Case Studies in Thermal Engineering 28 (2021) 101400

Contents lists available at ScienceDirect

Case Studies in Thermal Engineering

journal homepage: www.elsevier.com/locate/csite

Case studies on performance prediction of radial inflow turbine

under multiple rotor structures and organic fluids
Wenyu Li a, b, Yunyuan Li a, Xiang Ling a, b, *
a
School of Mechanical and Power Engineering, Nanjing Tech University, Nanjing, 211800, China
b
Jiangsu Key Laboratory of Process Enhancement & Energy Equipment Technology, Nanjing, 211800, China

A R T I C L E I N F O A B S T R A C T

Keywords: Due to the lack of selection criteria for the ORC radial inflow turbine’s loss models, and infor­
Radial inflow turbine mation about the models’ limitations. This paper presents a comprehensive analysis for perfor­
Performance prediction mance prediction of the ORC radial inflow turbine, and a novel turbine performance prediction
CFD simulation
program is proposed. This study can be divided into two parts, during the analysis part, 125 cases
ORC
of CFD simulation were conducted under 5 rotor structures and R245fa as the working fluid. The
performance of 6 loss models and their 18 combinations were compared. And in the examination
part, another three organic fluids from different categories were chosen to exam the reliability of
the proposed prediction program. The results show that loss model combinations E, O, and Q have
a good performance in rotor loss prediction. Besides, Δβ4 has an important influence on pre­
diction reliability and turbine performance, it should be lower than a certain value (about 40◦ in
this paper) during the analysis process to get relatively reliable results under different rotor
structures. The proposed program can realize an acceptable accuracy for turbine performance
prediction under multiple organic fluids and rotor structures, with an efficiency deviation within
±5% and an output power relative error within ±10%.

1. Introduction

Nowadays a significant challenge for the sustainable development of human society is caused by the shortage of traditional energy
and environmental pollution. Improving the efficiency of energy utilization and exploiting renewable energy has become one of the
most important research topics. The ORC system has been proven to be an effective method to convert low-temperature heat sources
into electrical power since it has the advantages of high efficiency, simple structure, and high reliability [1]. It is promising to be
applied to recover the industrial waste heat, or to utilize geothermal energy and low-temperature solar energy, etc. The basic com­
ponents of the ORC system are the evaporator, turbine, condenser, and pump. And there is a critical component that significantly
influences the efficiency of the ORC system, which is the turbine [2,3]. While among various turbines, the radial inflow turbine was
demonstrated as a suitable one to be used in ORC [4].
Generally, the turbine efficiency is assumed to be a constant value in previous ORC system analysis [5–7]. However, this may lead
to a huge deviation between prediction results and actual conditions [8]. Hence, in recent years, more and more researchers take the
turbine’s performance prediction models into account in their studies [9,10]. Song et al. [11] developed a one-dimensional analysis

* Corresponding authorSchool of Mechanical and Power Engineering, Nanjing Tech University, No. 30 Puzhu South Road, Nanjing, Jiangsu
province, 211800, China.
E-mail addresses: [email protected] (W. Li), [email protected] (Y. Li), [email protected] (X. Ling).

https://doi.org/10.1016/j.csite.2021.101400
Received 4 May 2021; Received in revised form 15 August 2021; Accepted 28 August 2021
Available online 31 August 2021
2214-157X/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

model for the ORC system. They coupled the radial inflow turbine performance prediction models in the system model. The influence
of the working fluid properties and the system operating conditions on the turbine performance was evaluated. Their results indicated
that the turbine efficiency had a notable influence on the working fluid selection and the system parameter determination. Li et al. [8]
developed a multi-objective optimization model that considered the thermodynamic performance and economic factors simulta­
neously was coupled with a dynamic turbine efficiency model. Their results revealed that the variation of net power output between
constant turbine efficiency and variable turbine efficiency was different. With constant turbine efficiency, R245ca was the optimal
working fluid. However, with variable turbine efficiency, R365mfc was the optimal working fluid. Li et al. [12] conducted research on
the optimization of the high-temperature ORC system based on radial-inflow turbine design. They concluded that the turbine power
output and cycle thermal efficiency were more sensitive to the expansion ratio, while the turbine size was more sensitive to the turbine
inlet total temperature and specific speed. Besides, Al Jubori et al. [13] researched the performance of turbine types and their con­
figurations under a low-temperature heat source and low-power ORC system (<20 kW). Their findings exhibited that the configu­
rations of the two-stage turbines showed a considerably higher turbine performance compared to the single-stage configuration at a
high-pressure ratio, and highlighted the potential of using a two-stage turbine configuration to enhance the performance of a
small-scale ORC system.
Recently, more and more studies have appeared in the literature which focus on the performance of the ORC radial inflow turbine.
Schuster et al. [14] provided a comprehensive overview of radial turbine performance modeling and a unified mean-line model that
was validated against experimental data and supported by CFD results. Zheng et al. [15] performed the preliminary design of a radial
inflow turbine using R134a and used the CFX to accomplish the three-dimensional CFD simulation of the designed turbine. They

Fig. 1. Schematic of the radial inflow turbine [22].

2
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

concluded that the model can generate reliable results under the nominal condition, through the comparisons to CFD analysis.
Alshammari et al. [16] utilized a mean-line off-design model to predict the performance characteristics of a radial inflow turbine,
which was validated against experimental data of a turbine with the R123 as working fluid. Persky et al. [17] compared 1.5 million loss
model configurations applied to a radial turbine working with R134a and concluded that only a few configurations were suitable to
predict the turbine efficiency at off-design conditions. Masi et al. [18] have studied the prediction of the maximum efficiency
achievable by turbines for the ORC system through similarity principles. They concluded that strict flow similarity does not hold if the
same design is used for different working fluids.
In addition, the research on the ORC turbine optimization method is also a very important topic. Li et al. [19] utilized a genetic
algorithm to optimize 10 key design parameters which are need in the turbine aerodynamic design process. The turbine designed
through their optimization method showed a good off-design performance, its maximum efficiency deviation is only 2.5% with
rotational speed variations among the range of 10%. Han et al. [20] developed a radial inflow turbine preliminary design method
which coupled with a particle swarm optimization method, six working fluids and eight critical design parameters were selected to
optimize. The results showed that the turbine with R245fa was determined to be optimal due to its small geometry size, high exergy
efficiency (0.929) and high load coefficient (1.1027), meanwhile also had a good off-design performance. Besides, Al Jubori et al. [21]
constructed a more systematic optimization method that including a 3D CFD MOGA optimization process. And based on the B-spline
composed of 20 control points, the blade geometry was parameterized. It was revealed from their 3D CFD MOGA optimization results
that the turbine performance was improved significantly compared with the baseline design results. The maximum enhancements in
turbines’ isentropic total-to-total efficiency and power output were 13.95% and 14% for the turbine with R245fa as the working fluid
and were 10.55% and 11.98% for the turbine with isopentane as the working fluid.
From the above, though many researchers had taken loss models of the radial inflow turbine into the ORC system analysis, to this
moment, doesn’t exist a broadly recognized loss model combination for predicting the performance of turbines with various organic
fluids. Besides, most of the studies were only focused on a certain organic fluid and a certain turbine, there doesn’t have enough
information about the influence of rotor structure and fluid type on turbine performance prediction.
So in this paper, the performances of 6 commonly used rotor loss models and their 18 combinations were compared based on high-
precision CFD simulation. To investigate the influence of working fluid type on prediction performance, 4 organic fluids from different
categories were selected (R245fa, R123, Pentane, and MDM). And to investigate the prediction performance under different rotor
structures, 5 kinds of rotor structures were analyzed. Besides, a novel turbine performance prediction program is present, which can be
used in both subsonic and transonic working conditions. This paper gives intuitive insights into the selection of loss models, and their
limitations in ORC turbines’ performance prediction.

