See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/260212848

Modelling thermostatic expansion valves

Article  in  International Journal of Refrigeration · February 2014

DOI: 10.1016/j.ijrefrig.2013.06.010

CITATIONS READS

12 3,529

3 authors:

Ian W. Eames Adriano Milazzo

University of Nottingham University of Florence
73 PUBLICATIONS   1,965 CITATIONS    53 PUBLICATIONS   616 CITATIONS   

SEE PROFILE SEE PROFILE

Graeme Maidment
London South Bank University
127 PUBLICATIONS   1,514 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

HPC 2012 View project

IIR Mitigation of Direct Emissions of Greenhouse Gases in Refrigeration View project

All content following this page was uploaded by Ian W. Eames on 10 October 2016.

The user has requested enhancement of the downloaded file.

 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 8 ( 2 0 1 4 ) 1 8 9 e1 9 7

Available online at www.sciencedirect.com

w w w . i i fi i r . o r g

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

Modelling thermostatic expansion valves

Ian W. Eames a,*, Adriano Milazzo b, Graeme G. Maidment c

a
Department of Mechanical, Materials and Manufacture, University of Nottingham, Nottingham, NG72RD, UK
b
Dipartimento di Energetica, Università degli Studi di Firenze via di Santa Marta, 3, 50139 Firenze, Italy
c
Faculty of Engineering, Science and the Built Environment, London South Bank University, 103 Borough Road,
London SE1 0AA, UK

article info abstract

Article history: This paper describes and evaluates the development of generalised steady state and
Received 18 July 2012 transient mathematical models for thermostatic expansion valves (TEVs) of the types used
Received in revised form in commercial refrigeration systems. The model is of a generalised nature, because it is not
20 May 2013 necessary to input performance or geometrical data for a particular valve to operate the
Accepted 23 June 2013 model. However, if required, the models can provide an accurate correlation of valve
Available online 2 July 2013 manufacturer’s data. Derivations are provided and validating data is presented. The
mathematical models described in the paper form part of computer software that simu-
Keywords: lates the thermal operation of whole refrigeration systems. The software is titled Vapour
Vapour compression refrigerator Compression Refrigeration System (VCRS) simulator, and by way of example the paper
Simulation presents results from the VCRS simulator which are used to aid a discussion of operating
Modelling faults, such as hunting and under damping and their possible causes, where these can be
Thermostatic expansion valve attributed to the expansion valve.
Software ª 2013 Elsevier Ltd and IIR. All rights reserved.

Modélisation thermostatique des détendeurs

Mots clés : Système frigorifique à compression de vapeur ; Simulation ; Modélisation ; Détendeur thermostatique ; Logiciel

1. Introduction optimised. Recent research lead the authors of this paper to

believe that computer models of refrigerator components,
The majority of commercial refrigeration systems, such as including TEVs, are more useful when they are non-specific; in
those used for food storage and processing, use a thermostatic other words where ever possible the need for empiric perfor-
expansion valve (TEV) to meter refrigerant flow between the mance data or geometric information about a specific valve as
condenser and evaporator in a way that causes the refrigerant input data to a model should be avoided.
to be superheated at the inlet to the compressor. In this way This paper is concerned with modelling the steady state
the compressor is protected from damage by liquid ingestion and transient performance of thermostatic expansion valves,
and the heat transfer performance of the evaporator is (TEVs). Published literature on thermostatic expansion valve

* Corresponding author. Tel.: þ44 (0) 1246238374.

E-mail addresses: [email protected] (I.W. Eames), [email protected] (A. Milazzo), [email protected] (G.G.
Maidment).
0140-7007/$ e see front matter ª 2013 Elsevier Ltd and IIR. All rights reserved.
http://dx.doi.org/10.1016/j.ijrefrig.2013.06.010
190 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 8 ( 2 0 1 4 ) 1 8 9 e1 9 7

Nomenclature d value of ðPb  PÞe when a valve is fully opened

2 s time constant (s)
A area (m )
r density of saturated liquid (kg m3)
Cn constants of proportionality
h enthalpy (J kg1) Subscript
MOP maximum operating pressure (Pa) b bulb
m_ refrigerant mass flow (kg s1) c condenser
P pressure (Pa) d valve diaphragm
Q_ e evaporator heat rate (J) e evaporator
R individual gas constant (J kg1 K1) f saturated liquid
T temperature ( C) g saturated vapour
MC the thermal mass of the sensing-bulb (J K1) n nominal
ks spring constant (N m1) mop maximum operating pressure
U overall heat transfer coefficient (W m2 K1) r reserve
z compressibility factor sat saturated
sc subcooling
Greek
sh superheat
a pressure equivalent of static superheat setting (Pa)
v valve
b valve flow area constant (m2 Pa1)

