IAPWS G6-03

The International Association for the Properties of Water and Steam

Vejle, Denmark
August 2003

Guideline on the Tabular Taylor Series Expansion (TTSE) Method

for Calculation of Thermodynamic Properties of Water and Steam
Applied to IAPWS-95 as an Example

2003 International Association for the Properties of Water and Steam

Publication in whole or in part is allowed in all countries provided that attribution is given to the
International Association for the Properties of Water and Steam

President:
Professor Koichi Watanabe
Department of System Design Engineering
Keio University
3-14-1 Hiyoshi, Kohoku-ku
Yokohama 223-8522, Japan

Executive Secretary:
Dr. R. B. Dooley
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94304, USA

This guideline contains 8 pages, including this cover page.

This guideline has been authorized by the International Association for the Properties of
Water and Steam (IAPWS) at its meeting in Vejle, Denmark, 24-29 August 2003, for issue by
its Secretariat. The members of IAPWS are: Argentina and Brazil, Britain and Ireland,
Canada, the Czech Republic, Denmark, France, Germany, Italy, Japan, Russia, the United
States of America, and associate member Greece.
The Tabular Taylor Series Expansion (TTSE) method has been authorized by the
International Association for the Properties of Water and Steam for issuance as a guideline.
In the judgment of IAPWS, the method provides users with a means of rapid and accurate
calculation of the thermodynamic properties of water and steam when used in conjunction
with the IAPWS Formulation 1995 for the Thermodynamic Properties of Ordinary Water
Substance for General and Scientific Use (IAPWS-95).
The supporting document for this guideline is: Miyagawa, K. and Hill, P. G., “Rapid and
Accurate Calculation of Water and Steam Properties using the Tabular Taylor Series
Expansion Method,” ASME J. Eng. Gas Turbines & Power, 123, 707-712 (2001).
Further information about this guideline and other documents issued by IAPWS can be
obtained from the Executive Secretary of IAPWS, or on the IAPWS Website at
http://www.iapws.org.
 2

1 Nomenclature

Thermodynamic properties and variables, units:

cp isobaric specific heat capacity, kJ/kg⋅K
cv isochoric specific heat capacity, kJ/kg⋅K
h specific enthalpy, kJ/kg
κ isentropic exponent
p pressure, MPa
s specific entropy, kJ/kg⋅K
T absolute temperature, K
v specific volume, m3/kg
w speed of sound, m/s
x independent variable
y independent variable
z dependent variable
Deviation:
∆ deviation of properties from IAPWS-95
Max maximum
RMS root-mean square
Indices:
i index for y-direction (specific enthalpy in the example)
j index for x-direction (pressure in the example)
Subscript:
sat saturation state
 3

2 Introduction

This guideline contains equations and additional information for using a method for rapid
and accurate calculation of the properties of water and steam.
The method employs a two-dimensional six-term Taylor Series expansion around selected
grid points on a suitable plane of independent variables. The calculations are fast because of
the small number of numerical operations required in using a table of stored properties and
their derivatives. The stored primitives and derivatives are precisely determined from a
fundamental equation so that the accuracy of the TTSE method depends only on grid spacing
of the storage tables.
This guideline contains a sample TTSE application based on a grid on the Mollier state
plane. The fundamental equation used for this example is the Helmholtz equation of state for
ordinary water substance, IAPWS-95. Like the IAPWS Industrial Formulation 1997
(IAPWS-IF97), this example well approximates the IAPWS-95 formulation; however, it is
not intended to replace IAPWS-IF97 for contractual purposes and does not replace IAPWS-
95 for general and scientific use.