2. Engineering methodologies

2.1. Overview of the radial inflow turbine

The radial inflow turbine consists of three main components as volute, nozzle, and rotor, as shown in Fig. 1 (a). The incoming fluid
is accelerated and distributed uniformly by the volute. Further acceleration in the circumferential component of velocity is attained via

Fig. 2. Enthalpy-entropy diagram of the fluid expansion [23].

3
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

the nozzle before the fluid enters the rotor. At last, the working fluid expands in the rotor and converts its internal energy to the
mechanical energy of the rotor. Fig. 1 (b) shows the schematic of velocities and angles at the rotor inlet and outlet.
The function of the turbine is to convert the working fluid’s energy to mechanical energy then drive the generator to power. Fig. 2
shows the energy change of working fluid during the expansion process in the turbine. The ideal enthalpy drop Δhid corresponds to an
isentropic expansion from the volute inlet total conditions to the rotor exit static pressure (P5). The energy of the working fluid that can
be converted to the mechanical energy of the rotor is Δhact, which is the total enthalpy drop between rotor inlet and outlet. Moreover,
the difference value between Δhid and Δhact is the energy loss caused by the entropy increase and kinetic energy loss. Then total-to-
static efficiency (ηts) is expressed by Eq. (1).
ηts = Δhact /Δhid (1)

2.2. Rotors and working conditions in this research

In this paper, the influence of rotor structure on performance prediction is discussed. The blade number of rotors (Nr) and the ratio
rs5/r4 are chosen as independent variables, and some simplified design criteria were also adopted according to worldwide researchers’
experience [20,22,24]. The structure parameters of rotors investigated in this paper are listed in Table 1, and the rotor design criteria
are expressed by Eq. (2)–(7). All of these rotors have the inlet radius r4 as 50 mm and the same blade setting angles. For Rotor 1, 2, and
3, they have different rs5/r4 as 0.7, 0.6, and 0.5 respectively. Besides, Rotor 4 and 5 have different numbers of the blade based on the
structure of Rotor 2.
rs5
rs5 = r4 (2)
r4

rh5 = 0.4rs5 (3)

b4 = 0.08r4 (4)

ΔZr = 1.5(rs5 − rh5 ) (5)

tb4 = 0.04r4 (6)

tb5 = 0.02r4 (7)

The selection of a suitable working fluid for ORC systems is critical and challenging, as it has a significant effect on the cycle ef­
ficiency. Organic working fluids have large molecular weights and a low boiling temperature and pressure and are usually heavy
mixtures [21].To emphasize the difference of working fluid, four common organic fluids from different categories were selected to
refer to the research of Aboelwafa et al. [25]. They belong to hydrofluorocarbons (HFCs), hydrochlorofluorocarbons (HCFCs), hy­
drocarbons (HCs), and siloxanes respectively. The basic information of them is listed in Table 2. During the analysis process of loss
models’ comparison, R245fa was chosen as working fluids for all simulation cases. After the suitable loss model combination is
decided, the other three working fluids were used to exam the reliability of the turbine performance prediction program. However, it
should be noticed that which fluid is more suitable for ORC was not discussed in this paper.
During the analysis part of loss models’ comparison, a total number of 125 simulation cases were conducted. The way to change the
working conditions is by adjusting the rotation speed of the rotor (N) and the pressure ratio (PR). In addition, the pressure at the
turbine outlet is keeping constant in the research, the change of PR is by adjusting the total pressure and total temperature at the
turbine inlet. The saturation pressure of R245fa at 30 ◦ C was selected as the turbine outlet pressure (P5). And the total temperature at
the turbine inlet is the saturation temperature under corresponding total pressure. The information on working conditions for the
analysis process in this paper is listed in Table 3.

Table 1
Rotors in this research.
Parameter Rotor 1 Rotor 2 Rotor 3 Rotor 4 Rotor 5

rs5/r4 0.7 0.6 0.5 0.6 0.6

Nr 15 15 15 13 17
r4 [ mm ] 50 50 50 50 50
rs5 [ mm ] 35 30 25 30 30
rh5 [ mm ] 14 12 10 12 12
b4 [ mm ] 4 4 4 4 4
b5 [ mm ] 21 18 15 18 18
ΔZr [ mm ] 31.5 27 22.5 27 27
γ4 [ ◦ ] 0 0 0 0 0
γs5 [ ◦ ] − 71.3 − 71.3 − 71.3 − 71.3 − 71.3
γh5 [ ◦ ] − 51.8 − 51.8 − 51.8 − 51.8 − 51.8
εr [ mm ] 0.5 0.5 0.5 0.5 0.5
εx [ mm ] 0.5 0.5 0.5 0.5 0.5

4
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

Table 2
The working fluids in this study.
Category Substance M [kg/kmol] Tc [K] Pc [MPa] ρc [kg/m3] Tbo [K]

HFCs R245fa 134.05 427.16 3.65 516.08 288.29

HCFCS R123 152.93 456.83 3.66 550.00 300.97
HCs Pentane 72.15 469.70 3.37 232.00 309.21
Siloxanes MDM 236.53 564.09 1.41 256.74 425.66

2.3. Loss models for the rotor

The loss in a rotor includes incidence loss, passage loss, trailing edge loss, clearance loss, windage loss, and exit energy loss. To
simplify the experiment that some secondary losses were not taken into account, including windage loss and trailing edge loss which
occupy little proportion in total loss [17,26]. In this study, 6 very commonly used loss models were selected for further comparison.

