Experimental study of flow patterns and pressure drops of heavy oil-water-gas vertical flow*

LIU Xi-mao (劉曦懋)

International Ltd., Chuanqing Drilling Engineering Co. Ltd., CNPC, Chengdu 610051, China,

E-mail:simonliu2007@sina.com

ZHONG Hai-quan (鐘海全), LI Ying-chuan (李穎川), LIU Zhong-neng (劉忠能)

State Key Laboratory of Reservoir Geology and Exploitation, Southwest Petroleum University, Chengdu 610500, China

WANG Qi (王琦)

School of Petroleum Engineering, Chongqing University of Science and Technology, Chongqing 401331, China

Experimental study of flow patterns and pressure drops of heavy oil-water-gas vertical flow*

LIU Xi-mao (劉曦懋)

International Ltd., Chuanqing Drilling Engineering Co. Ltd., CNPC, Chengdu 610051, China,

E-mail:simonliu2007@sina.com

ZHONG Hai-quan (鐘海全), LI Ying-chuan (李穎川), LIU Zhong-neng (劉忠能)

State Key Laboratory of Reservoir Geology and Exploitation, Southwest Petroleum University, Chengdu 610500, China

WANG Qi (王琦)

School of Petroleum Engineering, Chongqing University of Science and Technology, Chongqing 401331, China

(Received Janaury 17, 2013, Revised August 6, 2013)

A stainless steel apparatus of 18.5 m high and 0.05 m in inner diameter is developed, with the heavy oil from Lukeqin Xinjiang oil field as the test medium, to carry out the orthogonal experiments for the interactions between heavy oil-water and heavy oil-water-gas. With the aid of observation windows, the pressure drop signal can be collected and the general multiple flow patterns of heavy oil-water-gas can be observed, including the bubble, slug, churn and annular ones. Compared with the conventional oil, the bubble flows are identified in three specific flow patterns which are the dispersed bubble (DB), the bubble gas-bubble heavy oil (Bg-Bo), and the bubble gas-intermittent heavy oil (Bg-Io). The slug flows are identified in two specific flow patterns which are the intermittent gas-bubble heavy oil(Ig-Bo)and the intermittent gas-intermittent heavy oil (Ig-Io). Compared with the observations in the heavy oil-water experiment, it is found that the conventional models can not accurately predict the pressure gradient. And it is not water but heavy oil and water mixed phase that is in contact with the tube wall. So, based on the principle of the energy conservation and the kinematic wave theory, a new method is proposed to calculate the frictional pressure gradient. Furthermore, with the new friction gradient calculation method and a due consideration of the flow characteristics of the heavy oil-water-gas high speed flow, a new model is built to predict the heavy oil-water-gas pressure gradient. The predictions are compared with the experiment data and the field data. The accuracy of the predictions shows the rationality and the applicability of the new model.

heavy oil, multiphase flow, experiment, flow patterns, gradient prediction

Introduction

The oil-water-gas multiphase flow is a fundamental issue for an efficient recovery of oil and gas. The conventional oil as the main energy resource is quickly exhausted with its limited reserves. Consequently, the heavy oil as an unconventional resource is paid a wide attention both at home and abroad.

Multiphase flow experiments were extensively carried out in the past few decades, but most of them are conducted for the conventional oil of low viscosity. In China, heavy oil-water-gas multiphase vertical flow experiments were not paid adequate attention. José[1]developed a “core-annular flow” model to predict the heavy oil-water vertical flow based on the heavy oilwater laboratory experiments. Then Barnnward[2]presented a flow pattern map for different heavy oilwater superficial velocities in the vertical flow. Later he discussed the applicability of the conventional gasliquid models for the heavy oil multiphase vertical flow, but they all fail to predict the total pressure gradient[3,4]. Schmidt et al.[5]used polyvinylpyrrolidone (PVP) and water to simulate different high-viscosityfluids to obtain void fraction correlations, and it is found that the exiting methods could not predict the mean value properly. Two new correlations were presented. Wei and Gong[6]in China reported the structure change and the transition characteristics of the high-viscosity oil/water horizontal flow, and they presented the transition boundary for each flow pattern. Zhang et al.[7-13]studied the high viscosity oil-watergas flow. In their experiment ranges, the slug flow is generally the only vertical flow pattern, although an early unified model is found applicable to the test according to their experiment.

