LIQUID CARRYOVER IN GEOTHERMAL STEAM-WATER SEPARATORS

Rizaldy1*, Sadiq J. Zarrouk2 and Chris Morris3

1PT Pertamina Geothermal Energy, Skyline Building No. 9, Jakarta, Indonesia
2 Department of Engineering Science, The University of Auckland, Private Bag 92019, Auckland, New Zealand
3 Contact Energy Ltd. Wairakei Power Station, Private Bag 2001, Taupo 3351, New Zealand

*[email protected]

Keywords: Geothermal separator design, water carryover, Wet steam from the separator can lead to the problem of
liquid film entrainment, separator efficiency. scaling in the turbine as liquid becomes superheated and
leaves mineral deposits. In addition, water droplets in the
ABSTRACT steam will impact and erode the blades and rotor causing
Geothermal steam power plant uses saturated steam to run further damage in the turbine. Adiprana (2010) reported de-
the turbines. Therefore, it is important to ensure that the rating and below-optimal turbine behavior in the Gunung
steam is high in quality and purity to keep turbines and Salak power station, Indonesia was caused by solid particle
other equipment working properly. Mineral deposition and and ferrous iron deposition in the turbine, scrubber and
moisture damage have been reported in many geothermal demister. The analysis and physical inspection concluded
turbines around the world, affecting the turbine efficiency that those materials were brought by wet steam. Kubiak et
and leading to generation losses. al. (2005) presented an investigation of the failure of 110
MWe geothermal steam turbine blades in Mexico due to
Current design models for calculating the separator liquid carryover. Mubarok and Zarrouk (2016) reported
efficiency reports about 99.995% separation efficiency, similar moisture and mineral damage to the turbines in the
though it is often found that scaling and moisture damage Ulubelu geothermal power development.
continue to take place in newly constructed steam power
plants. Recent studies have noted the potential for volatile Figure 1 shows the typical mineral deposition on a turbine
silica to mainly contribute to this damage for cases of high rotor.
separation pressures. However, this is not the view of the
authors.

This work investigates water entrainment due to the

formation of a liquid film on the walls of the cyclone
separator, resulting in significant amount of entrained liquid
droplets in the steam. A new model was proposed for
addressing the liquid film entrainment. The model shows
that the actual efficiency of separator is expected to be
lower than the theoretical/calculated efficiency.

Liquid carryover analysis was presented to give better idea

on the factors controlling liquid film carryover in vertical-
cyclone separators. Liquid film entrainment modeling using
field data from the Wairakei geothermal field was reported.
The results show that the rate of entrainment increase as the
liquid loading fraction increases. It also reveals the impact
of liquid film thickness to the rate of entrainment since Figure 1: Mineral Deposition on a Turbine Diaphragm
higher liquid loading leads to thicker liquid film. In at Wairakei Power Plant (from Morris and
addition, inlet velocity is important in determining the rate Robinson, 2015).
of liquid entrainment. As the inlet velocity rises, the
entrainment rate increase causing more water carryover. The turbine is the most expensive piece of equipment in a
power plant. Its maintenance obviously comes at high cost
1. INTRODUCTION and causes lost generation, hence lost profit due to the unit
The dryness and purity of the steam entering the turbine are needing to be shutdown. Thus it would be more efficient to
important parameters in the long term operation of prevent potential damage and keep the turbine in a top
geothermal power plants. It is crucial to ensure the inlet operating condition by supplying dry and clean steam.
steam quality is higher than 99.5% in order to keep the
turbines, pipelines and other equipment operating properly. The main steam-water separation equipment in wet
The steam quality affects not only the turbine efficiency but geothermal fields is the separator, which can be a vertical
also its reliability and lifespan. Ideally, the steam fed into cyclone separator or a horizontal separator. However,
the turbine should have wetness of less than 1%, but in separators are not 100% efficient (Zarrouk and Purnanto,
practice many steam fields cannot meet this crucial 2014). There is always liquid carryover from the separator
requirement and this leads to turbine damage (Morris and and condensation in the steam pipeline that needs to be
Robinson, 2015). removed before the steam enters the turbine. The efficiency
of separators is typically reported to be about 99.995% by
design. Even when efficiency is that high, liquid carried in