3 Tabular Taylor Series Expansion (TTSE) Method

This IAPWS guideline presents a method of rapid and accurate calculation by using, as an
example, a specific version [1] of the TTSE method which has enthalpy h and pressure p as
the independent variables and is based on a Helmholtz function whose independent variables
are temperature T and density ρ. Previously, the method has been demonstrated with
pressure and temperature as independent variables for turbine stage-by-stage calculation [2].
It has also been used with the internal energy and density coordinates appropriate to the
Denton three-dimensional time-marching method for the turbine flow field [3]. In these
applications, the TTSE method achieves both high speed and high accuracy with modest
requirements for computer memory.
The basic source of the thermodynamic data used in formulating this example version of
the TTSE method is the wide-ranging Helmholtz equation included in the IAPWS
Formulation 1995 for the Thermodynamic Properties of Ordinary Water Substance for
General and Scientific Use (IAPWS-95) [4, 5].
The range of states included in the example may be defined by their locations on the
enthalpy-pressure plane. These cover enthalpies from 4.2563 to 4004.2563 kJ/kg and
pressures from 0.0006128 to 105.81 MPa. Excluded are those metastable or unstable states
located above an arbitrary boundary that is denoted as the upper limit of practical interest in
the metastable liquid region and a boundary in the metastable vapor region which
corresponds to approximately 5 % equilibrium liquid. The latter boundary may be considered
the lowest possible location of the Wilson line and we denote it as the lower limit of practical
interest. The definitions of these boundaries are given in detail in reference [1].
For TTSE calculations of functions of two variables, z(x,y), a grid is chosen composed of
rectangular cells on the x-y plane. The nodes at which the information on a thermodynamic
property z and its derivatives are stored is located at a point (xj,yi), usually the center, of the
cell. The Taylor series expansion may be expressed as:
 4

 ∂z  1  ∂2z   ∂z  1  ∂2z   ∂2z 

z = zi, j + (x − xj )  + (x − xj )2 2  + ( y − yi )  + ( y − yi )2 2  + (x − xj )(y − yi )  (1)
 ∂x i, j 2  ∂x i, j  ∂y i, j 2  ∂y i, j  ∂x∂y i, j

in which the subscripts i and j represent the locations of cells in the x-y plane. It has been
found that truncating the Taylor Series expansion at the level of second derivatives (6 terms)
provides high accuracy of property calculation with only modest memory capacity required
for storage of the primitive and derivative quantities associated with each node location. The
speed of the method is associated with only 6 coefficients for each property evaluation.
The variables x and y can be chosen so that they are the arguments of the functions most
frequently computed. With enthalpy and pressure as independent variables, Eq. (1) can be
used to calculate T(p,h), v(p,h), and s(p,h) if the appropriate primitive and derivative values
are tabulated for cell locations i and j. Table 1 shows how the appropriate first derivatives of
these functions may be obtained directly from the Helmholtz function, which has density and
temperature as independent variables.
The second and cross derivatives appearing in Eq. (1) could also have been obtained in
closed form from the Helmholtz equation but it has been much easier to obtain them by a
highly precise numerical differentiation. Since derivative quantities such as isobaric specific
heat capacity cp, isochoric specific heat capacity cv, speed of sound w, and isentropic
exponent κ are often of importance, it is useful to note that:

 ∂z   ∂z   ∂2z   ∂2 z 
  =   + ( x − x j ) 2  + ( y − y i )  (2)
 ∂x  y  ∂x  i , j  ∂x  i , j  ∂x∂y  i , j

Table 2 shows how specific heat capacities, isentropic exponent and speed of sound can be
evaluated using only the derivatives already stored for evaluation of T, v, and s.
Since Eq. (1) is quadratic, the inverse functions, in which p-T, p-s, and h-s are independent
variables, can be solved directly. An important feature is that these inverse functions are
fully consistent with the forward functions z(p,h).

Table 1. Derivatives of T, v, and s

 ∂T  1  ∂T  1  ∂h 
  =   =  
 ∂h  p c p  ∂p  h A  ∂ρ T
 ∂v  − 1  ∂p   ∂v  1  ∂h 
  = 2     = 2  
 ∂h  p Aρ  ∂T  ρ  ∂p h Aρ  ∂T  ρ
 ∂s  1  ∂s  −v
  =   =
 ∂h  p T  ∂p h T
where:
 ∂p   ∂h   ∂p   ∂h 
A =     −    
 ∂T  ρ  ∂ρ T  ∂ρ T  ∂T  ρ
 5

