in
MICRO phase, they could propose alternatives for the type of [10]
windows and doors and for ventilation techniques. On the other
hand, it would be possible to enrich the set of criteria by offering
[11]
new ones and to improve their assessment.
An important issue is the capacity of the tool to be used in a new [12]
case/context. Up to now, four experimentations were performed.
About ten new cases in Malartic district are planned in the
following months. The number of interested owners is important: [13]
there are more than seven hundred houses in Malartic and among
[14]
them more than 50 owners are members of the association ‘Des
Fourmis dans le compteur’. The tool can also be used in other dis-
tricts with rather close context. For instance, it was imagined that it [15]
could be used for another district close to Malartic which share the [16]
problematic and a close typology of house/owner. For different
contexts, it would be possible to make a new version of the tool. [17]
This, obviously, raises the question of the calculation of the energy [18]
consumption. In the current model, regarding the closeness of the
houses, a thermal simulation of a limited set of houses after a prior [19]
typological analysis was sufficient to provide a rough but accurate
[20]
enough assessment. The extension to a different or more diverse
asset is not straightforward. This raises necessarily the issue of
coupling the decision support tool with thermal simulation tools. [21]
Many points remain open for discussion, but the ARD-FOURMI [22]
tool has already allowed to help owners to formalize their
thoughts regarding the thermal renovation and so to help them to [23]
take informed decisions. It was designed and implemented in order
to develop knowledge, awareness and control over the renovation [24]
process. It integrates a more global process which aims, not only at
improving houses (technical perspective), but also at diffusing low- [25]
energy culture to residents. In this perspective, the association ‘Des
Fourmis dans le compteur’ has a major role by transforming indi-
[26]
vidual initiative in a share project for the Malartic district.
Acknowledgement [27]
[28]
The ARD-FOURMI tool is the collective result of the FOUR-
MINERGIE project team work. The authors thank all those who [29]
participated in the project, including VALOREM (project leader),
NOBATEK (thermal simulation), Michel Sarrazin (construction
[30]
economist), Alexandra Georgeoliani (architect) and co-operators of
the association “Des fourmis dans le compteur” who actively
participated in the creation and improvement of the tool, the [31]
Aquitaine region (France) that funded the project and the Creahd
cluster. [32]
References [33]
[1] P. Abrahamson, O. Salo, J. Ronkainen, J. Warsta, Agile Software Development
Methods: Review and Analysis, 2002, p. 478. [34]
[2] M. Almeida, A. Bencresciuto, M. Ferreira, A. Rodrigues, Cost-effective energy
133
and carbon emission optimization in building renovation e a case-study in a
low income neighbourhood, Energy Procedia 78 (2015) 2403e2408.
D. Anastaselos, E. Giama, A.M. Papadopoulos, An assessment tool for the en-
ergy, economic and environmental evaluation of thermal insulation solutions,
Energy Build. 41 (11) (2009) 1165e1171.
P. Arroyo, Exploring Decision-making Methods for Sustainable Design in
Commercial Buildings, University of California, 2014.
L.G. Caldas, L.K. Norford, A design optimization tool based on a genetic algo-
rithm, Autom. Constr. 11 (2) (2002) 173e184.
CGDD, Repe re e Chiffres clefs de l’e
nergie - Edition 2012, C. G. a. D. Durable,
2012.
F.P. Chantrelle, H. Lahmidi, W. Keilholz, M.E. Mankibi, P. Michel, “Develop-
ment of a multicriteria tool for optimizing the renovation of buildings, Appl.
Energy 88 (4) (2011) 1386e1394.
M. Cinelli, S.R. Coles, K. Kirwan, “Analysis of the potentials of multi criteria
decision analysis methods to conduct sustainability assessment, Ecol. Indic. 46
(2014) 138e148.
CSTB, Base nationale française de re
fe

rence sur les impacts environnementaux
et sanitaires des produits, e
quipements et services pour l’e
valuation de la
performance des ouvrages, 2015.
