Focused-beam reflectance method aids hydrate blockage prediction

Jan. 7, 2013
Examining the variance of hydrate particle size with focused-beam reflectance measurements (FBRM) yields a clearer understanding of the microscopic mechanisms guiding hydrate formation and plugging, with the square-weighted chord length method emerging as capable of predicting hydrate plugging fairly well.

Xiaofang Lv
Jing Gong
Wenqing Li
Yixuan Tang
China University of Petroleum
Beijing

Jiankui Zhao
China National Oil and Gas Exploration
and Development Corp.
Beijing

Examining the variance of hydrate particle size with focused-beam reflectance measurements (FBRM) yields a clearer understanding of the microscopic mechanisms guiding hydrate formation and plugging, with the square-weighted chord length method emerging as capable of predicting hydrate plugging fairly well.

Reducing pressure or raising temperature in a subsea natural gas pipeline can inhibit hydrate formation. If the flow rate is low enough, however, a natural gas pipeline with a low water cut (~10%) may still risk hydrate blockage.

A minimum safe flow rate exists under given operating pressure, ambient temperature, and antiagglomerate (AA) dosage. A lower flow rate will lead to hydrate blockage. If the flow rate exceeds this value, the hydrate slurry can maintain its mobility despite hydrate formation.

Higher system pressure, however, accelerates hydrate plugging. Hydrate formation temperature and blocking temperature both increase with pressure, while the time from formation to blockage decreases.

Heating an entire subsea pipeline is not feasible. But operators can pig and perform dehydration treatments before transporting gas. If a blockage does occur, neither increasing inlet pressure nor pigging are recommended solutions. Both can lead to increased damage. Decompressing the pipeline at its outlet, however, can help clear the blockage.

Background

A hydrate blockage took place offshore China in a subsea natural gas pipeline operating between the QK18-1 central platform and the Boxi onshore natural gas processing plant in southwestern Bohai. The 48-km (30-mile) pipeline was carrying wet gas at a flow rate of about 9,500 cu m/hr at the time of the blockage. Its operating temperature ranged between 5° and 45° C., and operating pressure between 15 and 21 bar (Fig. 1).

On the date of the blockage inlet pressure reached as high as 2.8 MPa, vs. a normal 2.1 MPa, and flow rate dropped to zero. Adjusting inlet and outlet pressure and injecting methanol from both upstream and the downstream directions allowed resumption of minimal gas flows. But transient flow rate fluctuated between 0 and 10,000 cu m/hr, greatly impairing operations.

On site analysis led to three possible reasons for the inlet pressure increase:

• Hydrate blockages in the subsea pipeline.

• Liquid accumulation large enough to cause a pressure drop.

• Pig blockage.

Pigging records showed no lost pigs but a great amount of liquid discharge at the processing plant. Analysis of the liquid found both gas condensate and free water, reminiscent of problems encountered years earlier in the West-East Gas Transmission pipeline (OGJ, June 22, 2009, p. 52). Gas hydrates emerged as the likely cause of the blockage, having formed under the pipeline's high pressure-low temperature operating environment, collided, aggregated, stuck together, and finally deposited to form a plug.

China National Offshore Oil Co. (CNOOC) and China University of Petroleum-Beijing (CUPB) carried out theoretical and experimental research on gas hydrate formation and slurry flow to investigate the hydrate formation mechanism leading to the blockage. This article provides reasonable hydrate management strategies based on this research, including the use of AA.

Gas sample

Three gas samples taken at the gas pipeline's inlet on platform QK18-1 defined the phase distribution of gas in the pipeline. Table 1 shows composition of the gas samples. Using the equation of state1 yields the phase envelope of gas samples and verifies part of the experimental data's calculated values.

Calculation results agreed with the experimental results from Fig. 2, allowing use of the phase envelope to predict the phase state of gas in the pipeline under various operating conditions. Pipeline operating conditions (temperature: 5° C.-45° C.; pressure: 1.5 MPa-2.1 MPa) in the two-phase region produced liquid hydrocarbons in the pipe, validated by the sample liquid.

Free water found in the liquid sample would form hydrate under the proper thermodynamic conditions. The pressure search method (OGJ, June 22, 2009, p. 52) applied to a sapphire autoclave determined points of hydrate formation.

A two-step hydrate formation mechanism proposed by Chen and Guo2 3 predicted the natural gas hydrate formation curve. Formation of a stoichiometric basic hydrate through a quasichemical reaction comprised the first step. The second step consisted of adsorption of gas molecules into the empty linked cavities of basic hydrate, leading to nonstoichiometric properties.

Equations 1-3 predict the conditions for hydrate formation in the Chen-Guo model.

