Professional Documents
Culture Documents
Fisher - CTF
Fisher - CTF
Published in ASHRAE Transactions 2004, Vol 109, Part 2. For personal use only. Additional distribution in
NA-04-9-1 either paper or digital form is not permitted without ASHRAEs permission.
Ipseng Iu is a graduate student and D.E. Fisher is an assistant professor in the Department of Mechanical Engineering, Oklahoma State Univer-
sity, Stillwater, Okla.
where
T ( x, )
q = k ------------------- (2)
x Ny Nx N
H out = Y n T is, n + X n T os, n + n q ko, n
Although the one-dimensional, transient conduction
n=1 n=1 n=1
problem can be solved analytically, the analytical solution is (3b)
immediately complicated when the analysis is extended to
multi-layered constructions. Analytical solutions for multi- Likewise, Equations 4a and 4b show the flux at the inside
layered slabs require special mathematic functions and surface.
complex algebra. Ultimately, numerical methods must be
employed at some level to solve the problem. Solution tech- q ki, = Z 0 t is, + Y 0 t os, + H in (4a)
niques include lumped parameter methods, frequency
where
response methods, finite difference or finite element methods,
and Z-transform methods (McQuiston et al. 2000). The toolkit Nz Ny N
implements Laplace and state-space methods for calculating
H in = Z n T is, n + Y n T os, n + n q ki, n
conduction transfer functions (CTFs) and provides an algo- n=1 n=1 n=1
rithm to derive periodic response factors (PRFs) from a set of (4b)
conduction transfer functions.
As indicated in Equations 3 and 4, the current heat fluxes
CTFs and PRFs are dependent only on material properties are closely related to the flux histories. The flux histories,
and reflect the transient response of a given construction for shown as constant terms in Equations 3 and 4, are not only
any set of environmental boundary conditions. Since material related to previous surface temperatures but also related to
properties are typically assumed to be constant in HVAC ther- previous heat fluxes. Equations 3a and 3b or Equations 4a and
mal load calculations, it is possible to pre-calculate these coef- 4b are usually solved iteratively with an assumption that all
ficients. Although CTF and PRF coefficients for typical previous heat fluxes are equal at the beginning of the iteration.
constructions are available in the ASHRAE Handbook The converged solution produces flux history terms (Hout and
Fundamentals (2001) and Spitler and Fisher (1999b), the Hin) that correctly account for the thermal capacitance of a
ASHRAE Loads Toolkit (Pedersen 2001), makes it possible to given construction.
quickly and accurately construct a stand-alone computer The temperatures operated on by the conduction transfer
program that will calculate CTFs and PRFs for any arbitrary functions may be either surface or air temperatures. Surface-
wall configuration. This paper presents an algorithm for pre- to-surface CTFs, which operate on surface temperatures and
calculating these coefficients using the toolkit modules. are required by the heat balance method, have the advantage
of allowing for variable convective heat transfer coefficients.
FORMULATIONS OF Air-to-air CTFs operate between either the sol-air tempera-
TRANSFER FUNCTION EQUATIONS ture or the air temperature on the outside and the air setpoint
temperature on the inside. Air-to-air CTFs include the appro-
The transfer function equations for conduction calcula- priate film coefficients as resistive layers in the wall assembly.
tion are formulated differently in load calculation methods. As shown in Figure 1, surface-to-surface CTFs are represented
The HBM uses conduction transfer functions (CTFs), while by the thermal circuit between Tos and Tis, while air-to-air
the RTSM uses periodic response factors (PRFs). In the HBM, CTFs are represented by the thermal circuit between To and Ti.
the instantaneous conduction flux is represented by a simple For constructions with the same material layer arrangement
linear equation that relates the current rate of conductive heat and properties, the surface-to-surface CTFs are always the
transfer to temperature and flux histories, while in RTSM, the same, while air-to-air CTFs differ depending on the selected
conduction flux is a linear function of temperatures only. values of the film coefficients.
The 1997 ASHRAE HandbookFundamentals presents factors) rather than CTFs to calculate conductive heat trans-
an air-to-air conduction equation that includes additional fer through walls and roofs. PRFs operate only on tempera-
simplifications. The b and c terms shown in Equation 5 operate tures; the current surface heat flux is a function only of
on the sol-air temperature and the constant room air tempera- temperatures and does not rely on previous heat fluxes, as
ture, respectively. shown in Equation 8.