modelling goes back several decades and includes work differential, (PbePe), and specific to the TEV valve used in the
covering theoretical and experimental research of both steady experiments. They found that their linear model was able to
state and transient operation. Often, studies focussed on capture stable dynamic changes in evaporator conditions but
modelling vapour compression refrigerators as a whole; with not hunting, which was predicted only when their non-linear
the TEV forming only a part of the formulation. An early model was used. The Mithrarartne and Wijevsundera’s model
example of transient refrigeration system modelling is pro- failed to predict the TEV flow characteristic when the suction-
vided by Chi and Didion (1982). Their model incorporated a line vapour temperature, measured close to the sensing bulb,
steady state linear model for a TEV. It is not clear from the was above about 16  C. On checking the specification for the
paper what input data was needed to run the Chi and Didion’s type of TEV used in the experiments the authors of this paper
(1982) TEV model. Experimental measurements and theoret- found that it was a MOP (maximum operating pressure), or
ical results are compared graphically for 6 min periods after gas-charged type, which may explain their unexpected result.
start-up. Up to the 2 min point there is significant difference The operation of a MOP valve is described later in this paper.
between the experimental data and model predictions for Aprea and Mastrullo (2002) reported a comparative study
evaporator inlet temperature. This may have been due to the between the performance of a TEV and electronic expansion
linear TEV assumption used in the formulation of the model. valve (EEV) when replacing a system’s refrigerant from R22 to
Also, measured data were taken at 30 s intervals making R407C. Under steady state operation the results show the two
impossible to know what was happening to the system’s types of valve produced similar results regardless of which
temperatures and pressures during the first 30 s after start-up. refrigerant was used in the system tested. During compressor
However, after about 2 min of simulated time following start- start-up though, when the pressures in the evaporator and
up the predicted and measured values were shown to be close. condenser were equalised, the EEV’s response was found to be
James and James (1987) described their transient model of a faster than that of the TEV, but when the start-up was initi-
TEV. This was a particularly interesting study. However, in ated with the condenser and evaporator pressures out of
order to run the model geometric data of the valve and bulb balance the opposite was found. It is not entirely certain why
was needed, which made it unsuitable for the purposes of a this reversal occurred, however, the authors of this paper
generalised refrigerator system model. believe that it may have been due to the type of EEV used in
Conde and Sutera (1992) described and validated their the tests, which relied entirely on temperature signals to
mathematical model to simulate the steady-state operation of initiate flow changes whereas a TEV uses both pressure and a
thermostatic expansion valves. The disadvantage of the temperature signals to meter the flow.
Conde and Sutera’s (1992) model was that it required an ac- Yu et al. (2006) investigated how TEVs in air-cooled
curate description of the throttling section geometry. While condenser systems can prevent COP values from rising as
the model described in this paper has similarities to that of condenser air-on temperature falls. Their analysis showed
Conde and Sutera’s model it does not require geometric data that in order to maintain a desired refrigerant flow through a
as inputs, which makes it more flexible. TEV it was necessary to maintain a high differential pressure
Mithrarartne and Wijeysundera (2002) carried out a theo- across the valve. This was achieved by shutting-down or
retical and experimental study of a refrigerator system. They slowing variable speed condenser cooling fans as the ambient
compared the results from both linear and a non-linear TEV air temperature fell. In their analysis Yu et al. (2006) used a
models. Both models were steady state and empirically based, linear TEV model similar to that proposed by ASHRAE (1994).
being curve fits of volumetric flow as a function of pressure The degree of scatter in their results may have been due to the
 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 8 ( 2 0 1 4 ) 1 8 9 e1 9 7 191