The experiments mentioned above are only focused on the flow pattern distribution or the frictional gradient of the heavy oil-water flow or the flow patterns of the heavy oil-water-gas horizontal flow. In China there are few published papers to discuss the heavy oil-water-gas three-phase flow patterns and the pressure gradient. In this paper, a stainless steel apparatus of 18.5 m high and 0.05 m in inner diameter is developed, and with the high-viscosity heavy oil from oil fields as the test medium, over 600 groups of data are obtained for the study of the heavy oil-water flow patterns and the pressure gradient changes. A new model is proposed to predict the heavy oil-water-gas pressure gradient.

1. Experiments

1.1 Experimental apparatus

As shown in Fig.1 the experiment is carried out in an inverse-U shape apparatus with the height of 18.5 m and the inner diameter of 0.05 m, wrapped with the heat insulation material. To observe the flow patterns, two tempered-glass observation windows of 0.7 m high are installed in the pipe system at the height of 10 m and 16.7 m from the ground, respectively. Between the windows is the measurement section of 6 m long to measure the fluid pressure change, with two differential pressure sensors and two pressure sensors. In the ground of the apparatus system, there are the mass flow meter, the oil-gas-water separation tank, the oil-water experimental close loop heating and injection tank, and the oil-gas-water experimental open heating mixing tank. The data of pressure, pressure drop, gas injection rate and oil/water flow rate are sent through a data transmission system to the computer terminal and the heavy oil viscosity data is shown in Table 1.

1.2 Test procedure

In the heavy oil-water experiment, the heavy oil and the water are first put into the heavy oil-waterheating and injection tank in the required proportions and heated to the desired temperature. Then by using an air compressor, the air is pushed into the test loop, while the heating system turns on. Thirdly, the mixture circulates thoroughly to make sure that the fluid is heated to the desired temperature and the thermal state is stable. Finally, the data transmission system is turned on to start the experiment.

In the gas-water experiment, the valves are first adjusted to let the heavy oil-water-gas heating and injection tank linked to the pipeline, instead of the heavy oil-water heating and injection tank. Secondly, the mixture of oil and water is heated and stirred to the required temperature, and the thermal state is made stable, while the oil-water mixture is injected into the experimental circulation line by the gas compressor. Thirdly, the data transmission system is turned on to start the experiment.

The heavy oil-water orthogonal tests are carried out in 25 groups and 432 sets of data are collected, as shown in Table 2. The heavy oil-water-gas orthogonal tests are carried out in 12 groups and over 160 sets of data are collected, as shown in Table 3.

1.3 Flow patterns

1.3.1 Heavy oil-water flow patterns

From the experimental observation, at a relatively low heavy oil-water mixture superficial velocity, the “core annular” flow pattern is observed, in which the water circulates and the heavy oil flows along the tube, and the heavy oil in the core is in a bamboo shape, as shown in Fig.2 (fw=40%,Vm=0.14 m/s andT=30oC, 40oC, 50oC, 60oC). However, other phenomenon is observed while the mixture superficial velocity rises. It is the mixture that flows in the vertical pipe line, instead of the “core annular” flow. And it is not only the water, but also the heavy oil-water mixture that is in contact with the wall. The flow patterns of the heavy oil-water in different water cuts and different mixture superficial velocities are shown in Fig.3. From the tests it can be seen that the “core annular” flow happens in a small range. That is when the superficial velocity is relatively small, and it tends to occur more often while the heavy oil ratio becomes larger. However, the heavy oil-water mix flow dominates the main flow patterns, which affects the frictional gradient greatly.

