Sei sulla pagina 1di 9

Pergamon PII: SO360-1323(96)000654

Building and Environmew, Vol. 32, No. 4, pp. 331-339, 1997 6% 1997 Elsevier Science Ltd. All riehts reserved Prmted m &eat Britain 036&1323/97 $17.00+0.00

Multicriteria Optimisation Energy-Saving Buildings


WOJCIECH MARKS*

of Shape of

(Received 18 December 1995; revised 6 April 1996; accepted 7 November 1996) The object is to determine the optimum dimensions of the shape of a building of volume V and height h, based on the following criteria: (1) minimum building costs, including the cost of the materials and construction; (2) minimum yearly heating costs. The solution to the problem is presented in two ways. In the first, it is assumed that the shape of the plan of the building is defined by two arbitrary curves bounding the south and north faces and that the windows on the southern side are defined by a continuous function as a percentage of the total waN area. In the second, it is assumed that the building is ofprismatic shape on polygonalplan, and using non-linear programming methods the proportions of wall lengths, wail angles and building height are determined. This problem was solved numerically by means of the CAMOS computer program. It is not the object of the paper to obtain a practical design. The results constitute information for designers on the optimum proportions of wall lengths, their angles and glazing parameters, taking into account the above-mentioned criteria. The degree to which these results can be applied in practice depends on many other requirements present in the design of buildings. 0 1991 Elsevier Science Ltd.

INTRODUCTION The architectural design of a building is influenced by the cost of the energy that will be needed during its service life. It is necessary to find a compromise between the classical elements of design (form, structure and function) and the requirements resulting from the introduction of an additional criterion [ 11. The optimisation problems of the shape of buildings with regard to the economical use of energy have been the subject of many publications. Those concerning unicriteria optimisation of building shape include, among others, references [2%5]. In Fokin [2], the shape of a building of given volume was optimised taking the minimum heat energy loss as the criterion. The solution was a spherical shape. Introducing an additional constraint that the building must be a rectangular prism, a cube was obtained. The problem of choosing the optimum dimensions of a building on rectangular plan with minimum heat requirement per m3 of volume was solved by Gadomski [3]. Heat gain due to insulation was not taken into account. The geometry of building shape was analysed in Menkhoff et al. [4]. The notion of geometric compactness was introduced as the quotient of the area of external walls to the volume of the building. Buildings of various shapes were set up using four identical cubes, giving geometrical compactness coefficients between 4/a and 14.1/a. A building having the shape of a rectangular prism was optimised in Petzold [5], taking into account heat gains due to insulation through transparent and opaque
partitions. Applying the criterion of minimum heat

requirement, the optimum relation between the lengths of building walls and the optimum number of floors were determined. Examples of the application of multicriteria optimisation in the solution of architectural problems can be found in references [&l 11.An illustration of the work in these papers is given by the example of optimisation of a prism-shaped multistorey office building in Australia [ 1 l] applying the following criteria:
l

l l

minimum thermal load ratio (the ratio of the total heating and/or cooling loads predicted for a building to that of a model building in Sydneys climate); minimum capital cost; maximum net usable area (the floor area within the external walls less the area taken up by lifts and staircases, circulation and toilet facilities). with their range of values were:

The design variables

0 aspect ratio ~ 1, 2 or 3; orientation - north, or 30 or 60 east of north; l number of storeys - 1, 2, 3, 4 or 5; l glazing fractions on each facade - 0.4, 0.5 or 0.6; l glazing type-clear, heat reflection or heat absorbing.
l

*Institute of Fundamental Technological Academy of Sciences, Swietokrzyska 21, Poland.

