You are on page 1of 8

A comparison of the behaviour of

refrigerant R404A in a scraper shaft heat


exchanger(SSHE) made from different
materials and used as an evaporator
Kunal Gupta1
TU Delft, Faculty of Aerospace Engineering, The Netherlands1
Dilip R. Sarda2
Synergy Agro Tech Pvt Ltd., Gujrat, India2
November 1, 2022

Abstract
This paper examines the flow behaviour of the refrigerant R404A in a scraper shaft
heat exchanger made out of different materials. The Gungor-Winterton correlation was
used to create a digital model which predicted the two-phase convective heat transfer
coefficient(htp ) of R404A along the length of the heat exchanger(evaporator), the over-
all heat transfer coefficient(U ) of the evaporator along its length, and finally the vapor
quality of the flow along the length of the evaporator for 3 different materials: Copper,
Brass(60/40) and Stainless Steel-304. The model was later validated with an experimental
run of the refrigeration system with a custom built SS-304 evaporator of the same dimen-
sions. The test and the model both showed that there was about 2.4% decrease in the
capacity of the evaporator for SS-304 material in comparison to Brass. With these results
the conclusion drawn was that the thermal conductivity of the material does not affect
the capacity of the evaporator a lot and can be compensated for by reducing the thickness
of the material or reducing the flow area to increase the mass velocity. Finally a linear
correlation was obtained for the vapor quality of the refrigerant flow over the length of the
evaporator.

Keywords: Two-phase flow, Two-phase convective heat transfer coefficient, Overall


heat transfer coefficient, heat flux, scraper shaft heat exchanger, vapor quality

Nomenclature G Mass velocity


g gravitational acceleration
ṁ mass flow rate
h convective heat transfer coefficient
µg viscosity gas(vapour)
Hf g ,hf g latent heat of vaporization
µl viscosity liquid
hf ,hlo liquid heat transfer coefficient
ρ density
hg , hv vapour heat transfer coefficient
ρg density gas(vapour)
ρl density liquid Hi inlet enthalpy

A Area Hl liquid state enthalpy

Bo Boiling number Ho outlet enthalpy

cp specific heat at constant pressure htp two-phase heat transfer coefficient

cv specific heat at constant volume ID inner diameter


D Diameter k thermal conductivity
Dh Hydraulic Diameter L Length
dl length of small section M Molecular mass
Fr Froude number OD outer diameter

1
p pressure A. Flow-Boiling
Pc critical pressure Much like most other liquids, as the refrigerants start to
heat up evaporation occurs, and as the difference between
Pr reduced pressure
the temperature of the cylinder wall and the refrigerant
Psat saturated pressure increases small bubbles start to form on the wall surface
in contact with the fluid. This bubbling phenomenon is
Pr Prandtl number called nucleate boiling and is helpful if the focus is on
q ′′ heat flux higher heat transfer rates. The bubbles reduce the ther-
mal resistance of the fluid and greatly increase the heat
Re Reynolds number transfer coefficient. As the heat flux increases, more and
U Overall heat transfer coefficient more liquid refrigerant turns to vapor, and the flow ve-
locity increases. This increase in the flow velocity also
x vapour quality increases the heat transfer coefficient of the refrigerant.
With an increase in vapor quality[x] a mist flow can be
Xtt Martinelli parameter
expected in the cylinder. Even though the htp increases
with vapor quality at small values of x, a sharp drop is
observed in the heat transfer coefficient as the flow dries
and the vapor quality gets near 70 to 90 percent.[7]