simplifying assumptions implicit in the linear model used. used in commercial refrigeration systems, Dossat (1978).
The authors recommended the use of an electronic expansion With regard to a refrigeration system, the functions of such
valve (EEV) to over come this problem. Another solution sug- valves are:
gested by the authors of this paper may have been to use a
pumped return from the condenser to the evaporator. (i) Maintain a required pressure difference between the
Ndiaye and Bernier (2009) modelled refrigerant flow condenser and evaporator.
through a TEV bleed port. A bleed port is required for systems (ii) Meter refrigerant flow to the evaporator at (ideally) the
that need to quickly equalise evaporator and condenser same rate as it is drawn off by the compressor.
pressures before compressor start-up. They used four types of (iii) Protect the compressor from liquid ingestion by main-
inlet/outlet steady-state flow condition and showed that their taining an acceptable degree of superheat in the suction
model’s predictions generally compared well with experi- line; normally set to be between 5 K and 10 K.
mental results.
The research described in this paper forms part of a wider The operation of this type of valve and other types of TEV
study aimed at the development of high fidelity models that are described by ASHRAE (1994). The same reference gives the
simulate the transient and steady-state operation of whole capacity of TEV to be,
refrigeration systems of the type used for the storage qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 
and processing of food. At the time of writing, the models Q_ e ¼ Co rc;f ðPc  Pe Þ he;g  hc;f ; (1)
are integrated within a software package titled Vapour
Compression Refrigerator System (VCRS) simulator. This has where Co is a constant dependant on valve geometry. Valves
been described by Eames et al. (2010, 2012) and was developed that use the same refrigerant in their sensing bulb as used in
as part of a larger study into energy use by refrigeration sys- the refrigerator are termed ‘straight-charged’. In practice it is
tems in the food chain carried out for the UK Government’s common to charge a sensing bulb with a refrigerant or a blend
Department for the Environment, Food and Rural Affairs of refrigerants that has a higher saturation pressure than the
as described by James et al. (2009). This simulator provides a refrigerant in the refrigerator evaporator at a lower saturation
fully integrated model that simulates food cooling processes, temperature. This prevents excessive compressor suction
cold store or room and refrigeration system, together, in a vapour superheat at low evaporator temperatures. Valves
single package. This allows practitioners to investigate the which use a different refrigerant to that in the refrigerator
implications of refrigeration system design choices in terms of system are know as ‘cross-charged’. To model cross-charged
energy usage and carbon generation. At the time of writing the valves at steady-state it is necessary to know the saturation
VCRS package can be downloaded free of charge from the pressureetemperature relationship of the refrigerant or blend
project website at; http://www.grimsby.ac.uk/industry/defra- of refrigerants contained by the sensing bulb. For commercial
refrigerationmodels.php or from the corresponding author. reasons this information is not normally made available by
In order to enable a wide range of refrigerator systems to be valve manufacturers and therefore, for the purposes of
easily modelled it was necessary that the TEV model was not refrigeration system modelling it is usual to assume that a TEV
complicated by the need for geometric descriptions of the is straight charged. However, as shown later in this paper it
valve or empiric performance data. This paper shows that the was interesting to observe that the model described here,
aim was achieved and in addition the steady-state model which assumes a straight-charged liquid valve, appears to
provides described here provides a useful correlating equation
for manufacturer’s data, if that is required. Together with a
transient model, a practical way is provided to simulate and
investigate operational problems sometimes encountered in
refrigeration system. These included such problems resulting
from over-sizing or under-sizing a TEV, or by incorrect
installation.
Section 2 provides descriptions of various types of TEV
covered by the research described in this paper. Sections 3 and
4 describe the derivation of a TEV steady-state model. Part-
load performance of a TEV is discussed in Section 5 and a
simplified transient model is derived in Section 6. Some vali-
dation results are presented in Section 7 and in Section 8 re-
sults for the VCRS software, which incorporates the models
described in this paper, are used to aid the discussion of TEV
operating faults.

2. Operation of a thermostatic expansion

valve

Fig. 1 shows a schematic view of an internally equalised,

liquid charged, thermostatic expansion valve (TEV) of a type Fig. 1 e Schematic view of TEV model.
192 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 8 ( 2 0 1 4 ) 1 8 9 e1 9 7