F.D. Davis, R.P. Bagozzi, P.R. Warshaw, User acceptance of computer tech-
nology: a comparison of two theoretical models, Manag. Sci. 35 (8) (1989)
982e1003.

narios Prospectifs
DGCE, Direction generale de l’energie et du climat-Sce
Energie, Climat, Air  a horizon 2030 Document de Synthe se. Paris, 2011.
EC, EUROPEAN COMMISSION - Communication from the Commission to the
European Parliament, the Council, the European Economic and Social Com-
mittee and the Committee of the Regions on Resource Efficiency Opportu-
nities in the Building Sector, 2014 (Brussels).
EC, EUROPEAN COMMISSION - Energy Prices and Costs in EUROPE, 2014
(Brussels).
M. Erlandsson, M. Borg, Generic LCA-methodology applicable for buildings,
constructions and operation servicesdtoday practice and development needs,
Build. Environ. 38 (7) (2003) 919e938.
C. Friedman, N. Becker, E. Erell, Energy retrofit of residential building enve-
lopes in Israel: a cost-benefit analysis, Energy 77 (2014) 183e193.
M. Gorgolewski, P.C. Grindley, S.D. Probert, Energy-efficient renovation of
high-rise housing, Appl. Energy 53 (4) (1996) 365e382.
K. Govindan, M.B. Jepsen, ELECTRE: a comprehensive literature review on
methodologies and applications, Eur. J. Op. Res. 250 (1) (2016) 1e29.
K. Gram-Hanssen, Retrofitting owner-occupied housing: remember the peo-
ple, Build. Res. Inf. 42 (4) (2014) 393e397.
A. Guitouni, J.-M. Martel, “Tentative guidelines to help choosing an appro-
priate MCDA method, Eur. J. Op. Res. 109 (2) (1998) 501e521.
Hickey, A. M. and A. M. Davis (2003). Elicitation technique selection: how do
experts do it? Requirements Engineering Conference, 2003. Proceedings. 11th
IEEE International.
A. Ishizaka, P. Nemery, Multi-criteria Decision Analysis: Methods and Soft-
ware, Wiley, 2013.
ISO, 14040: Environmental Management. Life Cycle Assessment - Principles
and Framework, 2006.
l. Itard, F. Meijer, E. Vrins, H. Hoiting, Building Renovation and Modernisation
in Europe: State of the Art Review, Delft University of Technology, 2008. Final
Report ERABUILD.
Y.-K. Juan, P. Gao, J. Wang, A hybrid decision support system for sustainable
office building renovation and energy performance improvement, Energy
Build. 42 (3) (2010) 290e297.
Y.-K. Juan, S.-G. Shih, Y.-H. Perng, Decision support for housing customization:
a hybrid approach using case-based reasoning and genetic algorithm, Expert
Syst. Appl. 31 (1) (2006) 83e93.
B. Katalinic, A. Civic, B. Vucijak, 24th DAAAM international symposium on
intelligent manufacturing and automation, 2013Multi-criteria optimization of
insulation options for warmth of buildings to increase energy efficiency,
Procedia Eng. 69 (2014) 911e920.
KBOB, Donne
es des e

cobilans dans la construction, 2014.
H.-H. Liang, M.-C. Ho, Toxicity characteristics of commercially manufactured
insulation materials for building applications in Taiwan, Constr. Build. Mater.
21 (6) (2007) 1254e1261.
P. Liu, E.N. Pistikopoulos, Z. Li, An energy systems engineering approach to the
optimal design of energy systems in commercial buildings, Energy Policy 38
(8) (2010) 4224e4231.
E. Mansour, C. Loxton, R.M. Elias, G.A. Ormondroyd, “Assessment of health
implications related to processing and use of natural wool insulation prod-
ucts, Environ. Int. 73 (2014) 402e412.
V. Martinaitis, E. Kazakevi cius, A. Vitkauskas, A two-factor method for
appraising building renovation and energy efficiency improvement projects,
Energy Policy 35 (1) (2007) 192e201.