2.3.1. Rotor incidence loss

Concerning rotor incidence loss. The first model (Rinc-I) is described by Eq. (14) which is widely used in papers [14,26,27]. The
expression of this model is very simple, which only regards the tangential component of inlet relative velocity as the factor to cause
loss.
1
ΔhR,inc,1 = Wt42 (8)
2
The second model (Rinc-II) takes the relative flow angle into account, which can be expressed as:
1 ( )
ΔhR,inc,2 = W42 sin 2 β4 − β4,opt (9)
2

where β4,opt is the optimum relative flow angle at rotor inlet, which is calculated by following correlation [28,29].
( ) − 1.98 tan(α4 )
tan β4,opt = (10)
Nr (1 − 1.98/Nr )

2.3.2. Rotor passage loss

For rotor passage loss, the first one (Rpas-I) was present by NASA [30].
1 ( ( ) )
ΔhR,pas,1 = Kp,NASA W42 cos 2 β4 − β4,opt + W52 (11)
2

where Kp,NASA is an empirical coefficient, for which a value of 0.3 was recommended.
The second one (Rpas-II) is the CETI model [31,32].
{( ) [ ( )2 ] }( )
Lhyd,rotor r5 cos(γ5 ) W42 + W52
ΔhR,pas,2 = Kp ,CETI + 0.68 1 − (12)
Dhyd,rotor r4 b5 /cR 2

where a default value of Kp, CETI = 0.11 is used for (r4-r5s)/b5 ≥ 0.2 and Kp, CETI = 0.22 for (r4-r5s)/b5 < 0.2. The Lhyd and Dhyd represent
the hydraulic length and hydraulic diameter of the rotor passage respectively. In addition, cR is the rotor blade chord, approximated by
ΔZr
cR = (13)
cosγ

1
tanγ = (tan γ4 + tan γ 5 ) (14)
2

Table 3
Working conditions for the analysis process.
Parameter R245fa

N [rpm] 10,000–30,000
PR 3–5
P02 [kPa] 533.37–888.95
T02 [K] 338.23–357.97
P5 [kPa] 177.79
Rotor 1, 2, 3, 4, 5

5
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

2.3.3. Rotor clearance loss

About rotor clearance loss, the first clearance loss model (Rcle-I) follows the empirical correlation proposed by Baines [23,31].

U43 Nr ( √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅)
ΔhR,cle,1 = Kx εx Cx + Kr εr Cr + Kxr εx εr Cx Cr (15)
8π

where Kx = 0.4, Kr = 0.75, Kxr = − 0.3. The parameter ε presents the size of the tip clearance. And Cx, Cr can be calculated by:
(1 − rs5 /r4 )
Cx = (16)
Cm4 b4

rs5 (ΔZr − b4 )
Cr = (17)
r4 Cm5 r5 b5
Another model (Rcle-II) was used in the paper [33,34], which was proposed by Spraker.
1 ṁL 2
ΔhR,cle,2 = U (18)
2 ṁ s5

where ṁL represents the leakage flow rate:

1
ṁL = ρ5 Us5 εr (2πrs5 )KL (19)
2

where Us5 is the blade velocity at the tip and KL is a coefficient whose value was determined as 1.5 by experiments.
In addition, these 6 commonly used rotor loss models are combined into eighteen loss combinations in this paper, as shown in
Table 4.

3. Turbine performance prediction program

This paper presents a novel turbine performance prediction program, and Fig. 3 illustrates an overview of the program calculation
process. The proposed program is based on the basic assumption of mean-line modeling that the properties of the working fluid are
uniform on a plane normal to its motion direction. The program mainly utilizes the continuity equation, real fluid state equation, loss
prediction models, and Euler’s turbomachine equation to calculate the flow features and fluid properties at the inlet and outlet of each
component. Besides, the program has been implemented in Matlab, and the properties of organic fluids source from the NIST REFPROP
library [35]. The input parameters of the turbine performance prediction program mainly include the geometric parameters of
components, the thermodynamic conditions of working fluid at the turbine inlet, and the outlet that including T01, P01, and P5.

3.1. Models of the volute, nozzle, and interspace

At the volute inlet, the velocity C1 is guessed initially. Static enthalpy is then calculated using the definition of total enthalpy. The
other thermodynamic properties are subsequently calculated by the functions, where f represents a call of REFPROP. And the process is
repeated until the convergence of C1.

Table 4
Combination method for loss models.
Combination Incidence Passage Clearance

A \ Rpas-I \
B \ Rpas-II \
C Rinc-I Rpas-I \
D Rinc-I Rpas-II \
E Rinc-II Rpas-I \
F Rinc-II Rpas-II
G Rpas-I Rcle-I
H \ Rpas-I Rcle-II
I \ Rpas-II Rcle-I
J \ Rpas-II Rcle-II
K Rinc-I Rpas-I Rcle-I
L Rinc-I Rpas-II Rcle-I
M Rinc-I Rpas-I Rcle-II
N Rinc-I Rpas-II Rcle-II
O Rinc-II Rpas-I Rcle-I
P Rinc-II Rpas-II Rcle-I
Q Rinc-II Rpas-I Rcle-II
R Rinc-II Rpas-II Rcle-II

6
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

Fig. 3. Flowchart of turbine performance prediction program.

C12
h1 = h01 − (20)
2

s1 = f (T01 , P01 , fluid) (21)

ρ1 = f (h1 , s1 , fluid) (22)

ṁ
C1 = (23)
ρ 1 A1
To calculate the parameters at the volute outlet (nozzle inlet), an iterative process is required to calculate the absolute velocity C2
until it satisfies the continuity equation. Volute loss is calculated using Eq. (24) as introduced in Ref. [31], and the value of kvol takes as
0.1 [27].
1
Δhvolute = kvol C22 (24)
2

h2s = h2 − Δhvolute (25)

P2 = f (h2s , s1 , fluid) (26)

[ρ2 , s2 ] = f (P2 , h2 , fluid) (27)

7
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

ṁ
C2 = (28)
ρ 2 A2
To calculate the parameters at the nozzle outlet, an iterative process is required to calculate the absolute velocity C3 until it satisfies
the continuity equation. The calculation process is similar to the above for volute. At first, the initial value of s3 and C3 are assumed.
And the absolute flow angle at the nozzle outlet is calculated by Eqs. 29 and 30, where the coefficient c0, c1 and c2 are 6, 0.96 and − 3.4
respectively [16].