6 6 6
23
q e, = b n T e, n d n q e, n T rc c n (5)
n=0 n=1 n=0
q = P j ( T e, j T rc ) (8)
j=0
It should be noted that Equation 5 is suitable only for load
calculations. Historically, it was used in the Transfer Function This formulation is premised on the steady, periodic
Method (TFM) (McQuiston and Spitler 1992) and can be used nature of the sol-air temperature over a 24-hour period (Spitler
without loss of generality in the Radiant Time Series Method et al. 1997). Although the number of PRFs may vary, the 24
(RTSM). PRFs shown in Equation 8 correspond to 24 hourly changes in
Although Equations 3 through 5 are solutions to the tran- the sol-air temperature for a single diurnal cycle. It is clear
sient, one-dimensional conduction problem, it is useful to
from Equation 8 that the overall heat transfer coefficient, U, is
consider the steady-state limit of these equations. Under
represented by the sum of the periodic response factors as
steady-state conditions, the exterior and interior heat fluxes
are equal and the following identities are readily apparent shown in Equation 9.
(Equation 6):
23
Nx Ny Nz 6 6 U = Pj (9)
Xn = Yn = Zn or b n = c n (6) j=0
n=0 n=0 n=0 n=0 n=0
The periodic response factor directly scales the contribu-
In combination with the standard formulation for steady-
tion of previous fluxes (in the form of temperature gradients)
state heat transfer through a wall ( q = U T ), an expression
to the current conductive heat flux. As a result, the periodic
for U, the overall heat transfer coefficient, in terms of conduc-
response factor series provides a visual representation of the
tion transfer functions can be derived as shown in Equation 7.
thermal response of the wall. As shown in Figure 2, wall 17 has
Ny 6 a slower thermal response then roof 10 because it is a more
Yn bn thermally massive construction.
n =0 n =0 -
U = ------------------------
N
- or Uf = ---------------
6
(7) PRFs are directly related to CTFs as shown in Equation 10
IMPLICIT ASSUMPTIONS OF
TRANSFER FUNCTION EQUATIONS
The assumptions behind the transfer function equations
come from the CTF calculation methods. Two widely used
CTF calculation methods are the Laplace method (Stephenson
and Mitalas 1971; Hittle 1979) and the state-space method
(Ceylan and Myers 1980; Seem 1987; Ouyang and Haghighat
1991). A brief overview of these two methods is included in
the following sections.
Figure 2 Periodic response factors for Roof 10 and Wall Laplace Transform Method
17. Hittle (1979) introduced a procedure to solve the conduc-
tive heat transfer governing Equations 1 and 2 by using the
Laplace transform method. The system in the Laplace domain
is shown in Equation 14.
P0 P5 P4 P3 P2 P1
D ( s ) 1
P1 P0 P5 P4 P3 P2
q ki ( s ) ----------- -----------
B(s) B(s) Ti ( s )
= ------------
- (14)
P2 P1 P0 P5 P4 P3 1 A ( s ) To ( s )
q ko ( s ) ----------- --------------
P = P3 P2 P1 P0 P5 P4 (11) B(s) B(s)
Xn = Yn = Zn = U ( 1 n ) (15)
b0 0 0 b3 b2 b1 n=0 n=0 n=0 n=1
b1 b0 0 0 b2
where
b2 b1 b0 0 0
b = (13) Nx = N y = N z
The Xn, Yn, and Zn are exterior, cross, and interior CTFs,
respectively. They are equivalent to CTFs shown in Equations
b3 b2 b1 b0 3 and 4. The number of CTF terms will increase to satisfy the
criteria shown in Equation 15. Heavyweight (long thermal
As shown in Equation 10, the PRFs are related to the cross response time) constructions require more CTFs than light-
and flux CTF terms. The first column of the P matrix is the weight constructions.
resulting PRFs, P0, P1, P2, ., P23. Since the sol-air temper- The number of CTF terms can be determined in different
ature is used in RTSM conduction calculations, the b and d ways. Mitalas (1978) suggests that the number of CTF terms
matrices must be filled with air-to-air CTFs. This eliminates should be
the surface heat balance calculations in HBM. However, if
conductive heat transfer is an isolated concern, the PRFs can Nx = Ny = Nz = 1+N, (16)
1 0 2.95287743E 02 3.90083134E01
3.90083134E01 1 0 0 2.95287743E02
2.95287743E02 3.90083134E01 1 0 0
d = 0 2.95287743E02 3.90083134E01 1 0 0
0. 0. 2.95287743E02
. 3.90083134E01
. 1. . 0.
. . . . . . .
. . . . . . .
0 0 0 2.95287743E02 3.90083134E01 1
Table 4. Required Data Values in the Toolkit Input File for CTF and PRF Calculation
MATERIALLAYER Layer names; thickness, thermal conductivity, density, specific heat or thermal resistance
Structure of the Algorithm the application of a graphical user interface as discussed in the
next section.