Literatures guided evaluation of Antoine constants Ai', Bi', Ci', and binary interaction coefficients, Aij.2 Equation 5 formulates the Langmuir constant, Cj, of component j.

Fig. 3 shows the simulation results and experimental results in good agreement, allowing the model to be used to predict hydrate formation conditions in the subsea gas pipeline. Commercial software PIPESIM simulated pressure and temperature along the pipeline. Applying the Chen-Guo model then allowed calculating the hydrate formation pressure value under the simulated temperature value of every pipeline section (Fig. 4).

Results showed most of the simulated pressures as above those of hydrate formation, prompting the formation of hydrate with any free water in the pipeline and making it reasonable to ascribe the blockage to hydrate formation.

Experimental loop

Researchers used a high-pressure hydrate experimental loop devoted to flow assurance studies to investigate the hydrate's formation further. The loop separately injects gas and liquid by a plunger compressor (2,200 cu m/hr) and uses a custom-made magnetic centrifugal pump to maintain a flow rate of up to 12 cu m/hr.

Two sight glasses sit in the test sections. The gas injection point is the test section inlet. At the outlet of the test section gas and liquid collect in an insulated separator and are re-directed toward the test section compressor and pump. Several tanks allow maintenance of loop and separator pressure as hydrate forms.

The 30-m stainless steel test section consisted of two rectilinear horizontal lengths joined together to form a pipe with 1-in. OD. A 2-in. diameter jacket circulating a water-glycol blend surrounded the test section. Process temperature control ranged –20° to 80° C.

Thermocouples lay along the pipe, inside the separator, inside the water-glycol system, and on the different gas utilities. A Coriolis flowmeter measured liquid mixture density and flow rate. Two FM1000 gamma ray densitometers were also available to measure the multiphase fluid's mean density. Differential pressure sensors along the loop followed evolution of the linear pressure drop. Rapid data acquisition permitted detection of quickly occurring events.

A focused-beam reflectance measurements (FBRM) probe allowed monitoring evolution of objects—droplets, bubbles, solid particles—carried inside the flow. The FBRM consisted a low-intensity rotating laser beam reflected when intercepting a particle.4 5 Measuring the reflection time allows deduction of a chord length.

Assignment of a chord-length distribution (CLD) and a mean chord length followed every measurement duration equal to 10 sec. The CLD gives an idea of particle size distribution (PSD) of objects carried by the flow. A representative sampling of the particle size distribution recommended installation of the particle size analyzer on the straight vertical pipe ahead of the experimental loop's inlet. The analyzer's probe window cuts the stream lines at a 45° angle beginning at the center of the pipe.

The FBRM probe estimated initial water droplet (Dp) size inside the fluid and followed the hydrate particles' agglomeration with time. The mean square-weighted chord length gives more weight to the longer chord length and is particularly well adapted to agglomeration phenomena. Equation 6 gives the mathematical expression of the square-weighted mean chord lengths.6

Fluids

Testing used deionized water, civil natural gas, and diesel (Tables 2-3). Combined Span20 served as the AA agent. The 77.78 l. of test liquid loaded in the loop had a 10% water cut. An electronic balance weighed AA quantity, with measuring error ±0.01, and a high-pressure piston pump adjusted concentration of AA in the water phase to 0 wt %, 1 wt %, 2 wt %, and 3 wt %. The preliminary Chen-Guo model determined the curve of hydrate formation (Fig. 5) for the defined natural gas composition.

Experiment

Cooling experiments occurred under constant pressure in accordance with actual operating conditions as much as possible. Researchers first investigated the blockage process of oil-water emulsion without AA. Experiments with different AA doses followed to test the feasibility of adding them to the transport stream followed. Contrasting the two series of experiments allowed understanding of the hydrate formation and plugging mechanism and investigation of AA use.

Results

Experiments used the same initial flow rate but different pressures: 3.2 and 4.1 MPa (Figs. 6-7).

In Fig. 6, as the relative time proceeded, temperature decreased to the hydrate formation equilibrium point, Te (7.4° C.), then reached the hydrate formation point, Tf (4.4° C.), and finally the hydrate plugging point, Tb (2.9° C.). The rate of temperature drop was sharp from Te to Tf and remained constant from Tf to Tb. Flow rate and density also dropped sharply at first then more moderately in the end.

The large amount of heat released by initial crystallization of the hydrate particles led to significantly increased fluid viscosity and reduced fluid flow rate. As hydrate volume fraction grew, hydrate formation rate gradually slowed and finally became constant, eventually plugging the pipe.