α3 = c0 + c1 cos− 1 (o3 / esse3 ) Ma3 < 0.3 (29)

α3 = c0 + c1 cos− 1 (o3 / esse3 ) + c2 (Ma3 − 0.3) Ma3 > 0.3 (30)

The nozzle loss model was built by Rodgers [36], which is an empirical correlation as shown in Eq. (37). It proposes a loss based on
the nozzle exit Reynolds number, the nozzle blade solidity (cn/esse3), and the throat aspect ratio (o3/b3). The γ3 is the setting angle of
the blade at the nozzle outlet.
[ ]
0.05 3 tan(γ3 ) o3 1 2
Δhnozzle = 0.2 + C (31)
Reb esse3 /cn b3 2 3

ρ3 C3 b3
Reb = (32)
μ3

o3 ≈ esse3 sin γ3 (33)

Once the nozzle loss is calculated, the isentropic enthalpy can be calculated by Eq. (34), then other parameters at the nozzle outlet
can be calculated based on the conservation of mass, real fluid state equation.
h3s = h3 − Δhnozzle (34)
The interspace is a very small space between the nozzle trailing edge and the rotor leading edge. It is small enough to avoid leakage
and losses [32]. Therefore, the angular momentum equation is adopted, which is expressed by:
Ct3 r3 = Ct4 r4 (35)

s4 = s3 (36)

3.2. Checking of Ma3

After the calculation of the nozzle outlet, the Mach number (Ma3) should be checked. If Ma3 = 1, the flow is chocked and the mass
flow rate remains constant for any higher PR. Therefore the mass flow rate needs to be revised when Ma3 > 1. And for different values
of Ma3, two different calculation processes are adopted for the rotor. It should be noticed that the real minimal area in the nozzle is at
the throat where the chock happens. There is a deviation in Mach number between the nozzle outlet and nozzle throat sections.
However, to simplify, the Mach numbers were only examined at the nozzle outlet in this paper.
Ma3 = C3 /a3 (37)

3.3. Model of the rotor

To calculate the parameters at the rotor inlet, an iterative process is required to calculate the velocity Cm4 until it satisfies the
continuity equation, which is the same as the calculation process for the volute inlet.
To calculate the parameters at the rotor outlet, the efficiency should be guessed initially. Then two different processes are adopted
based on the value of Ma3. If Ma3 < 1, here is called the unchocked condition. The rotor relative outflow angle (β5) is a parameter that
relates to the structure of the rotor, therefore which needs to be adjusted for different rotors [16,37].
h5ss = f (P5 , s1 , fluid) (38)

Δhid = h01 − h5ss (39)

Δhact = Δhid ηts (40)

h05 = h04 − Δhact (41)

Ct5 = (U4 Ct4 − Δhact )/U5 (42)

Wt5 = Ct5 − U5 (43)

8
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

Cm5 = Wt5 /tan(β5 ) (44)

C52
h5 = h05 − (45)
2

ρ5 = f (P5 , h5 , fluid) (46)

The incidence loss, passage loss, and clearance loss models of the rotor have been introduced in section 2.3, the total loss of rotor is
expressed by Eq. (48).

C52
Δhexit = (47)
2

Δhrotor = Δhincidence + Δhpassage + Δhclearance + Δhexit (48)

Δhact
ηts = (49)
Δhact + Δhvolute + Δhnozzle + Δhrotor
The new efficiency is calculated as shown in Eq. (49) and compared with the initial guess. The process needs to repeat until
convergence is achieved. After the convergence of efficiency, the new mass flow rate can be calculated, then another iterative loop runs
until the convergence of mass flow rate.
If Ma3 = 1, here is called the chocked condition. To calculate the parameters at the rotor outlet, an iterative process is required to
calculate the velocity Cm5 until it satisfies the continuity equation. And firstly, s5, h05, and Cm5 should be guessed.
W5 = Cm5 /cos(β5 ) (50)

C52 = W52 + U52 − |2W5 U5 cos(90 − |β5 |)| (51)

Wt5 = W5 sin(β5 ) (52)

Ct5 = Wt5 + U5 (53)

C52
h5 = h05 − (54)
2

s5 = f (P5 , h5 , fluid) (55)

Δhact = U4 Ct4 − U5 Ct5 (56)

h05 = h04 − Δhact (57)

Δhact
ηts = (58)
Δhact + Δhvolute + Δhnozzle + Δhrotor

ṁ
Cm5 = (59)
ρ 5 A5

4. CFD methodology

The accurate performance of turbines should be tested through experiments. However, it is difficult to conduct experimental
research for multiple organic working fluids. Besides, it is also hard to measure or observe the flow characteristics of the fluid in
turbines effectively during the experiment process. Therefore, ANSYS CFX, a commercial CFD software was chosen to evaluate the
performance of turbines in this study.

4.1. Numerical model

The organic fluid vapor can’t be seen as an ideal fluid like the air, the thermodynamic and transport properties of them cannot be
accurately predicted by a simple state of the equation. In this study, to realize an accurate calculation of the turbine performance, the
thermodynamic and transport properties for different fluids are incorporated into the CFX solver through a user-generated lookup
table (RGP file), with fluids’ properties sourced from the NIST REFPROP [35].
ANSYS CFX provides the wall treatment model for the mesh with various y+ values. It is an automatic wall treatment model that
switches between the low-Reynolds number formulation at low y+ values and the wall function approach at higher y+ values. The SST
turbulent model [38] in CFX with the automatic wall treatment function has these advantages, therefore many researchers use this
model in their studies [33,39,40]. So the SST turbulence model was also selected in this research.

9
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

4.2. Mesh

The turbine stage is modeled as a single passage for the stator and rotor. For high computational accuracy of numerical results, fine
hexahedral grids are generated for the rotor and nozzle model using ANSYS TurboGrid. The CFX solver recommends a range of 30 < y+
< 300 for the selected SST turbulence model [3,34,39]. So in this study, the values of y+ were examined for each simulation case to
ensure they were in the recommended range. And the sensitivity analysis of the element number was also conducted, the ranges of the
element count for the meshes of components are listed in Table 5. The schematic of meshes is shown in Fig. 4.

4.3. Boundary conditions

The total pressure at the nozzle inlet (P02) and the total temperature (T02) were set at the nozzle inlet. Also, the medium level of
turbulence intensity was selected. At the rotor outlet, the average static pressure (P5) was set. The hub, shroud, rotor blade, and nozzle
blade were defined as smooth adiabatic walls. Concerning the lateral interfaces, which were set as rotational periodicity. For steady-
state simulations, the rotor to stator interface was coupled using the mixing-plane model which performs a circumferential averaging
of the fluxes through bands on the interface [33,34,39]. Second-order numerical schemes were used for all the solution processes. At
last, the mass average properties of working fluids were measured at each section of turbine passage respectively in this paper, as
shown in Fig. 5.