Proceedings 38th New Zealand Geothermal Workshop

23 - 25 November 2016
Auckland, New Zealand
the steam line is a substantial factor in steam field The design of horizontal separators has been improved in
operation. Morris and Robinson (2015) showed that for 100 order to achieve higher efficiency. Many designs claim to
t/h steam flow at 99.995% separator efficiency, there is 5 have better separation efficiency than vertical separators.
kg/h carryover. Improvement can be seen in the inlet of recent designs of
separator which has used an inflow distributor to break the
The estimation of new separators efficiency is often based two-phase flow and effectively separate steam and water
on sizing and design calculations (Zarrouk and Purnanto, before the outlet chamber through the vane-plate demister.
2014). The actual efficiency can be calculated after the The steam outlet points towards the end cap in order to
construction and installation through field trials, which reduce the suction effect. The outlet chamber covers the
usually involve steam condensate or mass flow sampling. steam outlet to prevent any liquid re-entrainment while the
The actual efficiency of separators has been shown to be vane-plate demister eliminates small size mists/droplets that
lower than that produced by calculation due to mineral could not be separated in vertical separators. Two inlets
deposition and turbine damage as the result of liquid droplet with symmetrical design are used to improve efficiency in
carryover. the case of higher flow rates. The larger liquid surface area
in horizontal separators reduces the effect of brine surges.
2. STEAM-WATER SEPARATOR Schematics of the designs are shown in Figure 3 below.
The separator is the main equipment to separate water from
steam in wet geothermal fields. In dry steam geothermal
fields separators are not required, though a moisture Two- Steam Outlet
removal system (MRS) is required near the power plant to phase
Vane-plate
ensure the dryness and the cleanliness of the steam entering Demister
the turbine. Most conventional geothermal fields worldwide Inflow
are liquid-dominated reservoirs producing a mixture of Distributor
steam and water and therefore separators are required
Liquid Level
(Zarrouk and Moon, 2014). Currently, there are two general
types of separators used in the geothermal industry:
horizontal and vertical.
Brine Outlet
2.1 Horizontal Gravity Separator
A horizontal separator relies on gravitational force to Figure 3: Design Development of a One-Inlet Horizontal
separate water from steam. Water, which has higher Separator (after Josefsson, 2016).
density, will fall to the bottom of separator and will be
separated from the steam flow. A horizontal separator is
commonly used for high gas-liquid ratio fluid whereas a 2.2 Vertical Cyclone Separator
vertical separator is used for low gas-liquid ratio fluid
This design also known as the Webre separator (Figure 4),
(Arnold and Stewart, 2008). Worldwide, horizontal
the design dates back to early 1950s and was first used in
separators are used less frequently in geothermal steam
the Wairakei field, New Zealand. Since then, it has been
fields. Iceland, Japan and Russia are the main countries that
used in many parts of the world to separate steam from
use horizontal separators when designing the steam field
brine in liquid dominated geothermal fields. The vertical
(Zarrouk and Purnanto, 2014). Horizontal separators are
bottom outlet cyclone (BOC) separator uses centrifugal
often equipped with a mist eliminator at the steam outlet to
force to separate steam and water. Steam, which has lower
improve the separation process (Figures 2-3). In general, for
density than water, tends to flow in the middle area of the
the same flow rate a horizontal separator requires a larger
vessel whereas liquid flows on the separator wall, causing
vessel than a vertical separator (Zarrouk and Purnanto,
the liquid to lose its momentum and fall to the bottom of
2014) and gravity separation alone has lower efficiency
separator under gravity.
than cyclone separation (Perry et al., 1997). On the other
hand, horizontal separators are simpler to construct and The efficiency of a vertical BOC separator is claimed to be
operate compared with a vertical separator. It also has higher than 99.9% (Lazalde-Crabtree, 1984) when
higher flow rate range. Horizontal separators do not require simulated using computational fluid dynamics (CFD)
a water vessel at the brine outlet and less complex seismic (Purnanto et al., 2013). A vertical BOC separator offers the
design. The typical arrangement of horizontal separator is advantages of high efficiency and smaller size but the
shown in Figures 2 and 3 below: efficiency of separation is only achieved in a narrow flow
rate range (Zarrouk and Purnanto, 2015), and it also
Two-phase Steam Outlet
involves some pressure drop. Large-sized vertical BOC
Inlet separators are also difficult to construct and operate.

Vane-plate
Demister

Liquid Level

Brine Outlet

Figure 2: Typical Configuration of a Horizontal

Separator (after Swanborn, 1988).

Proceedings 38th New Zealand Geothermal Workshop

23 - 25 November 2016
Auckland, New Zealand
 η𝑠𝑠 = η𝑚𝑚 × η𝐴𝐴 (2)

where η𝑠𝑠 is the separator efficiency, η𝑚𝑚 is the centrifugal

efficiency, and η𝐴𝐴 is the annular efficiency.

The separator efficiency determined above is considered the

Two-phase
Inlet
theoretical separator efficiency. It does not really represent
the actual efficiency since other mechanisms such as liquid
re-entrainment is not adequately addressed in that
correlation. Once steam and water is separated in a vertical
BOC separator, most of the water drops down and goes out
into the brine line (𝑚𝑚̇𝑙𝑙 ), though some water droplets stays
with the steam and falls down the bottom of the pipelines
(𝑚𝑚̇𝑏𝑏 ) as given in equation (1). However, a layer of liquid
(brine), referred to as ‘thin liquid film’ form on the wall of
separator. As the lighter steam moves upwards in a spiral, it
Brine contacts the surface of the film and causes the droplets to
Outlet enter the steam flow as carryover (𝑚𝑚̇𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ). The
Steam Outlet
simplified flow mechanism inside the separator can be seen
in Figure 5.
Figure 4: Vertical Bottom Outlet Cyclone Separator.
This re-entrainment mechanism was proposed by Foong
Most steam field system design nowadays involves the use
(2005), which is based on similar liquid film behaviour.
of larger-size centralized BOC separators. This seems
Foong (2005) noted that the water creeps along the
preferable because it involves relatively low capital and
separator wall and reaches the top before coalescing and
maintenance costs. The main reason for choosing a vertical
falling into the steam outlet pipe. This mechanism will
BOC separator is its simplicity, since it is preferable for
necessarily reduce the effective efficiency of the separator.
steam to be removed at the bottom near ground for simpler
It should be noted that while 𝑚𝑚̇𝑏𝑏 can be quantified as will
pipe support (Bangma, 1961). It also does away with
be described next. The 𝑚𝑚̇𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 is not measured or
interior baffles and funnels, which could corrode or be
damaged by the deposition of minerals. quantified. There are two different opinions as to the nature
of the mineral carryover (Arifin and Zarrouk 2015)
2.3 Steam-Water Separation Efficiency suggested that 𝑚𝑚̇𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 takes place through micro scale
liquid partials/droplets while (Addison et al, 2016) related
Separator performance is a measure of the proportion of that to volatile silica in the separated steam.
brine that is carried over with the steam:

𝑚𝑚̇𝑠𝑠
η𝑠𝑠 = × 100% (1) 𝑚𝑚̇ 𝑠𝑠 +𝑚𝑚̇𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 + 𝑚𝑚̇𝑏𝑏
𝑚𝑚̇𝑠𝑠 + 𝑚𝑚̇𝑏𝑏
Liquid
where η𝑠𝑠 is the separator efficiency and 𝑚𝑚̇𝑠𝑠 and 𝑚𝑚̇𝑏𝑏 are the Film
flow rate of steam and brine carryover, respectively. Droplet Carryover

The efficiency of a cyclone separator depends on several

design parameters, such as the dimensions of the separator, 𝑚𝑚̇𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡
particle density and operating temperature. The physical
properties of the fluid, namely the density and viscosity,
and operating parameters such as the inlet velocity of the
fluid into the cyclone and the outlet condition also affect the
separation efficiency (Gimbun et al., 2004).

Combining these factors, Leith and Licht (1972) suggested

that the efficiency of a BOC separator depends on three
dimensionless parameters: a cyclone design number 𝑚𝑚̇ 𝑙𝑙
depending upon the physical shape (not size); a modified
type of impaction parameter depends upon operating 𝑚𝑚̇ 𝑠𝑠 − 𝑚𝑚̇𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
condition; and the exponent in the modified form of the + ̇
vortex law for tangential velocity distribution. Leith and
Licht’s (1972) approach was based on the concept of
continual radial back-mixing of the uncollected particles, 𝑚𝑚̇𝑏𝑏
coupled with the calculation of an average residence time
for the gas in a cyclone separator having a tangential inlet. Figure 5: Simplified Flow Mechanism inside a Vertical
This was developed and modified by Lazalde and Crabtree BOC Separator.
(1984) for the geothermal vertical BOC separator. Lazalde
and Crabtree (1984) defined the separator efficiency as a The separation efficiency achieved by the cyclone separator
product of mechanical or centrifugal efficiency and annular is quite sufficient for most applications. However, in
efficiency: geothermal power plants, the cumulative effects of small

Proceedings 38th New Zealand Geothermal Workshop

23 - 25 November 2016
Auckland, New Zealand
brine carryover with associated dissolved chemicals can be Sodium concentrations taken from samples at the outlet of
quite large (Foong, 2005). drain pots are shown in Figures 7 and 8, respectively.
Sodium measurements were made at various wash water
The current method used to determine the actual efficiency rates upstream of the demister, downstream of the demister,
of a separator (1) requires the accurate measurement of the and before the turbine inlet. The increase of Sodium
brine in steam (𝑚𝑚̇𝑏𝑏 ). Unfortunately, it is not possible to concentration is caused by a larger amount of brine
directly measure this since it is a small component of the carryover passing over the drain pot as it flow with the
total flow in the steam line. Moreover, the brine carryover steam.
is already diluted by steam condensate along the pipeline.
However, the chemical signatures of geothermal brine can
be used to indirectly measure the brine carryover.

Bangma (1961) and White (1983) included Sodium (Na)

and Chloride (Cl) as the chemical signatures to measure
carryover and separator efficiency. Sodium is preferable
since it has a higher concentration limit which is detectable
(> 1ppm) in the steam. The equation below calculates the
brine carryover (𝑚𝑚̇𝑏𝑏) :

̇ Figure 6: Simplified flow diagram of the steam field

𝐶𝐶𝑆𝑆 𝑁𝑁𝑁𝑁 setup
𝑚𝑚̇𝑏𝑏 = (3)
𝐶𝐶𝑠𝑠𝑠𝑠 𝑁𝑁𝑁𝑁
Sodium brought by the brine from separator is diluted by
𝑁𝑁𝑁𝑁 wash water and accumulates in the demister. This shows
Where 𝐶𝐶𝑆𝑆̇ is the mass flow rate of Sodium in separated that the performance of the separator is not always within
steam in mg/s and 𝐶𝐶𝑠𝑠𝑠𝑠 𝑁𝑁𝑁𝑁 is the Sodium in separated brine design. Significant concentrations of Sodium are also found
(Zarrouk & Purnanto, 2015). In many geothermal fields, downstream of the drain pot after the demister and the
Sodium analysis of samples in the first drain pot right after turbine inlet drain pots.
separator is used to compute the efficiency of separation
through brine carryover.