Table 2. Properties derived from the derivatives of basic TTSE functions

 ∂T  v 2  ∂s 
c p = 1 /  w=  
 ∂h  p B  ∂h  p

− BT v  ∂s 
cv = κ=  
D B  ∂h  p
where:
 ∂v   ∂s   ∂v   ∂s 
B =     −    
 ∂h  p  ∂p  h  ∂p  h  ∂h  p
 ∂T   ∂v   ∂T   ∂v 
D=    −    
 ∂h  p  ∂p  h  ∂p  h  ∂h  p

4 Grid and Cell Structure

The grid of points (i,j) on the Mollier plane is of size 200×200 with uniform spacings of
20 kJ/kg in the enthalpy (i) direction where the enthalpy ranges from 4.2563 to 4004.2563
kJ/kg. In the pressure (j) direction the spacings are of equal pressure ratio of 1.06215 as the
pressure ranges from 0.0006128 to 105.81 MPa. The critical point occupies the location
(104,173) where the pressure is 22.064 MPa and the enthalpy is 2084.2563 kJ/kg.
The primitives of the temperature T, specific volume v, entropy s, and its derivatives have
been evaluated at a node point in each cell. As a general rule, the node point has been located
at the center of the cell, as long as the Helmholtz equation is valid there. At the column
j=173, and when the center is out of the valid region, the node point is set on the saturation
line. In addition, at cells where j=172 and which include the liquid saturation line, the node
point is set on the saturation line. The exceptions are required if the center is in an invalid
region or outside the defined region of practical interest.
The evaluated values are used as TTSE coefficients after being stored in double precision
and in binary form. Three 6×200×200 arrays include all coefficients needed for the
calculation of the properties T, v, and s. The size of the binary file is about 1.92 megabytes
for each property.
For the saturation function Tsat(p), a truncated Taylor expansion is applied. With a single
dependent variable, only 3 coefficients are required for the primitive and first and second
derivatives of temperature with respect to pressure; these have been determined from a
subroutine of IAPWS-95. The same pressure scale as for z(p,h) has been used.
The time to find the indices of the cell in which the values of the input variables h-p, p-T,
p-s, or h-s are located is significant and fast cell location strategies have been developed [1].

5 Accuracy

To indicate the accuracy of the method (relative to the underlying Helmholtz function) in
determining the functions T(p,h), v(p,h), and s(p,h), tests of the method have been performed
over 6 regions (see Figs. 1 and 2) defined by:
Region 1: p = 0.0006128 to 100 MPa, T = 275 to 623.15 K, stable liquid (almost
equal to Region 1 of IAPWS-IF97)
 6

Region 2: p = 0.0006128 to 100 MPa, T = 275 to 1000 K, stable vapor (almost

equal to Region 2 of IAPWS-IF97)
Region 3: Region 3 of IAPWS-IF97, except regions near critical point defined
below
Region 4: p = 22.064 to 24 MPa, T = 641.77 to 652 K
Region 5: p = 20.7 to 22.064 MPa, T = 641.77 K to saturation temperature
Region 6: p = 20.7 to 22.064 MPa, T from saturation temperature to 652 K
Deviations from the basic equation were calculated using random inputs with uniform
distribution in p and T. The number of test points was 10,000,000 for regions 1 to 3 and
300,000 for regions 4 to 6. The maximum and root-mean-square of deviations are listed in
Table 3.
Corresponding results of random access tests for the functions cp(p,h), cv(p,h), and w(p,h),
i.e., functions requiring second derivatives, for regions 1 to 3 are given in Table 4. In the
vicinity of the critical point, this example does not accurately simulate these properties.
Reference [1] provides additional information on accuracy tests of this example, as well as
for the saturation function Tsat(p). In general it may be said that this example represents
IAPWS-95 much better than the accuracy specification of IAPWS-IF97 [6] or at least (for
specific heat capacities and speed of sound) satisfies the IAPWS-IF97 specification.

