A procedure for multicriteria selection of building assemblies

1
A procedure for multi-criteria selection of
building assemblies

• K. Nassar,, Walid Thabet, Yvan Beliveau

• 18 January 2003

• Prepared by
• sultan al mutairi

2
Outline

• 1. Introduction
• 2. The assembly selection problem
• 3. The proposed procedure
• 4. Computer implementation
• 5. Conclusions and recommendations for
  future research

3
Introduction

• One of the aspects of building design is to find
  trade-offs that satisfy a multitude of
  performance objectives.
• most design variables in these studies, like the
  building or openings shapes and space sizes, are
  usually governed by other design constraints that
  are not easily expressed or quantified (such as
  aesthetics and social design constraints).

4
Introduction

• On the other hand, one of the design variables
  that can significantly affect the performance of
  the building is the choice of the construction
  materials/components used in the various
  assemblies in the building.

5
Introduction

• Recently, the process of modeling buildings in
  the state-of-the-art CAD tools such as Revit,
  ArchiCad and Architectural Desktop are based on
  drawing various elements of the building such as
  walls, roofs and windows and then assigning
  specific assemblies to each of these elements.
  Building analysis tools are being integrated with
  these CAD tools to evaluate the performance of
  the building with respect to a number of criteria
  such as energy performance.

6
Introduction

• In this paper, we describe a procedure for the
  selection of the assemblies that best suit the
  designers performance criteria. The procedure is
  implemented in a prototype computer program that
  allows the user to model the building, choose
  specific performance criteria and assign their
  relative weights.

7
Introduction

• The software then helps the user to select the
  best assemblies that meet the user-specified
  criteria and their relative weights based on the
  developed selection procedure.
• In the next section, the assembly selection
  problem is discussed, and in Section 3, we
  describe the developed selection procedure. The
  computer implementation is then presented, and
  finally, an example is provided.

8
2. The assembly selection problem

• Any building can be broken down to a number of
  building elements j 1,2,3. . .,m (walls, roof,
  floors, etc.), and for each building element j,
  the objective is to find the best assembly ij
  1,2,3. . .,nj.
• The definition of best assembly construction
  is dependent on a set of designer-selected
  criteria C1,C2,C3. . .,CN.
• There are several criteria that affect the
  selection of one assembly construction versus the
  other (Table 1 shows a partial list of some
  criteria).

9
2. The assembly selection problem

• We can divide these criteria into two categories
  building level criteria and assembly level
  criteria. Building level criteria are those that
  are related to the building entity as a whole
  (e.g. annual thermal load) while assembly level
  criteria are related to each assembly type by
  itself (e.g. external wall durability).

10
2. The assembly selection problem

• Examples of assemblies used in various building
  elements are shown in Table 2, along with some
  assembly level criteria and their scores. Some of
  these criteria are physical properties of the
  assemblies, while others are subjective scores
  given based on experience. In Table 2 for
  example, serviceability and ease of maintenance
  are subjective scores for the roof assemblies
  from Ref.13.

11

• Depending on the particular design situation,
  some
• of these criteria might be considered while
  others will
• not, and the relative importance for these
  criteria will
• also be assigned. Therefore, the best assembly
  combination
• is the one that provides the best trade-off
• between the user-selected criteria according to
  the
• assigned relative importance.

12

• When performing the selection, one has to
  consider
• all the combinations of the applicable
  assemblies.
• This will lead to a combinatorial explosion.
• For example, in particular design situations, if
  there
• are 10 different applicable external wall
  assemblies,
• 5 internal wall assemblies, 5 different door and
• window assemblies and 8 roof assemblies, we would
• have to consider 2000 combinations (1055
• 8 2000). Furthermore, if more criteria were
  added,
• the time to perform the selection increases
  substantially
• and becomes unmanageable.

13

• This problem can
• be tackled by breaking down the assembly
  intostages and states as will be described in the
  selection
• procedure below.

14
(No Transcript)
15
3. The proposed procedure

• A flow chart of the assembly selection
  procedure is shown in Fig. 1. The selection
  procedure can be divided into five main steps
• Building design definition
• Criteria selection and assigning importance
  weights
• Normalizing criteria scores
• Formulating the best solution
• Performing the selection

16
Building design definition

• The first step in the selection procedure is to
• breakdown the building design into the different
• building elements. Any design can be modeled by
• abstract building elements such as in Fig. 2.

17
Criteria selection and assigning
importanceweights

• The next step is to select the performance
  criteria that are important from the perspective
  of the designer in the particular design case and
  assign relative importance weights to those
  selected criteria (e.g. external wall
  construction permeability, 20 annual thermal
  load, 20 roof construction maintenance, 20
  and material cost, 40).