satisfactorily correlate the performance of a proprietary cross- valve. This occurs both when the evaporator saturation temper-
charged valve within its normal operating temperature range. ature increases or decreases from its design-point condition and
Cross-charged valves, like ‘gas-charging’ valves discussed when the refrigerant flow through the TEV does not equal that
later, provide the added benefit of limiting the maximum oper- through the compressor. This flow mismatch causes the evapo-
ating pressure (MOP) in the evaporator. This can help to protect rator to become either flooded or starved of refrigerant.
compressor drive motors from being overloaded during start-up From Equation (1) the refrigerant flow through a TEV is
by preventing excessively dense vapour being drawn from the given by,
evaporator when the air entering the evaporator might be qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
significantly warmer than the design condition, (Dossat, 1978). _ ¼ C1
m rc;f ðPc  Pe Þ (2)
Under normal operation the quantity of refrigerant con-
tained in the sensing bulb of a gas-charged or MOP (maximum Referring to Fig. 1, the constants of proportionality, Co and
operating pressure) valve is significantly smaller than that C1, in Equations (1) and (2), are functions of the flow area, Av.
contained by a liquid-charged valve, whether it is cross- To determine the nature of this function a force balance is
charged or straight-charged. As the sensing bulb of a MOP applied to the spring-bellows system:
valve is heated its rising temperature causes the saturation
ðPb  Pe ÞAd ¼ ks ðxo  xÞ
pressure of the liquid within the valve to increase. Thus,
within the normal control range of the valve the bulb- Where,
pressure, (Pb) and bulb-temperature are both saturated and
related logarithmically and the valve behaves like a conven- xo ¼ length of spring when fully extended
tional cross- or straight-charged valve. However, above a x ¼ length of compressed spring
predetermined saturation temperature the valve is designed
so that all the liquid within the sensing bulb will have evap- In the above force balance the effect of pressure differen-
orated. With only vapour in the sensing bulb the controlling tial across the valve port is assumed to be negligible compared
pressure becomes almost directly proportional to the absolute with that of the bellows.
temperature of the bulb ðPb fTb Þ as the containment volume of
the valve is approximately constant. The effect this produces
is to close the valve when the evaporator temperature is Letting Dx ¼ ðxo  xÞ; when the valve is closed the force
greater than that determined by MOP setting of the TEV. With reaction of the valve seat ¼ ks Dxo
the valve closed the saturation temperature in the evaporator
falls rapidly when the compressor is started and this results in Where the subscript ‘o’ is used here to denote values at the
a significant reduction in start-up torque. point of valve opening.
For gas-charged or MOP valves it important that the bel- For the valve to be open,
lows is always warmer than the sensing-bulb in order to ðPb  Pe ÞAd > ks Dxo
prevent any liquid contained in the bulb from migrating to the
bellows causing the valve to malfunction. Therefore, for the valve to just begin to open,
Another cause of malfunction can result from the use of
highly dynamic evaporators such as those that contain rela- ðPb  Pe ÞAd ¼ ks Dxo
tively small amounts of refrigerant, such as plate-type evap- The bulb pressure at which the valve just begins to open
orators, or systems that can have short ON-OFF cycling times. can be adjusted by tensioning the control spring. This has the
Such evaporators might be found in air conditioning systems. effect of setting the minimum degree of superheat in the
The response of a refrigeration system to changes in air-on compressor suction line where the sensing bulb is located, as
temperature, or chilled water-on temperature, is normally shown in Fig. 1. The superheat temperature required to just
dampened to a greater or lesser extent by the thermal mass cause the valve to begin to open is known as the ‘static
and liquid volume of the evaporator. However, if this ‘natural’ superheat setting’ (SSS).
damping is insufficient, because of a small liquid volume for Therefore, when ðPb  Pe ÞAd > ðPb  Pe Þo Ad then, ks Dx > ks
example, then ‘artificial’ damping may be introduced through Dxo
the addition of ‘ballast’ to the sensing bulb. This increases the Therefore,
time constant (s) of the valve by raising its thermal mass.  A
ðDx  Dxo Þ ¼ ðPb  Pe Þ  ðPb  Pe Þo (3)
ks
Fig. 2 shows the geometry of the valve stem in relation to its
3. A theoretical mass flow model seat when the valve is open. Referring to Fig. 2, the flow area,
Av, is given approximately by,
The following describes theoretical models for liquid or gas,
straight or cross charged thermostatic expansion valves that Av ¼ p:d:a
are either internally or externally equalized.
Equation (1) adequately models the capacity of a TEV when the The distance ‘a’ is proportional to the valve stem position,
degree of evaporator superheat is fixed, which may be assumed to ðDx  Dxo Þ, and therefore,
be for steady-state performance at a given system design condi-
Av ¼ C2 ðDx  Dxo Þ
tion. However, under part-load or transient operation the degree
of superheat can change and doing so will alter the capacity of the Where C2 ¼ constant of proportionality
 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 8 ( 2 0 1 4 ) 1 8 9 e1 9 7 193

The value of a is determined by the minimum degree

of superheat set by the valve, and d is the pressure equiva-
lent to the degree of superheat needed to fully open the
valve. When a TEV is full-open the mass flow becomes a
function of (rf (Pc e Pe))1/2 as given in the ASHRAE Handbook
(1994).