̇
𝐶𝐶𝑆𝑆 𝑁𝑁𝑁𝑁
η𝑠𝑠 = �1 − � × 100% (4)
𝐶𝐶𝑠𝑠𝑠𝑠 𝑁𝑁𝑁𝑁

Nevertheless, the Sodium analysis in the drain pot outlet

does not perfectly reflect the actual efficiency of separator
because not all of the water carryover can be separated by
the drain pot. As suggested in Figure 5, brine carryover
(𝑚𝑚̇𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ) as droplets in the steam. Their size varies
from very small to larger ones that can possibly hit the
inner wall or settle down at the bottom of the pipe. Those
that are very small have little chance of settling down and
will continue to flow along with the steam. The efficiency
of the drain pot is also an important factor to determine how Figure 7: Sodium Concentration Taken from Drain
it can effectively remove the water. The same observation Pots.
given above also applies if the (𝑚𝑚̇𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ) takes place as Generally, separator efficiency decreases as Sodium
volatile silica. However, while volatile silica may be concentration increases (equations 1-3) in the samples taken
plausible for high separation pressures, Morris & Mroczek from the drain pots. Any increase in separator loading
(2015) show it isn’t viable at lower pressures. If there is means an increase in the inlet velocity and this will escalate
sodium present in the carryover, then it is fairly certain that the Sodium concentration downstream of the separator and
volatile silica isn’t a significant mechanism. vice versa.
3. LIQUID CARRYOVER ANALYSIS Total dissolved solid (TDS) concentrations from the
The liquid carryover (𝑚𝑚̇𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ) is a common problem in samples reveal similar information about the brine
geothermal power plants. Even though geothermal steam carryover (Figure 8).
fields have moisture removal systems (MRS), it is often
found that the steam flow entering the turbine contains high
amounts of liquid. High concentrations of measured
Sodium and Chloride in steam condensate indicate that
there is significant liquid carryover. In several cases,
turbine damage due to liquid carryover still occurs even
though sampling results show that the steam which entered
the turbine was clean (Morris and Robinson, 2015).

A set of field experiments from the Wairakei geothermal

field, New Zealand shows that the Sodium concentration at
the demister increases with steam washing. The simplified
diagram of the steam field setup is shown in Figure 6.

Proceedings 38th New Zealand Geothermal Workshop

23 - 25 November 2016
Auckland, New Zealand
 Steam Flow Droplets

Separator wall
a)

Steam Flow Droplets

Separator wall
b)

Figure 9. Droplet Breakage Mechanism (a. Undercut

Figure 8: TDS of Samples Taken from Drain Pots, Re-Entrainment – Low Film Reynolds Number; b) Roll
before and after washing. Wave Re-Entrainment – Higher Film Reynolds
Number) (after Ishii and Grolmes, 1975).

Figure 8 shows a significant increase in TDS concentration

at the outlet of the demister which is about three times the It is important to know the rate of liquid entrainment since
concentration at the upstream drain pot. Figure 8 also shows it affects the quality of steam sent to the power station. The
that water washing dilutes the TDS measured at the drain term ‘entrainment fraction’ (Fe) is used to describe a
pots, higher water washing rate results in more dilution. quantification of the rate of liquid entrainment to the steam
flow. It is defined as the fraction of the total liquid flow in
the form of droplets in the steam core. By definition, the
value of the entrainment fraction between 0.0 and 1.0. The
3.1. Liquid Entrainment Modeling
entrainment fraction results from the dynamic equilibrium
Entrainment is an important factor in liquid carryover between the rate of deposition of drops from the steam core
analysis. In geothermal steam field operations, entrainment to the liquid film and the rate of droplet formation (also
can take place in the separator or in the steam line. Based called atomization) at the vapour-liquid interface caused by
on the droplet-forming mechanism in two-phase annular waves occurring on the film surface (Pan and Hanratty,
flow, entrainment is caused by the thin liquid film formed 2002a,b; Mantilla et al., 2009; Al-Sarkhi et al., 2011a).
at the separator vessel or pipe wall. High steam velocity
Authors have used different correlations in their models, all
creates high interfacial shear stress and instability on the
of which vary in their prediction of the entrainment fraction
film surface. At some points, it may create a wave effect
and are often limited to certain conditions. Ishii and
causing continuous entrainment of droplets into the steam
Mishima (1989) proposed an entrainment fraction
flow.
correlation that is limited to a certain Reynolds number.
There are two relevant mechanisms of droplet re- Wallis’s (1969) correlation is very simple and
entrainment at the separator or pipe wall: roll wave re- straightforward but results in higher error for entrainment
entrainment and undercut re-entrainment (Ishii and fractions greater than 0.5. Moreover, some correlations
Grolmes, 1975). They are determined by the liquid film have narrow coverage (e.g. only air-water systems). Pan
Reynolds number which is affected by the steam velocity: and Hanratty (2002 a, b) presented an entrainment fraction
correlation in annular flow for vertical and horizontal pipes.
These correlations are considered to be accurate but the
model is very complex. Furthermore, Sawan et al. (2008)
𝜌𝜌𝑙𝑙 × 𝑢𝑢𝑙𝑙 × 𝛿𝛿 and Liu et al. (2001) have been actively involved in vertical
𝑅𝑅𝑅𝑅𝑙𝑙 = (5) two-phase annular flow simulation using computational
𝜇𝜇𝑙𝑙
fluid dynamics (CFD) which takes droplet entrainment and
droplet behaviour into consideration. Their model has been
where 𝑅𝑅𝑅𝑅𝑙𝑙 is the liquid film Reynolds number, 𝛿𝛿 is the adopted by fluid dynamic analysis software.
thickness of the film (mm), 𝑢𝑢𝑙𝑙 is its mean velocity (m/s), 𝜌𝜌𝑙𝑙
is the density (kg/m3) and 𝜇𝜇𝑙𝑙 is the viscosity of the liquid Sawant et al. (2008) divided the process of entrainment into
(kg/m.s). three parts as shown in Figure 10. The first part of
entrainment is dependent only on the Weber number. The
Figure 9 shows the droplet re-entrainment mechanism at the second part is dependent on both the Reynolds number and
separator inner wall. The droplet entrainment mechanism is Weber Number, while the last part depends only on the
difficult to predict. No analytical solution exists due to the Reynolds number only.
complexity of the flow and interfacial structure of
entrainment. Yet, modelling attempts in the past have
shown that it can be done with varying degrees of accuracy
and complexity (Al-Sarkhi et al., 2011a). Many authors
have proposed models for droplet entrainment in annular
two-phase flow but no general entrainment rate correlation
satisfies all the experimental information that has been
developed.

