Académique Documents
Professionnel Documents
Culture Documents
BY
G.P. MITALAS AND J.G. ARSENEAULT
OTTAWA
JUNE 1972
CP 33
I
•
I;"
by
OTTAW A
June 1972.
Fortra n IV- Progra m to Calcul ate z~Transfer
Functi ons for the Calcul ation of Transi ent
Heat Trans fer throug h Walls and Roofs
1
G. P. Mitala s and J. G. Arsen eault
e0 A B 8.
1
Q
0
c D
Q D -I 8
0 0
B
Q. -A 8.
1 1
e0
e.1
the coeffi-
The prograJ TI presen ted in this paper evalua tes
of a set of z-tran sfer functio ns that are equiva lent to the
cients
1/C and D/C.
Laplac e transf er functio ns D/B, 1/B, A/B, A/C,
zwtran sfer functio ns relate to the z-tran sforrn s of the surfac e
These
in the same way as their counte rpart
tempe rature s and heat fluxes
e transf er functio ns relate to the expres sions above,
Laplac
633
The program will evaluate z-transfer functions that are
exact for either a unit step input, a ramp type input or a periodic
input with specified harmonic components, The user can choose,
therefore, the option that beat suits a particular problem,
The heat transmission matrix for a wall or roof has elements A, B, G and D, i.e.,
::1 c
The elements A, B, G and D are functions of the Laplace parameter, s, and of the thermal properties,
thickness and position of materials in the wall. When the boundary conditions are of the first kind
(i.e. when 9 and 6, are given), the fluxes are given by
0 1 .
and when boundary conditions are of the second kind, the equations invert to
-I Q
0
-D Q.
1
The program presented in this paper evaluates the coefficients of a set of z-transfer functions
that are equivalent to the Laplace transfer functions D/B, 1/B, A/B, A/G, 1/G and D/C. These
z-transfer functions relate to the z-transforms of the surface temperatures and heat fluxes in the
same way as their counterpart Laplace transfer functions relate to the expressions above.
The program will evaluate z~transfer functions that are exact for either a unit step input, a
ramp type input or a periodic input with specified harmonic components. The user can choose,
·therefore, the option that best suits a particular problem.
The z-transfer functions for a multilayer wall can be calculated in two ways:
634
Method 2 involves solving a set of simultaneou s linear algebraic equations
to obtain the coefficients of a z-transfer function whose frequency response
matches the exact frequency response of the multilayer slab at certain selec-
ted frequencies ,
The z-transfer function correspondi ng to any one of the Laplace transfer functions can be ex-
for all
pressed as the ratio of two finite polynomials , N{z)/D(z). The denominato r, D{z}, is the same
denominato r in their Laplace transfer function equivalents ,
the transfer functions that have a common
is the same for Method l and Method 2. The procedure for finding the coefficients of the denom-
and
inator polynomial involves two steps.
The elements of the transmissio n matrix for a wall have an infinite set of roots, which lie
along the negative real axis in the s-plane. The position of the roots depends on the di-
mensions and thermal properties of all the layers, and cannot be expressed in any simple
way. The necessary ·poles can be found numerically , however, by evaluating the functions
B or C for a sequence of negative real values of s. This program evaluates the roots of B
between zero and -30/fi, and the roots of C between zero and -450/t:., where /::,is the
specified sampling inter~al of the z-transform .
-~ 6
n -1
D(z) =IT (1 - e z )
when the parent Laplace transfer function has the element B in the denominato r, or
-1 . -Y 6
n -1
D(z) = (1 - z ) IT (1 - e z )
Methods l and 2 differ only in the way the numerator polynomial is determined . Method 1
input)
requires the evaluation of the time function that correspond s to 1/s (step input) or to l/s2 {ramp
times the appropriate Laplace transfer function, for t :: fi, 26, 3£.1,...... The coefficients of O(z)
are evaluated by finding the residues of the Laplace transfer function at the previously determined
poles.