1.3.2 Heavy oil-water-gas flow patterns

Similar flow patterns occur between the heavyoil multiphase flow and the conventional gas-liquid flow, along with different ones. Specific flow patterns are as follows:

(1) Bubble flow

From the experimental observations, the heavy oil dispersed bubble flow (DB) is identified with the gas distributing as small bubbles in the mixture of the heavy oil and water, the bubble gas-bubble heavy oil flow (Bg-Bo)is identified with the heavy oil granularly distributing in the continuous aqueous phase andwith the gas as bubbles; and the bubble gas-intermittent heavy oil flow (Bg-Io)is identified with the heavy oil phase being elongated intermittently in the continuous aqueous phase and with the gas as bubbles, as shown in Fig.4.

(2) Slug flow

From the experimental observations, the intermittent gas-bubble heavy oil flow (Ig-Bo)is identified with the heavy oil granularly distributing in the aqueous phase and with the gas as slugs, and the intermittent gas-intermittent heavy oil(Ig-Io)is identified with the heavy oil distributing as a non-continuous phase in the aqueous phase and with the gas as slugs as shown in Fig.5.

(3) Churn transitional flow

When the mixed phase velocity is large, the Taylor bubble is unstable and will rupture. The continuous liquid phase appears in an irregular state, and the heavy oil is granularly or massively distributing in the liquid phase as shown in Fig.6.

(4) Annular flow

The gas phase is a continuous phase distributing in the central tube, and the heavy oil-water mixture is around the tube wall as shown in Fig.7.

Testing results of the heavy oil-gas-water flow pattern map is shown in Fig.8 referring as the Aziz dimensionless plate.

2. Pressure gradient and discussions

2.1 Frictional gradient analysis

When the miscible flow velocity is large, the heavy oil-water mixed phase moves along the vertical tube, and it is not the water, but the mixture that is in contact with the tube wall. The accuracy of the frictional gradient ΔPfprediction is affected by the flow patterns. And the difference is mainly reflected on the value of the fluid viscosityμ.

As the Reynolds number is defined as

whereRe is the Reynolds number,ρis the density of the fluid,vis velocity,dis the pipe diameter andμis the viscosity.

Many conventional liquid-gas pressure gradient models are based on the Moody Chart to determine the frictional gradient. In heavy oil-water experiments, due to the high viscosity, the value of the Reynolds number will be significantly reduced and goes beyond the scope of the Moody Chart prediction. If the calculation is made according to the Moody Chart fitting formula from Nikurades, Colebrook or Zigrang and Sylvester etc., the value range of the coefficient friction will exceed the Moody Chart’s forecast range, which leads to a large error between the predicted value and the measured value.

The kinematic wave theory is applied to analyze the interface wave velocity between the viscous oil and the water, to provide predictions in accordance with the experimental data. Based on the kinematic wave theory, a model of the total frictional pressure drop ΔPfcan be built composed of the flow friction ΔPfhand the buoyancy term ΔPb, where

in which ρwis the density of the water,ρois the density of the heavy oil,g is the gravitational acceleration,fwis the water cut,mis the viscosity ratio between the heavy oil and the water,f(fw,m)is expressed as

and the flow friction term ΔPfhis defined as

a,nandkare the parameters to be determined from the experiments,μwis the viscosity of the water, vmis the superficial velocity of the mixture phase.

Based on the core-annular flow, José[1]proposed a model using the water phase viscosityμwto calculate the miscible viscosityμ, and to obtain relevant values ofa,n,k . However, as mentioned above, the experimental observation shows that it is not the case. When the heavy oil-water mixture superficial velocity is large, the test data do not agree with the model results.

Assuming that the mixture flows close to the wall of the vertical pipe, a new heavy oil-water frictional pressure gradient model can be established, where μlis used to calculate the ΔPfof the mixture instead of μw. And the mixture physical property parametersa, nandkare redetermined with the aid of the MatLab software, and the correlation degree is 0.9408, as shown in Fig.9.

The new frictional model is expressed as

2.2 Heavy oil-water-gas gradient

In the heavy oil-water-gas experiments, the mixture’s pressure gradient is sensitive to the velocity change when the total velocity is relatively large. It is found that near the annular flow, the total pressure gradient decreases with the increase of the superficial velocity. Further calculations indicate that the frictional gradient decreases dramatically, implying that the structure of the mixture changes with the high speed.