Proceedings 38th New Zealand Geothermal Workshop

23 - 25 November 2016
Auckland, New Zealand
 The first task is to determine 𝑅𝑅𝑅𝑅𝑙𝑙 , the liquid film Reynolds
number. Under the influence of the upward centrifugal
steam flow, the film moves upward along the cyclone wall
at an angle (α) to the horizontal (Figure 11). We can
therefore take the wetted perimeter as equation below:
𝑦𝑦
𝑃𝑃𝑤𝑤 = (9)
𝑠𝑠𝑠𝑠𝑠𝑠 𝛼𝛼

Wetted
Perimete
y Liquid
Film

Figure 10: Sawant et al. (2008) Correlation

Methodology. Imaginary
Two-phase y
Separator Wall
Inlet Area

Departing from the above methodology, Al-Sarkhi and Separator

Sarica (2011a) developed an entrainment model that is α Inner Wall
simple yet accurate and covers a wider range of
2πD
applications. The model is equated to the fouling growth Brine
model for a conventional heat exchanger which has similar Steam Outlet >2πD
asymptotic trend characteristics (Figure 10). By analogy,
the entrainment fraction equation leads to the following
model: Figure 11: Wetted perimeter and inflow exit angle.
𝑊𝑊𝑊𝑊𝑆𝑆𝑆𝑆 We define the wetted perimeter (𝑃𝑃𝑤𝑤 ) as the width of the
𝐹𝐹𝑒𝑒 = 𝐹𝐹𝑒𝑒,𝑚𝑚𝑚𝑚𝑚𝑚 �1 − exp � �� (6)
𝑊𝑊𝑒𝑒 ′𝑆𝑆𝑆𝑆 film measured normal to the direction of flow. In fact, the
steam moves at certain angle from the horizontal, where α
𝜌𝜌𝐺𝐺 𝑣𝑣𝑆𝑆𝑆𝑆 2 𝐷𝐷 𝜌𝜌𝑙𝑙 − 𝜌𝜌𝐺𝐺 1/4 is the exit angle from the separator inlet. As an
𝑊𝑊𝑊𝑊𝑆𝑆𝑆𝑆 = � � (7) approximation, we assume that the steam moves at the same
𝜎𝜎 𝜌𝜌𝐺𝐺 angle with liquid film (α). The wetted perimeter and inflow
exit angle are described in Figure 11.
where 𝐹𝐹𝑒𝑒 is the entrainment fraction, 𝑊𝑊𝑊𝑊𝑆𝑆𝑆𝑆 is the superficial
gas Weber number, 𝐹𝐹𝑒𝑒,𝑚𝑚𝑚𝑚𝑚𝑚 is the asymptotic value of the If it is assumed that the liquid loading fraction on the
entrainment fraction, 𝑣𝑣𝑆𝑆𝑆𝑆 is the superficial gas velocity, 𝐷𝐷 is separator wall (ƒ) moves in the liquid film, we can get 𝑅𝑅𝑅𝑅𝑙𝑙 .
pipe diameter, 𝜌𝜌 is density, and 𝑊𝑊𝑒𝑒 ′𝑆𝑆𝑆𝑆 is the analogous
time constant. 𝑄𝑄𝑙𝑙 × ƒ × 𝜌𝜌𝑙𝑙
𝑅𝑅𝑅𝑅𝑙𝑙 = (10)
𝑃𝑃𝑤𝑤 × 𝜇𝜇𝑙𝑙
The time constant in this case is dimensionless and is the
Weber Number when the entrainment fraction reaches
63.2% of its asymptotic value (Al-Sarkhi and Sarica, where 𝑄𝑄𝑙𝑙 is the total liquid flow to the cyclone, ƒ is the
2011a). Since all the thermodynamic properties remain captured fraction, and 𝑃𝑃𝑤𝑤 is the wetted perimeter (equation
constant, the Weber Number ratio can be represented by the 9).
ratio of superficial gas velocities. The maximum
entrainment fraction can be calculated using the equation Theoretically, the entrainment rate (𝑄𝑄𝑒𝑒,𝑠𝑠 ) in the separator
(8) below: can be calculated after the maximum entrainment fraction
(𝐹𝐹𝑒𝑒,𝑚𝑚𝑚𝑚𝑚𝑚 ) and entrainment fraction (𝐹𝐹𝑒𝑒 ) are known (equation
𝑅𝑅𝑅𝑅𝑆𝑆𝑆𝑆 0.6 11). The 𝐹𝐹𝑒𝑒,𝑚𝑚𝑚𝑚𝑚𝑚 has a value of little less than one (Assad et
𝐹𝐹𝑒𝑒,𝑚𝑚𝑚𝑚𝑚𝑚 = 𝐹𝐹𝑒𝑒,𝑚𝑚𝑚𝑚𝑚𝑚,𝑙𝑙𝑙𝑙𝑙𝑙 �1 − exp �− � � �� (8)
𝑅𝑅𝑒𝑒 ′𝑆𝑆𝑆𝑆 al., 1998; Sawant et al., 2008). However, it is also
acceptable to use the value of 1. Since the 𝐹𝐹𝑒𝑒,𝑚𝑚𝑚𝑚𝑚𝑚 is the
where 𝑅𝑅𝑅𝑅𝑆𝑆𝑆𝑆 is the superficial liquid Reynolds number, ultimate amount of entrainment, the value of 𝐹𝐹𝑒𝑒 is either
𝐹𝐹𝑒𝑒,𝑚𝑚𝑚𝑚𝑚𝑚,𝑙𝑙𝑙𝑙𝑙𝑙 is the asymptotic or limiting value of the lower than or approaching the value of 𝐹𝐹𝑒𝑒,𝑚𝑚𝑚𝑚𝑚𝑚 .
maximum entrainment fraction (0.98 – 1.0) and 𝑅𝑅𝑒𝑒 ′𝑆𝑆𝑆𝑆 is the
analogous time constant in the form of a Reynolds number. 𝑄𝑄𝑒𝑒,𝑠𝑠 = 𝐹𝐹𝑒𝑒 × 𝑄𝑄𝑙𝑙 × ƒ (11)

