Nord Stream dry air purge improves nitrogen-slug use

Oct. 7, 2013
Purging of precommissioning dry air from the Nord Stream offshore natural gas pipeline demonstrated the reliability of a simple nitrogen-slug shortening calculation model in estimating seal volume requirements and the validity of inert slug purging in large-diameter, long transmission pipelines.

Alessandro Terenzi
Matteo Paglialunga
Daniela Zenobi

Saipem SpA
Fano, Italy

Marco Casirati
Jarleiv Maribu
Vladimir Borovik

Nord Stream AG
Zug, Switzerland

Purging of precommissioning dry air from the Nord Stream offshore natural gas pipeline demonstrated the reliability of a simple nitrogen-slug shortening calculation model in estimating seal volume requirements and the validity of inert slug purging in large-diameter, long transmission pipelines.

Commissioning a natural gas pipeline includes displacing the residual dry air used during precommissioning for dewatering and drying. This must occur before gas is introduced to begin normal transport. Pipeline companies use several methods to purge their lines, from direct purge, in which air is displaced by gas without pigs, up to injection of a complete pipeline-volume of inert gas before natural gas' introduction.

Purging must avoid formation of a flammable air-gas mixture at the interface between the two products. A contamination region can form due to turbulent diffusion. Direct purging minimizes the contamination interface volume by maintaining sufficiently high gas velocity to ensure a sharp interface during travel along the pipeline, limiting stratification close to pipe wall and the subsequent fluid mixing.

The American Gas Association's "Purging Principles and Practice" provides minimum recommended velocities to limit stratification.1 For example, an air-light gas interface flowing in a 48-in. OD pipeline must maintain a minimum velocity of about 1.83 m/sec. The same publication addresses inert slug purging, in which a batch of nitrogen acts as a barrier between the flammable gas and air. The slug's dimensions prevent contact between the two boundary gases, so long as sufficient velocity is maintained.

This method has the advantage of increasing operational safety and reducing nitrogen volume requirements as compared with whole-pipeline inertization. Calculating the slug shortening that occurs during air displacement as mixing takes place at the two interfaces determines the amount of nitrogen required.

AGA "Purging Principles and Practice,"1 recommends nitrogen seal volumes for pipelines up to 36-in. OD and 15 km long based on modeling and field tests performed by Southwest Research Institute (SWRI).2 Data sources addressing slug shortening relevant to direct application of the inert slug purging method are limited; few data are available for pipelines of 32-in. OD and 260-km maximum length.3

This article describes the model for nitrogen slug sizing calculation with reference to available tuning data from past experiences as well as the application of this method to the gas-in operation of the second Nord Stream gas pipeline.

Nord Stream consists of two 48-in. OD subsea pipelines running 1,224 km (about 760 miles) from Russia to Germany through the Baltic Sea, the world's longest single-section offshore pipelines. The pipelines export gas from Vyborg, Russia, across the Gulf of Finland and the Baltic Sea, to a receiving terminal in Lubmin (near Greifswald), Germany (OGJ, May 6, 2013, p. 100).

Slug-shortening model

Calculating inert slug shortening during propagation inside the pipeline involves modeling the diffusion process at the leading and trailing edges. Turbulence is the primary cause of diffusion, with molecular diffusion negligible in the considered operating conditions.

Fig. 2 shows the evolution of the slug's nitrogen concentration profile while travelling. Turbulence-generated mixing progressively transforms the sharp interfaces present at the beginning into S-shaped profiles. The following simplifying assumptions allow development of a mathematical model of the diffusion process:

  • Slug propagation velocity is roughly constant.
  • Turbulent diffusivity is constant.

Equation 1 expresses the one-dimensional Fick's law in terms of nitrogen concentration, c, in the infinite bar, with initial conditions given by Equation 2. Equation 1 is written in a frame moving at the slug velocity, representing the contribution of only turbulent diffusion. The initial condition is the step function, indefinitely extending the initial discontinuity upstream and downstream and representing the initial status of the nitrogen slug's rear interface.