For the heavy oil-water-gas vertical flow, the total pressure gradient is well expressed[14,15]as

Therefore,μwis considered to be the parameter for the pressure gradient calculation instead of μlin the annular flow, to reflect the actual fluid property of the heavy oil-gas-water high-speed miscible flow.

For predicting the heavy oil-water-gas flow,ΔPfis computed by using the new frictional model mentioned above in Eq.(7), and Aziz, HK shows a great agreement with the experimental data when calculating ΔPgand ΔPa. And the results of each flow pattern’s minimum error are shown in Table 4.

2.3 Evaluation of the new model

2.3.1 Evaluation based on flow patterns

There are generally four flow patterns observed in the heavy oil water gas flow: the bubble, the slug, the churn and the annular ones. The gas/liquid ratio, the liquid superficial velocity and the relevant parameters are shown in Table 5.

From over 160 sets of data, every flow pattern’s proportion is shown in Table 6, and both the average error and the absolute error of each flow patterns are also shown.

It can be seen that for the bubble, slug and annular flows, the new model give predictions in good agreement with the test data. As for the churn flow, due to the different gas/liquid ratios and superficial velocities, the mixture structure is complicated, where some part of the mixture is the continuous heavy oil phase, and some part is the continuous water phase. Though the error of the prediction in the churn case is a little higher than others, it is still acceptable.

2.3.2 Evaluation based on water cut

The water cuts are mainly in four categories, that is, 30%, 50%, 70%, 90%. To evaluate the new model, the data proportion, the average error and the absolute error of each water cut are shown in Table 7.

From the results, it can be seen that, for the 30% water cut, the new model’s prediction is within 7% in the absolute error, which is close to the actual data. For the 50% and 90% water cuts, the absolute error is around 8%, in which the new model shows a high accuracy in a wide range of water cuts. In the 70% water cut, the mixture structure is easy to change under different flow rates, and the pressure gradient varies within a larger range, so the error is slightly higher.

2.3.3 Evaluation based on total test data

There are over 100 sets of experimental data. All test data are compared with the new model’s prediction, as shown in Fig.10. In both sides of the diagonal line show the error limits of minus/plus 20%. And the average error of the total test data is 9.84%.

2.3.4 Evaluation based on field data

The new model of the heavy oil water gas flow is also applied to the field production wells. The field data of the heavy oil wells are from the Tahe oil field in Xinjiang. 15 sets of data of the heavy oil wells are collected from the Tahe oil field and the related parameters are shown in Table 8.

The predicted pressure gradients and those from the field heavy oil well data are compared in Fig.11. The average error of the new model is 5.65%, and at the same time, the absolute error of the new model is 7.97%.

3. Conclusions

A stainless steel apparatus of 18.5 m high and 0.05 m in inner diameter is built with observation windows and data collecting and transmission system, for the tests of the high viscosity heavy oil from Xinjiang oil field. Orthogonal experiments are carried out for both the heavy oil-water flow and the heavy oil-watergas flow, and over 600 sets of data are obtained in all. The following conclusions can be drawn:

(1) For the heavy oil-water flow, the core-annular flow is observed at a relatively low superficial velocity. Once the velocity gets higher, that flow pattern disappears, and the heavy oil and water mixture becomes the dominating flow patterns. The distributions of the flow patterns are obtained.

(2) General multiple flow patterns of the heavy oil-water-gas are observed, which are the bubble, the slug, the churn and the annular patterns. Compared with the conventional oil, the bubble flows are identified in three specific flow patterns which are the dispersed bubble (DB), the bubble gas-bubble heavy oil (Bg-Bo), the bubble gas-intermittent heavy oil (Bg-Io), and the slug flows are identified in two specific flow patterns which are the intermittent gas-bubble heavy oil(Ig-Bo)and the intermittent gas-intermittent heavy oil(Ig-Io).