N ( z) = :c:_D.,_,(z'-')--;~--"-o,_,(z"-)
I(z)
6
or l(z) = for a ramp input.
-1 2
z( 1 - z )
635
the wall at s = iw, and the
Method 2 requires the evaluatio n of the Laplace transfer function of n
1'w 1::.
angular velocity at which the
calculati on of the denomin ator D(z) at z = e n , where Wn is the
gives a pair of equations for each
z-transfe r function is to match the exact frequenc y response . This
real and imaginar y parts are equated s_eparate ly) except at Wn = 0. 0 (i.e. steady
value of Wn (i.e.
The resulting set of equations for a series of
state) where only the real part of the equation exists.
values of Wn can be expresse d in matrix form, viz.:
ao X(O)
2. 1 Input
Card 2 and 3 Descript ion of the slab for title purpose only.
Format: (80 Al)
on of the
Cards 4 to I Groups of cards giving thermal propertie s, thicknes s, and descripti
e, the first card of the group contains values of
layers. Wheneve r applicabl
thicknes s of layer,
thermal conductiv ity,
density,
636
speci fic heat, and
resist ance of radia tion path.
al resist ance of a layer that has
Other wise, the first card conta ins the therm
neglig ible heat stora ge capac ity,
Form at: (SF 10.4)
the group can be used for the descr ip-
The secon d, third _.,, or more cards of
ted in Coiur rm One.
tion of the layer if an integ er is inser
Form at: (30 Al)
of therm al pr-op erties and their de-
Card I+ 1 Blank card to termi nate the above input
scrip tions .
frequ encie s, NW, to be fitted when
Card I+ 2 Code numb er, !CAS E, and the numb er of
Meth od 2 is to be used (see Table 1).
Form at: (11, 11)
or 5, It speci fies the perio ds of the
Card I+ 3 This card is read only when !CAS E = 2
in frequ ency respo nse calcu lation s.
harm onics to be used
~
n Meth od 1 Meth od 2
Inval id
First Comb i-
natio n 1 2
Kind
Secon d
3 4 5
Kind
3. Refer ence
funct ions for
Calcu lation of heat condu ctioil trans fer
[1] Steph enson , D. G. and Mital as, G.P. , ry 1971.
AE for prese ntatio n Janua
multi layer slabs , Subm itted to ASHR
637
INPUT SUBROUTINE
* SUBROUTINE
*
(START FROM;/
DT -,.
\ POLES r \ MATRIX
CARDS
'o.o
TERMINATE
PROGRAM
CALCULATE '2 OR 5
THE COEFFICIENTS I CASE
OF D !ZI
•1,30R4
METHOD I
EVALUATE
INVERSE
LAPLACE
CALCULATE
RESIDUES AT
CALCULATE
RESIDUES AT / SUBROUTINE
*
TRANSFORMS THE POLES THE ORIGIN \ ORIGIN
AT t•A,2A ...