5. Results and discussions

5.1. The comparison between different models

The flow loss predicted by a reliable loss model should be linearly related to the actual losses. Therefore, the correlation analysis
was conducted firstly, and the results of Pearson’s correlation coefficients (CC) for different models are listed in Table 6. In this
research, a value of correlation coefficient bigger than 0.95 was regarded as a significant linear correlation. According to the results,
model Rinc-II and combinations E, F, O, P, Q, R have good performances in correlation. Among these models, combination E has the
biggest correlation coefficient of 0.9708.
Next, the statistic parameters mean (μ) and standard deviation (σ) were chosen to reflect the prediction performance of these seven
models, and the results are listed in Table 7. A lower σ means that the prediction accuracy is more stable under different working
conditions. And a μ closer to zero means that the average prediction performance is more accurate. From the results, model Rinc-II has
the lowest value of σ , and all of the other six combinations have also combined model Rinc-II in them. This seems to illustrate that model
Rinc-II plays a critical role in prediction stability. Comparing combinations E and F, the difference between them is only the type of
passage loss model, while better performances both in σ and μ have been achieved by combination E than F. This seems to illustrate
that model Rpas-I has a better performance than Rpas-II in loss prediction, within the scope of this research. And the same conclusions
also can be demonstrated by comparing combinations O and P and combinations Q and R respectively. Besides, it seems that there isn’t
a significant difference between the performances of combinations E, O, and Q. Therefore, combinations E, O, and Q were selected for
further analysis.
δ = (Δhmodel − ΔhCFD )/Δhid (60)

1∑ n
μ= δi (61)
n i=1

√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
1∑ n
σ= (δi − μ)2 (62)
n i=1

5.2. The prediction error of models

Fig. 6 (a) shows the prediction errors of models under different cases. It can be seen that the variation tendencies for prediction
errors of these four models are very similar. This also illustrates that model Rinc-II plays a critical role in prediction performance, which

Table 5
Mesh number of components.
Component Element count

Nozzle 509,106
Rotor 1 787,964
Rotor 2 724,341
Rotor 3 739,062
Rotor 4 865,991
Rotor 5 718,908

10
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

Fig. 4. Grid partition results of components.

Fig. 5. Measuring section in the turbine passage.

Table 6
Different models’ correlation coefficients.
Model CC Model CC Model CC

Rinc-I 0.8996 C 0.9204 K 0.8467

Rinc-II 0.9646 D 0.9271 L 0.8884
Rpas-I 0.6646 E 0.9708 M 0.9125
Rpas-II 0.8680 F 0.9621 N 0.9237
Rcle-I − 0.7944 G 0.2169 O 0.9534
Rcle-II − 0.6951 H 0.6013 P 0.9491
A 0.6646 I 0.7582 Q 0.9698
B 0.8680 J 0.8601 R 0.9617

is mentioned in section 5.1. With the increase of Δβ4 the prediction error decrease firstly, then turns to out-of-order when Δβ4 exceeds a
certain value (about 40◦ in this paper). To further explore this phenomenon, the results of combination Q are taken as an example and
the prediction error under different rotors is shown in Fig. 6 (b). It can be seen that the difference in prediction error under different
rotor structures is small when Δβ4 is lower than about 40◦ . However, a notable difference in prediction error under different rotor
structures appears when Δβ4 exceeds about 40◦ . When Δβ4 exceeds about 40◦ , the prediction error under Rotor 1 decreases consis­
tently, while for the other rotors the prediction error turns to raise with different gradients. The biggest value of δ is 8.3% when the
structure is Rotor 4 and Δβ4 equal to 66.4. The smallest value of δ is − 5.6% under Rotor 1 and Δβ4 equal to 65.6.
The main distinction in combinations E, O, and Q is the clearance loss model part, that combination E doesn’t include a clearance
loss model, combination O includes model Rcle-I, and combination Q includes model Rcle-II. From Fig. 6 (a), a notable deviation in the

11
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

Table 7
Statistical character of models.
Model σ μ
Rinc-II 1.96% − 8.44%
E 2.54% − 0.70%
F 5.09% 4.11%
O 2.99% 0.92%
P 4.58% 5.73%
Q 2.48% − 0.31%
R 4.85% 4.50%

Fig. 6. Prediction errors of models under different cases.

prediction error can be seen between combinations E, O, and Q when Δβ4 is relatively small. The prediction error of combination Q is
slightly higher than that of combination E. Meanwhile, the prediction error of combination O is obviously higher than that of the other
two combinations. The biggest deviation between O and E is 6.59% when the structure is the Rotor 5 and Δβ4 equal to − 29.5.
Δβ4 = β4 − βopt (63)

The performances of a turbine are mainly reflected in the efficiency and output power of two parts. Fig. 7 shows the performances of
turbines under different cases. As shown in Fig. 7 (a), the turbine efficiency decreases sharply for all turbines when Δβ4 exceeds about

12
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

Fig. 7. Performance of turbines under different cases.

13
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

40◦ . Meanwhile, the efficiencies of different turbines are similar when Δβ4 exceeds about 40◦ . Besides, when Δβ4 is lower than 40◦ ,
turbines’ efficiencies fluctuate slightly with the change of PR. In this situation, it should be seen that the efficiency of Rotor 3 is
obviously lower than other rotors, as shown in Fig. 7 (b). Fig. 7 (c) shows the output power of turbines under different cases. As same as
the variation of the efficiency, the output power also decreases sharply when Δβ4 exceeds about 40◦ . In addition, when Δβ4 is lower
than 40◦ , the output power of Rotor 3 is lower than other rotors under the same rotation speed and pressure ratio, as shown in Fig. 7
(d).
From the above discussion, the range of Δβ4 should be strictly restricted to avoid poor turbine performance. Among five rotors,
Rotor 3 has worse performance in both efficiency and output power than others under the same rotation speed and pressure ratio.
Therefore, it seems that the structure of Rotor 3 isn’t a suitable design. Based on the above factors, the simulation data has been sieved,
that the cases under Rotor 3 and the cases with Δβ4 higher than 40◦ are excluded. Fig. 8 shows the after sieved data of combination O, it
can be seen that the prediction performance under different rotor structures is near the same under a certain Δβ4. Besides the statistical
characters of these sieved data for each model are listed in Table 8. It can be seen that σ is reduced for each model under these restricted
conditions. The σ of combination E reduces from 2.54% to 1.16%, σ reduces from 2.99% to 2.31% for combination O, and σ reduces
from 2.48% to 1.41% for combination Q.

