CFD Simulation of a Vertical Induced Static Flotation Vessel English Letter

7/28/2019 CFD Simulation of a Vertical Induced Static Flotation Vessel English Letter

1/11

1www.exterran.com

1.0 INTRODUCTION

A better understanding o the uid dynamics within a separation vessel or tank can

allow or identifcation o design aws and potential solutions or optimization.

Gas otation separators (IGFs and ISFs) are even more complex and it is difcult, i

not impossible, to optimize design by construction and operation o equipment alone.

An additional tool to assist in optimizing oil in water removal is to utilize Computational

Fluid Dynamic (CFD) modeling tools to simulate ow and phase behavior within these

vessels or tanks to rapidly evaluate perormance at a much lower cost and less time.

While use o CFD or modeling uid dynamics is not new, the techniques and code

developed by GLR Solutions uniquely allow or modeling o the complex multiphase (oil,

water, solids, and gas) conditions within otation processes. By numerically modeling

the process, it is much easier to manipulate conditions (geometry, ow rates, inlet

concentrations, droplet size, etc.) to determine their individual impact on perormance

or both existing equipment or to prove perormance or any proposed modifcations to

vessels/tanks.

This paper presents one approach or CFD modeling o multi phase ow within a

vertical Induced Static Flotation (ISF) vessel o a common design and identifes a

number o design issues that negatively impact separation perormance.

2.0 APPROACHES TO CFD

Physical processes within the otation vessel are very complex. Thereore, many

assumptions must be adopted in the deriving o the model equation and accuracy

o the solution depends on the validity o those assumptions. In this case, it was

assumed that the ow is incompressible and properties are constant. In addition, the

gas bubbles and oil droplets are treated as spherical particles. The ree surace is

assumed to be stationary and at.

This CFD model is based on Euler/Lagrange approach or time-dependent calculationso three-dimensional ow o water with macro gas bubbles and oil droplets. The water

ow was calculated on the Euler approach by solving a three-dimensional Reynolds-

averaged conservation equation (Navier Stokes equation) with a k-e turbulent model.

Two-way coupling was accounted or by adding dispersed phase source terms in all

conservation equations o the continuous phase. Gas bubble and oil droplet motion

was calculated by 15th Annual Produced Water Seminar Houston, 2005 pg. 2 solving

an equation o motion taking into account drag orce, pressure, lit orce and buoyancy.

TECHNICAL PAPER

D. Lee and A. A. Mohamad

1GLR Solutions Ltd.

Suite S, 1338 36th Ave. NE

Calgary, AB

Canada T2E 6T6.

Tel: +1.403.219.2210

CFD Simulation of a Vertical Induced

Static Flotation VesselReprint o paper presented at the 15th Annual Produced Water Society Seminar - League City, Texas - 2005


2/11

2www.exterran.com

For the case presented, the ollowing inlet conditions were established and remained

constant throughout:

Water Flow Rate = 600 m3/hr

Oil in Water Concentration = 300 ppm

Oil Viscosity = 0.0531 N.s/m2

Oil Specifc Gravity =0.96

Oil Droplet Size = we examined rom 3 m to 200 m

Water Temperature = 80 oC

Water density = 980 kg/m3

Gas Volume Flow Rate= 0.032 m3/s (68 cm)

Gas Bubble Size = 150-300 m

Gas Density =0.657 kg/m3

Gas Viscosity =1.12x10-5 kg/(m.s)

Figure 1 shows a schematic diagram o the vessel utilized or this particular model.

The ow enters the vessel through a vertically oriented nozzle located at bottom.

The resulting inlet water jet is confned within a vertically oriented cylindrical shell to

prevent short circuits with the vertically oriented outlet at the bottom o the vessel.

At the top o the vessel there is a skimming weir that runs the circumerence which

contains several outlet nozzles where skimmed oil can be taken out.

The physical dimensions o the vessel are 3.6 m diameter x 3.8 m height. The inlet and

outlet nozzle diameter is 0.254 m (10 in.). The jet confning cylinder dimension was 3.5

m in height and 3.05 m in diameter. Based on the ow conditions established, it was

determined that there was approximately a 4 minute retention time across the vessel.