Research, Polish 00-049 Warsaw,

The optimum solution was the building oriented towards the north (in Australia), with a relative window area of 0.4, heat reflection glazing and an aspect ratio of unity. The number of storeys depends on the method of choosing the preferred solution. The problem of optimisation of energy-saving buildings was most comprehensively presented by Owczarek [ 121. He presented a model of solar radiation heat energy gained across the windows. In this paper, the thermal resistance of outer walls and the percentage of glazing of walls for the fixed geometry of building shape, linear dimensions of the hexagonal building perimeter at the 331

332

W. Marks
pressed in the form of a vector xT = (u,. x2,.,... u,J in an n-dimensional space called the decision space. Every point in that space corresponds to a building with it decision variables. In the optimisation of structures or buildings, unconstrained extrema of the objective function are seldom looked for. A great number of constraints are usually imposed, defining the allowable solution space (feasible region). Constraints imposed on decision variables determine the boundary of the feasible region. The feasible region usually constitutes a part of the n-dimensional decision space. The constraints occur in the form of equalities or inequalities describing certain conditions to be satisfied by a structure or building, imposed directly on the decision variables by limiting certain quantities which depend on the decision variables. They are of the form: h,(x)=0 g,(x)<0 i= 1,2 ,.... r.

imposed angles and wall angles were optimised. A bicriteria1 optimisation problem for a building of hexagonal shape in plan was also solved for the criteria of minimum cost of the materials and minimum yearly heating energy costs for the building. A multicriteria optimisation problem [13, 141 was presented for buildings of a given volume and octagonal plan. Minimum construction costs and minimum yearly running costs were adopted as optimisation criteria. The decision variables were wall lengths. building height. wall angles, window sizes and thermal resistance of individual external partitions. The influence of the number of service years of the building and the parameter defining the shape of its plan on utility function were also investigated [ 141. Adamski [ 151 formulated and solved a particular case of optimisation of a building with vertical walls and the plan defined by two arbitrary curves, adopting the same criteria as in [ 131. There are many review papers highlighting the need to find a solution to the problems of shape definition of energy-saving buildings (e.g. [ 161). In this paper the problem of optimisaton of the shape of buildings is discussed, taking into account (1) the minimum building costs including materials and construction costs, and (2) the minimum yearly heating costs. In order to ensure that the building shapes so defined can form a basis for further design, it is necessary to take into account ~ in the form of constraints - functional and construction requirements of these buildings, or, alternatively, to search for optimum solutions depending on certain parameters which shall be defined subsequently in such a way that those requirements are met. The object of this paper is to determine the optimum dimensions of a building of the volume V. The optimum values of thermal resistances of walls, roof and floor can be determined independently [ 12, 131. The solution to the problem is presented in two ways. In the first, it is assumed that the shape of the plan of the building is defined by two arbitrary curves defining south and north faces and that the south facade windows are defined by a continuous function as the percentage of wall area. This problem was solved using calculus of variations. In the second it is assumed that the building is of rectangular prism shape and polygonal in plan, and the proportions of individual wall lengths, their angles and the building height are determined using non-linear programming methods. It is not the object of this paper to obtain a practical design. Its results present information for design engineers on optimum proportions of wall lengths, their angles and glazing parameters, in view of the above criteria. The degree to which these results can be applied in practice depends on many other requirements of building design. THE BASICS OF THE MULTICRITERIA OPTIMISATION PROBLEM The basic notions in the formulation of a multicriteria optimisation problem are decision variables, constraints and optimisation criteria, also called objective functions. Decision variables are quantities describing structures or buildings subjected to variations during the optimisation process. Decision variables are usually ex-

j = 1, 2 ,__...s.

In an n-dimensional decision space. the constraints form a hypersurface containing the points fulfilling these constraints in the form of equalities. In optimisation it is accepted to call a mathematical expression describing a certain property of the structure or the building *the objective function. The property under examination can be described in the form of a function or a functional, depending on the way the optimisation problem is formulated. The assessment of the objective function is the basis of the selection of the structure or material from many other possible solutions