K. Mjo € rnell, A. Boss, M. Lindahl, S. Molnar, A tool to evaluate different reno-
vation alternatives with regard to sustainability, Sustainability 6 (7) (2014)
4227.
A.N. Nielsen, R.L. Jensen, T.S. Larsen, S.B. Nissen, Early stage decision support
for sustainable building renovation e a review, Build. Environ. 103 (2016)
165e181.
W. Ott, R. Bolliger, Pitfalls in the economic and ecological evaluation of energy
related building renovation strategies and measures, Energy Procedia 78
134 F. Taillandier et al. / Building and Environment 109 (2016) 121e134
(2015) 2340e2345.
[35] A. Pasanisi, J. Ojalvo, A multi-criteria decision tool to improve the energy ef-
ficiency of residential buildings, Found. Comput. Decis. Sci. 33 (1) (2008)
71e82.
[36] P. Pelletier, Comite
ope
rationnel « re
novation des ba ^timents existants » :
rapport d'e
tape, ministre de l'e
cologie, du de
veloppement et de
l'ame
nagement durables. d. d. d. Ministe re de l'e

cologie, 2008.
[37] B. Peuportier, I.B. Sommereux, Simulation tool with its expert interface for the
thermal design of multizone buildings, Int. J. Sol. Energy 8 (2) (1990)
109e120.
[38] R. Ramírez-Villegas, O. Eriksson, T. Olofsson, Assessment of renovation mea-
sures for a dwelling area e impacts on energy efficiency and building certi-
fication, Build. Environ. 97 (2016) 26e33.
[39] B.C. Roberts, M.E. Webber, O.A. Ezekoye, Development of a multi-objective
optimization tool for selecting thermal insulation materials in sustainable
designs, Energy Build. 105 (2015) 358e367.
[40] B. Roy, The Outranking Approach and the Foundations of Electre Methods.
Readings in Multiple Criteria Decision Aid, C. Bana e Costa, Springer Berlin
Heidelberg, 1990, pp. 155e183.
[41] B. Roy, Robustness in operational research and decision aiding: a multi-
faceted issue, Eur. J. Op. Res. 200 (3) (2010) 629e638.
[42] E.M. Ryan, T.F. Sanquist, Validation of building energy modeling tools under
idealized and realistic conditions, Energy Build. 47 (2012) 375e382.
[43] T. Saaty, The Analytic Hierarchy Process, McGraw-Hill, New-York, 1980.
[44] S. Schiavoni, F. D‫׳‬Alessandro, F. Bianchi, F. Asdrubali, Insulation materials for
the building sector: a review and comparative analysis, Renew. Sustain. En-
ergy Rev. 62 (2016) 988e1011.
[45] A.A. Stec, T.R. Hull, Assessment of the fire toxicity of building insulation ma-
terials, Energy Build. 43 (2e3) (2011) 498e506.
[46] F. Taillandier, S. Aries, A. Ndiaye, B. Leblanc, C. Fernandez, F. Labat, Comment
ame
liorer les performances thermiques de ma maison ? Une affaire d'expert !
31emes Rencontre de l’AUGC, 2013 (Cachan, France).
[47] P. Taillandier, F. Taillandier, Multi-criteria diagnosis of control knowledge for
cartographic generalisation, Eur. J. Op. Res. 217 (3) (2012) 633e642.
[48] T. Tervonen, J.R. Figueira, R. Lahdelma, J.A. Dias, P. Salminen, A stochastic
method for robustness analysis in sorting problems, Eur. J. Op. Res. 192 (1)
(2009) 236e242.
[49] V. Venkatesh, H. Bala, Technology acceptance model 3 and a research agenda
on interventions, Decis. Sci. 39 (2) (2008) 273e315.
[50] G. Verbeeck, H. Hens, Life cycle inventory of buildings: a calculation method,
Build. Environ. 45 (4) (2010) 1037e1041.
[51] L.G. Wallace, S.D. Sheetz, The adoption of software measures: a technology
acceptance model (TAM) perspective, Inf. Manag. 51 (2) (2014) 249e259.