I. Introduction
B. Existing models for htp calculation
The refrigeration cycle is based on the state change of the
As mentioned earlier there are some models which were
coolant from liquid to gas and back. Even though this
created with experimentally obtained data from certain
multi phase system is used in many refrigeration systems,
refrigerants and are specific to those refrigerants such
the two-phase state change model is still not completely
as the Shah model[2], the Kandlikar model[7], the Wat-
understood and is debated upon by many researchers and
telet model[6], the Gungor-Winterton model[6]. How-
institutions. The existing models for the two-phase con-
ever, a comparative study performed by O.E. Turgut and
vective heat transfer coefficients[htp ] such as Kandlikar[1],
M. Asker[6] provided an overview of how various models
Shah [2], Gungor-Winterton[3], etc do not all yield the
compare with the experimentally obtained data from a
same results for the same fluid under the same conditions.
set of 9 refrigerants: R22, R134a, R290, R404a, R410a,
This can be observed in several research papers which plot
R600, R507, R717, and R744. In their research pa-
the htp predicted by the existing models along with the ex-
per titled ”Flow Boiling Behaviours Of Various Refrig-
perimentally obtained data.[4][5][6]
erants Inside Horizontal Tubes: A Comparative Research
It is important to note that all previously mentioned
Study” we can see that for R404A and most other refrig-
papers stop at the comparison of ht p correlations. How-
erants experimented with, the Gungor-Winterton model,
ever, this research paper takes it a step further and analy-
lies within ±20 percent of the experimentally obtained re-
ses how to apply the predicted models and correlations to
sults throughout the vapor quality distribution(from 0.1
the refrigeration systems. More specifically, how can the
to 0.9). The equations or the Gungor-Winterton model
overall heat transfer coefficient[U ] be obtained for various
are as follows:
parts in a refrigeration system such as an evaporator and
weather the design can be further optimised?
G2
This research paper attempts to answer the questions F rl = (1)
above with the help of an extensive literature study, a dig- ρ2l ·g·D
ital creation of the Gungor-Winterton, development of a If F rl is less than 0.05 and the tube is horizontal then the
”vapour quality vs cylinder length” model, a comparison E2 and S2 are calculated using Equation 2 otherwise E2
between various metals and refrigerants in use, the exper- and S2 both equal 1.
imental results from a ”Continuous Ice-Cream Freezer”
provided by Synergy(also the location of my internship), (0.1−2·F rl ) 1/2
E2 = F rl S2 = F rl (2)
and finally a conclusion based on the obtained results.
The Reynolds number and the Prandtl number can
be calculated with the formulas stated in Equation 3.
II. Literature review
G · (1 − x) · D µl · cp
The models mentioned in this paper were created in an Rel = P rl = (3)
µl kl
attempt to aid the refrigeration industry in optimizing the
design of their heat exchangers (especially the evaporator). The next step in the process is to determine the
The models were created for smooth horizontal tubes and Lockhart-Martinelli parameter(Xtt ) and the Boiling num-
include a htp vs vapour quality plot based on the Gungor- ber(Bo). The Lockhart-Martinelli parameter is defined as
Winterton correlation, an over all heat transfer coefficient the square root of the ratio of the pressure gradient of liq-
vs vapour quality and the vapour quality vs cylinder length. uid to the pressure gradient of gas or vapor and the boiling
In order to understand how these models were created, it number is the ratio of the actual heat flux to the maxi-
is important to first understand the behavior of the flow. mum heat flux achievable by complete evaporation of the

2
liquid.[8][9] A. Model Inputs

1−x
0.9  0.5  0.1
ρv µl In order to calculate the htp , the model requires certain
Xtt = · · (4) inputs:
x ρl µv
– Input 1: The inlet temperature of the refrigerant.
q ′′
Bo = (5) [deg Celsius]
G · hf g
The two-phase heat transfer coefficient takes into ac- – Input 2: Length of the evaporator [m]
count both, the liquid heat transfer coefficient at zero
vapour quality hlo and the pool boiling term hp proposed – Input 3: The total heat load on the refrigerant [W]
by Cooper[10]. Both can be calculated with Equation 6
and Equation 7 respectively. – Input 4: The diameter of the tube(s) [m]

kl – Input 5: The mass flow rate of the refrigerant [kg/s]


hlo = 0.023 · Re0.8
l · P rl0.4 · (6)
D
The total heat load on the refrigerant can be calcu-
hp = 55 · Pr0.12 · M −0.5 · q ′′0.67 · (− log10 Pr )−0.55 (7)
lated from the heat lost by the ice cream mix to turn
where Pr is the reduced pressure and can be calcu- into soft ice cream. For a mix with 12% fat, 11% non-
lated using the following formula: milk solids(NMS), 16% sugar, and 0.3% stabilizer the spe-
cific heat is estimated to be 3.35 kJ/kg ◦ C. Similarly,
Psaturated
Pr = (8) the specific heat for semi-frozen ice cream of this mix is
Pcritical 2.72 kJ/kg ◦ C. Furthermore, the latent heat of water is
The final step before obtaining the htp value for the 334.2 kJ/kg for a water percentage of 60.7% and a freez-