5.3. The examination of the turbine performance prediction program

Due to the inherent limitation of the CFD method, some loss factors were not embodied during the simulation process such as
surface roughness. Al Jubori et al. [41] conducted experiments to validate the performance of the turbine 3D CFD simulations. From
their research, both turbine efficiency and output power calculated by CFD are overestimated to a small extent, which compares to the
experiment results. So the turbine flow loss calculated from CFD is underestimated than that under actual working conditions [42].
Therefore, in this paper, the combination O is selected to use in the turbine performance prediction program, which produces little
higher loss prediction values than CFD results, as illustrated in Table 8.
After the determination of the suitable rotor loss model combination, the turbine performance prediction program was examed. To
investigate the potential influence of the type of working fluid on prediction performance, another three organic fluids MDM, R123,
and Pentane were chosen. The working conditions for the examination process are listed in Table 9. Besides, the deviations of effi­
ciency and output power are expressed by:
Δηts = ηCAL − ηCFD (64)
/
δpower = (PowerCAL − PowerCFD ) PowerCFD (65)

Fig. 9 (a) shows the turbine efficiency prediction deviation between the results of the program and the CFD. It can be seen that most
values of deviation are within the range from − 5% to 5%. Besides taking turbines with Pentane as examples, Fig. 9 (b) shows the
turbine efficiencies calculated from both CFD and program (CAL), the variation tendencies of ηts are matching well.
From Fig. 9 (a), it should also be noticed that the difference in rotor structure leads to a difference in prediction performance under
a certain working fluid. Besides, The working fluid type leads to a difference in prediction performance under a certain rotor structure
too. This is due to the prediction program only taken simple structure parameters and velocity distribution into account, while the real
rotor structure is complex and induces a complex 3D flow field. Table 10 lists the standard deviation (σ ) and mean (μ) of Δηts under
different cases. Among cases, the biggest μ of Δηts is − 5.21% when working is MDM and under Rotor 5. Under a certain rotor, the values
of both σ and μ for cases with R123 and Pentane are very close.

Fig. 8. Data after sieved.

14
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

Table 8
Statistical character of the sieved data.
Model σ μ
E 1.16% − 1.71%
O 2.31% 1.38%
Q 1.41% − 0.84%

Table 9
Working conditions for program examination.
Parameter MDM R123 Pentane

N [rpm] 30,000 30,000 45,000

PR 3–5 3–5 3–5
P02 [kPa] 303.99–506.65 328.74–547.90 303.99–506.65
T02 [K] 483.49–510.29 343.10–362.73 350.77–371.30
P5 [kPa] 101.33 109.58 101.33
Rotor 1, 2, 4, 5 1, 2, 4, 5 1, 2, 4, 5

Fig. 9 (c) shows the turbine output power prediction error between the results of the program and the CFD. Most deviation values
are within the range from − 10% to 10%. The reason for a higher prediction deviation in output power than that of efficiency is that
there also exists a deviation in mass flow rate. The boundary layer exists in real flow in turbines and reduces the effective flow area.
However, the current turbine performance prediction program hasn’t taken this into account. This is the inherent limitation of the
mean-line analysis and illustrates the deficiency of the prediction program presented in this study.
As same as the efficiency prediction, the rotor structure and the fluid type also influence the output power prediction performance.
The σ and μ of δpower under different cases are listed in Table 11. Among cases, the σ of δpower under Rotor 1 is notably higher than in
other cases, with the biggest value of 8.11%. This also illustrates the influence of rotor structure on turbine performance prediction.
Fig. 9 (d) shows the turbine output power calculated from both CFD and program (CAL), the variation tendencies are matching well.

6. Conclusions

In this study, a comprehensive analysis for the prediction performance of the ORC radial inflow turbine was present. The prediction
performance of different loss models, and the influence of both rotor structure and working fluid type on prediction performance were
discussed. The main conclusions are as follows.
According to the comparison of loss models, combinations E, O, and Q have a relatively good performance in loss prediction, among
them combination E has the biggest correlation coefficient of 0.9708. It also can be seen that, in the scope of this study, the perfor­
mance of loss model Rinc-II is better than Rinc-I, and Rpas-I is better than Rpas-II. Besides, model Rinc-II has a critical impact on loss
prediction stability.
When Δβ4 exceeds a certain value (about 40◦ in this study), notable differences in loss prediction performance appeared for cases
with different rotor structures. The biggest value of δ is 8.3% when the structure is Rotor 4 and Δβ4 equal to 66.4. The smallest value of
δ is − 5.6% when the structure is the Rotor 1 and Δβ4 equal to 65.6. While after the restriction on Δβ4 and rotor type, the σ of com­
bination E reduces from 2.54% to 1.16%, reduces from 2.99% to 2.31% for combination O, and reduces from 2.48% to 1.41% for
combination Q.
Different rotor structures lead to a difference in prediction performance under a certain working fluid, meanwhile, the different
working fluid types also lead to a difference in prediction performance under a certain rotor structure. Relatively close results of the
prediction deviation for cases with R123 and Pentane, but there is a notable difference compared to cases with MDM. The biggest μ of
Δηts is − 5.21% when working is MDM and under Rotor 5, and the biggest σ of δpower is 8.11% under Rotor 1 and with MDM as the
working fluid.
The proposed program can predict the performance of radial inflow turbines with multiple organic fluids and rotor structures with
acceptable accuracy in total. The Δηts roughly ranges from − 5% to 5%. And δpower roughly ranges from − 10% to 10%. However, It
should be noticed that there are also some obvious limitations in this study. All of the rotors investigated in this study have a radius
equal to 50 mm. Therefore, the conclusions in this paper may not suitable to conditions that the rotor size is far from this study. And the
program hasn’t taken the boundary layer into account, and this will lead to a deviation in the effective flow area. At last, to realize a
more reliable and fast turbine performance prediction under a larger scope of structures, working conditions, and working fluids, more
efforts need to be done by worldwide researchers in the future.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to
influence the work reported in this paper.

15
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

Fig. 9. The examination of the prediction program.

16
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

Table 10
The σ and μ of Δηts under different cases.
Rotor R123 MDM Pentane

σ μ σ μ σ μ
1 1.16% 3.39% 1.88% 1.51% 1.28% 3.12%
2 0.99% − 0.73% 0.36% − 4.12% 0.94% − 1.36%
4 1.25% 0.68% 1.17% − 2.38% 1.11% 0.06%
5 0.66% − 1.72% 0.73% − 5.21% 0.64% − 2.52%

Table 11
The σ and μ of δpower under different cases.
Rotor R123 MDM Pentane

σ μ σ μ σ μ
1 7.44% 6.18% 8.11% 2.58% 7.42% 6.00%
2 0.86% 5.83% 4.77% − 1.90% 1.15% 4.56%
4 1.25% 8.38% 2.52% 1.58% 2.06% 6.84%
5 0.99% 3.98% 3.30% − 2.90% 1.50% 2.54%

Acknowledgments

The authors gratefully acknowledge the financial support of the Postgraduate Research & Practice Innovation Program of Jiangsu
Province (No. KYCX20_1005).