3/11

3www.exterran.com

2.1 Basic Conservation Equations (Dispersed Phase):

The critical equations utilized within our model are defned below:

2.1.1 Momentum Equation:

The momentum equation or a droplet o mass md

is:

(1)

(2)

(3)

(4)

(5)

(6)

(7)

2.1.2 Mass Equation:

(8)

(9)

(10)


4/11

4www.exterran.com

3.0 SIMULATION RESULTS

Figure 2 shows trajectories o gas bubbles and oil droplets. It is clear that the jet ater

issuing rom the inlet nozzle expands and bubbles coalesce and disperse. The jet

expansion is controlled by the confning cylinder. The presence o the confning cylinder

is essential to prevent the gas bubbles and oil droplets direct ow to the outlet.

As the ow rises through the confning cylinder, we see airly even ow velocities about

1 m rom the jet inlet. However, we observe a preerential pathway or both gas

bubbles and oil droplets along the interior wall o the confning cylinder. Yet we see the

highest water velocities at the center o the confning cylinder.

As the water reaches the top o the confning cylinder, it is ree to ow in both a

horizontal and downward vertical axis. Since the outlet nozzle can not be at center o

the vessel it has been placed o-center on one side. As a result we observe the ow

becoming asymmetric and three dimensional.

Figure 3 shows velocity vectors at the mid-cross section o the vessel. There is evidence

o a donut-type vortex ormation around the inlet jet at the bottom o the vessel. A closer

look at ow patterns on other sections o the vessel are shown in Figure 4-6. The main

reason or ormation o the vortex is that the jet issues with high velocity and pressure

decreases around the jet which results o ow toward the jet uid entrainments.

Figures 7-9 show ow pattern at dierent horizontal sections. Figure 7 is a ow pattern

at the bottom o the vessel. Figure 8 shows the velocity vectors at the section near

the skimming outlets. Figure 9 shows velocity vectors at the section where the jet ow

leaves the confning cylinder. The ow becomes almost uniorm about one meter aterthe jet inlet. In act, when testing various inlet ow rates, it is ound that as the ow

rate decreases the ow becomes almost uniorm ater a short distance rom the jet

inlet. It is also observed that the confning cylinder enhances coalescence rates as

the cylinder walls increase the probability o bubble collision and result in rapid growth

in bubble size.

Figure 10 shows velocity vectors at the ree surace o the vessel. The radial ow

is clearly evident and we fnd the ow to be airly symmetrical. However, the ow

becomes asymmetric in the annular gap between the vessel and the jet confning

cylinder because the o-set location o the outlet nozzle at the bottom o the tank.

The results o this simulation reect the physics o ow under these specifc conditions.It is worthy to mention that a simulation o water ow without gas bubbles and/or oil

droplets is very dierent than the ow with gas bubbles and/or oil droplets. It is

intended to say that it is not appropriate to extrapolate data or simulation results

without gas bubbles or IGF/ISF simulation.


5/11

5www.exterran.com

Figure 1 Schematic diagram o the skim vessel. Figure 2 Gas bubbles and oil droplets

trajectories. Figure 3 Velocity vectors at mid-cross section o the vessel. Figure 4Velocity vectors at the middle section o the vessel. Figure 5 Right-hand section o

the vessel. Figure 6 Let-hand section o the vessel. Figure 7 Flow pattern at the

bottom o the vessel near the inlet and outlet. Figure 8 Velocity vectors near the

skimming outlets. Figure 9 Velocity vectors at section where the jet ow leaves the

confning cylinder.

Figure 1 Schematic diagram o the skim vessel.


6/11

6www.exterran.com

Figure 2 Gas bubbles and oil droplets trajectories.

Figure 3 Velocity vectors at mid-cross section o the vessel.


7/11

7www.exterran.com

Figure 4 Velocity vectors at the middle section o the vessel.

Figure 5 Right-hand section o the vessel.


8/11

8www.exterran.com

Figure 6 Let-hand section o the vessel.

Figure 7 Flow pattern at the bottom o the vessel near the inlet and outlet.


9/11

9www.exterran.com

Figure 8 Velocity vectors near the skimming outlets.

Figure 9 Velocity vectors at section where the jet ow leaves the confning cylinder.


10/11

10www.exterran.com

Figure 10 Velocity vectors at the ree surace o the vessel.

4.0 CONCLUSIONSThe results o this simulation show that it is possible to investigate the eect o

dierent controlling and operating parameters on the perormance and efciency o a

separation vessel.

The physics o the ow is quite complicated; hence it is impossible to determine the

precision o the resulting data in particular due to the number o modeling assumptions

and difculty in calculating the rate o coalescence. However, we eel confdent that as

a qualitative tool this is extremely useul or design and operation purposes.

The results clearly show the eectiveness o the jet confning cylinder or initial contact

o gas and oil droplets while preventing short circuiting to the outlet. The results

also indicate that the confning cylinder also acts to establish a desirable suraceow pattern or skimming o oil/roth. However, we see that the surace turbulence

allows a signifcant amount o water over the skimming weirs when skimming is set

up in a continuous operation. Thereore, it is advisable to operate this particular

vessel geometry with a batched skimming philosophy at timed intervals to allow or

oil accumulation. The model also shows that the walls o the confning cylinder act

to rapidly coalesce inlet gas bubbles which has the impact o increasing their rise

velocity and turbulence within the surrounding water. This is viewed as a negative

design aspect given the reduced probability o oil-bubble attachment and ultimately

lower oil removal efciency.


11/11

11www.exterran.com

1. GLR Solutions Ltd. was acquired by Ex terran Holdings Inc. in January, 2008

05PROD-WTS-09-03-EN-MR12-F1-US

Global Sales

16666 Northchase DriveHouston, Texas 77060

[email protected]

Canada Sales

1721 27th Ave NE

Calgary, Alberta Canada

T2E 7E1

[email protected]

Latin America Sales

Talcahuano 833 piso 11A

C1013AAQ

Buenos Aires, Argentina

[email protected]

Middle East Sales

P.O. Box 293509

East Wing 5B, 4th Floor

Dubai Airport Free Zone

Dubai, UAE

[email protected]

Due to the design o this vessel, we observe a very limited window or contact o

gas bubbles and oil droplets. When larger gas bubbles are entrained (>75 microns),essentially these two phases are only in intimate contact with each other during the

period o retention within the confning cylinder. Once these large bubbles reach the

surace they are released into the headspace and dont sink and there is no urther

contact or lit with the oil. The particle trajectories, observed in Figure 1, going down to

the outlet are predominantly those o oil alone as these larger gas bubbles cannot sink

at these water velocities. This is in contrast to much smaller bubbles (< 75 microns)

when they are modeled which are able to descend or a short distance prior to ully

coalescing and rising to the surace. These conclusions were impacted urther by the

walls o the confning cylinder which acts as a coalescing surace or the gas bubbles,

i.e. rapidly exaggerating their size because they are trapped in that zone ree to coalesce

on those walls. While coalescing eect cannot be eliminated, the perormance o oilremoval can be improved by entraining smaller bubbles into the inlet stream so that a

reasonable bubble size is maintained even ater wall coalescing eects.

The o set designed location o the outlet nozzle on this vessel creates an asymmetric

and three dimensional ow patterns. The result is a higher velocity vortex down one

side o the vessel. We observe that smaller oil droplets are reely caught in this

higher velocity zone and become ree to leave in the outlet water. Better design can

be achieved by replacing the single outlet o the vessel with multiple outlets evenly

distributed at the bottom. This conclusion becomes more signifcant as the droplet

size o the oil and bubbles decreases.

As a summary, we fnd this CFD modeling gives clear insights o ow distributionwhich helps in better understanding optimal design and operation conditions or

this particular vessel under these specifc conditions. Our intent is not to criticize

this design but rather to demonstrate the value o using CFD as a support to the

engineering design process or complicated multi phase conditions.

The model developed is quite exible and can simulate a wide range o inlet conditions

and ow rates. We would like to point out that dierent inlet conditions may result in

alternative conclusions.

Documents

CFD Simulation of a Vertical Induced Static Flotation Vessel English Letter