refrigerant is the determination of the enhancement and ing point of -2.63 C for the mix. This slush formation
suppression factors defined as E & S respectively.[11] freezes roughly 48% of the water content in the mix.[12]
0.86
E = 1 + 2.4 · Bo1.16 · 104 + 1.37 · (1/Xtt ) (9)
Sensible heat of mix = [5 − (−2.63)] · 3.35 [kJ/kg]
1
S= (10)
1 + (1.15 · 10−6 · E 2 · Re1.17 )
l = 25.56 [kJ/kg] (12)
The complete form of the Gungor-Winterton correla-
tion for the two-phase convective heat transfer coefficient Latent heat of fusion = 334.2 · 0.607 · 0.48 [kJ/kg]
is as follows:
= 97.37 [kJ/kg] (13)
htp = E · E2 · hlo + S · S2 · hp (11)
Sensible heat of slush = [−2.63 − (−5)] · 2.72 [kJ/kg]

III. Digital Models = 6.45 [kJ/kg] (14)

A digital model was created on PYTHON based on the Total energy lost/kg = 25.56 + 97.37 + 6.45 [kJ/kg]
previously mentioned Gungor-Winterton correlation. This
model calculates the htp of the refrigerant/fluid flow- = 129.38 [kJ/kg] (15)
ing through a smooth horizontal tube. It was then ap-
plied to a simplified model of a scrapper shaft heat- As the machine produces 108 kg of ice cream/hr, the
exchanger(SSHE) where the refrigerant flows between the cooling load is obtained by multiplying Equation 15 with
outer and inner cylinder with the inlet and outlet being the machine capacity giving approximately 3.9 kW. This
the smaller tubes on left and right end of the cylinder calculation is an estimate of the heat load, hence a slightly
respectively and the secondary fluid(in this case an ice- higher value of 4.14 kW was chosen for the model.
cream mix) flowing within the inner cylinder in counter Furthermore, the flow of refrigerant is in between two
flow to the refrigerant. cylinders instead of being within a cylinder, this results
in the need for hydraulic diameter(Dh ) to be used for the
diameter input part of the model.

π
· IDoutercylinder 2 − ODinnercylinder 2

4· 4
Dh =
π · (IDoutercylinder + ODinnercylinder )
(16)
In addition to the user inputs, there is also a database
that consists of all the thermodynamic properties of the
refrigerant in use(R404A). This database is a txt file which
Figure III.1: Simplified design of the SSHE. - contains properties such as the refrigerant’s density, vis-
Mr. Rahul K. Vachheta cosity, thermal conductivity, enthalpy, entropy, etc for
both liquid and vapour form.[7]

3
B. Assumptions – Mass flow rate of the refrigerant: 0.041 [kg/s]
As challenging as it is to model and analyse a two-phase We see that according to the Gungor-Winterton cor-
flow for the refrigerant, it is even more challenging to per- relation, the htp in general drops with increasing vapour
form the same activity for the ice-cream mix(liquid) which quality for the given inputs. From x = 0 to x = 0.8 the heat
slowly turns to solid as it moves along the evaporator. transfer coefficient drops by almost 67% after which the
Hence a few assumptions were made to simplify the model flow becomes dry and therefore causes the rapid fall of htp .
a little and reduce the required computing power. However, it is important to note that at the smaller values
of x, the convective heat transfer coefficient increases for
– Assumption 1: A constant heat flux is assumed high mass flow rates as seen in Figure III.3 where the mass
through out the length of the cylinder. flow rate is 20 times higher. This is consistant with the
– Assumption 2: The average convective heat transfer explanation in Section II.A.
coefficient of the ice-cream mix is used for the calcula-
tions and assumed to be constant (280 W/m2 K).[13]

– Assumption 3: There is no heat loss/gain to/from the


atmosphere.

– Assumption 4: It is assumed that there is flow tran-


sition within the evaporator.

– Assumption 5: The model does not account for pres-


sure drops, hence it is assumed that there will only
be very small pressure drops if at all.