The condition is symmetrical for the front of the slug, allowing a doubling of the calculated interface spread to evaluate the full shortening. The nitrogen volume is assumed large enough that the two mixing interfaces don't touch during propagation and a pure nitrogen region is always present inside the slug sufficient to address seal sizing.

Equation 3 provides the solution to Equation 1 and can be used to evaluate the spread of nitrogen concentration over a certain time, t, by evaluating the argument of the error function, Φ at the propagating front location expressed by Equation 4. The approximate value of the solution is 2.3, resulting in the coordinate of the front given by Equation 5.

Symmetry dictates that the shortening distance at the rear is the same, resulting in a total spread length given by Equation 6. Equation 7 guides evaluation of the turbulent diffusivity coefficient as a function of turbulent velocity, derived by the Dreissler equation of turbulent velocity radial profile in a pipe cross section. The main fluid property affecting this expression is density, requiring it be used individually to evaluate differences between gases that might be used.

Doubling Equation 6's value and including a correction factor, k, to address "real life" factors express total slug shortening length. Factors addressed by k include acknowledgement that:

  • Interfaces are not completely sharp at any time.
  • Slug velocity is not constant during motion.
  • Pressure and density are not constant during motion, making diffusivity a variable as well.
  • Different tail and front gases have different properties, making front and rear diffusivities slightly unequal.

Equation 10 defines the resulting expression.

Diffusivity variations due to gas density, time, and space differences are quite limited. Table 1 reports the values of a typical natural gas, air, and nitrogen density in a typical pipeline purging operating pressure range. Purging operations normally occur at low pressures because they start from a post-drying condition, and pressurization takes place slowly during air displacement. Low-pressure operation also helps maintain steady flow at higher velocities.

Table 1 also shows the average density at inert slug interfaces and density-dependent diffusivity (Equations 7-8). Percent deviations of average density-dependent diffusivity with respect to both single interface values and pressure variations appear at the bottom of Table 1, confirming its relatively low effect on normal engineering practices.

Model tuning

Available data on estimated or measured minimum nitrogen slug length needed to avoid contamination between air and natural gas during pipeline purging provided the basis for model tuning. AGA-recommended values for nitrogen-slug volumes,1 in pipes larger than 16-in. OD and with a maximum length of 15.2 km allow evaluation of constant k (Equation 10). Hydraulic simulations of pipeline purging carried out with SPT's OLGA 7.1 software determined relevant characteristics to consider when evaluating diffusion-model properties.

Table 2 provides a calculation-data summary. Each investigation used Equation 10 to estimate slug shortening length and determine calibration constant k to match AGA-specified nitrogen volumes. A value of 1 kg/cu m represented average gas density in expressions for turbulent velocity (Equation 8), being an average of air, nitrogen, and natural gas densities at the considered operating pressure. The last column of Table 2 shows k values ranging between 4 and 5 with an average of 4.7.

A more significant model tuning considered measured nitrogen slug contamination length during purging of different sections of the Bolivia-Brazil Gas Pipeline.3 Reference data for this article came from operations of the 32-in. OD north branch of the pipeline, starting at Corumba and ending at Replan.

With a total length of about 1,200 km, purging occurred separately on different 122-256 km sections separated by launching and receiving pig trap stations. A constant inlet pressure during air displacement and gas filling provided progressive pipeline pressurization, with outlet pressure increasing in time.

Purging injected 2,000 standard cu m of nitrogen in each section ahead of the entering gas. An average value of 3.7 kg/cu m represented average gas density during the operation, lying between nitrogen and natural gas densities at operating pressure. Measured slug contamination lengths reported only for the natural gas–nitrogen interface required that only half the length evaluated by Equation 10 be considered. Table 3 provides a summary of the relevant calculation data, with k values between 4 and 5 with an average of 4.34 appearing in the last column.

All available data for diffusion model tuning referring to long, large-OD pipes confirm a calibration-constant k value near 4.5. This factor also applied to purging Nord Stream.