(3) According to experimental tests and observations, the mixture of the heavy oil and water flows close to the wall. Based on the kinematic wave theory, a new method is developed to predict the heavy oil water pressure gradient. Furthermore, a new pressure gradient model for the heavy oil-water-gas vertical flow is established.

(4) The prediction accuracy of the new model is shown by comparing with experiment data, in different flow patterns, water cuts and the total test data. The prediction of the new model is also shown by comparing with the field data, and both comparisons show rationality and applicability of the present model.

[1] JOSé W., VANEGAS P. and BARNNWART A. C. Modeling of vertical core-annular flows and application to heavy oil production[J]. Energy Resources Technology, 2001, 123(3): 194-199.

[2] BARNNWART A. C., RODRIGUEZ M. H C. and De CARVALHO H. M. et al. Flow patterns in heavy crude oil-water flow[J]. Energy Resources Technology, 2004, 126(3): 184-189.

[3] BANNWART A. C., CARVALHO C. H. M. and OLIVEIRA A. P. Water-assisted flow of heavy oil and gas in a vertical pipe[C]. SPE/PS-CIM/CHOA International Thermal Operations and Heavy Oil Symposium. Calgary, Alberta, Canada, 2005.

[4] BANNWART A. C. Experimental investigation on liquid-liquid-gas flow: Flow patterns and pressure-gradient[J]. Journal of Petroleum Science and Engineering, 2009, 65(1-2): 1-13.

[5] SCHMIDT J., GIESBRECHT H. and Van Der GELD C. W. M. Phase and velocity distributions in vertically upward high-viscosity two-phase flow[J]. International Journal of Multiphase Flow, 2008, 34(4): 363-374.

[6] WANG Wei, GONG Jing. Flow regimes and transition characters of the high viscosity oil-water two phase flow[C]. International Oil and Gas Conference and Exhibition in China. Beijing, China, 2010.

[7] ZHANG H. Q., SARICA C. Unified modeling of gasoil-water pipe flow-basic approaches and preliminary validation[J]. SPE Project Facilities Construction, 2006, 1(2): 1-7.

[8] ZHANG H. Q. Identification and classification of new three-phase gas/oil/water flow patterns. Cengizhan Keskin[C]. SPE Annual Technical Conference and Exhibition. Anaheim, California, USA, 2007.

[9] WANG S., ZHANG H. Q. and SARICA C. Experimental study of high viscosity oil/water/gas three phase flow in horizontal and upward vertical pipes[C]. Offshore Technology Conference. Houston, Texas, USA, 2012.

[10] VUONG D. H., ZHANG H. Q. and LI M. Experimental study on high viscosity oil/water flow in horizontal and vertical pipes[C]. SPE Annual Technical Conference and Exhibition. New Orleans, Louisiana, USA, 2009.

[11] AKHIYAROV D. T., ZHANG H. Q. and SARICA C. High-viscosity oil-gas flow in vertical pipe[C]. Offshore Technology Conference. Houston, Texas, USA, 2010.

[12] ZHANG H. Q., VUONG D. H. and SARICA C. Modeling high viscosity oil/water flows in horizontal and vertical pipes[C]. SPE Annual Technical Conference and Exhibition. Florence, Italy, 2010.

[13] ZHANG H. Q., SARICA C. A model for wetted wall fraction and gravity center of liquid film in gas-liquid pipe flow[J]. SPE Journal, 2011, 16(3): 692-697.

[14] CHEN Jia-lang, CHEN Tao-ping. Petroleum engineering gas-liquid two-phase flow in pipelines[M]. Beijing, China: Petroleum Industry Press, 2010, 48-59(in Chinese).

[15] LI Ying-chuan. Petroleum production engineering[M]. Beijing, China: Petroleum Industry Press, 2009, 17-35(in Chinese).

10.1016/S1001-6058(14)60071-8

* Project supported by the National Science and Technology Major Project (Grant No. 2008ZX05049-004-006HZ).

Biography: LIU Xi-mao (1989-), Male, Ph. D. Candidate

ZHONG Hai-quan,

E-mail: swpuzhhq@126.com