This model was assessed and compared with other The entrainment becomes liquid carryover once it enters the
entrainment models. Several experimental data sets from steam line. Some will be deposited at the pipe wall because
previous authors are used to check whether the model can of turbulence, while those droplets of very small size stay in
predict entrainment curve in general. The result shows that the core of the steam flow. The same entrainment
the model proposed can satisfy all the models with mechanism also occurs in the steam line. The relative
reasonable average error (Al-Sarkhi and Sarica, 2011a). For velocity difference between steam and liquid films will
this reason, this model will be used to predict the liquid potentially create droplets and re-entrain them back to the
entrainment fraction in geothermal separator.

Proceedings 38th New Zealand Geothermal Workshop

23 - 25 November 2016
Auckland, New Zealand
steam flow as shown in Figure 9. Thus the entrainment rate
in the pipe line can be predicted using the same model.

𝑄𝑄𝑒𝑒,𝑝𝑝 = 𝐹𝐹𝑒𝑒 × 𝑄𝑄𝑒𝑒,𝑠𝑠 × ƒ (12)

where 𝑄𝑄𝑒𝑒,𝑝𝑝 is the entrainment rate in the pipeline and 𝐹𝐹𝑒𝑒 is

the entrainment fraction in the pipeline.

It is important to note that it is difficult to exactly measure

how much of the liquid fraction forms a film layer on the
pipe wall. In this case, all the carryover is assumed to be
deposited in the liquid film. Thus the value of ƒ=1 is
considered for simplicity. A more accurate prediction of the
liquid carryover in the separator and scrubbing line might
be obtained if the critical parameters (such as separator inlet
flow angle, liquid film thickness, and steam-liquid relative Figure 13: Liquid Entrainment in Separator at
velocity) are provided through CFD modelling. Different Inlet Velocity.

A set of field data from a geothermal steam field in New

Zealand was used to test the model. The entrainment rate in The amount of liquid carryover will necessarily reduce
a vertical cyclone separator and steam line was simulated separator efficiency. For this reason, the actual efficiency of
and compared with the actual data. The dimensions of the the separator can be expressed by the equation below:
separator are shown in Table 1.
ηactual = ηeffective − ɸ (13)