4. Determination of the valve constants

4.1. Values of a and b for a model valve

For a theoretical model of a valve, the value of a the pressure

equivalent to the static superheat setting) in Equation (4a) is
set to equal the pressure difference, (Pb  Pe), required for
Fig. 2 e Flow area through valve, Av.
zero mass flow at the required minimum degree of super-
heat temperature. At that condition the valve is just closed
Substituting this result into Equation (3) and collecting or just starting to open. In terms of temperature, this pres-
together all the constant terms together gives, sure difference will equate to the static superheat setting.
  The value of the valve area flow constant, b can be deter-
:
Av ¼ b ðPb  Pe Þ  ðPb  Pe Þo mined using values of Q e , ðPb  PÞe , rf and ðPc  Pe Þ, taken at a
In this paper b is referred to as the valve flow area constant. refrigerator’s known design-point conditions. If a sensing
Increasing b causes the capacity of a valve to increase and vice bulb is charged with the same refrigerant as the evaporator
versa. In other words, if a system has a TEV fitted that has a b- it controls then the value of Pb in Equation (4) equals the
value larger than what should normally be fitted then the saturation pressure of the refrigerant at the desired super-
valve would tend to always over feed evaporator with refrig- heat temperature at the location of the sensing bulb in the
erant. This would cause the evaporator temperature to compressor suction line.
constantly vary; rising and falling in a sinusoidal way which is
referred to as hunting. 4.2. Values a and b for a real valve
Substituting the above result for the constant, C1, in
Equation (2) and letting a ¼ ðPb  Pe Þo gives the mass flow If a model of a proprietary valve is required then values for a
through the valve: and b can be determined using information provided by
manufacturer’s data sheets. Fig. 3 shows comparative data for
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
_ ¼ b½ðPb  Pe Þ  a
m rc;f ðPc  Pe Þ for a  ðPb  Pe Þ  d; (4a) a Danfoss TS2; an internally equalised valve with a cross-
charged bulb using Danfoss capsule 03 and working with
where a is the pressure equivalent of static superheat setting R404a. The values of a and b were calculated using a linear
(SSS) and d is the value of ðPb  Pe Þ when the valve is fully _
regression curve of pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
m
against ðPb  Pe Þ. The resulting
rf ðPc Pe Þ
opened.
characteristic equation in this example is,

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 
_ ¼ 3:04939  107 ½ðPb  Pe Þ þ 20:12477 rc;f ðPc  Pe Þ kg s1 for a  ðPb  Pe Þid
m (5)

In this case, the results in Fig. 3 show that Equation (4a)

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi provided an accurate correlation of the manufacturer’s data.
_ ¼ b½d  a rc;f ðPc  Pe Þ for ðPb  Pe Þid
m (4b)
It is thought reasonable to assume that this would be true in
any other case, although the correlation should be checked for
_ ¼ 0 for ðPb  Pe Þ  a
m (4c) each. In Fig. 3 the error in mass flow between the correlation
and model is less than þ/ 2%.
If a gas-charged or MOP valve is to be modelled, once the
liquid in the sensing bulb has evaporated and if further vol-
ume change in the bellows is neglected, then,
5. Part-load performance characteristics of a
  TEV
Pb ¼ Pb rb ; Tb for Pb  Pmop ; (4d)

where, rb ¼ saturated vapour density at a pressure equal to The capacity of a thermostatic expansion valve is defined by
Pmop . ASERCOM, (2005) as,
194 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 8 ( 2 0 1 4 ) 1 8 9 e1 9 7

180 The maximum capacity of a valve may be greater than the

160 nominal capacity at the rating conditions, thus providing an
Mass flow (kg hour-1)

140 amount of reserve capacity. ASERCOM (2005) does not specify

a limit on reserve capacity. However, the characteristic in
120
Fig. 4 suggests this might be as much as 50%.
100
The significance of the characteristic shown in Fig. 4 is the
80
linear relationship between capacity and opening superheat
60 within the control region and the reserve capacity above the
Te = -30 degC Te = -20 degC
40 Te = -10 degC Te = 0 degC
nominal capacity. Any model of a TEV must, therefore, reflect
Te = 5 degC Te = -30 degC (model)
20 Te = -20 degC (model) Te = -10 degC (model) both these characteristics. Fig. 5 shows the results for the
Te = 0 degC (model) Te = 5 degC (model)
0 theoretical steady-state model, defined by Equations (4a)e(4c),
400 600 800 1000 1200 for an internally equalised, straight-charged, TEV operating
Presssure drop across valve (kPa) on R134a with a nominal capacity of 10 kW at the rating
conditions. The results in Fig. 5 show that the correlation
Fig. 3 e Comparison between manufacturer’s data for a satisfies this requirement.
Danfoss TS2 (internally equalised, cross-charged bulb),
Capsule/Orifice 03, working with R404a, and the model of a
TEV defined by Equation (5). 6. Transient responses of a TEV to changes
in superheat