[52] H.-J. Wang, Z.-T. Zeng, A multi-objective decision-making process for reuse
selection of historic buildings, Expert Syst. Appl. 37 (2) (2010) 1241e1249.
[53] W. Wang, R. Zmeureanu, H. Rivard, Applying multi-objective genetic algo-
rithms in green building design optimization, Build. Environ. 40 (11) (2005)
1512e1525.
[54] J.A. Wright, H.A. Loosemore, R. Farmani, Optimization of building thermal
design and control by multi-criterion genetic algorithm, Energy Build. 34 (9)
(2002) 959e972.
[55] J. Xu, J.-H. Kim, H. Hong, J. Koo, A systematic approach for energy efficient
building design factors optimization, Energy Build. 89 (2015) 87e96.
[56] J. Zagorskas, E.K. Zavadskas, Z. Turskis, M. Burinskiene, _ A. Blumberga,
D. Blumberga, Thermal insulation alternatives of historic brick buildings in
Baltic Sea Region, Energy Build. 78 (2014) 35e42.
[57] S.H. Zanakis, A. Solomon, N. Wishart, S. Dublish, Multi-attribute decision
making: a simulation comparison of select methods, Eur. J. Op. Res. 107 (3)
(1998) 507e529.
[58] Z. Zhang, X. Wu, X. Yang, Y. Zhu, BEPASda life cycle building environmental
performance assessment model, Build. Environ. 41 (5) (2006) 669e675.
[59] E. Znouda, N. Ghrab-Morcos, A. Hadj-Alouane, Optimization of Mediterranean
building design using genetic algorithms, Energy Build. 39 (2) (2007)
148e153.
[60] D. Calì, T. Osterhage, R. Streblow, D. Müller, Energy performance gap in
refurbished German dwellings: lesson learned from a field test, Energy Build.
127 (2016) 1146e1158.
[61] P. Hoes, J.L.M. Hensen, M.G.L.C. Loomans, B. de Vries, D. Bourgeois, User
behavior in whole building simulation, Energy Build. 41 (3) (2009) 295e302.
[62] O. Guerra Santin, L. Itard, H. Visscher, The effect of occupancy and building
characteristics on energy use for space and water heating in Dutch residential
stock, Energy Build. 41 (11) (2009) 1223e1232.
[63] G.A. Guagnano, P.C. Stern, T. Dietz, Influences on attitude-behavior relation-
ships: a natural experiment with curbside recycling, Environ. Behav. 27 (5)
(1995) 699e718.
[64] P.C. Stern, New environmental theories: toward a coherent theory of envi-
ronmentally significant behavior, J. Social Iss. 56 (3) (2000) 407e424.
[65] J. ThØgersen, Understanding of consumer behaviour as a prerequisite for
environmental protection, J. Consum. Policy 18 (4) (1995) 345e385.
[66] M. Moezzi, L. Lutzenhi, What's missing in theories of the residential energy
user, in: ACEEE Summer study on energy efficiency in buildings, 2010.
[67] E. Shove, Beyond the ABC: climate change policy and theories of social change,
Environ. Plann. A 42 (6) (2010) 1273e1285.
[68] P. Jugert, K.H. Greenaway, M. Barth, R. Büchner, S. Eisentraut, I. Fritsche,
Collective efficacy increases pro-environmental intentions through increasing
self-efficacy, J. Environ. Psychol. 48 (2016) 12e23.
[69] J. Desmedt, G. Vekemans, D. Maes, Ensuring effectiveness of information to
influence household behaviour, J. Clean. Prod. 17 (4) (2009) 455e462.
[70] H.R. Bernard, G.W. Ryan, Analyzing qualitative data: systematic approaches,
SAGE, 2010.
[71] A. Davis, O. Dieste, A. Hickey, et al., Effectiveness of requirements elicitation
techniques: empirical results derived from a systematic review, in: Proc. 14th
IEEE Int'l Conf, 2006.