# Separator optimization

Why is it important?

Why do we want to maximize oil production in our processing plants? Do we get the same value whether we sell our produced hydrocarbons as part of a liquid phase or a gas phase? The short answer is no! For any given component, a mole of that component sold as part of the stock tank oil has more value than the same mole sold as part of the gas (2-10x, depending mainly on gas and oil price differential). This is the driving force to maximize stock tank oil (condensate) production for a given stream feeding our process facilities. The first SPE paper using an electrical computer was in 1949 by Muskat and McDowell – specifically about finding separator conditions to maximize surface oil production.

Stock tank oil (condensate) rates are dependent on the surface process (“path dependence”) and particularly on the so-called equilibrium K-values.  For a given temperature, pressure and overall composition (zi), the K-values dictate the relative amount of a given component that partitions into equilibrium gas and equilibrium oil at each separation stage.

Furthermore, the amount of gas phase separated in the first primary stage of separation will influence how many moles of a component partition into the final surface gas and oil products. That is, the lighter the processed stream (i.e. the higher the stream gas-oil ratio), the less efficient a given process will be in partitioning components into the more valuable surface oil product.

K-values are defined as the ratio of equilibrium gas composition yi to the equilibrium liquid composition xi (Ki≡yi/xi). K-values larger than 1 indicate a relative preference for the component to remain in the equilibrium gas phase, while Ki<1 indicates a preference for the component to remain in the equilibrium oil phase. Fig. 1 shows an example of the classical log-log plot of K-value vs. pressure at a given temperature (T=100oF in this case).

At common separator conditions Ki≈pvi(T)/p where pvi is the vapor pressure of a component at the separator temperature. As Standing (197x) shows, this relation is valid up to about 1,000 psi (shaded are in the figure below), with slight deviations for the heavier C7+ components that deviate at a lower pressure from the simple straight-line behavior with -1 slope (dashed line in figure below) for Ki α 1/p (log Ki = -1 log p).

What is important?

For a given stream coming into the processing facilities, there are two main things that can control the amount of stock tank oil produced: (1) number of separation stages and (2) separator (p,T) conditions. We show the effect of each of these in the amount of stock tank oil obtained when processing four different fluid streams classified as*: low-GOR black oil (330 scf/STB), volatile oil (2000 scf/STB), rich gas condensate (6650 scf/STB or 150 STB/MMscf) and lean gas condensate (50,000 scf/STB or 20 STB/MMscf).

* GORs given are for a 2-stage flash of fluid composition Psep1 = 300 psia, Tsep1 = 100 F, Psep2 = 14.7 psia and Tsep2= 60 F.

• Effect of number of stages.

In this first example we compare the most inefficient separator process conceivable – a single-stage at atmospheric conditions – with a two-stage separator process with 300 psia and 100oF primary separator conditions.

The figure below shows the % difference in stock tank oil volume produced for a given stream when comparing the 2-stage process with a single-stage process directly to stock tank conditions. It is clear that having an additional stage yield larger stock tank oil volumes for all cases, but this is most relevant for lean gas condensate system.

• Effect of separator (p,T) conditions

Since the separation process is mainly controlled by the K-values of each component, the volume of stock tank oil will be maximized at the conditions at which K-values are lower, indicating a preference for a given component to be in the liquid phase. K-values are lower at higher pressures (as shown in Fig. 1) and lower temperatures (lower vapor pressure pvi at lower temperature). Figs. 3 and 4 summarizes the impact of primary separator conditions on surface oil volume produced (relative to primary separator conditions of 300 psi and 100oF).

The leanest gas condensate has monotonic improvement in surface condensate produced at increasing primary stage pressure. All fluid streams from low-GOR oil to high-GOR gas condensate have improvement of surface oil volumes as primary stage temperature decreases.

The two oils and rich gas condensate show non-monotonic improvement in surface oil volume with primary separator pressure. This behavior can be attributed to the relative importance of (1) K-value dependence on (p,T) and (2) the relative amount of gas liberated at first and second stages. The gas liberated at the second stage will have much higher K-values than the first stage separator, and consequently move more heavy components to the less-valuable surface gas product.

When is it important?

As seen in the figures above the largest impact was observed in leaner fluid systems, although richer gas condensate systems and volatile oils can also see improvements of 10-20% additional stock tank oil volumes from separator optimization. The reason leaner fluid systems will generally be affected more by the conditions of the separation process is the amounts of light-intermediate component (C3-C7) that are generally present in these types of systems. These components are the components that will have K-values crossing the unity line (Ki = 1) and will be the components that will “switch” from preferring to be in the gas phase, to preferring to be in the liquid phase (resulting in more stock tank oil volume).

How can it be done?

First, it is necessary to have an EOS model should that is tuned to surface process data for a wide range of GOR fluids. Then with the tuned EOS model and the help of a PVT or process simulator one should:

• Pick representative input compositions (zi) to the processing facility. These will change over time, and could come, for example, from reservoir simulation or a depletion experiment (either from lab or simulated).
• For a given zi, surface process calculations should be made with the EOS:
• With varying combinations of psp and Tsp for a given separator
• Range of psp and Tsp should be selected knowing the surface process constraints i.e. minimum line pressure, well head pressure, etc.
• With varying number of separator stages – usually 1-2 are good enough in addition to the stock tank
• Calculating total GOR, stock tank oil volumes and other important properties / constraints (e.g. liquid API).
• Optimum surface process is selected giving the maximum stock tank oil volume.

Optimum process conditions will likely change over time due to changing zi with depletion, gas injection, or tie-in of a new reservoir.

The optimal primary (Tsp,psp) conditions will obviously depend on the type of stream being processed – i.e. the “GOR” of the stream. In all fields the processed streams will change over time (due to e.g. depletion, gas injection or tie-in of a new reservoir) and optimal separator conditions will also change over time. However, the largest total wellstream molar rates will always come earlier in a field’s life, making the early optimal separator conditions most important to the design of surface facilities.

###

Global
Curtis Hays Whitson
curtishays@whitson.com

Asia-Pacific
Kameshwar Singh
singh@whitson.com

Middle East