The structure of the algorithm for CTF and PRF calcula- The function GetNumObjectsFound and the subroutine
tions is shown in Figure 4. The USE statement is a Fortran GetObjectItem in this module are used in the CTF and PRF
90 keyword that makes the subroutines in a FORTRAN 90 calculations. GetNumObjectsFound returns the number of
module available to a program, subroutine, or another module. surfaces in the calculations, while GetObjectItem returns
USE is followed by the toolkit module name. Subroutines in two arrays containing all data values (numeric and alpha) for
one FORTRAN 90 module cannot be called by another the keyword Surface. Although there will be a number of
module unless the calling routine or module uses the target surface data values returned from this subroutine, only the
surface name and the construction ID are useful for the calcu-
module. The module InputProcessor handles all input data.
lations. The argument variables of this subroutine are
The data are organized under keywords with a one-to-one
described in the toolkit documentation (Pedersen 2001). In a
correspondence between each definition and data value. The similar way, the data associated with the keywords
keywords for calculating CTFs and PRFs are SURFACE, CONSTRUCTION and MATERIALLAYER are read into the
CONSTRUCTION, and MATERIALLAYER. Table 4 shows program.
the required data for each keyword. By changing the data The program can be configured to use either the Laplace
values of these keywords, the toolkit generates different CTFs or the state-space CTF module. Figure 4 shows the program
and PRFs. Although manual construction of a toolkit input configured to use the StateSpaceCTFCalc module. It
data file is tedious, the structured input format is conducive to contains subroutines that import the construction and material
Figure 5 Dialog box used for creating toolkit input file (IP).
The default layer type of fully specified properties, can be Evaluation of the Program Outputs
changed by clicking the Edit button. For input convenience, The PRF generator can efficiently and conveniently
an ASHRAE material database, which contains the data calculate the desired CTFs and PRFs for any construction.
shown in Table 22, chapter 29, 2001 ASHRAE Handbook This section establishes the validity of the program in terms of
Fundamentals, is included as shown in Figure 6. This database the self-consistency of the program outputs, the agreement of
can be modified and saved for future reference. The interface program output with previously published data, and the
conservation of energy based on a steady-state test.
can handle up to 100 surfaces, which may be specified in either
Consistency of PRFs and CTFs. Since the U-factor is
SI or IP units. All IP unit data are converted to SI units by the
unique for any construction and since it can be calculated
interface before writing the input file, since input to the toolkit either using CTFs or PRFs, the U-factor check can be used to
modules must be in consistent SI units and in accordance with evaluate the program algorithm that converts CTFs to PRFs.
toolkit conventions. Figure 7 compares the U-factors calculated by CTFs and PRFs
for the wall and roof database used by Spitler and Fisher Therefore, the total thermal resistance is calcuated using
(1999b). The diagonal line represents a zero percent difference Equation 22.
between the two. The results show nearly perfect agreement
n
between the CTF and PRF U-factors. The differences show up
RT = Ri (22)
in the fifth decimal place and are probably due to round-off
i=1
error.
Comparison with Published Data. The second step in If the calculated U-factor from the steady-state Equation
evaluating program output is based on the fact that the PRF 19 is equal to that from the CTFs and PRFs, energy is
series is unique for any construction regardless of whether it conserved. Figure 9 shows the comparison of the program
is derived from Laplace or state-space CTFs. Since a PRF output U-factors with the steady-state U-factors based on the
database for the walls and roofs in the ASHRAE Handbook same construction database used by Spitler and Fisher
Fundamentals is already available in the literature (Spitler and (1999b). The U-factors shown on the vertical axis are the PRF
Fisher 1999b), the validity of the program outputs is evaluated U-factors from Figure 7.
by a term-by-term comparison of calculated and published Figure 9 shows that use of the program outputs to calcu-
PRFs for each construction, as shown in Figure 8. The small late U-factors is satisfactory. The steady-state U-factors agree
differences shown in the figure could be caused by slight with the calculated U-factors to within 3.4%. Although some
numerical differences in the input data, the type of CTFs used round-off error is expected, the root cause of the differences is
to derive PRFs, as well as the round-off error in both calcula- primarily due to the convergence criteria used in the Toolkit
tions. In addition, if the convergence criterion used in the CTF module. The CTFs calculated from the computer program are
calculation is different from that used by Spitler and Fisher based on the default Toolkit settings. A better result is obtained
(1999b), it would also cause the numerical difference in the if one tightens the convergence criteria in the Toolkit CTF
PRFs. In general, the program outputs agree very well with the module.