C. Model Output & Analysis


Based on the aforementioned inputs and assumptions, the
model firstly outputs a curve of convective heat transfer
coefficient(htp ) with respect to the vapour quality of the
flow(x). Figure III.3: A graph of htp [W/m2 K] vs x with
a mass flow rate of 0.82 Kg/s.

The second output of the model is the overall heat


transfer coefficient U of the evaporator with respect to the
vapor quality. One of the many goals of the internship was
to increase the overall heat transfer coefficient of the evap-
orator. This part of the model analyses the effectiveness
of 3 different materials used in making the inner cylin-
der. Copper, Brass(60/40), and Stainless Steel 304 were
compared to understand how the design could be made
more affordable while still performing adequately. Alu-
minium was also considered in the analysis but it suffers
from pitting corrosion as the evaporator is cleaned with
water after a production cycle causing many customers
to suggest against it. Therefore Aluminium was removed
from the list of potential materials.
U can be split into 3 parts, convection in ice-
Figure III.2: A graph of htp [W/m2 K] vs x. cream(CIce−cream ), convection in R404A(CR404A ) and
conduction in the metal(CM etal ).This calculation was
The model inputs for Figure III.2 are as follows: done over the entire vapor quality range(x = 0 to x =
1) with an increment of x = 0.001.[14]
– Refrigerant: R404A

– Inlet temperature: −25◦ [Celsius] 


1

CIce−cream : (17)
– Heat load: 4140 [W] hicecreamavg · AIDinnercylinder

– Inner diameter of the outer cylinder: 0.1065 [m] 


1

CR404A : (18)
– Inner diameter of the inner cylinder: 0.077 [m] hR404Atp · AODinnercylinder

– Outer diameter of the inner cylinder: 0.083 [m]  


ODinnercylinder − IDinnercylinder
– Length of the evaporator: 0.63 [m] CM etal : (19)
2 · π · kmetal · l · IDinnercylinder

4
thickness of the inner cylinder to 2.5 mm and changing
the ID of the outer cylinder from 0.1065 mm to 0.103 mm
(CIce−cream + CR404A + CM etal )−1 as seen in Figure III.6.
U= [W/m2 K]
AODinnercylinder
(20)

Figure III.6: Graph of U [W/m2 K] vs x for


IDoutercylinder = 0.103 m & ODinnercylinder = 0.082
Figure III.4: A graph of U [W/m2 K] vs x for m.
cylinder thickness of 3 mm.

D. Vapour quality vs Cylinder length


The final output of the model is the Vapour quality vs
the Cylinder length curve. This plot together with the htp
and U curves can provide either the htp of the refrigerant
or the U of the evaporator along the length of the evapo-
rator. This model provides an estimate of the state of the
refrigerant and its thermodynamic properties at any given
length for the desired model inputs.
The cylinder is first split horizontally into several
small sections and the inlet enthalpy of the fluid is set
to its enthalpy in its liquid state. Now for every section
following the starting section, the inlet enthalpy is set to
equal the outlet enthalpy of the previous section.

Figure III.5: A graph of U [W/m2 K] vs x for


cylinder thickness of 5 mm.

Figure III.4 shows that even though the thermal con-


ductivity of Copper is 3 times that of Brass(60/40), the
difference in U between Copper and Brass(60/40) is less
than 1% for the given model inputs. It is also observed
that as the thickness of the metal cylinder increases, the
thermal conductivity plays a more important role causing
the U to drop further and an increase in the difference in U Figure III.7: Cylinder sectioned into multiple
among the three metals. This can be seen in Figure III.5 parts.
where all the inputs stay the same except the ID(inner
diameter) of the inner cylinder which changes to 73 mm.
This makes Brass a more feasible material. Hence q ′′ · dl · π · ODinnercylinder
Synergy is currently making use of Brass cylinders for the Ho = H i + (21)

evaporator. However, a move to SS-304 is being considered
as the decrease in U from Brass to SS-304 is only about 2% Ho − Hl
x= (22)
for the given model inputs and cylinder thickness of 3 mm Hf g
but can drastically improve the lifetime of the evaporator. The outlet enthalpy Ho and the vapour quality at that
This 2% loss in U can easily be recovered by reducing the point can be calculated with Equation 21 and Equation 22

5
above. When performed over 1000 sections with the inputs
provided in Section III.A, a graph of vapour quality vs
cylinder length is obtained. The correlation between x
and cylinder length seems to be linear and from linear
interpolation follows the equation:

vapourquality ≈ 0.8566 · lengthcylinder (23)

Figure IV.1: Process flow diagram for the


refrigeration system.