The flow predicted by Equations (4a)e(4d) represent steady-

:   : state values for given condenser and evaporator conditions.
Q e ¼ m he;g  he;i ; (6)
In practice the valve response will not be instantaneous and
where he;g is the enthalpy of saturated vapour leaving an there will be a delay in response between a change in super-
evaporator and he;i is the enthalpy of wet-vapour entering. heat temperature and a corresponding change in refrigerant
Fig. 4 shows how the capacity of a TEV changes with flow through the valve whenever the saturation conditions in
superheat temperature. In Fig. 4, the static-superheat the evaporator and condenser alter. It is reasonable to assume
setting is normally set at the rating condition by the valve that for most cases (excepting those requiring the use of a
manufacturer, although this can be adjusted by varying the ballast type sensing bulb discussed earlier) the response time
tension in the control spring, shown in Fig. 1. Based on the of a valve will be much shorter than that of evaporator and
ASERCOM (2005) proposals, static-superheat would be set condenser to changes in operating conditions. This transient
between 3 and 4 K. The ‘opening-superheat’ is the superheat behaviour of the evaporator-valve system is assumed here to
needed to open the valve between the just-closed position be due only to the thermal resistance between the bulb and
and that needed to ensure the capacity is equal to the suction pipe and thermal capacity of the bulb.
valve’s nominal capacity at the rating conditions. ASERCOM
dTb
(2005) propose that this degree of superheat should not be MC ¼ UAb ðTe;sh  Tb Þ (7)
dq
greater than 5 K. Therefore, for the valve to be fully open the
degree of superheat should not exceed 8e10 K. ASERCOM Integrating Equation (7) between time q and q þ Dq, and
(2005) standard rating conditions are, Te,sat ¼ 4  C, assuming ðTe;sh  Tb Þq is known at q ¼ q gives,
Tc,sat ¼ 38  C with 1 K of sub-cooling. At these conditions the  
Dq
rated differential pressure across a TEV should equal the ðTb ÞqþDq ¼ ðTe;sh ÞqþDq  ðTe;sh  Tb Þq exp  ; (8)
s
difference between the saturation pressures in the
condenser and evaporator.
18 static superheat
linear control region maximum capacity = 15kW
region
16
14
Capacity (kW)

design-point
Data:
12 Refrigerant = R134a
superheat = 8 C

valve fully
10 Valve type: internal
open here
equalization, straight
8 charged bulb
Tc = 38 C superheat at
6 Subcooling = 1 K which valve
fully open =
4 Te = 4 C
11.25 C
Qn = 10 kW TEV model results
2
0
0 5 10 15
Degrees of superheat (K)

Fig. 5 e Results of the theoretical model for a TEV operating

Fig. 4 e Capacity-superheat operating characteristic of a on R134a with a nominal capacity of 10 kW at the
typical TEV (ASERCOM, 2005). ASERCOM (2005) rating conditions.
 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 8 ( 2 0 1 4 ) 1 8 9 e1 9 7 195

where s ¼ MC
UA
is the response time for the valve. The result 12
shown in Equation (8) assumes: compressor start-up here

7 Static superheat setting

τ = 40 s
(SSS) = 2K
1) Valve stem stiction, friction and inertia are negligible. Fully open superheat
2 setting = 10K

Te,sat (oC)
2) The temperature of the liquid within the bulb and the body τ = 0s τ = 20s τ = 80s
of the bulb are equal. -3
3) Momentum changes of the liquid in the liquid-line leading
to the TEV are negligible. -8

Therefore, at any time, ðq þ DqÞ, the bulb temperature may -13

be calculated using Equation (8). The bulb pressure can be
-18
determined using a saturation pressure-temperature corre-
34 36 38 40 42 44
lation and thus the refrigerant mass flow through the TEV can
Time (mins)
be determined from Equations (4a)e(4d), depending on the
type of valve and the degree of superheat. Fig. 7 e Showing the VCRS simulator predictions for the
evaporator temperature of a large chill-store during start-
up showing the effects of changing the time constant of a
correctly sized straight liquid charged, internally
7. Model validation
equalised, thermostatic expansion valve.