published data. The steady-state evaluation is necessary but not sufficient
Steady-State Evaluation. The program outputs for a to guarantee the accuracy of the transient calculation. A more
given construction can also be verified at the steady-state limit rigorous transient evaluation would apply a sinusoidal temper-
by comparing the U-factor predicted by the CTFs or PRFs ature variation to the outside surface and a constant tempera-
with that predicted by the steady-state calculation. At steady ture to the inside surface and compare the resulting heat flux
state, the U-factor is calculated as shown in Equation 19. to the analytical solution presented in ASHRAE RP-1052
(Spitler et al. 2001). However, for most standard construc-
1
U = ------ (19) tions, the steady-state test is a good indicator of CTF and PRF
RT
accuracy.
The thermal resistance for each layer is calculated using
either Equation 20 or 21. SUMMARY AND CONCLUSIONS
Transfer function equations continue to provide a robust,
1
R = --- for air resistance layer (20) accurate, and tractable approach to calculating conductive
h
heat gains in cooling load procedures. This paper illustrates
L two commonly used formations in conduction calculation,
R = --- for thermal mass layer (21)
k i.e., CTF and PRF formulations. The application and the rela-
tion between CTF and PRF are discussed. Although care must NOMENCLATURE
be taken to consistently apply CTFs and PRFs, depending on a, b, c, d = coefficient matrices that depend on material
whether they were generated with or without a convective properties and/or film coefficients
resistance layer, the application of boundary conditions and b = 2424 air-to-air cross CTF coefficient
solution techniques is straightforward and consistent. matrix (Equation 13)
Implicit assumptions of the two CTF calculation methods d = 2424 air-to-air flux CTF coefficient matrix
(Laplace and state-space) imply that CTFs can be different for (Equation 12)
the same material construction. The differences are in terms of
C = material capacitance, JC-1 (BtuF-1)
number of CTF terms and the CTF numeric values. Hower-
bn = air-to-air cross CTF coefficient, Wm-2K-1
ever, the predicted overall heat transfer coefficients are the
(Btuh-1ft-2F-1)
same. PRFs represent the thermal response of a material
construction and therefore are unique regardless of the calcu- cn = air-to-air interior CTF coefficient,
Wm-2K-1 (Btuh-1ft-2F-1)
lation methods. An example comparing the Lapalce and state-
space methods showed that the more thermally massive dn = air-to-air flux coefficient, dimensionless
construction (Wall 17) carries more CTF terms and has slower A(s), B(s), D(s) = overall transmission matrices that depend on
thermal response than Roof 10. material properties and/or film coefficients
In the past, the most serious drawback to the use of trans- hi = inside film coefficient, Wm-2K-1
fer function and response factor methods was the complexity (Btuh-1ft-2F-1)
of the computer code required to generate the coefficients. The ho = outside film coefficient, Wm-2K-1
ASHRAE Loads Toolkit addresses this problem by providing (Btuh-1ft-2F-1)
the source code required to generate conduction transfer func- k = thermal conductivity, Wm-1K-1
tions and periodic response factors for arbitrary wall or roof (Btuh-1ft-1F-1)
constructions. The computational algorithm required to n = number of layers
implement the toolkit modules in a CTF/PRF generator Nx = number of exterior CTF terms
program was presented in this paper. In addition, input/output Ny = number of cross CTF terms
and interface issues were discussed. The CTFs and PRFs
Nz = number of interior CTF terms
calculated by the Toolkit algorithms can be directly applied to
heat balance and radiant time series load calculation proce- N = number of flux CTF terms
dures. P = 2424 PRF matrix (Equation 10)
The outputs of the computer program were evaluated Pj = periodic response factors, Wm-2K-1
based on the physical significance of the CTFs and PRFs and (Btuh-1ft-2F-1)
were compared to the published literature. A simple method q = heat flux, Wm-2 (Btuh-1ft-2)
for checking the steady-state accuracy of CTFs and PRFs was qki = heat flux at interior surface, Wm-2
also used in the evaluation. The results showed that based on (Btuh-1ft-2)
the steady-state check, program outputs are within 3.4% of qko = heat flux at exterior surface, Wm-2
the calculated U-factor. (Btuh-1ft-2)
Iu, I., D. Fisher. 2004. Application of Conduction Transfer Functions and Periodic
Response Factors in Cooling Load Calculation Procedures. ASHRAE Transactions,
110(2): 829-841.
Reprinted by permission from ASHRAE Transactions (Vol. #110, Part 2, pp. 829-841).
2004 American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc.