J. Phys. D: Appl. Phys. 29 (1996) 1469–1472.

Printed in the UK

Low-pressure vortex tubes

B Ahlborn†, J Camire† and J U Keller‡
† Department of Physics, UBC Vancouver, V6T 1Z1, Canada
‡ Institut für Thermodynamik, Universität Siegen, Germany

Received 25 October 1995, in final form 29 February 1996

Abstract. The temperature separation effect, which is well known from standard
vortex tubes that are driven by compressed air, has been observed in a device
where air is sucked into the tube by applying vacuum at the hot and the cold exit
ports. Rather than depending on the absolute inlet pressure, the temperature
separation was found to be a linear function of the normalized pressure drop
X = (p0 − pc )/p0 between the inlet and the cold end of the vortex tube.

1. Introduction 2. Background

A vortex tube is a very simple mechanical device without In a vortex tube a stream of air is injected tangentially at
any moving components, which separates a stream of gas high speed through one or several nozzles. The air flow
of inlet velocity u0 into a hot and a cold component. separates into a cold component issued from a small hole
From a gas flow with an inlet temperature T0 ≈ 290 K on the axis close to the inlet plane and a hot component that
available at an upstream inlet pressure of p0 = 3 bar one leaves through a large orifice at the other end. All reported
can typically obtain a hot stream of Th = 320 K and a vortex tubes have been given by a gauge pressure in the
cold stream of Tc = 265 K. For a given vortex tube the range of 2–10 atm.
temperature splitting changes with the inlet pressure and This ‘temperature separation’ has baffled investigators
the ratio of the cold and the hot mass flows jc and jh . The for over 60 years since heat seems to be travelling in the
pressures in the hot and cold flows arrange themselves as wrong direction, namely from the cold to the hot part of
p0 > ph > pc , where the cold gas pressure is generally the the flow. The working fluid starts in a plenum chamber at
ambient pressure pc = 1 atm. practically zero velocity. It gains kinetic energy and loses
The vortex tube was discovered by Ranque [1] temperature as it passes through the inlet nozzle into the
and first described in detail by Hilsch [2]. Vortex vortex tube. It splits into a cold and a hot component which
tubes are now commercially used [3] for low-temperature each circle around in the tube and eventually emerge either
applications, e.g. to cool parts of machines, to set solders,
on the hot side or on the cold side of the device. Typically
to cool electronic control cabinets, to chill environmental
each flow component then passes through a slow down
chambers, to cool food, to test temperature sensors, and
section where it must lose all its kinetic energy. Assuming
they are also applied to dehumidify gas samples [4].
adiabatic and frictionless flow of an ideal gas, each one
Recently it has been proposed that vortex tubes could
of these flow components should emerge with the same
be used as components in refrigeration systems replacing
temperature, namely the entrance temperature T0 .
the conventional expansion nozzle in order to increase the
efficiency [5]. The coolants in refrigeration systems pass Of course no process is truly adiabatic. For instance as
through a thermal cycle in which the pressure may well the gas speeds up in the entrance nozzle and its temperature
drop below atmospheric. However all previously reported sinks below ambient some heat will flow into the gas.
vortex tubes were operated at high entrance pressures, We previously modelled this step as an isothermal process
which ranged from about 2 to 10 atm. In these experiments [6] and derived limits for the temperature separation. To
the pressure never dropped below one atmosphere. It is test if an adiabatic or an isothermal model represents the
therefore not experimentally proven that vortex tubes will inlet process more correctly we have recently measured the
indeed yield the temperature splitting effect in the intended pressure and the temperature directly in the nozzle throat,
low-pressure environment. Based on our recent model and we observed that the flow is more closely adiabatic than
calculations [6] we conclude that the effect depends on the isothermal. These results will be discussed in a forthcoming
normalized pressure ratio X = (p0 − pc )/pc rather than on publication.
the absolute values of the entrance pressure p0 and exhaust However, many reported measurements and our own
pressure pc . It should hence be possible to operate a vortex experiments indicate that there is a definite flow of energy
tube as a suction device with atmospheric intake pressure, from the cold to the hot flow component. This anomalous
provided pc and ph are kept below one atmosphere. This heat flow is indeed hard to understand. Several different
has been verified by experiments with a suction vortex tube, qualitative theories have been offered to explain this strange
arranged as shown in figure 1. temperature splitting effect: internal friction with various

0022-3727/96/061469+04$19.50
c 1996 IOP Publishing Ltd 1469
B Ahlborn et al

Figure 1. Vortex tube driven by vacuum pump. The insets show the inlet plenum chamber and the hot end.