The theoretical model described by Equations (4a)e(4d) and

Equation (8) was incorporated VCRS simulator, (Eames et al.,
2010). Fig. 6 shows a comparison between experimental and qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 
theoretical results for the VCRS software validation case study _ ¼ 1:775605  106 ½ðPb  Pe Þ þ 34:9 rc;f ðPc  Pe Þ kg s1
m
on a 6.2 m3 air blast refrigerated food cooler described by
This equation is based on the following design point
Eames et al. (2012). The time-constant value for the TEV model
conditions:
in this case was 20 s. The refrigerator supplied 13 kW of
cooling at its design condition. The system included a 7 kW
Refrigerant ¼ R404A
defrost heater which was timed to come on for 20 min every
Evaporator heat rate ¼ 13 kW
4 h. The ‘saw-tooth’ characteristic of the graph lines in Fig. 6 is
Evaporator temperature ¼ 5  Csat
due to the 1.5 kW evaporator fans remaining on when the
Evaporator superheat ¼ 5 K
compressor was shut-down by a thermostat in the cooler
Minimum superheat ¼ 2 K
chamber. The comparative results shown in Fig. 6 serve to
Condenser temperature ¼ 40  C
indicate the potential usefulness of the VCRS model.
Condenser subcooling ¼ 5 K
As the VCRS software was designed to model transient
responses of refrigeration systems to changes in inputs, such
as ambient weather, refrigerator door opening, loading the
refrigerator with food stuff and so on, the system was 8. Results and discussion
particularly useful for testing the TEV model described here.
The characteristic equation for the steady state operation The TEV model described in this paper was used in the VCRS
of the valve was, software to study the performance effects of under-damping,

14
τ = 40s Static superheat setting = 2K
Degrees of superheat (K)

Fully open superheat setting = 10K

12 Capacity Factor = 1

10 τ = 20s

4
τ = 0s

0
35 37 39 41 43 45
Time (mins)
Fig. 6 e Comparative experimental and model data from
the VCRS software: Variation in evaporator saturation Fig. 8 e Showing the effect on the variation in superheat
temperature for a 13 kW (rated), 6.2 m3 refrigerated air degrees (K) with time caused by changes in the time
blast cooler. constant of a TEV.
196 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 8 ( 2 0 1 4 ) 1 8 9 e1 9 7

0.120
Static superheat setting = 2K
0.100 Fully open superheat setting = 10K
TEV flow τ = 20s
Mass flow (kg/s)

Capacity Factor = 1
0.080

0.060

0.040 compressor
flow
0.020

0.000
35 37 39 41 43 45
Time (mins)

Fig. 9 e Showing the variations in valve and compressor Fig. 11 e VCRS predication for the variation in evaporator
flows (kg sL1) with time for a design-point capacity TEV temperature under ‘normal’ and hunting conditions.
with a time constant of 20 s.

evaporator of refrigerant. A system with a correctly fitted TEV

caused by incorrect sizing of the TEV and inappropriate loca- would quickly come into balance so that the degree of su-
tion and fitment of the sensing bulb, which are thought to perheat is held constant and flow through the TEV equals that
increase the time constant of the valve. through the compressor.
Fig. 7 shows that as the time constant of a valve increases Mithraratne and Wijeysundera (2002), reporting the work
the control system becomes progressively under-damped. of Broersen (1982), suggest a time-constant of 20 s for a TEV.
The results show that the initial amplitude and wave length Although this value may be acceptable for modelling a whole
of the temperature oscillations both increase with the time- refrigerator, in practice there may be occasions when less
constant. However, these oscillations gradually disappeared responsive valves are better, such as for evaporators using in
as the system came into balance in the manner of an under- air conditioning systems (Danfos, 2005).
damped control system. In this case the reason for the tem-
perature oscillations is thought to be repeated increases and
decreases in liquid level in the evaporator, which is reflected 8.1. TEV hunting
by the variation in super-heat temperature as shown in Fig. 8.
When the degree of super-heat is large the TEV over-feeds the TEV hunting can be caused by number of operational, design or
evaporator causing refrigerant to enter faster than the manufacturing faults. The most common faults are thought to
compressor can remove it, which causes the saturation be incorrect positioning and over-sizing of the valve. Fig. 10
pressure in the evaporator to rise as shown in Fig. 9. If the compares the predicted results for a valve which has twice the
liquid content increases too rapidly the TEV control to over- required flow capacity (Capacity Factor ¼ 2) with those for a
shoots the balance condition, reducing the degree of super- correctly sized valve (Capacity Factor ¼ 1) and an undersized
heat and closing the TEV completely and starving the valve (Capacity Factor ¼ 0.5). These results show that over sizing
a TEV causes the flow control to be under-damped, which causes
the evaporator to be alternately over and under-fed with liquid
15 refrigerant. As the liquid level rises the degree of superheat falls,
Static superheat setting = 2K
causing the valve to move towards its closed position and
Fully open superheat setting = 10K
10 slowing the flow into the evaporator. If the compressor draws
capacity factor = 0.5
refrigerant from the evaporator at a greater rate that it is sup-
5
Te,sat (oC)