Nomenclature

Symbols
A area (m2)
b blade width (m)
C absolute velocity (m/s)
c blade chord (m)
esse blade pitch (m)
f functions in REFPROP
h enthalpy (J/kg)
Δhid ideal enthalpy drop (J/kg)
Δhact actual enthalpy drop (J/kg)
ṁ mass flow rate (kg/s)
ns specific speed
N rotational velocity (rpm), blade number
P pressure
PR total to static pressure ratio
r radius (m)
Re Reynolds number
s entropy (J/kg K)
T temperature (K)
U rotor blade velocity (m/s)
W relative velocity (m/s)
ΔZr rotor axial length (m)

Greek letters
α absolute flow angle to meridional (◦ )
β relative flow angle to meridional (◦ )
γ setting angle to meridional (◦ )
δ relative error
η efficiency
ν velocity ratio
ρ density (kg/m3)
ω angular speed
ε tip clearance (m)

17
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

Ω reaction degree
μ mean
σ2 variance

Subscripts
0 total condition
1–5 stations within the turbine
bo boiling point
c critical state
h hub
hyd hydraulic
m meridional direction
s shroud/tip, isentropic
sat saturation state
sup superheat degree
t tangential direction
ts total to static
tt total to total

Acronyms
ORC organic Rankine cycle
CC correlation coefficient
PEN Pentane

Author statement

Wenyu Li: Conceptualization, Methodology, Software, Data curation, Writing - original draft, Validation.
Yunyuan Li: Methodology, Software.
Xing Ling: Resources, Supervision.

References

[1] J. Xia, K. Zhou, Y. Guo, J. Wang, J. Lou, Y. Dai, Preliminary design and CFD analysis of a radial inflow turbine and the turbine seal for an organic Rankine cycle
using zeotropic mixture, Energy Convers. Manag. 209 (2020) 112647, https://doi.org/10.1016/j.enconman.2020.112647.
[2] Y. Li, X. Ren, Investigation of the organic Rankine cycle (ORC) system and the radial-inflow turbine design, Appl. Therm. Eng. 96 (2016) 547–554, https://doi.
org/10.1016/j.applthermaleng.2015.12.009.
[3] D.Y. Kim, Y.T. Kim, Preliminary design and performance analysis of a radial inflow turbine for organic Rankine cycles, Appl. Therm. Eng. 120 (2017) 549–559,
https://doi.org/10.1016/j.applthermaleng.2017.04.020.
[4] D. Fiaschi, G. Manfrida, F. Maraschiello, Design and performance prediction of radial ORC turboexpanders, Appl. Energy 138 (2015) 517–532, https://doi.org/
10.1016/j.apenergy.2014.10.052.
[5] A.S. Panesar, An innovative organic Rankine cycle approach for high temperature applications, Energy 115 (2016) 1436–1450, https://doi.org/10.1016/j.
energy.2016.05.135.
[6] R. Kong, T. Deethayat, A. Asanakham, N. Vorayos, T. Kiatsiriroat, Thermodynamic performance analysis of a R245fa organic Rankine cycle (ORC) with different
kinds of heat sources at evaporator, Case Studies in Thermal Engineering 13 (2019) 100385, https://doi.org/10.1016/j.csite.2018.100385.
[7] T. Deethayat, T. Kiatsiriroat, C. Thawonngamyingsakul, Performance analysis of an organic Rankine cycle with internal heat exchanger having zeotropic
working fluid, Case Studies in Thermal Engineering 6 (2015) 155–161, https://doi.org/10.1016/j.csite.2015.09.003.
[8] P. Li, Z. Mei, Z. Han, X. Jia, L. Zhu, S. Wang, Multi-objective optimization and improved analysis of an organic Rankine cycle coupled with the dynamic turbine
efficiency model, Appl. Therm. Eng. 150 (2019) 912–922, https://doi.org/10.1016/j.applthermaleng.2019.01.058.
[9] P. Klonowicz, F. Heberle, M. Preißinger, D. Brüggemann, Significance of loss correlations in performance prediction of small scale, highly loaded turbine stages
working in Organic Rankine Cycles, Energy 72 (2014) 322–330, https://doi.org/10.1016/j.energy.2014.05.040.
[10] A. Meroni, M. Robertson, R. Martinez-Botas, F. Haglind, A methodology for the preliminary design and performance prediction of high-pressure ratio radial-
inflow turbines, Energy 164 (2018) 1062–1078, https://doi.org/10.1016/j.energy.2018.09.045.
[11] J. Song, C. Gu, X. Ren, Influence of the radial-inflow turbine efficiency prediction on the design and analysis of the Organic Rankine Cycle (ORC) system, Energy
Convers. Manag. 123 (2016) 308–316, https://doi.org/10.1016/j.enconman.2016.06.037.
[12] Y. Li, W. Li, X. Gao, X. Ling, Thermodynamic analysis and optimization of organic Rankine cycles based on radial-inflow turbine design, Appl. Therm. Eng.
(2020) 116277, https://doi.org/10.1016/j.applthermaleng.2020.116277.
[13] A.M. Al Jubori, R. Al-Dadah, S. Mahmoud, New performance maps for selecting suitable small-scale turbine configuration for low-power organic Rankine cycle
applications, J. Clean. Prod. 161 (2017) 931–946, https://doi.org/10.1016/j.jclepro.2017.05.177.
[14] S. Schuster, C.N. Markides, A.J. White, Design and off-design optimisation of an organic Rankine cycle (ORC) system with an integrated radial turbine model,
Appl. Therm. Eng. 174 (2020) 115192, https://doi.org/10.1016/j.applthermaleng.2020.115192.
[15] Y. Zheng, D. Hu, Y. Cao, Y. Dai, Preliminary design and off-design performance analysis of an Organic Rankine Cycle radial-inflow turbine based on mathematic
method and CFD method, Appl. Therm. Eng. 112 (2017) 25–37, https://doi.org/10.1016/j.applthermaleng.2016.10.036.
[16] F. Alshammari, A. Karvountzis-Kontakiotis, A. Pesiridis, P. Giannakakis, Off-design performance prediction of radial turbines operating with ideal and real
working fluids, Energy Convers. Manag. 171 (2018) 1430–1439, https://doi.org/10.1016/j.enconman.2018.06.093.
[17] R. Persky, E. Sauret, Loss models for on and off-design performance of radial inflow turbomachinery, Appl. Therm. Eng. 150 (2019) 1066–1077, https://doi.org/
10.1016/j.applthermaleng.2019.01.042.