Figure III.8: Graph of vapor quality vs


cylinder length. The experiment was run for a duration of 15-16 mins
and the steady state results were recorded after 7 mins
had passed. Three separate measurements were taken
The Figure III.9 is an image of the current evapora- at 3 minutes intervals each after steady state had been
tor. The inlet conditions are the same as mentioned in achieved. Table 1 shows the list of data collected. A list
Section III.A with an addition of the inlet vapor quality of reference data with Brass evaporator is also presented
of 0.5. This means that the refrigerant at the outlet is in in Table 2.
a superheated vapor form as it then enters the compres-
sor because the length of the evaporator is larger than the
difference in length between that at x = 1 and x = 0.5.

Table 1: 1st run with SS-304 evaporator.

Time [mins] +7 +3 +3 +3
Density
of Ice-cream Mix 1111
[g/Ltr]
Density
of Ice-cream - 525 535 540
[g/Ltr]
Temperature
of Mix 8.5 11 11 11
Figure III.9: An image of scraper shaft heat
[Celsius]
exchanger.
Ice-cream
outlet temp - -4.1 -4.3 -4.5
[Celsius]
IV. Experimental Data Time for 1 Ltr
- 19.6 19.8 19.5
Ice-cream
An SS-304 evaporator was manufactured with the same Ice-cream
dimensions as the old brass evaporators to test its feasi- capacity - 183.67 181.82 184.62
bility and validate the model. The stainless steel evap- [Ltr/hr]
orator(freezing cylinder) was mounted onto a simple re- Product factor
1.09
frigeration system with a condenser, a compressor, and a for Ice-cream*
thermostatic expansion valve. The process flow diagram Product factor
0.86
for the experiment is presented in Figure IV.1. for Mix**

6
Table 2: Reference data with the current Brass the vapor quality of the refrigerant along the length of the
evaporator. evaporator appears to have a linear correlation suggesting
a constant amount of vapor formation for every section of
Time [mins] +7 +3 +3 +3 length dL in Figure III.7.
Density This model can be further improved by accounting
of Ice-cream Mix 1116 for the variation in heat flux along the length of the evap-
[g/Ltr] orator. It would give a more accurate and realistic result.
Density It is also recommended to account for heat losses to the
of Ice-cream - 558 560 563 environment for an even better result.
[g/Ltr]
Temperature
of Mix 5 5 5 5 VI. Acknowledgement
[Celsius]
Ice-cream This research was supported by Synergy Agro Tech Pvt
outlet temp - -4.9 -4.9 -4.9 Ltd. I would like to thank the entire Synergy team for pro-
[Celsius] viding great support throughout the span of my internship
Time for 1 Ltr and this research and also for assisting with the experi-
- 18 18.1 18 mental setups and testing. I would also like to take this
Ice-cream
Ice-cream opportunity to thank Mr. Yogesh Lavingia, Mr. Ameet
capacity - 200 199 200 K. Kaila and Mr. Harshal D. Sarda for providing techni-
[Ltr/hr] cal support and sharing their wisdom. Finally I would like
Product factor to express my gratitude to Mr. Dilip R. Sarda for his ex-
- pertise, guidance and insights without which this research
for Ice-cream*
Product factor probably would not have taken place.
-
for Mix**