Table 1: Separator parameters used in the entrainment 𝐹𝐹𝑒𝑒 × ƒ × 𝑚𝑚̇ 𝑙𝑙

ɸ= × 100% (14)
test. 𝑚𝑚̇ 𝑙𝑙
μl 0.000168 kg/m.s where ηactual is the actual efficiency of separator, ηeffective
ρl 905.50 kg/m3 is the theoretical efficiency of separator (using equations 1-
A 1.13 m2 3), ɸ is the deficiency of separator due to the liquid
entrainment from liquid film inside the separator, 𝐹𝐹𝑒𝑒 is the
D separator 3.60 m
entrainment fraction, ƒ is the captured fraction, and 𝑚𝑚̇𝑙𝑙 is
D inlet 1.20 m the separated water flow rate (Figure 5).
D outlet 1.05 m
More entrainment from the separator will increase the
amount of liquid carryover, which increases the deficiency
Figures 12 and 13 show the entrainment fraction and the and reduces the separator efficiency. The model simulation
amount of liquid carryover in the separator based on several also predicts the effect of inlet flow angle in the separator.
liquid fractions loaded on the wall of separator. The inlet As shown in Figure 14, as the angle increase it will result in
angle of the separator is represented as α=30ᵒ. It is clear that less deficiency compared to a smaller angle. This proves
the rate of entrainment gets higher as the liquid loading that a bigger angle will lead to a longer width of steam flow
fraction increases. This result also show the impact of liquid path in the separator which enhances contact between steam
film thickness on the rate of entrainment as higher liquid and liquid film to form more droplets.
loading will lead to thicker liquid film. In addition, the
result also shows that inlet velocity is important in
determining the rate of liquid entrainment. As the inlet
velocity rises, the entrainment gets worse.

Figure 14: The Effect of Inlet Angle on Separator

Deficiency.

Figure 12: Entrainment Fraction in the Separator

To make sure that the model is applicable, the next stage
will be to match the results from the model with the actual

Proceedings 38th New Zealand Geothermal Workshop

23 - 25 November 2016
Auckland, New Zealand
data taken from the field. In this case, the liquid carryover annular two-phase flow’’ by Sawant, P., Ishii, M.,
prediction should be compared with the actual brine flow Mori, M. Nucl. Eng. Des. 238, 1342–1352.
analysis taken from the first drain pot after the separator.
Unfortunately, the actual brine flow data needed is not Arifien, B.N., Zarrouk, S. J., & Kurniawan W., (2015)
currently available for comparison. Nevertheless, the model Scrubbing Lines in Geothermal Power Generation
simulation result seems to be a satisfactory representation Systems. Proceedings the 37th New Zealand
of the liquid entrainment in geothermal separator and steam Geothermal Workshop, 18-20 November, Taupo,
pipeline. New Zealand.

CONCLUSIONS Arnold, K., and Stewart, M. (2008). Surface Production

Several geothermal power developments have reported Operation Design of Oil Handling System facilities
moisture and mineral scaling damage in their turbines as a 3rd ed. Gulf Professional Publishing. Burlington.
result of the separator breakdown (carryover). USA.

The BOC vertical separator has potential problems in some Assad, A., Lopez de Bertodano, M., Beus, S. (1998). Scaled
of the cases presented that results in a reduction of the Entrainment Measurements in Ripple-annular flow in
overall efficiency of geothermal steam field operation. The a small tube. Nucl. Eng. Des. 184, 437–447.
horizontal separator, on the other hand, has the potential to
Bangma, P. (1961). The Development and Performance of a
avoid some of these problems and thus it is reasonable to
Steam-Water Separator for Use on Geothermal Bores.
give consideration to the testing and utilization of a
(pp. 60–77). UN Conf. New Sources Energy Rome 3
horizontal separator in new geothermal developments.
(G/13).
There are two different arguments as to the cause of the
Foong, K.C. (2005). Design Concept for a More Efficient
minerals carryover. The first due to small brine droplets
Steam-Water Separator. Proceedings World
carried with the steam, which cannot be quantified using
Geothermal Congress 2005 (pp. 24–29). Antalya,
existing separator efficiency models. The second is due to
Turkey.
volatile silica carried with the steam. While both arguments
are plausible for high separation pressures, we feel that Freeston, D. (1981) Condensation Pot Design: Model Tests.
droplets carryover is likely to be the main cause. This is Trans. Geothermal Resource Council (United States),
supported by recent field testing data from the Wairakei. A 5(CONF-811015).
new separator efficiency model is presented that account for
this carryover. Gimbun, J., Choong, T., Fakhru'l-Razi, A., & Chuah, T.
(2004). Prediction of the effect of dimension, particle
The new separator efficiency model considered liquid field density, temperature, and inlet velocity on cyclone
entrainment on the walls of the BOC separator through a collection efficiency. Jurnal Teknologi, 37–50.
deficiency correction parameter. However, this model
requires more testing and field validation. Ishii M., Grolmes M.A. (1975). Inception criteria for
droplet entrainment in two-phase concurrent film
flow. AIChE Journal, 21:308–318.
ACKNOWLEDGEMENTS Ishii, M., Mishima, K. (1989). Droplet entrainment
The authors would like to thank the New Zealand correlation in annular two-phase flow. Int. J. Heat
Government for the NZAID Scholarship, PT Pertamina Mass Transfer 32, 1835–1846.
Geothermal Energy for the support to this research, and
Contact Energy Ltd. for their support and kind permission Josefsson V.A. (2016). Steam Separation System
to publish this work. Development. Icelandic Geothermal Conference.
Iceland, 26-29 April, 2016 Reykjavik, Iceland.