4 METHOD 2
638
LAST • 0 R2• Rl +
IROOT •I
SET Rl
I ~ ! ~.r
20.0
AND R3
CHECK FOR
TWO CHANGES
IN SIGN
FPI ~ FP2
STORE: ROOTS <0
FUNCTIO NS
AND DERIVATIVES I -21
IN ITS PROPER
ORDER ~0 NO YES
REPLACE EITHER
RI OR R2 BY ,.---_ _,;l._
RTEMP SUCH c¢ 1 •I
r----f THA T Rl ANO R2 LAST
ARE OPPOSITE **NOT E
IN SIGN •0
~_L~·
SUBROUTINE
YES IF b4ATRIX
HEMP
I0-14 Rl + Fl
Rl ~ROOT (LAST-I
R3•ROO T (LAST)
YES
CHECK FOR ~-- ~"'
SUBROUTINE
NO >--~--J CHANGE IN MATRIX
SIGN FROM R2-+F2
F I+ F2
-----~---__J
J-NN~~-~-----------~----
~0
1- 25 '>--+---1
>0
SUBROUTINE POLES
639
START) I • I """"" T
ELEMENTS OF
MATRIX FOR
LAYER I
fl Ill
ARE
DERIVATIVES
ANTED
NO
!'It I
CALCULATE
~;>
DERIVATIVES
F •0.0 OF ELEMENTS ~0
OF MATRIX ~-M
. •0 F2 (IJ
F3(JI~FUJJ
>0
~
F=F+F3tJ-U
* F3!Jl
I • I
F" J ' I
FF • 0.0 " M
•0 ' I
F31Ji=F
YES ARE F: F I CI I
- F3!J)= F2UJ DERIVATIVES FF=f2111
WANTED
NO
FF•FF+F
J=J+I
( RETURN
~0
J-M
~ r
>0 J, I+ I -:>- M I= 2 F •0.0 1--
. >0
f:ftfi(I-1)
>0 ~ * Fl III
~
1•1+1 Flili'F
0
SUBROUTINE - MATRIX
Figure A-3, Flow diagram for the eva:tuation of A, B, C, D, and their derivatives for any real
negative value of s,
640
•0
TEr.lPI"'
"'
W!--------~-----:>--
··---·-~~-------· ---- ------ ---
SUBRROUTINE ORIGIN
641
CALCULATE
J =I TRANSMISSION ¢0
, MMMM=O
( START ) ,
I=l
MATRIX ELEMENTS
FOR W IJI AND
I-1
LAYER I
=0
MMMM=
MMM=MM MMMM+
MM*MMM
(RETURN)
>0
MMM=MMMM
fO
J-LW
SUBROUTINE - FREQUE
Figure A-5. Flow diagram for the evaluation of A, B, C and D for pure imaginary. arguments,
642
APPENDIX A A-1
.
00022C'OCC CAR!:I(3)*
00023G OCC
0002U0QCC C1.PD(4) *
Ot025COCC "
*
XL [I) , XK(I) ,D {I), SH (I) , RES (I) , {TEXT {I ,J) ,J=1, 30) WHERE
."'
~·
00026C'OCC
0002700CC
0002800CC
OOC29'JOCC
0003000CC
CARD {I+3)
"
"
CARD {H+3}
~HE?.E
.
* I ISDICATES T!-iE I'TH LAYER OF !BE SLAB (5F10.4,30A 1) •
*
Xl=THICKN fSS 0? LAYEP..
0003100CC XK=TRER~AL CON'JUCT!V I:Y.
0003200CC D=~E'NS!'!'Y·.
0003300CC Sa=S?ECI~IC HE\~.
0 0034( occ RES=P.ESIS7 A:.lCE •); P.ADIATION P'-TH 'liF.E~·EVER APPLIC.\BLE
0003500CC OP. TtlER:'IAL lESISTA:lCE OF LA!IR '!il'gEN" THERE IS
0003600CC NEGLIGIBLE 3~rtT SfO?.AGE.
00037COCC TEXT=!)ESC RIPTIO!! 0? LAYER, A SECOND CARD AND SO ON CAN BE
0003800CC USE:U 3Y I:lS~P.'n!:G A~Y INTEGER IN CO:.UMN ONE n).
OC039CIJCC I:'I=!IU~3:::R OF LHE?.S 'IHE SLAB IS CCM?OSE~ OF.
OO(I!.I.OCCCC CARD (:'1+4) :SLJ.:-t!< CA!t;) !0 S~CP ABOVE IS PUT ..
0004100CC CARD{~+S) ICASE,llirl {I1,I1)
0004200CC WIH.:?.::::; :;:.:=:;u:'!B:=:~ C~ F?.EQUE~:CIES TO BE USEOJ iHE.N
0004300CC FREQUEliCY E£S~ONSE IS I!iVOLVED.