References
The experiment showed that the ice cream capacity
had been reduced by 8-9%. However, since neither the [1] Kandlikar, S. “A General Correlation for Satu-
mix nor the outlet ice cream was at the required tempera- rated Two-Phase Flow Boiling Heat Transfer In-
tures: 5◦ C and -5◦ respectively, correction factors needed side Horizontal and Vertical Tubes”. In: Journal of
to be applied. The Product factor of the ice cream is based Heat Transfer-transactions of The Asme - J HEAT
on the amount of ice cream production expected at -5◦ C, TRANSFER 112 (Feb. 1990), pp. 219–228. doi: 10.
200 Ltr/hr with the Brass evaporator. So if the ice cream 1115/1.2910348.
outlet temperature is -4.5◦ C, the machine should be pro-
ducing a higher volume of ice cream to match the heat [2] Shah, M. “Chart correlation for saturated boiling
load. heat transfer: Equations and further study”. In:
Similarly, the Product factor of the mix is also based vol. 88. Jan. 1982.
on 5◦ C in a Brass evaporator. If the mix temperature is [3] Gungor, K. and Winterton, R. “A general correla-
higher than 5◦ C, 11◦ C in this case, then the machine is tion for flow boiling in tubes and annuli”. In: In-
expected to produce a smaller quantity of ice cream. ternational Journal of Heat and Mass Transfer 29.3
This results in a net product factor of roughly 0.93, (1986), pp. 351–358. issn: 0017-9310. doi: https:
giving an expected capacity of 187.5 Ltr/hr. Since the / / doi . org / 10 . 1016 / 0017 - 9310(86 ) 90205 - X.
achieved capacity with the SS-304 evaporator was approx- url: https://www.sciencedirect.com/science/
imately 183 Ltr/hr, a reduction of about 2.4% is observed article/pii/001793108690205X.
which is similar to the reduction predicted by the model. [4] Kwang-Il Choi Nguyen-Ba Chien, J.-T. O. “Heat
Transfer Coefficient during Evaporation of R-1234yf,
R-134a, and R-22 in Horizontal Circular Small
V. Conclusion Tubes”. In: Advances in Mechanical Engineering
2013 (Aug. 2013).
While almost every refrigeration system relies on the two-
[5] A. Greco, G. V. “Flow-boiling of R22, R134a, R507,
phase state change of the refrigerant, the flow of the refrig-
R404A, and R410A inside a smooth horizontal
erant through the evaporator as it changes from a liquid
tube”. In: International Journal of Refrigeration 1
state to a vapor state is still not very well understood.
(Apr. 2005). doi: 10.1016/j.ijrefrig.2005.01.
This paper shows that the flow of the refrigerant through
008.
the evaporator is governed by many factors including its
temperature, the heat load, its thermodynamic proper- [6] O.E. Turgut, M. A. “Flow boiling behaviours of var-
ties, and most of all by its mass velocity(mass flow rate ious refrigerants inside horizontal tube: a compara-
per unit area). Interestingly, the thermal conductivity of tive research study”. In: Eskisehir Technical Uni-
the metal has little effect on the overall heat transfer co- versity Journal of Science and Engineering 1 (Jan.
efficient of an evaporator due to the small thickness. This 2022). doi: 10.18038/estubtda.749040.
allows for a wider variety of metals to be used in the evap- [7] ASHRAE. 2021 ASHRAE Handbook: Fundamentals.
orator with only minor changes to their geometry. Finally, Mark S. Owen, 2021.

7
[8] Lockhart, R. and Martinelli, R. “Proposed Cor-
relation of Data for Isothermal Two-Phase, Two-
Component Flow in Pipes”. In: Chemical Engineer-
ing Progress 45 (1949).
[9] Thome, J. R. Engineering Data Book 3. Wolverine
Tube, Inc, 2004.
[10] Cooper, M. G. “Saturation nucleate pool boiling: a
simple correlation”. In: 1st UK National Conference
on Heat Transfer. Vol. 2. 1984, pp. 785–793.
[11] Sadik Kakac, H. L. and Pramuanjaroenkij, A. Heat
Exchangers: Selection, Rating, and Thermal Design.
CRC Press, Taylor and Francis group, 2012.
[12] Robert T. Marshall H. Douglas Goff, R. W. H. Ice
Cream. Kluwer Academic/Plenum Publishers, 2003.
[13] Byun, H.-W., Lee, J.-W., and Kim, N.-H. “Heat
Transfer Characteristics in the Evaporator of a
Soft Ice Cream Maker”. In: Journal of the Korea
Academia-Industrial cooperation Society 13 (Apr.
2012). doi: 10.5762/KAIS.2012.13.4.1466.
[14] A.D. Parekh P.R. Tailor, T. P. “Numerical Sim-
ulation of R410a-R23 and R404A-R508B Cascade
Refrigeration System”. In: International Journal of
Mechanical and Mechatronics Engineering 4 (Oct.
2010).

You might also like