Time duration from Te to Tf, induction time, measured about 1.37 hr. Subcooling (Tc), the temperature gap between Te and Tf, was about 3° C.,7-9 and the plugging time from beginning to Tb about 11.4 hr. Fig. 7 shows identical results, with the differences between it and Fig. 6 listed in Table 4.

The quantitative comparison in Table 4 shows the relative time of hydrate blockage is related to pressure. Hydrate formation temperature, Tf, and plugging temperature, Tb, rose with pressure, but the relative time between them decreased with pressure, making hydrate blockage easier.

Without AA, hydrate particle formation significantly increased fluid viscosity and flow resistance, with fewer solid particles entrained in the flow. More hydrate particles deposited in the pipeline, blocking the flow channel and further reducing flow rate until eventually leading to a hydrate plug.

AA influence

Study of the influence of AA dosages on hydrate plugging also occurred under constant conditions, with the same oil-water emulsion, a 4.1-MPa system pressure, a 20-hz initial pump speed, and an 18° C. temperature. Figs. 8 and 9 show the results.

All cooling rates of the water-oil emulsion are basically the same before Tf point (Fig. 8). Once past Tf, the temperature of emulsion without AA declined more slowly, while temperature rose at first then slowly dropped in emulsions with 1 wt % and 2 wt % AA.

The water phase could be well dispersed into the oil phase after AA addition, magnifying the water-oil (water-natural gas) contact areas, leading to more hydrate formation, and a higher water conversion ratio and hydrate volume fraction in the slurry. A large amount of heat escaped during hydrate crystallization, not only compensating for the cooling effect from the water bath jacket, but also increasing slurry temperature in a short time.

Fig. 8 also shows the three relative times at Tf as identical, differing from the formation temperature. Specifically, Tf = 5.5° C. (1 wt % AA) and Tf = 5.6° C. (2 wt % AA) were both less than Tf = 6.3° C. (without AA), meaning the hydrate induction time was less affected by the AA dosage than by forces like pressure. AA addition could, however, slightly improve the subcooling of hydrate formation, promoting hydrate formation to some extent.

The density curves with 1 wt % and 2 wt % did not show the v-shaped trend in hydrate formation seen in the samples without AA (Fig. 9). The well-dispersed water-oil emulsion following AA's addition prevented large-scale hydrate aggregation and the associated sudden boost in liquid volume. Formed hydrate particles instead scattered in the oil, the fluid density drop, or gradually increased liquid volume, stemming from the continuous transition of water droplets to hydrate particles.

Fig. 9 also shows the plugging time with AA as more rapid than without. Adding AA augments hydrate formation, raising the slurry's viscosity and accelerating hydrate plugging at the same initial pump speed.

The phenomena displayed in Fig. 9 do not, however, repudiate the flow assurance benefits of adding AA. A "minimum safe flow rate" concept, however, must be defined. When the initial pump speed is less than this minimum value, adding AA prompts earlier plugging. But the numerous hydrate particles will maintain good fluidity in the slurry even with a higher flow-shear rate as long as the initial pump speed exceeds the minimum value. The definition "minimum safe flow rate" can also apply to operating conditions without AA, even with the less uniform distribution and aggregation of hydrate particles in the slurry.

Process analysis

Analysis of particle chord changes in hydrate formation and plugging happened under these experimental conditions: system pressure, 4.1 MPa; initial pump speed, 20 hz; AA dosage, 1 wt %.

Fig. 10 shows little disparity in the chord length distributions (CLD) collected by FBRM at each time interval, the majority of particles measuring 5-20 μm. Similar results emerged with 2 wt % AA dosage. These results suggest that hydrate particles convert directly from water droplets, which could be described with the shell model and is consistent with studies by Turner, et al.10 Water interfacial tension, oil properties, shear rate, water cut, pressure, and temperature might also influence CLD scope.

Fig. 11 shows the variance of larger particle sizes (square-weighted mean chord) tracked by FBRM during hydrate formation, a significant characteristic of the hydrate plugging accident. The square-weighted mean chord measured about 55 μm, fell quickly, then soared and kept stable for a time at about 52 μm. The square-weighted mean chord plummeted on restart following the blockage, demonstrating that the larger particle size changed significantly as a result of particle aggregation in hydrate plugging.

The following procedure can interpret the square-weighted mean chord decline from A to B. During hydrate formation a layer of hard hydrate shell grows on the surface of water droplets and is smashed into smaller hydrates pieces soon afterwards by the high pump shear rate. The soar-up after point B, meanwhile, depended on aggregation and flocculation between hydrate particles during the plugging.

Phenomena observed before the hydrate formation also confirm this process. The water-oil emulsion was flowing before the hydrate formation. A small amount of white powder-like particles emerged at formation. Finally, a lot of white granular solid attached to the inner wall of the pipe during blockage.

Feasibility study

Hydrate plugging occurred in experiments of varying pressure and AA dosage but constant water cut (10%) and initial pump speed (20 hz). Adding AA accelerated time-to-plugging (Tb). But increasing pump speed improved slurry fluidity. A minimum safe flow rate (Qin) should, therefore, exist given certain water cut, pressure, and ambient temperature guidelines, and with or without AA. A flow rate lower than Qin will lead to hydrate blockage, but hydrate slurry will remain in motion at flow rates higher than Qin.

A slurry with 10% water cut, 4.1 MPa pressure, and 3 wt % AA, still leads to plugging at an initial pump speed of 35 hz (1,684 kg/hr), while the same slurry maintains a constant flow rate at 40 hz (1,760 kg/hr). Fig. 12 illustrates these using temperature and flow rate curves during hydrate formation.

The temperature began to rise while flow rate began to decline at hydrate formation. But the temperature remained steady when the hydrate phase reached equilibrium curve at a flow rate around 523 kg/hr and no plugging of the pipeline occurred.

Preventing hydrate blockage at fixed parameters of cooling, system pressure, water cut, AA dosage, etc., requires determining the critical minimum safe flow rate (Qin). But reducing either the system temperature or flow rate could break this equilibrium, causing blockage.

References

1. Peng, D.Y., and Robinson, D.B., "A New Two-Constant Equation of State," Industrial & Engineering Chemistry Fundamentals, Vol. 15, No. 1, pp. 59-64, February 1976.

2. Chen, G.J., and Guo, T.M., "A New Approach to Gas Hydrate Modeling," Chemical Engineering Journal, Vol. 71, No. 2, pp. 145-51, Dec. 2, 1998.

3. Chen, G.J., and Guo, T.M., "Thermodynamic Modeling of Hydrate Formation Based on New Concepts," Fluid Phase Equilibria, Vol. 112, No. 1, pp. 43-65, November 1995.

4. Pauchard, V., Darbouret, M., and Palermo, T., "Gas Hydrate Slurry Flow in a Black Oil. Prediction of Gas Hydrate Particles Agglomeration and Linear Pressure Drop," 13th International Conference on Multiphase Production Technology, Edinburgh, June 13-15, 2007.

5. Boxall, J., Greaves, D., Mulligan, J., Koh, C., and Sloan, E.D., "Gas Hydrate Formation and Dissociation from Water-in-oil Emulsions Studied Using PVM and FBRM Particle Size Analysis," International Conference on Gas Hydrate, Vancouver, July 6-10, 2008.

6. Darbouret, M., Le Ba, H., Cameirao, A., Herri, J.M., Peytavy, J.L., and Glenat, P., "Lab Scale and Pilot Scale Comparison of Crystallization of Hydrate Slurries from a Water-in-oil Emulsion Using Chord Length Measurements," International Conference on Gas Hydrate, Vancouver, July 6-10, 2008.

7. Sun, C.Y., Chen, G.J., and Yue, G.L., "The Induction Period of Hydrate Formation in a Flow System," Chinese Journal of Chemical Engineering, Vol. 12, No. 4, pp. 527-31, August 2004.

8. Skovoborg, P., Ng, H.J., and Rasmussen, P., "Measurement of Induction Times for the Formation of Methane and Ethane Gas Hydrate," Chemical Engineering Science, Vol. 48, No. 3, pp. 445-53, February 1993.

9. Natarajan, V., Bishnoi, P.R., and Kalogerakis, N.," Induction Phenomena in Gas Hydrate Nucleation," Chemical Engineering Science, Vol. 49, No. 13, pp. 2,075-87, July 1994.

10. Turner, D.J., Kleehammer, D.M., and Miller, K.T., "Formation of Hydrate Obstructions in Pipelines: Hydrate Particle Development and Slurry Flow," International Conference on Gas Hydrate, Trondheim, June 13-16, 2005.

The authors

Xiaofang Lv ([email protected]) is a PhD student in oil and gas transportation at China University of Petroleum-Beijing. He is a member of SPE.

Jing Gong ([email protected]) is a professor at China University of Petroleum-Beijing. She holds a PhD (1995) from China University of Petroleum, Beijing and is a member of SPE.

Wenqing Li ([email protected]) is a PhD student at China University of Petroleum-Beijing.

Yixuan Tang ([email protected]) is a graduate student in chemical engineering at China University of Petroleum-Beijing.

Jiankui Zhao ([email protected]) is an engineer at China National Oil and Gas Exploration and Development Corp., Beijing. He holds a PhD (2009) from China University of Petroleum-Beijing.