0004·4( CCC CARD{M+6) 'ii'{2) ,¥:(3) .............. ........ w(l:ir:+1) {8F10 .. 4)
0004500CC AB07E CA3D 0~~~ REA~ iriHER F&EQUE~CY FFSPOJSE IS
occu.toocc l~lVOlVE:l. (ICASE=2 0?. ICAS!=S} i (1) IS SET TO 0.0
OOC470'JCC iol {I) ~s A2E ::G: ?E?.IODS ..
000413COCC
OOC49COCC NOl'iE,CLATU ?.E:
ooosocccc R?.=1RICKNZSS/:E~3~A: CON~Uc;:VI!Y (Xl/XY.)
COOS 1 ~ C!:"C 0?. TEE?.f.A:::. ?.EZIS7A:;cz OF lAY::? ~;;Ell ':P.EP.E IS
0005200CC NEGLIGIBLE BEAT S~03AGE. THE~ P-ES=O.O
cocs;c.occ B!:i' A.*3E7 A= XI. *XL*!i* S:i/XK
?OR ICASE=~.
OC·05.:tC·CCC C0=1'3E?.~.;!. CONDUC'!Al' CE AN~ USED AS 1/C'A.':: THE POLE
A-l
0008300C 1,2X,' 1
,2X, 1 '.,4X,'DESCRIPTION
0008400C 20P' LA YE3. 1 / }
OC08500C CO=O.O
ooossso I=O
OCOS€0 0 DO 10 :1=1,1+0
0008650 I=I+ 1
0008700C B=!
0008800C XL {I) ,U': (I) ,0 (I} ,58 {I}, RES (I)., (TEXT (I,J) .,J=1,30)
E.EA.O {CARD, 6)
0008900C 6 FO?.~AT( 5?10.4,30!1)
OOC9000C I?{RES(I).EQ.O.O.AND.SB(I).EQ.O.O.AND.XL(I).NE.O.O} GO XO 20
OC09100C I?,RES{I).EQ.O.O.A::OD.XL(I}.:EQ.O.O) GO TO 30
occnooc I? {XI. (I).:.::::. O. 0) GO ':0 40
00093C0C R?. (I) =RES (I) '
OOC9UOQC BE?A (I} =0. 0
00095COC GO :o 50
OC09600C 40 P.P. (I) =XL {I) /XY. (I)
0009700C E~~A (I) =XL (I) *DSQRT (D (I) *SH (I) jXK(I))
OOC9800C CO=C({+P.:.:s {i)
0009900C 50 CO=CO +P.r. {I)
0010000C li'R.ITE (P3I~':', "1) I ,XL (I) , :Y:K {I) , D (I) , SH (I) , E:E.S {I} , (TEXT {I, J) ,J= 1, 30}
0010100C 7 ~0~~1':{1X,I3,P'10.~,F15.C,!18.4,F15.4.,F20.4,13X,30A1)
0010200C GC TO 10
0010300C 20 w::r:~{P?.I~'.:',S) {TEXT(t,~-r} ,J=1,30)
OC1Q400C 8 FOR~;). T (9SX,30.d1)
0010500C I=I-1
0010600C 10 cat:::r:wz
A~3
0010700c rl:=21
00Hl800C 30 H=M-1
001090CC DO 60 I:::1,M
0011000c IY(XL(I}.N E.O.O) GO TO €0
0011100C RES(!)=O.O
0011200C £0 COS"TINU:g
001j300C co=t.o;co
001 400c W?.::;::l'E (P?.I!:l', 1)
00115C'OC WRI':'? (?RI!f'!, 1}
0011500C ~?.IT~{P3I~T,9) CO
ootnooc 9 :F'Ot~J..T(30X, 1
'!'HER!'l.AL CONDUCTAN CE, U>=',F6.3, 2X,'
001 1800C 1 'l
00t1900C WRITE (PRIN'!';.1)
001200(\C WUI73 {PUHlT, , ) D1'
001~100C 11 ?OP.!1A1:...(39X. 1 $AI::.PI.ING TillE I.Nl'~P.V!L, DT"'-',!'9.3 ,' ')
0012200C rT:g_I~3 {?RI:O:T, 1)
0012300C P.:SAD (CARD, tl) ICA.SE,Ntl'
0012400 12 FORaA!{ ~1,11)
0012500C Nli=:n+1
00126C'OC MY=:'/'11'*2-1
0012700C I? {IC AS:E:. '!tE. 2. AND. !CASE. NE·. 5) GO TO 6.5
0012800C F!E~D {CA?.), 13) {W (I) ,:L=2,N«)
0012900C 13 YO?.f:AT( BF10.q)
00130')1)( DO 61 I=2,~lol'
00131COC 61 t.; {I) =2.0*3.1L~.1592.65/W (I)
00132C0C 65 IF(!CASE.L T.3) GO TO 70
.,. 0013300C IX=2
...
