L. Oranje
Gasunie Research
Groningen, The Netherlands
Full-scale performance tests of four types of gas-liquid separators have indicated a cyclone-type separator can have a catch-efficiency rate which approaches 100%. The tests were carried out by Gasunie Research, Groningen, The Netherlands, under operating conditions.
The tests were prompted by recurring condensate formation in Gasunie's gas-transmission system. Preliminary findings indicated that the condensate troubles resulted on several occasions from separators which failed to meet manufacturers' performance specifications on catch efficiency under Gasunie's operating conditions
THOUSANDS IN USE
In natural-gas production and transmission systems, the number of gas-liquid separators installed is considerable. In the Dutch gas systems, between 1,000 and 2,000 separators are in use.
Gas-liquid separators can be divided into two main groups: separators installed in gas-treatment installations and separators to protect vital installations (compressor stations, gas-metering stations, etc). Moreover, many industrial gas customers use inlet separators to protect their gas equipment.
Clearly, proper performance of the gas-liquid separators installed is a condition for undisturbed gas delivery. Operators of gas production and transmission systems must be sure that all separators installed function according to performance specifications.
In the past, in the Dutch gas systems, however, Gasunie has been faced with condensate problems in pipelines and installations several times. Investigations of these problems indicated that quite often the problems were caused by separators that failed to meet Gasunie's requirements.
This experience led to the decision to test the various types of separators in use under operating conditions. When a separator failed to meet the performance specifications, the problem was investigated.
In this way we gained a great deal of experience and insight into the various separation principles and their advantages and disadvantages.
Having decided that the currently available separators were inadequate for our purposes, we applied the knowledge gained towards designing a new separator which would meet our requirements.
Reported here are the results of this research and development work.
TEST METHODS
The performance of gas-liquid separators has been investigated in field tests and in a special test installation at Gasunie's high-pressure laboratory.
For the field tests the following methods were used:
- Injecting a well-known liquid flow upstream from the separator to be tested and measuring the quantity caught in the separator.
- Measuring the liquid quantity formed upstream from the separator and comparing it with the liquid caught. This method has proven to be useful when a pressure governor is situated between the sample-point and the separator tested and all the liquid is formed in the governor.
- Isokinetic sampling upstream and downstream from the separator and simultaneously measuring the liquid caught in. This method can only be used when complete mist flow occurs in the inlet and outlet of the separator tested.
These methods have been used to measure the catch efficiency for liquids. The first and the last method could probably also be used when measuring the catch efficiency for dust.
Although the reliable measurement of catch efficiency can be difficult, in most field tests the error could be limited to 5%.
Field tests on separators have given very useful information. Most of these tests dealt with troubleshooting of a treatment installation in which malfunctioning of a separator proved to be the cause.
The apparatus for the high-pressure laboratory experiments is shown schematically in Fig. 1. It functions as follows.
The gas flow to the city of Groningen can be passed through the test set-up. The gas pressure in the system can be varied from 700 to 3,900 kPa.
INSPECTION WINDOWS
A large high-pressure vessel with inspection windows is part of the test loop giving the possibility for visual observation of the separation process in a transparent model.
These visual observations have proven to be especially useful for discovering causes of malfunctioning.
Apart from a study of the separation process, the catch efficiency for liquids could be measured accurately by injection of a liquid flow upstream from the separator. The liquid volumes caught in the separator tested and in the control filter downstream from it were measured during the experiment. In all tests the mass balance was checked carefully.
To ensure the experiments are relevant for the field, the droplet-size distribution in the gas passing through the test set-up should be the same as it is in practice.
Because the droplet size in a gas pipeline is a function of gas flow, gas pressure, and liquid properties, the experimental conditions were adjusted to be those occurring in pipelines in practice.
The catch efficiency (n) of gas-liquid separators which are installed in vessels can be expressed as a function of the gas load factor (L) in which:
[SEE FORMULA]
In this formula, Vgas is the volumetric gas flow over the inner cross sectional area of the separator vessel.
The mass densities of gas and liquid are pgas and Pliquid respectively. From the formula, it can be seen that the separation process of liquid and gas is mainly a function of the velocity head of the gas flow. For vanes or meshpad separators, the Vgas is defined as the gas flow over the area of the gas separator package.
VANE SEPARATOR
This type of liquid separator consists of vertical vanes. These vanes force the gas flow passing the separator into a zig-zag flow path. Droplets impinge on the surface by inertial forces and the liquid film formed flows into vertical slits, through which it is drained by gravity.
Vane separators are used to catch mist and spray from a gas flow.
The catch efficiency of a commercially available vane package has been tested in Gasunie's laboratory.
The result of the test carried out at 20 bar is given in Fig. 2.
It can be seen that the vane separator functions well up to a gas load factor of 0.25.
At higher flow rates, carryover occurs.
At a later stage a field test was done with another make of vane separator. These tests showed the same value for Lmax.
Vane separators are not suitable for catching slugs. This is understandable because the slits through which the liquid drains will be overfilled when a slug passes through the system.
The smallest catchable droplet size in vane separators (dcrit) depends on the gas pressure. For water droplets the dcrit has been calculated for the maximum admissible gas flow (at lambda = 0.25). The results are given in Fig. 3.
The smallest catchable droplet size is about 40 m at a gas pressure of 5,000 kPa.
Vane separators can only handle relatively small amounts of liquid in a gas flow, particularly when the liquid separated is relatively viscous (e.g., glycol).
Further, a vane package should not be too high because this reduces the maximum liquid load that can be handled.
Vane separators appeared unable to handle a combination of a water-glycol mixture and condensate.
KNOCKOUT SEPARATION
A gravity or knockout separator is essentially an empty vessel. It is usually employed in systems where liquid slugs are expected.
Two different sizes of vertical knockout separators were tested at operating pressure. The catch-efficiency curves of both appeared to be the same (Fig. 2).
A knockout vessel is obviously applicable up to a gas load factor of = 0.8.
MESHPAD
Most meshpads are installed in the top of a vertical vessel at least at 0.5 vessel diameter from the gas inlet. The system has the advantage over the knockout vessel in that it is able to catch droplets.
A commercially available meshpad (0.1 m thickness) was tested in Gasunie's laboratory. These tests indicated that a meshpad functions well up to a gas load factor of 0.1. This is in accordance with the results found by others.1
The measured efficiency curve is given in Fig. 2. Here too the minimum catchable droplet size depends on the gas pressure.
At atmospheric pressure dcrit = 10 m, but for a gas pressure of 8,000 kPa, dcrit = 50 m.
Dcril as a function of the gas pressure at maximum gas flow (at lambda = 0.1) is given in Fig. 4.
TEST INDICATIONS
The many studies carried out on various types of gas-liquid separators and some dust separators lead to the following conclusion.
The process of droplet separation as a function of particle size can be reasonably predicted by calculation.
Gas-liquid separation, however, is a complex process. The catch efficiency is only partly influenced by the droplet-size distribution. Other effects almost always dominate the efficiency.
The only way to determine reliably the catch efficiency for liquids is to conduct tests under conditions as close as possible to those prevailing in the pipeline system.
Further, factors that negatively influence the efficiency for gas-liquid separation are the following:
- Spray formation in the separator by high gas velocities and sharp edges
- Occurrences of secondary flow (in cyclones) resulting in liquid being led to the center of the cyclone
- Under dimensioning of the separation space because of which both phases do not separate adequately
- Too high gas velocities in the reservoir causing reentrainment and even erosion
- Restrictions encountered by the liquid to flow unhindered into the reservoir of the separator. If this occurs, slugs will prefer to flow to the gas outlet instead of into the gas reservoir.
AN IMPROVED SEPARATOR
Making use of the acquired insight, Gasunie research has developed a separator in which the negative flow effects have been nearly eliminated.
It is a mono-cyclone separator suitable for catching dust and liquids (mist and slugs). An outline sketch is given in Fig. 5.
Features of the separator are: no sharp edges at critical points, and secondary flow effects have been suppressed.
Liquid that still flows to the center of the cyclone is forced to the outer wall again.
Additional features are: movement of liquids and dust in the reservoir is suppressed by carefully developed baffles and vortex spoilers and the separator has been made suitable for catching slugs with high efficiency.
The catch efficiency tests carried out with the high-pressure test set-up in the laboratory showed an efficiency of practically 100% over a wide flow range at gas pressures varying from 100 to 4,000 kPa. In these tests a gas load factor was applied of up to 0.5.
Moreover, the separator performed excellently as a slugcatcher even at the highest gas-inlet velocities we could create.
An especially severe test of the cyclone separator was performed in the following way.
The inlet diameter of the cyclone was reduced to 50% of the original diameter, causing a high gas velocity at the inlet. This caused severe spray formation of the liquid entering the separator.
The results of this test are given in Table 1.
Finally, tests were done at gas load factors up to lambda = 1, but even then the measured efficiency was still 99.5%. This was better than any separator tested before.
The smallest catchable droplet size of the new separator as a function of the gas load factor is given in Fig. 6.
COMPARING PERFORMANCES
The production costs of a separator are mainly determined by the cost of the pressure vessel. For comparing the performance of the various separator principles, it is reasonable to install the different systems in vessels of the same diameter.
We then calculated the maximum gas flows that could be cleaned by the various separator types under this assumption. The results are given in Table 2.
From Table 2 it can be concluded that the cyclone separator is much more economical than the other separator types of the same size. Moreover, it is the only separator that can handle slugs at each gas velocity.
Because of its small size and weight, the cyclone separator is very suitable for application on platforms.
REFERENCES
- Burkholz, A., "Trophenabscheidung an Drahtfiltern," Chem. Ing. Technik, 42. Jahrgang, 1970, pp. 1314-1321.
- Nonhebel, G., Gas Purification processes, George Newness Ltd. London.
Copyright 1990 Oil & Gas Journal. All Rights Reserved.