u71.
Multicriteria optimisation consists of choosing the best solution from many possible variants on the basis of many criteria. i.e. an objective function vector fT = (,f;, f;,...,,/i). A multicriteria optimisation problem can therefore be treated as an optimisation problem of the objective function vector, which is different from the single criterion optimisation, which can be considered as the optimisation of a scalar objective function. The objective functions space is k-dimensional. Every point in that space corresponds to one objective function vector ,/;(s,). In that space, the feasible region R is represented by the region f(O) (Fig. 1). The solution xl which makes every objective function reach its extremum independently of the remaining functions is called the ideal solution of multicriteria optimisation. In the case of the search for the minimum f(x), xld is therefore the ideal solution of a multicriteria problem if x& and f(x) 5 f(x) for every x&I. As objective functions are usually in conflict, the ideal solution does not exist in most cases. The solution in which none of the objective functions can be improved without simultaneous deterioration of at least one of the remaining objective functions is called the non-dominated solution. x* is a non-dominated solution when no x exists such thatJ;(x)<.f;(x*) at igK = { 1, 7 _,.__,k) and,f;(x)<f;(x*) for at least one ~EK. The search for non-dominated solutions is called optimisation in Pareto sense. In general, the Pareto solution is not unique. Many x* vectors usually exist, forming an effective curve of decision variables in the Q space to

Optimisation
X

of Shape of Energy-Saving

Buildings

333

feasible region

CT
X*

ideal point

\ the set of compromises

Fig. 1. Transformation

of the feasible region into the objective

region.

which the vector P = f(x*) corresponds, constituting the set of compromises (Fig. 1). In view of the great number of non-dominated solutions, it is necessary to select the best solution on the basis of an additional criterion. Such a solution is called the preferred solution. Thus, the preferred solution x P is a non-dominated solution selected on the basis of an additional criterion. It corresponds to the values of f(xld) contained within the objective region and is considered to be the best solution. A solution of a multicriteria optimisation problem therefore includes objective quantities - to which belong the set of compromises and the ideal point ~ and the quantities which depend on additional preferences - the preferred solution, i.e. the vector of objective functions fP and the corresponding vector of decision variables xpr. If there are no additional preferences, the preferred solution is assumed to be the point belonging to the set of compromises situated nearest to the ideal point and the corresponding vector of decision variables [8]. The problem of multicriteria optimisation can be formulated as follows: yEig f(x), where f: R=t=Rk the objective is function vector given by

HEAT LOSSES AND GAINS THROUGH EXTERNAL WALLS OF BUILDINGS The physical environment surrounding defined by the following values: the building is

SD - number of degree-days in a year (day K); Q,, 0, - average sums of the total solar radiation on the vertical east, south or west walls during the heating period (kWh/m) [12]; CC, conductivity coefficient for the transmission of heat from the outside to the inside (W/(m K)); CQ- conductivity coefficient for the transmission of heat from the inside to the outside (W/(mK)). Heat losses in a building include losses through the outer walls and those resulting from ventilation. Heat gains are due to solar radiation through the windows. The difference between losses and gains constitutes the part of energy that has to be supplied by the heating system installed. In the problem under consideration, only those constituent parts have been included which have an important bearing on the solution, i.e. heat losses through walls, floor slabs and roof, as well as heat losses and gains through the transparent partitions. Yearly losses through the walls, roof and floor slabs have been obtained from the formula: E, = Es+ E,+E, = 24SD.

IT(x) = {fi(x),f2(x),...,fk(x)}, and R = {xdt'lh(x) = 0, g(x)<O} is the feasible domain defined by the equality and inequality constraints. The componentsJ: R-R, i = 1, 2,..., k, are called the partial criteria of optimisation, and x is the vector of design variables. The problem of multicriteria optimisation can be solved in two stages. The first stage consists of determining the compromise set. In the second stage the preferred solution will be found. A number of methods exist allowing the compromise set to be generated. They are discussed in various publications (e.g. [18-201). The weighted objective method, the min-max method and the constrained objective functions method are often used to generate the compromise set. A few methods for the selection of the preferred solution are discussed in references [l8-201. Utility functions or matrix functions methods, constrained objective functions methods and lexicographic methods are often used to select the preferred solution.

The following notation is used: A, = area of walls; A, = area of windows; Ad = roof area; A, = floor area; R,, R,, Rd and R, are thermal resistances of the walls, windows, roof and floor, respectively;
(Pd = ctdr tdw)/(fsz kv>; pp = t&z tpnv)/(t,, tzw); t, t,,

and t,, are average temperatures of the external surface of the roof, walls and floor, respectively; and fdw, t,, and t,, are average temperatures of the internal surface of the roof, walls and floor, respectively. Daily heat gains through a vertical window of an area A, inclined at an angle c(, to the N-S axis, due to solar radiation of intensity J at an angle CC, the N-S direction to are equal to E, = AJcos(cr,a,). by 1 mz of

Annual solar gains that may be obtained vertical window have the following form:

334

W.

Marks
It is assumed that the heat gain due to the solar radiation during the heating period will be calculated from the relation:

where p is the inclination angle of the south-east relative to the N-S direction.

wall

OPTIMISATION OF A BUILDING ARBITRARY SHAPE


Formulation

OF AN
4 = 0, - ~(0, - t3,) arctan

qf the problem
Since dl = m we obtain F, = 2

y,.

(l) = dy ,

The subject under consideration is a building with vertical walls, of constant volume V and height h. The base of the building is described by two arbitrary curves y,(x) and y*(x) (Fig. 2). Heat gains through the north-facing windows will be disregarded and the building assumed to be symmetric along the N-S axis. The aim of the present considerations is to determine the form of the curves y,(x) and y2(x) using two criteria: (1) minimum building cost; (2) minimum annual cost of heating. The construction cost is defined by the following relation:
F, =2~;:{[I-p(x)]c,+p(x)c,)hdl,+2~;;hcqdl,+D,;

dx, dl co@ = dx and dl si$

lid{[ 1 -p(x)]c, s0

+p(x)c,)h,/m

dx

., +2 c0

hc,JFjpdx+D,,

(2)

Fz = 48SDc,

xh,/wdx+

where c, and c, are the cost per m* of a wall and window, respectively; p(x) is the ratio of window area to the total area of the wall; dl, and dl, are the lengths of elements of they, and y2 curves, respectively; and D, represents other costs, independent of the decision variables. The function expressing the annual heating cost is assumed to be of the form:
F2 = 48SDc,

2c,

11
0

QI- $ (0, - t3,) arctan

y I

x p(x)hJl+y;2

dx + Dz.

(3) are the functions

The decision variables in this problem y,(x), y*(x). It is assumed that [I 51:

1. v,(x) and y>(x) are continuous functions of C class within the range [0, x,]; 2. the shape of the building is symmetric in relation to the 0 Y axis, i.e.

Y!(O) = 0,
where R, and R, are thermal resistances of the wall and window, respectively; SD = number of degree-days in a year; c, = unit cost of energy; and Dz represents other costs independent of the decision variables.
y2(0) = 0; 3. the functions

(4) (5)

J?,(X), y2(x) bound a region of area V/h,

i.e.

1
Y

r
N

Y t

1B,(O>Y,(O))

,o
WP)
O=O

s,.s,= s

B2(O>y2KV)

Fig. 2. The form of the building.

Optimisation

of Shape of Energy-Saving
to the form

Buildings

335

rS:.[,.,(*_)-l.,(x+ =V/h;
Yl(X.) = Y*(x,) = 0; the ratio of window stant.

(6)
----2/I, dx On integrating equation

dv(x)

= 0. (14) we obtain

(14)

the functions y,(x) and y*(x) equal zero at the point, with the abscissa x,: (7)

v(x) = 21,x+ K. From condition K = 0. Hence (4) we find the integration

(15) constant

area in the south wall ~(1,) = con-

Solution of the optimisation problem A set of compromises can be determined by the method of weight coefficients. We are seeking the minimum of the objective function F=iF,+(l-1)F,, where IL < 0, 1> , with the assumption is satisfied. On the basis of equation substitute functional F* = A :~~dx+B~arctan2y,vi+)i;idx s +C s -21,V/h+D, where A = 2/2h[( 1 -p)cs +pc,] + 2( 1- l)h24SDc, $1 -p)+ $ -2h(l -i)pc,Q,, (9) ;Jmdx+21,[;[y,(x)-yz(x)]dx that condition (8), we obtain (8) (6) the

x = f (a_~,+2Barctany,

+2Barctan2y,)-----

Jl+v;z (16)

Since equation (16) cannot be solved with regard to y,, we will proceed in the following way. We denote x = $(Yl) = 4(P), dY, =pdx, dx = f(p) dYl = (A+2B+Barctan2YI 1 y, = I(--A+2Barctany,+Barctany,) 1 x------f-. J_ M 2A dp, Y; I )(,ll+y;2)1.

0I

The integration constant M should be determined from condition (7). Similarly, on integrating equation (11) we have
Y2W

B = 2(1 -i)hc,p&,

-Q,), Jmthat is s

21 I x C

(17)

C = 2hic, + 2( 1- /l)h24SDc, ;, D = iD,+(l-1)D,. This is an isoperimetric problem of variational calculus [21]. The conditions (4)-(7) enable us to determine the integration constants and the constant 2,. The functional [equation (9)] reaches its extreme value if Eulers equations [21] are satisfied, that is (A +2B+Barctany,) 1 (J=.Y% y,+2& = 0,

h(x) = J&p I I
Since y2(x) > 0 for x@O, x, > , we have

(18)

y2(x) = J&. Hence, on integrating, we have cz.

(19)

YZ(X) = - J_+

(20)

(10)

From condition

(6) we find the constant c, = JW.

c
(JG% Equation

y,+2E,,

= 0.

(11)
Hence

(4) can be reduced,

by substituting yz(x) = -J_+J_. (21) 1 Jm (12) We obtain: [y&Y(c/2n,)-x,]+x = (A/U,).

u(x) = (Ay,+2Barctar1y,+By,arctan~y,)~

__ dx

W-4 = (A+2B+

Barctany,)

1 (.&+r (13)

This is an equation of a circle with its centre at the point O2 (0, C,) and radius Rz = C/2L,. The area of the segment of that circle bounded by the OX axis is, for y2(x) < 0,

336

W. Marks The ideal point C and the preferred solution D have been found as the point within the set of compromises, nearest to the ideal point [20]. The shapes of building plans corresponding to various values & 0 <i 5 1, are shown in Fig. 3. The coefficient i can be interpreted as the service life of the building. ). = 1 corresponds to point zero of the service life, i.e. it takes into account the construction cost only. i. = 0 corresponds to an infinitely long service life.

[J,m-Jm]dx i II - x~J~.

(22)

= (C/21,) arcsin (x.i,/C)

The curve y,(x) was determined in parametric form. A suitable segment of the area has been determined numerically. The constant I, has been determined from equation (4) solving the system of equations

SHAPE

( 1

4 n ;Y.+Yt+ (C/U,)

arcsin

( 1
7 = 0, (23)

2x*4

OPTIMISATION POLYGONAL

OF BUILDINGS PLANS

ON

-x,Jm-1
1,x, =

(Ay,,+2Barctany,,-By,,arctan2y,,)

x
where

&,

i =O,l,...,n,

(24)

1 J:,, = I(-A+2B.y,,arctanyl,-Barctany,,)

xJ&-;I
x (-A +2By,, arctanylnBarctan~,~) J_ 1

The system of non-linear algebraic equations (23) and (24) has been solved by the CAMOS computer system. Numerical example The data used in the computation were as follows: c, = 40 PLZ/m, c, = 400 PLZ/m, c, = 0.135 PLZ/kWh, SD = 4000, l/R, = 0.72 W/m2 K, l/R0 = 1.6 W/m2 K, p = 0.3, 0, = 250kWh/m, 0, = lOOkWh/m, V = 1 m3, h = 1 m, x, = 0.8 m. The results are shown in Fig. 3, i.e. the compromise set AB on the normalised coordinate system, where the functions 4, and 42 are
4,

Formulation qfthe problem The subject of our consideration is a building with vertical walls, constant volume V and height h. The base of the building is octagonal (Fig. 4). It is assumed that the building is symmetric along the N-S axis. Heat gains through the windows in the south, south-east, east, southwest and west walls are taken into account. The aim is to determine the relationship between the lengths of the walls and their angles. The shapes of the building obtained in Owczarek [12] and Adamski and Marks [ 131 cannot in many cases form a basis for further design work, because of the contradiction with other functional and structural requirements. In view of this, an additional constraint was imposed, concerning the minimum area of the rectangle inscribed in the octagon (Fig. 4). Yearly heat gains have been calculated from the formula 0 = 0,cosb+&sinfi The two following optimisation criteria adapted: minimum construction cost, F, = i I,h(l -p,)c,i+ /= I and minimum F2 = i +l,h(l /=I .,i i pJ,hc,,+B,; have been

,= I

(25)

yearly heating cost, -p,) + i $p,l,h i=1 sz

=A

F,(i = 0)

42 =

Fz

24SDc,

F,(3 = 1)
+ ~P,M~zI~, + B2, (26)

where c, = unit cost of energy; c, and c, are costs/m2 of wall and window, respectively; B, and B, are other costs, independent of the decision variables adopted. The decision variables are: lengths of walls, I,, i = 1,...,5; @)20.9

0.81

Fig. 3. The compromise set, the ideal point, the preferable point and the corresponding shapes of the plans of buildings.

Fig. 4. Shape of the building and notation adopted.

Optimisation of Shape of Energy-Saving Buildings


ratios of window areas to wall areas,p,; inclination angles, Jl and y, of south-east and north-west walls, respectively, to the N-S direction (Fig. 4). The variables must fulfil the following constraints: condition of closure of the polygon, I, + 21, sin b - I, +- 21, sin y = 0; condition of the constant volume of the building, (27)

337

Altering the weight coefficiens /I within the range 0 to 1, we obtain the compromise set. The coordinates of the optimum point within the objective space are Fl = Fi,inr
F2 = F2mm

In order to find the preferred tion (28)

solution,

the utility func-

h[(l, + I2sin /I)& cos /?+ (Ii + 21, sin /I)& +(Z,+I,siny)l,cosy]condition concerning the magnitude area inscribed in the octagon, V= 0; F= F,+NF2 may be formed, in which N denotes the modified number of years of service life of the building, which is the number of years multiplied by a coefficient taking account of the interest rate and inflation. The following is the relationship between the modified number of the service life years and the weight coefficient: N=IZ. (31) 1-n

of the rectangular

(I, + 21, sin p)13 = I>; constraints regarding 1,20, constraints the linear and angular O<Br71/2, oIyI7c/2;

(29) dimensions (30)

concerning

the size of the windows = I,...) 5.

pisp,<g,,i

Solution of the optimisation problem. The set of compromises can be derived analytically, e.g. using the method of weighted coefficients [13, 221, creating a substitute objective function (equation (32)) or by numerical methods, e.g. using the optimisation program CAMOS [23]. The substitute objective function, taking account of the equality constraints (27)-(29) and the inequality constraints (30) and (3 l), has the following form: F= nF,+(l-1)F,+~,h(l,+21,sinP-1,-21,siny) + pz{ [(I, + Iz sin P)& cos B + (I, + + (& + l4 sin y)& cos y]h - V} + P~[(&+ 21, sin 8Y3 - Dl
+ dpl -/d-t

The preferred solution corresponds to the minimum construction and heating cost during N years. In this paper the problem was solved numerically using the computer system CAMOS [23]. It is an interactive system for the solution of single criterion and multicriteria optimisation problems. The system allows the solution of problems of the following dimensions: 10 objective functions, 20 decision variables, 100 inequality constraints and 15 equality constraints. NumericaI example The numerical example of the optimisation of a building was solved using the CAMOS computer system. The following numerical data were used: c,, = 27 PLZ/m, C o, = c, = 0.025 PLZ/kWh, 20 PLZ/m, SD = 4000, k,, = 1/R,, = 0.75W/m2K, k,, = 2.6W/m2K, p, = 0.1, pi = 0.4, 0, = 350 kWh/m*, e2 = 120 kWh/m2, V = 1 m3, h=lm. The problem was solved assuming, in turn, N = 0 (construction cost only), N = 10, 25, 50 and N (heating costs only). For each N, two cases of the ratio of the rectangle area D to the plan area of the building A (Fig. 4) were considered: D/A = 0.7 and 1.0. For each of these two cases the set of compromises and the optimum point were defined (Figs 5 and 6). The corresponding shapes of the

21,sin BY3

+ W - 742) V,(Y +
v,(p, -P*)

742)

VdPI --I%)+ -Pd

+ VdP, -a21 + Gp3

+ VdP, -831,

(32) objective theorem

The necessary conditions for the minimum function (32) resulting from the Kuhn-Tucker (221 have the following form: 3F x--o, 1ax,

x, LPI, =

P>Y?PI.

P2> o,,

i=l,..., l?F z >O,


I

5,

u=l,...,

8,

(33)

x,, = LP,> P>Y>

8F ax-,
l,>O, i = l,..., 5,

CO, /I>O,

x. = vu, y>O, v,>O, u = l,..., 8. that the

From the form of relations (32), it follows ratios of window areas to wall areas are PI = el orpi = Pi,
p2 = p2 orp2 = B2,

1
a

a
a

Fig. 5. The compromise


p3 = p3, orb = p3

responding

set, the ideal point and the shapes of building plans for D/A = 0.

338

W. Marks

0.8 -

Al

The relationships between the particular partitions in a building depend on the unit prices of non-transparent partitions, windows, heating energy, number of degreedays, total sum of the solar radiations and the period of service of the building. Introducing the factor of the cost/m of construction and heating of the partition W
in the form

Y = [(I -~,k,+wo,l+

NC, (1 -P,)~
~

24SD V

-U,cosfl,-&sin/?,

Fig.

6. The compromise set, the ideal point and the corresponding shapes of building plans for D/A = 1.O.

plans of buildings are also shown in these figures. Dimensionless wall lengths q( = l,/,,&.

CONCLUSIONS The optimisation problem of the shape of a building with an arbitrary base has been solved by variational methods. The solution obtained is composed of a circular segment bounding the northern part of the building and a curve described in parametric form describing its southern part. The ratio of the area of the southern part to the northern part depends on the size of the windows, the density of solar radiation energy and the ratio of unit costs of the windows and walls. They increase with the number N, which determines the modified service time of the building.

it can be stated that the size of the,jth partition should increase with the diminishing factor IV,. Depending on the data input, the plan of the optimum shape of the building can be an octagon or a polygon with a fewer number of sides. In the case of short heating periods, construction costs are of great importance and the shape of the building approaches a regular octagon. Depending on the unit prices of walls and windows, and on total solar radiation, the ratios of window areas to the areas of the relevant walls assume the highest or the lowest allowable values, As a result of the optimisation of the shape of the building, construction and heating costs for the N year period can be lowered by several to several dozen per cent. In an area close to the minimum, the function of the increase of costs is flat and the differences are of the order of several per cent. The gradient of the cost function becomes steeper as the distance from the minimum increases.

REFERENCES
1. Klosak, D. and Klosak, A., Architectural problems in the design of energy saving buildings. In Problems of Design, Erection and Maintenance c~/ Buildings with Low Energy Demand [in Polish]. Krakow-Mogilany, Krakow University of Technology, 1993, pp. 101~109. Fokin, K. F., Thermal Technology ofthe External Building Elements [in Russian]. Stroyzdat, Moskwa, 1934. Gadomski, J., Analysis of influence of the architectural concepts on heat loss amounts in buildings and some predictions of this field. Institute of Building Technology, Warszawa, 1987. Menkhoff, H., Blum, A., Trykowski, M. and Aapke, W., Energetisches Batten. Energiewirtschaftliche Aspekte zur Planung und Gestaltung von Wohngebauden. 04.086/1983, Schriftenreihe Bau-und Wohnforschung des Bundesministers fur Raumordnung, Bauwesen und Stadtebau, Bonn, 1983. Petzold, K., Zum EinfluB von Form und GroBe der Gebaude auf den Heizenergiebedarf. Lgft-und Kaltetechnik, 1983, 19(3), 130-135. Gero, J. S. and Radford, A. D., The place of multi-criteria optimization in design. Design Policy, Royal College of Arts, London, 1983. Marks, W., Problems in the multicriteria optimization of the energy conservation buildings [in Polish]. Zesz~tv Naukowe WSI w Zielonej G&e, 1989, 88, 15525. Markus, T. A., Cost-benefit analysis in building design: problems and solutions. Journal of Architectural Research. 1976, S(3), 22-23. Radford, A. D. and Gero, J. S., Tradeoff diagrams for the integrated design of the physical environment in buildings. Building and Environment, 1980, 18, 3-15. Radford, A. D., Gero, J. S. and Murthy, N. S., Designing Optimal Multtfimctional Surface Materials, ACMSMB. The University of Newcastle, Newcastle, 1982. Radford, A. D., Gero, J. S. and Cruz. N., Energy conservative design in context of the use of multicriteria decision methods. In Energy Conservation in the Design of Multi-storey Buildings, ed. M. J. Covan. Pergamon Press, Sydney, 1984, pp. 75-88. Owczarek, S., Optimization of the shape of energy-saving buildings on the plan of a polygon [in Polish]. Studio z Zakresu &,ynierii, Warsaw, 1993. Adamski, M. and Marks, W., Multicriteria optimization of shapes and structures of external walls of energy conservation buildings. Archives of&i1 Engineering, 1993, 39(l), 77-91. Jedrzeiuk, H. and Marks. W., Analvsis of the influence of the service life and shape of buildings on the cost of their construction and maintenance. Archives qfCizil Engineering, 1994; 40(3/4), 5071518. Adamski, M., Optimization of the form a building with an arbitrary base. Engineering Transactions. 1994, 42(4), 359-376.

2. 3. 4.

5. 6. 7. 8. 9. 10. 11.

12. 13. 14. 15.

Optimisation
16. 17. 18.

of Shape of Energy-Saving

Buildings

339

19. 20. 21. 22. 23.

Engstrom, L., Energy in the built environment. Swedish Council for Building Research, Stockholm, 1988. Brandt, A. M., ed., Criteria and Methods of Structural Optimization. PWN and M. Nijhoff Publishers, Warsaw, 1984. Hwang, C. L. and Masud, A. S. M., Multiple objective decision making-methods and applications a state-of-art survey. Lecture Notes in Economics and Mathematical Systems. Springer-Verlag, Berlin, 1979. Jendo, S. and Marks, W., On multicriteria optimization of structures [in Polish]. Archives of Civil Engineering, 1984, 30(l), 3-2 1. Jendo, S. and Marks, W., Problems in the multicriteria optimization of civil engineering structures [in Polish]. Engineering Transactions, 1986, 34(4), 457461. Gelfand, I. M. and Fomin, S. V., Calculus of Variations, trans. R. A. Silverman. Prentice-Hall, Englewood Cliffs, NJ, 1963. Findeisen, W., Szymanowski, J. and Wierzbicki, A., Theory and Computing Methods in Optimization [in Polish]. Polish Scientific Publishers. Warsaw. 1977. bsyczka A., Computer Aided Multicriterion Optimization System (CAMOS). International Software Publishers, Krakow, 1992.

Potrebbero piacerti anche