~
0013400C J'X=1
0013500C GO 'l'O 80
0013600C 70.IX=1
0013700C J'X=2
00138QCC 80 CALL POL:S(2R,BZTA,RES,P.OO~,DEP.,?ONC,~,IX,JX,DT,IP.OOT,ICASE)
0013900C DO 90 I=1, 100
0014COOC P0:.1 (1)=0.0
00141C'')C P0!.2 {I) =0. 0
0014:"00C F0!.3(I)=O. O
00-14300C POL4 (I} =0. 0
00144CCC P01.5 (I) =C. 0
001U.50CC 90 P016(I)=O. O
00145C'OC !'!:1:1=1
ooHnooc N:l:!=O
001480CC POI.1 (1)=1.0
001:.91)0( DO 100 I=1,IROOT
0015CCC.C POL2{1)=1 .0 .
001~100C POL2{2)=-~EXP{-P.OOT(I}*DT)
0015200C IF{DABS{P OI.2f2)} .LT.1.0D-1 6)GO TO 110
00153(:0( CALL PO:.Y1~ (POL 1, POL2, Nt:l~, ~!1i'1)
0015UOOC 1 oo co~::nsu:.:
O~t::;t0C 110 I?(j:C:OSE.L 'T'.3) GO 70 12C
OOlSCOOC FC12{1)=1 .0
0015700C rotz (2l =-1.0
c-o rs~~cc CA !.L ?C!. Y:'! (POL 1, ?OL2.,..ti~!!l ~:'!r:l'!}
0015900C 120 ID=N~~+1 '·
001COOOC IF(ICASE.E oJ.2.0R.ICA SE .. E:Q.S} GO TO 130
A-4
"'
:1>
>'
A-8
FORTRAN IV G LEVEl 18 MAIN DATE = 70132 14/14/56 PAGE OC'Ol
c SUBROUTI~E POLES(RR,BETA,RES,ROOT,OER,FUNCtM,I!,!J,OT,IROOT)
c
c SUBROUTINE TO CALCULATE THE ROOTS Of THE HEAT TRANSFER MATRIX
c AND WILL STORE THE VALUE OF THE FUNCTIONS AND THE FIRST DERIVATIVE
c AT THE ROOTS.
c
c THE MAXIMUM NUMBER OF ROOTS THAT CAN BE OBTAINED IS SET
c AT ONE HUNOREOtlOOI
c
c THIS METHOD WILL FIRST FIND A ROOT BETWEEN 30.0/0T AND
c 108.0/DT, BEING ASSUMED THAT A ROOT EXIST IN THIS INTERVAL. THIS
c ROOT IS ALSO LARGE ENOUGH TO GIVE SUFFICIENT ACCURACY TO EVALUATE
c THE RESPONSE FACTORS.