18

• The choice of criteria and their relative
  importance will change for each building design.
  For instance, the cost can be of utmost
  importance when compared to the sound performance
  in one design case while this can be switched in
  another design.

19

• it is important to define the importance weights
  for the criteria by considering the importance of
  one criterion over the other in a pair-wise
  comparison versus a comparison of all the
  criteria together.
• This kind of analysis can be accomplished using
  the analytical hierarchy process (AHP) 14. The
  AHP method has been praised 15 for facilitating
  exact analysis of criteria and producing more
  accurate importance weights than other similar
  methods.

20

• The goal of the AHP here is to come up with a
  relative importance vector for the selected
  criteria that reflects the perception and
  judgment of the designer.
• First, a weight matrix A(akl) is evaluated by
  comparing each criterion k with the other
  criterion l of the set of considered criteria
  C1,C2,C3. . .,CN.

21

• For example, for three criteria C1,C2 and C3,

22

• The importance of one criterion (Wk) over the
  other (Wl), akl(Wk/Wl), is determined by
  utilizing a preference scale shown in Table 3.
  Then, the relative importance weight vector WA is
  calculated as the eigenvector corresponding to
  the maximum eigen value of matrix A. It is
  important to note that the importance weight
  vector can be saved so that it can be used for
  similar design situations.

23
Normalizing criteria scores

• Once the relative importance of the criteria is
  established, the next step is normalizing the
  criteria scores.
• The different criteria score can be determined in
  different ways. Some criteria values can be
  retrieved from a database of the different
  assemblies and their performance scores.

24

• For example, the permeability of the external
  walls is stored as a value with each external
  wall assembly construction as shown in Table 2.
  Other criteria can be calculated by using
  specific analysis techniques. The annual cooling
  load is an example of a criterion that cannot be
  retrieved directly from a database (since this is
  one of the building level criteria considered in
  the software developed here, an external analysis
  tool is used to calculate the annual cooling load
  as will be described in the next section)

25

• In any case, the scores of these criteria will
  have a different unit of measurement. Therefore,
  it is necessary to first normalize the scores of
  the various criteria so that they can be
  aggregated in a combined score. When normalizing
  the criteria scores, it is important to provide a
  flexible way to map the different criteria scores
  to the normalized scale. In most cases for
  example, a linear interpolation between the
  minimum and maximum scores is not correct.

26

• In the computer implementation of the procedure
  described here, a normalization curve is assigned
  to each criterion as shown in Fig. 3. This curve
  maps between the raw criteria score and the
  normalized score. The user interactively adds
  points to the curve, and in return, the tools
  automatically fit the best curve. By
  interactively adding points to the curve, the
  user can refine the shape of the curve to
  describe the most appropriate interpolation
  between the raw and the normalized criteria
  scores.

27

• Some criteria, on the other hand, do not have
  explicit values such as designers preference,
  aesthetics, etc. For such criteria, the rank of
  the assembly in terms of the performance criteria
  will determine a normalized criteria score.
  Therefore, it is important to transform the rank
  into a normalized score. This is done using
  either the rank sum weight method or the rank
  reciprocal method, depending on the user
  satisfaction requirements. The rank sum weight is
  given as

28
(No Transcript)
29
(No Transcript)
30
(No Transcript)
31
(No Transcript)
32
(No Transcript)
33
(No Transcript)
34
(No Transcript)
35
(No Transcript)
36
(No Transcript)
37
(No Transcript)
38
(No Transcript)
39
(No Transcript)
40
(No Transcript)
41
(No Transcript)
42
(No Transcript)
43
(No Transcript)
44
(No Transcript)
45
(No Transcript)
46
(No Transcript)
47
(No Transcript)
48
5. Conclusions and Recommendations for Future
Research

• In this paper, a procedure for the selection of
  building assemblies was described.
• The designer provides an assembly-based abstract
  representation of the building.
• The procedure is then used to select the best
  building assemblies for the given design based on
  a set of user-specified criteria and their
  relative importance weights.

49
5. Conclusions and Recommendations for Future
Research

• This provides a practical tool that allows the
  designer to make an informed decision about the
  trade-offs in the building performance criteria.
• The multi-criteria facet of the problem was
  discussed and incorporated in the defined
  procedure.
• The different methods to calculate an aggregated
  criteria score were also described.

50
5. Conclusions and Recommendations for Future
Research

• The developed method for selecting and generating
  building assemblies complements the current
  manual
• Future work is needed to extend the capabilities
  of the developed tool. For example, new
  assemblies and criteria can be added. Another
  possible future development is incorporating
  machine learning in the process.

51
5. Conclusions and Recommendations for Future
Research

• The developed tool can also store design cases,
  learn and use them for future design situations.
• The computer can be taught to recognize which
  criteria are applicable and which are not, as
  well as which criteria are more important than
  others for particular design situations.

52
Thank you