assumptions about the radial and axial velocity distribution The pressures were measured with the help of strain
[7]; turbulent heat transfer in a stratified flow [8]; Goertler gauge sensors whose electrical signals were recorded by
vortices [9]; compressibility of the working fluid [10]; computer. The cold gas pressure was measured a few
turbulent transfer of thermal energy in an incompressible centimetres downstream of the cold exit port. The hot gas
flow [11]; effects associated with the particular working pressure tap was located between a detwister and the hot
fluid [12]; or acoustic streaming processes [13] leading exhaust valve. We also measured the pressure ppl in the
to increased flow velocities which therefore enhance the plenum chamber that supplied the four entrance nozzles,
kinetic energy available for conversion into heat. and recorded the pressure p0 inside the vortex tube very
For the use of vortex tubes in refrigeration and close to the entrance nozzles. The pressure at the intake of
dehumidification applications the question of how the the vortex tube was always the ambient barometric pressure
vortex tube functions is less important than the question pa ≈ 101.3 kPa. However the pressure in the plenum
of if it works at all below atmospheric pressures. This chamber is somewhat lower. In order to vary the mass
question has been positively answered by the experiments flow fraction Y the pressure ph must be adjusted relative
reported here. to the cold exit pressure pc . This was accomplished with
the help of the valve shown in figure 1. The vortex tube
may hence be considered as a branched flow system where
3. Experiments the mass flow ratio Y = jc /j0 can be adjusted by a single
valve in the hot branch. Of course a change of the valve
For the experiments we used a modular vortex tube. It setting also alters the total resistance of the two parallel
consisted of a Pyrex tube of 25.4 mm inner diameter branches, so that the pressure in the plenum chamber ppl
and 60 cm length, which was held between brass ends and inside the tube p0 must also vary when Y is altered.
containing the inlet–cold gas exit and the hot gas nozzle (see For the desired variation of the mass flow ratio in the range
figure 1). Air was drawn through the device by a vacuum 0.05 ≤ Y ≤ 0.95 the plenum chamber pressure adjusted
pump with plenum chamber, connected directly to the cold itself in the range 91 to 94 kPa, while the hot and the
end and affixed through a throttle valve to the hot end. This cold end pressures varied from 33 to 67 and 33 to 58 kPa
valve allowed one to change the hot gas pressure ph relative respectively.
to the cold gas pressure, thereby permitting adjustment of The temperatures at the locations shown in figure 1
the cold gas fraction Y = jc /(jh + jc ) to any desired value were measured with iron–constantan thermocouples, type
between 0.05 and 0.95. For a given vortex tube the cold and J gauge 24, inserted into the measuring volumes through
the hot exit temperatures are distinct functions of p0 and Y . Wilson vacuum seals. An ice bath was used at the
By applying suction at the exit holes of the vortex tube air of reference junction of each thermocouple. Data were
atmospheric pressure was drawn through an entrance hole recorded with a signal processing unit on an XT computer
into the plenum chamber which supplied the four equally via a DT-2805 12-bit A/D board. The board allowed
spaced nozzles of the vortex tube, three of which are shown programmable gains of 1, 10, 100, 500 and an input range
in the inset of figure 1. of ±10 V. A computer program was written to operate

1470
 Low-pressure vortex tubes

Figure 2. Exit temperatures for vortex tube with

atmospheric inlet pressure.

the board. The program allowed real time recording of

the measurements on the screen, averaging of the data,
selecting of gain and it included user control to save the
desired measurements.
Temperature measurements in moving fluids are not as
simple as pressure measurements [14]. The actual gas
temperature T would be found by a device that moves
at the velocity u with the flow, so that it is in thermal
equilibrium with the fluid locally. When a probe is inserted
into a moving fluid, some of the flow stagnates and the
pressure increases on the upstream side of the probe. If the Figure 3. Total temperature separation for vortex tube
entire flow came to rest, a probe would measure the ‘total’ operated with compressed air.
temperature Tt = T + u2 /2cp , where cp is the specific heat
at constant pressure. The component u2 /2cp is sometimes
given in figure 3. For these sets the plenum chamber
referred to as the ’dynamic temperature’ Td . A probe
pressure ppl was held constant. In these measurements the
inserted into a flow only stops the flow partially. To account
vortex tube was operated with compressed air from a tank
for this effect a dynamical correction factor K = 0.75 for
that was replenished by a small compressor. Unfortunately
cylindrical temperature sensors has been introduced [14]
this supply had a limited capacity. For the highest pressures
which permits one to obtain the actual gas temperature T
of this series (40 psig) there was not enough time to reach
from the temperature Tp measured by the probe using
thermal equilibrium at the hot end of the tube, owing to the
T = Tp − KTd = Tp − Ku2 /2cp . large thermal mass of the hot end valve. For that reason
no steady state temperatures Th could be obtained.
For accurate temperature measurements it is hence It was predicted from the analysis of reference [6] that
necessary to know the local flow velocity u at the locations the temperature splitting should depend on the normalized
of the thermocouples. Approximate values of uc and uh pressure drop X = (p0 −pc )/pc , rather than on the absolute
were derived from the known mass flows jc and jh [15]. pressure p0 . For each experimental run the inlet pressure
Typically uc was found to be ≈ 50 m s−1 , while uh was p0 (measured inside the vortex tube very close to the
about an order of magnitude smaller. entrance nozzle) and pc were recorded, so that X was
The vortex tube with atmospheric inlet pressure was known. A good figure of merit for the temperature splitting
operated for a variety of cold gas fractions Y = jc /j0 . The is the maximum (total) temperature difference Tth − Ttc =
temperature measurements are shown in figure 2. 1Thc,max between the hot and the cold component. X and
Clearly the temperature splitting effect is obtained. The 1Thc,max are given in table 1 for several entrance pressure
cooling is well developed, but it appears as if the heating is conditions. In order to show how well the vacuum values
not quite as pronounced. We have lately become aware that fit into the general scheme 1Thc,max is plotted in figure 4
varying the moisture content of the ambient air will change as a function of X.
the thermal capacity of this working fluid significantly and The value of X = 0.19 for ‘vacuum’ operation falls in
hence contribute to variations of the temperature change the middle of the range of the standard ‘compressed air’
1T . This effect could easily amount to several degrees and values. The temperature splitting for vacuum operation,
it should be more pronounced in the hot flow component Thc,max = 23 ◦ C, lies well on the line established for the
than in the cold stream. standard experiments.
To put the temperature measurements into perspective In conclusion, a vortex tube behaves identically in the
several high-pressure curves for the same vortex tube are above and below atmospheric pressure regimes. Essential