Kubiak J.A. and Perez J. (1989). Developments in

REFERENCES
Geothermal Energy in Mexico – Part Twenty-Two.
Addison, S., Richardson, I., Marsh, A., Hernandez, D. Causes of Erosion of the Rotor Blades in a
&Azwar, L. (2016). Challenges with High Separation Geothermal Turbine. Heat Recovery Systems and
Pressures, Iceland Geothermal Conference, 26-29 CHP 9 (2) 159–168.
April, 2016 Reykjavik, Iceland.
Lazalde-Crabtree, H. (1984). Design approach of steam-
Adiprana, R., Izzuddin, Yuniarto, E. (2010). Gunung Salak water separators and steam dryers for geothermal
geothermal power plant experience of scaling/deposit: applications. Geotherm. Resour. Counc. Bull., 11–20.
analysis, root cause and prevention. In: Proceedings
of the World Geothermal Congress, Bali, Indonesia. Leith, D., Licht, W. (1972). The collection efficiency of
cyclone type particle collectors: a new theoretical
Al-Sarkhi, A., Sarica, C. (2011a). Comment on: approach. AICHE Symp. Ser. 68 (126), 196–206.
‘‘Correlation of entrainment for annular flow in
horizontal pipes’’ by Pan, L., Hanratty, T.J., Int. J. Liu, Y., Li, W.Z., Quan, S.L. (2001). A self-standing two-
Multiph. Flow, 28(3), pp. 385–408. Int. J. Multiph. fluid CFD model for vertical upward two-phase
Flow 37, 535–536. annular flow. Nucl. Eng. Des. 241, 1636–1642.

Al-Sarkhi, A., Sarica, C. (2011b). Comment on: ‘‘Droplet

entrainment correlation in vertical upward co-current

Proceedings 38th New Zealand Geothermal Workshop

23 - 25 November 2016
Auckland, New Zealand
Lopez de Bertodano, M.A., Assad, A., Beus, S.G. (2001). Steam-Water Separators. IPENZ Transactions Vol.
Experiments for entrainment rate of droplets in the 40.
annular regime. Int. J. Multiph. Flow 27, 685–699.
Sawant, P., Ishii, M., Mori, M. (2009). Prediction of
Mubarok M.H. & Zarrouk, S. J. (2016). Steam field design amount of entrained droplets in vertical annular two-
overview of the Ulubelu geothermal Project, phase flow. Int. J. Heat Fluid Flow 30, 715–728.
Indonesia. Proceedings the 38th New Zealand
Geothermal Workshop, 23-25 November, Auckland, Sawant, P., Ishii, M., Mori, M. (2008). Droplet entrainment
New Zealand. correlation in vertical upward concurrent annular two-
phase flow. Nucl. Eng. Des. 238, 1342–1352.
Magrini, K. (2009). Master’s Thesis. University of Tulsa.
Swanborn R.A. (1988). A new approach to the design of
Mantilla, I., Gomez, L., Mohan, R., et al. (2009). gas-liquid separators for the oil and gas industry. PhD
Experimental investigation of liquid entrainment in Thesis, Technical University Delft, Netherlands.
gas in horizontal pipes. Proceedings of FEDSM2009
ASME 2009 Fluids Engineering Division Summer Wallis, G.B. (1969). One Dimensional Two-Phase Flow.
Meeting, August 2–5, Vail, Colorado, USA. McGraw-Hill. New York.

Morris, C., and Robinson, A. (2015). Geothermal Turbines: White, B.R. (1983). The Performance of the Bottom Outlet
A Maintainer’s Perspective. Proceedings of the World Cyclone Separators (Wairakei-Type). University of
Geothermal Congress 2015. Auckland, New Zealand.

Morris, C. & Mroczek E., (2015). Turbine Scaling, Williams, L.R., Dykhno, L.A., Hanratty, T.J. (1996).
Proceedings the 37th New Zealand Geothermal Droplet flux distributions and entrainment in
Workshop, 18-20 November, Taupo, New Zealand. horizontal gas-liquid flows. Int. J. Multiph. Flow 22,
1–18.
Pan, L., Hanratty, T.J. (2002a). Correlation of entrainment
for annular flow in vertical pipes. Int. J. Multiph. Zarrouk, S.J. and Moon, H. (2014). Efficiency of
Flow 28, 363–384. geothermal power plants: a worldwide review.
Geothermics 51, 142–153.
Pan, L., Hanratty, T.J. (2002b). Correlation of entrainment
for annular flow in horizontal pipes. Int. J. Multiph. Zarrouk, S. J., & Purnanto, M. H. (2014). Geothermal
Flow 28, 385–408. steam-water separators: Design overview.
Geothermics, 53, 236-254.
Perry, H.P., Green, W.D., and Maloney, J.O. (1997).
Perry’s Chemical Engineers Handbook 7th ed.
McGraw-Hill. New York.

Purnanto, M. H., Zarrouk, S. J., Cater, J.E. (2013). CFD

Modeling of Two-Phase Flow inside Geothermal

Proceedings 38th New Zealand Geothermal Workshop

23 - 25 November 2016
Auckland, New Zealand