18
W. Li et al. Case Studies in Thermal Engineering 28 (2021) 101400

[18] M. Masi, L. Da Lio, A. Lazzaretto, An insight into the similarity approach to predict the maximum efficiency of organic Rankine cycle turbines, Energy 198
(2020) 117278, https://doi.org/10.1016/j.energy.2020.117278.
[19] Z. Han, X. Jia, P. Li, Improved thermodynamic and aerodynamic design method and off-design performance analysis of a radial inflow turbine for ORC system,
Int. J. Energy Res. 43 (2019), https://doi.org/10.1002/er.4830 er.4830.
[20] Z. Han, X. Jia, P. Li, Preliminary design of radial inflow turbine and working fluid selection based on particle swarm optimization, Energy Convers. Manag. 199
(2019) 111933, https://doi.org/10.1016/j.enconman.2019.111933.
[21] A.M. Al Jubori, R. Al-Dadah, S. Mahmoud, Performance enhancement of a small-scale organic Rankine cycle radial-inflow turbine through multi-objective
optimization algorithm, Energy 131 (2017) 297–311, https://doi.org/10.1016/j.energy.2017.05.022.
[22] H. Ronald, Aungier. Turbine Aerodynamics: Axial-Flow and Radial-Inflow Turbine Design and Analysis, ASME Press, New York, 2005.
[23] S. Lee, H. Gurgenci, A comparison of three methodological approaches for meanline design of supercritical CO2 radial inflow turbines, Energy Convers. Manag.
206 (2020) 112500, https://doi.org/10.1016/j.enconman.2020.112500.
[24] L. Da Lio, G. Manente, A. Lazzaretto, A mean-line model to predict the design efficiency of radial inflow turbines in organic Rankine cycle (ORC) systems, Appl.
Energy 205 (2017) 187–209, https://doi.org/10.1016/j.apenergy.2017.07.120.
[25] O. Aboelwafa, S.E.K. Fateen, A. Soliman, I.M. Ismail, A review on solar Rankine cycles: working fluids, applications, and cycle modifications, Renew. Sustain.
Energy Rev. 82 (2018) 868–885.
[26] C.A.M. Ventura, P.A. Jacobs, A.S. Rowlands, P. Petrie-Repar, E. Sauret, Preliminary design and performance estimation of radial inflow turbines: an automated
approach, Journal of Fluids Engineering, Transactions of the ASME 134 (2012) 1–13, https://doi.org/10.1115/1.4006174.
[27] H.E. Bekiloğlu, H. Bedir, G. Anlaş, Multi-objective optimization of ORC parameters and selection of working fluid using preliminary radial inflow turbine design,
Energy Convers. Manag. 183 (2019) 833–847, https://doi.org/10.1016/j.enconman.2018.12.039.
[28] A. Whitfield, N. Baines, Design of Radial Turbomachines, 1990.
[29] Y. Gao, P. Petrie-Repar, Validation of meanline performance prediction method for radial and mixed flow turbine, in: Institution of Mechanical Engineers - 13th
International Conference on Turbochargers and Turbocharging 2018, 2018, pp. 357–372.
[30] C. Wasserbauer, A. Glassman, FORTRAN Program for Predicting Off-Design Performance of Radial-Inflow Turbines, NASA Technical Report, 1975.
[31] H. Moustapha, M. Zelesky, N. Baines, D. Japikse, Axial and Radial Turbines, Concepts NREC, 2003.
[32] S. Lee, G. Yaganegi, D.J. Mee, Z. Guan, H. Gurgenci, Part-load performance prediction model for supercritical CO2 radial inflow turbines, Energy Convers.
Manag. 235 (2021) 113964, https://doi.org/10.1016/j.enconman.2021.113964.
[33] A.L. Espinosa Sarmiento, R.G. Ramirez Camacho, W. de Oliveira, Performance analysis of radial-inflow turbine of ORC: new combined approach of preliminary
design and 3D CFD study, J. Mech. Sci. Technol. 34 (2020) 2403–2422, https://doi.org/10.1007/s12206-020-0517-5.
[34] A.L. Espinosa Sarmiento, R.G. Ramirez Camacho, W. de Oliveira, E.I. Gutiérrez Velásquez, M. Murthi, N.J. Diaz Gautier, Design and off-design performance
improvement of a radial-inflow turbine for ORC applications using metamodels and genetic algorithm optimization, Appl. Therm. Eng. 183 (2021) 116197,
https://doi.org/10.1016/j.applthermaleng.2020.116197.
[35] E.W. Lemmon, M.L. Huber, M.O. Mclinden, NIST Standard Reference Database 23: Reference Fluid Thermodynamic and Transport Properties - REFPROP,
Version 9.1, Standard Reference Data Program, National Institute of Standards and Technology, 2010. Nist Nsrds.
[36] C. Rodgers, Mainline performance prediction for radial inflow turbines. Technical report on Von Karman Institute for Fluid Dynamics, Lecture Series on Small
High Pressure Ratio Turbines, 1987.
[37] F. Alshammari, A. Karvountzis-Kontakiotis, A. Pesyridis, I. Alatawi, Design and study of back-swept high pressure ratio radial turbo-expander in automotive
organic Rankine cycles, Appl. Therm. Eng. 164 (2020) 114549, https://doi.org/10.1016/j.applthermaleng.2019.114549.
[38] F.R. Menter, Two-equation eddy-viscosity turbulence models for engineering applications, AIAA J. 32 (1994) 1598–1605, https://doi.org/10.2514/3.12149.
[39] J.A. Keep, I.H.J. Jahn, Numerical loss investigation of a small scale, low specific speed supercritical CO2 radial inflow turbine, J. Eng. Gas Turbines Power 141
(2019) 1–10, https://doi.org/10.1115/1.4043430.
[40] M. Kumar, D. Panda, R.K. Sahoo, S.K. Behera, Performance prediction, numerical and experimental investigation to characterize the flow field and thermal
behavior of a cryogenic turboexpander, Heat Mass Tran. 56 (2020) 1015–1036, https://doi.org/10.1007/s00231-019-02777-w.
[41] A.M. Al Jubori, F.N. Al-Mousawi, K. Rahbar, R. Al-Dadah, S. Mahmoud, Design and manufacturing a small-scale radial-inflow turbine for clean organic Rankine
power system, J. Clean. Prod. 257 (2020) 120488, https://doi.org/10.1016/j.jclepro.2020.120488.
[42] F. Alshammari, A. Pesyridis, A. Karvountzis-Kontakiotis, B. Franchetti, Y. Pesmazoglou, Experimental study of a small scale organic Rankine cycle waste heat
recovery system for a heavy duty diesel engine with focus on the radial inflow turbine expander performance, Appl. Energy 215 (2018) 543–555, https://doi.
org/10.1016/j.apenergy.2018.01.049.

19