1471
B Ahlborn et al

Table 1.

Vacuum Compressed air

Entrance/plenum pa = 2.47 × 105 ppl = 2.47 × 105 3.47 × 105 4.47 × 105
pressure (Pa) (= 1 atm) (= 10 psig) (= 20 psig) (= 30 psig)
1Thc ,max (◦ C) 23 15 33 45
X = (p0 − pc )/p0 0.19 0.16 0.24 0.29

[3] Swirl Tubes Norgren 5400 South Delaware, Littletown,

Colorado; Vortex Tubes Exair products, 1250 Century
Circle North, Cincinnati, OH, 45246, USA; Vortex Tubes
Vortec Corp, Cincinnati OH, USA
[4] von Linde R 1950 Einrichtung zum abkühlen eines
verdichteten gases German Patent 926729 (24 October
1950)
[5] Keller J U 1993 J. Kl Klima Luft Kälte Heizung 21 300–4
[6] Ahlborn B, Keller J U, Staudt R, Treitz G and Rebhan E
1994 Limits of temperature separation in vortex tubes
J. Phys. D: Appl. Phys. 27 480–8
[7] Silbulkin M 1962 Unsteady viscous circular
flow—application to the Ranque Hilsch vortex tube
Figure 4. Maximum temperature separation 1Thc ,max as a J. Fluid Mech. 12 269–93
function of the normalized pressure drop X = (p0 − pc )/p0 . Dreissler R G and Perlmutter M 1960 Analysis of the flow
and energy separation in a turbulent vortex Int. J. Heat
Mass Transfer 1 173–91
for the temperature splitting is the normalized pressure drop Pengelly C D 1957 Flow in viscous vortex J. Appl. Phys.
28 86–92
X and not the absolute entrance pressure. This experimental van Deemter J J 1952 On the theory of the Ranque Hilsch
observation allows safe incorporation of vortex tubes in cooling effect Appl. Sci. Res. A 3 174–96
closed cycles that may go below atmospheric pressures for [8] Schultz-Grunow F 1951 Turbulenter wärmedurchgang im
cooling and dehydration applications. zentrifugalfeld Forsch. Ing. Wes. 17 65–76
[9] Stephan K, Lin S, Durst M, Huang F and Seher D 1983 An
investigation of energy separation in a vortex tube Int. J.
Acknowledgments Heat Mass Transfer 26 341–8
[10] Amitani T, Adachi T and Kato T 1983 A study on
This work was supported in part by a research grant from temperature separation in a large vortex tube Japan Soc.
Mech. Eng. 49 877–84
the National Science and Engineering Research Council of [11] Lindstrom Lang Cu 1971 The three dimensional
Canada. We thank the referee whose questions prompted us distribution of tangential velocity and total temperature
to find the quantified relation between 1Thc and X, shown in vortex tubes J. Fluid Mech. 45 161–87
in figure 4. [12] Erdélyi J 1962 Wirkung des zentrifugalfeldes auf den
wärmezustand der gase, erklärung des Ranque effekts
Forsch. Ing. Wes. 28 181–6
References [13] Kurosaka M 1982 Acoustic streaming in swirling flow and
the Ranque Hilsch vortex tube effect J. Fluid Mech. 124
[1] Ranque G J 1933 Expérienesd sur la detente giratoire avec 139–72
productions simultanes d’un echappement d’air chand et [14] Benedict R P 1984 Fundamentals of Temperature, Pressure
d’un echappement d’air froid J. Phys. Radium IV (7) and Flow Rate Measurements (New York: Wiley)
112–14 [15] Camire J 1995 Experimental investigation of vortex tube
[2] Hilsch R 1946 Die expansion von gasen im zentrifugalfeld concepts MA Sc. Thesis University of British Columbia,
als kälte prozess Z. Naturforsch. 1 208–14 Vancouver

1472