Definition of Test Regions

100
90
80
70
Pressure / MPa

60 Region 1 Region 3
50
Regions 4,5,6
40
30 Region 2
20
10
0
250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 1050 1100
Temperature / K

Figure 1. Definition of test regions

 7
Definition of Test Regions (Detail of Regions 4, 5, and 6)

30
Region 3
641.77 K 652 K
Region 4
Region 1 24 MPa
Pressure / MPa

20 Region 5 20.7 MPa

Region 6
Region 2
Critical Point

10
600 650 700
Temperature / K

Figure 2. Definition of test regions (detail of regions 4, 5 and 6)

Table 3. Deviation of TTSE functions T(p,h), v(p,h), and s(p,h) from IAPWS-95

∆T / mK ∆v / % ∆s / J/(kg⋅K)
Max. RMS Max. RMS Max. RMS
Region 1 1.78 0.03 0.00208 0.00003 0.0030 0.0002
Region 2 9.21 0.25 0.00413 0.00103 0.0055 0.0014
Region 3 9.29 0.38 0.03826 0.00057 0.0044 0.0005
Region 4 16.21 1.31 0.08945 0.00557 0.0051 0.0006
Region 5 37.02 3.59 0.11248 0.01027 0.0053 0.0006
Region 6 0.88 0.21 0.00466 0.00132 0.0044 0.0011

Table 4. Deviation of TTSE functions cp(p,h), cv(p,h), and w(p,h) from IAPWS-95

∆cp / % ∆cv / % ∆w / %
Max. RMS Max. RMS Max. RMS
Region 1 0.32 0.004 0.22 0.004 0.51 0.013
Region 2 0.52 0.010 0.61 0.029 0.26 0.072
Region 3 8.33 0.032 5.32 0.043 2.95 0.076
 8

6 Computing Time

The details of calculation speed tests of this example relative to IAPWS-IF97 are given in
reference [1]. This example and IAPWS-IF97 are comparable in computing speed; both of
them are 100 to 200 times faster than IAPWS-95 for the functions in this example.

7 Further Information about this Example

This example was examined by a evaluation task group established in IAPWS. Detailed
information about accuracy, computing time, and applicability is seen in the test report of the
task group [7].

8 References

[1] Miyagawa, K., and Hill, P. G., Rapid and Accurate Calculation of Water and Steam
Properties using the Tabular Taylor Series Expansion Method, ASME J. Eng. Gas
Turbines & Power, 123, 707-712 (2001). NOTE: Information about downloading the software
and related data for this application can be obtained at http://www.iapws.jp.

[2] Miyagawa, K., and Hill, P. G., A Tabular Taylor Series Expansion Method for Fast
Calculation of Steam Properties, ASME J. Eng. Gas Turbines & Power, 119, 485-491
(1997).
[3] Hill, P. G., Miyagawa, K., and Denton, J. D., Fast and Accurate Inclusion of Steam
Properties in Two- and Three-dimensional Steam Turbine Flow Calculations, Proc.
Inst. Mech. Eng. C, 214, 903-919 (2000).
[4] IAPWS (International Association for the Properties of Water and Steam), Release on
the IAPWS Formulation 1995 for the Thermodynamic Properties of Ordinary Water
Substance for General and Scientific Use, Fredericia, Denmark, 1996.
[5] Wagner, W., and Pruß, A., The IAPWS Industrial Formulation 1995 for the
Thermodynamic Properties of Ordinary Water Substance for General and Scientific
Use, J. Phys. Chem. Ref. Data, 31, 387-535 (2002).
[6] Wagner, W., Cooper, J. R., Dittmann, A., Kijima, J., Kretzschmar, H.-J., Kruse, A.,
Mareš, R., Oguchi, K., Sato, H., Stöcker, I., Šifner, O., Tanishita, I., Trübenbach, J.,
and Willkommen, Th., The IAPWS Industrial Formulation 1997 for the
Thermodynamic Properties of Water and Steam, ASME J. Eng. Gas Turbines &
Power, 122, 150-182 (2000).
[7] Weber, I., Kodl, I., Kretzschmar, H.-J., and Knobloch, K., Test Report of the
Documentation and Software of TTSE Method applied to IAPWS-95 as an Example,
IAPWS Report, 2001. See note on [1].