c THE METHOO CHECKS THE INTERVAL BETWEEN THE ORIGIN AND THIS
c FIRST ROOT AND WHEN ANOTHER ROOT IS FOUND THE INTERVAL NEXT TO BE
c CHECKED BECOMES THE INTERVAL BETWEEN THIS NEW ROOT AND THE NEXT
c LARGEST ROOT AND SO ON. WHEN NO ROOT EXIST IN AN· INTERVAL THE NEXT
c SMALLEST INTERVAL IS SELECTED AND SO ON WORKING TOWARDS THE ORIGIN
c UNTIL ALL ROOTS ARE FOUND.
c TO CHECK FOR A ROOT THE METHOD SUBDIVIDES THE INTERVALS IN
c RELATIVELY LARGE SEGMENTS AND CHECKS FOR BOTH A CHANGE IN SIGN OF
c THE FUNCTION AND FOR TWO CHANGES IN DIRECTION OF THE SLOPE OF THE
c FUNCTION. lF A ROOT EXIST, BY MAKING THESE TWO CHECKS, IT IS
~
c INDICATED SO IN A RELATIVELY SHORT TIME. ONCE IT IS INOICATEO THAT
~
0
c A ROOT DOES EXIST IN A CERTAIN SEGMENT OF AN INTERVAL, THIS
c SEGMENT IS FURTHER SUBDIVIDED AND USING A SIMILAR ROUTINE AS ABOVE
c EXCEPT CHECKING FOR A CHANGE IN SIGN OF THE FUNCTION ONLY. IF ON
c THE FIRST PASS A CHANGE IN SIGN IS NOT FOUNO THE SEGMENT IS FURTHER
c SUBOIVIDEO INTO EVEN SMALLER PARTS UNTIL A CHANGE IN SIGN DOES
c OCCUR. ONCE A CHANGE IN SIGN OCCURS THE ROOT IS ARRIVED AT BY
c SPLITTING THIS INTERVAL SUCCESSIVELY IN HALF USING THE NEW SEGMENT
c WITH FUNCTION VALUE OF OPPOSITE SIGN UNTIL A ROOT IS REACHED
c WITHIN AN ACCURACY OF 10-14.
c THE SPLITTING OF THE SEGMENTS TO ARRIVE AT A ROOT IS USED
c BECAUSE A RELATIVELY CONSTANT NUMBER OF ITERATIONS ARE REQUIRED
c TO OBTAIN THE ACCURACY WANTED. IN THE CASE OF THE REGULA FALSI
c METHOD IT WAS FOUND THAT THE NUMBER OF ITERATIONS VARIED FROM AS
c LOW AS FIVE (5) TO MORE THAN THREE HUNDRED (30CJ ITERATIONS. IN THE
c LONG RUN IT WAS FOUND THAT THE SPLITTING OF THE POINTS REQUIRED
c LESS RUNNING TIME.
c
c NOMENCLATURE:
c RR=TH!CKNESS/THERMAL CONDUCTIVITY IXL/XKI
c OR THERMAL RESISTANCE OF LAVER WHEN THERE IS
c NEGLIGIBLE HEAT STORAGE
c BETA*BETA=Xl*XL*D*SH/XK
c WHERE O=DENSITY.
c SH=SPECIFIC HEAT.
c RES=RESISTANCE OF RADIATION PATH WHENEVER APPLICABLE.
c
c ROOT=CONTAINS THE ROOTS OF THE HEAT TRANSFER FUNCTIONS
c ON RETURN.
c DER=CONTAINS THE DERIVATIVE OF THE HEAT TRANSFER FUNCTIONS
A-9
FORTRAN IV G LEVEL 18 MAIN DATE = 70132 14/14/56 PAGE 0002
Note: Input and output are in consistent units. These examples are in the International System of
Units (SI).
663
A-21
NUMERICAL DATA FOR EXAMPLE WALL.
SLAB COMPONENTS.
LAYER THICKNESS CONDUCTIVITY DENSITY SP HEAT RESISTANCE
DESCRIPTION OF LAYER
~
SAMPLING TIME INTERVAL, OT= 3600.000
~