plied by the TEV then the liquid level will fall raising the degree of
0 capacity factor = 1 superheat in the suction-line and, in turn, increasing the flow
capacity factor = 2
through the TEV. This causes the liquid level in the evaporator to
-5 rise causing the TEV to close again and so on. Similar effects
occur when a TEV is incorrectly positioned so that its response
-10 time is increase. Fig. 11 shows results produced by the VCRS for a
correctly sized TEV and an oversized, unresponsive TEV. It is
-15
interesting to note that Mithraratne and Wijeysundera (2002)
34 36 38 40 42 44
have found that the amplitude and frequency of the hunting
Time (mins)
oscillations could be reduced by increasing the time constant of
Fig. 10 e Showing the variation in evaporator temperature the TEV bulb. A possible reason for this apparently contradictory
with time for a range of valve flow capacities. The results result may lie in the type of evaporator used by Mithraratne and
show that if a valve is oversized the evaporator Wijeysundera in their experiments. As stated at the end of
temperature will ‘hunt’. Section 2, if an evaporator is highly dynamic it can be necessary
 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 8 ( 2 0 1 4 ) 1 8 9 e1 9 7 197

to slow the response of the controlling TEV by adding thermal ASERCOM, 2005. Statement on Capacity Rating of Thermostatic
ballast at the sensing bulb in order to prevent hunting. Expansion Valves.
By varying the response time and capacity for the TEV Broersen, P.M.T., 1982. Control with a thermostatic expansion
valve. Int. J. Refrigeration 5 (4), 216e221.
model and other design-point settings in the VCRS, other ef-
Chi, J., Didion, D., 1982. A simulation model of the transient
fects can be simulated. These include an undersized TEV, a performance of a heat pump. Int. J. Refrigeration 5, 176e184.
punctured sensing bulb, and a partially blocked condenser. Conde, M.R., Sutera, P., 1992. A mathematical simulation model
for thermostatic expansion valves. Heat Recovery Syst. CHP 12
(3), 271e282.
9. Conclusions Danfos, 2005. Thermodynamic Expansion Valves.
DKRCC.PF.A00.A1.02/520H0337.
Dossat, R.J., 1978. Principles of Refrigeration, second ed. John
The development of both steady-state and transient models for
Wiley and Sons, ISBN 0-471-03550-5.
thermostatic expansion valves, of the types commonly found in Eames, I.W., Brown, T., Maidment, G.G., Evans, J.A., 2012.
commercial refrigeration systems, have been derived. Both Description and validation of a computer based refrigeration
models can be used without the need for manufacturer’s data, system simulator. COMPAG. ISSN: 0168-1699 85, 53e63.
which makes them useful when writing generalised transient or j.compag.2012.03.010.
steady state models of vapour compression refrigerator sys- Eames, I.W., Brown, T., Maidment, G.G., Missenden, J., Evans, J.A.,
Swain, M.J., James, S.J., 2010. An interactive refrigeration
tems. Notwithstanding the general nature of the models
system simulator software. Paper No 246. In: 1st IIR Int. Conf.
described by Equations (4a)e(4c), these were tested against
on Sustainability and the Cold Chain, Selwyn College,
manufactures data and the results were found to provide a close Cambridge UK, 29-31 March 2010, ISBN 978 2 913149 75 5. ISSN
correlation, (Fig. 3). The equations were also tested against the 0151.1637.
ASERCOM (2005) TEV characteristic and again close similarity James, K.A., James, R.W., 1987. Transient analysis of the
was found. The fidelity of the models within a complex refrig- thermostatic expansion valves for refrigeration system
eration system, under realistic loadings, was tested using the evaporators using mathematical models. Trans. Inst. Meas.
Control 9 (4), 198e205.
VCRS software, Eames et al. (2012). Comparison between the
James, S.J., et al., 2009. Improving the energy efficiency of food
prediction and measurements were excellent. The VCRS soft- refrigeration operations. Proc. Inst. Refrigeration, 5th February
ware results enabled the influence of TEV response time, over 2009.
capacity and hunting to be investigated. Mithraratne, P., Wijeysundera, N.E., 2002. An experimental and
numerical study of hunting in thermostatic-expansion-valve-
controlled evaporators. Int. J. Refrigeration 25, 992e998.
references Ndiaye, D., Bernier, M., 2009. Modelling the bleed port of a
thermostatic expansion valve. Int. J. Refrigeration 32,
826e836.
Yu, F.W., Chan, K.T., Chu, H.Y., 2006. Constraints of using
Aprea, C., Mastrullo, R., 2002. Experimental evaluation of
thermostatic expansion valves to operate air-cooled chillers
electronic and thermostatic expansion valves performances
at lower condensing temperatures. Appl. Therm. Eng. 26,
using R22 and R407C. Appl. Therm. Eng. 22, 205e218.
2470e2478.
ASHRAE, 1994. Refrigeration Handbook (SI). (Chapter 44).

View publication stats