Nord Stream

The gas-in operation for the Nord Stream twin pipeline system used different methods for the two lines in terms of nitrogen filling. The first pipeline, commissioned in 2011, was completely filled by nitrogen before gas was introduced. The second pipeline, commissioned in 2012, was only partially filled by nitrogen, using the inert slug method even though the resulting nitrogen seal was oversized with respect to AGA minimum requirements.

Saipem's preliminary calculations of nitrogen slug volume used the approach presented here. Measurements collected during purging allowed assessment of the accuracy of slug shortening predictions: a first for a pipeline of this scale.

Filling the second pipeline with dry air (residue from previous drying operation) at atmospheric pressure followed its precommissioning. Nitrogen pressurized two 32-in. OD onshore pipelines connecting the Russian gas compressor station to the offshore pipelines to 80 barg, providing a barrier between air and natural gas in the 48-in. OD pipeline and avoiding low temperature and possible liquid dropout during initial gas fill of the 32-in. pipelines.

Nitrogen discharge into the 48-in. pipeline occurred immediately before its gas fill, resulting in a 120-km nitrogen batch at atmospheric pressure in the 48-in. line. Air-nitrogen slug displacement lasted 3.4 days, during which inlet and outlet pressures, as well as the inlet gas flow rate, rose gradually to roughly 5 barg, 1.7 barg, and 100,000 std. cu m/hr, respectively. Average flow velocity measured 4.2 m/sec, well above the minimum AGA-specified velocity requirements of 1.83 m/sec.1

The inlet flow control system determined the hydraulic parameters' inlet and outlet time trends, respecting requirements dictated by the passage of gas through heaters and the discharge system characteristics at pipeline outlet in Germany. OLGA simulated the hydraulic transient by imposing monitored pipeline outlet pressure and inlet flow rate trends and obtaining the pipeline inlet pressure trend. Fig. 3 shows hydraulic parameter time trends.

Fig. 4 shows the calculated pipeline filling zones 54 hr before beginning the purging operation. Fig. 5 shows the oxygen time trend measured at pipeline arrival during passage of the air-nitrogen interface. Oxygen dropped to 0% in 7 min., corresponding to a nitrogen-air interface 2 km long.

The same result emerged while monitoring air displacement by nitrogen of the first Nord Stream pipeline. A time trend of methane content during the nitrogen-natural gas interface arrival is not available with the same detail. The transit, however, lasted 16 min, corresponding to a total length of 4.6 km.

Mixing in this zone is actually not fully ascribable to turbulent diffusion during purging of the 48-in. OD pipeline. Some mixing also took place during discharge from the 32-in. pipelines in Russia (high pressure, low velocity) into the 48-in. pipeline. Assuming that each factor played an equal role in mixing establishes a turbulent diffusion mixing length in the nitrogen-natural gas interface of 2.3 km.

Table 4 summarizes the calculated data relevant to nitrogen slug shortening during the second Nord Stream pipeline purging, with reference to both tail and front interface, as well as total shortening. Relevant calibration constants confirm modeling developed on the basis of past data tuning, with an average constant value of 5.0.

References

1. AGA Report No. XK0101, Purging Principles and Practice, Third Edition, Washington, 2001.

2. SwRI Report No. GRI-97/0104, "Pipeline Purging Principles and Practices Research," 1997.

3. Frisoli, C., Carvalho de Faria, J.A., and Ennes Senna, F.J., "Purging and Load Operations of Bolivia-Brazil Gas Pipeline," International Pipeline Conference, Calgary, Oct. 4-8, 2004.

4. Tikhonov, A.N., and Samarskii, A. A., "Equations of Mathematical Physics," Dover Publications, 1990; reprint of a translation of the original publication by Pergamon Press Ltd., Oxford, 1963.

The authors

Alessandro Terenzi ([email protected]) is a flow assurance lead at Saipem SpA, Fano, Italy. He has been a member of European Consortium for Mathematics in Industry and is currently a member of the scientific committee of Italian Association of Multiphase Fluid Dynamics in Industrial Plants. Terenzi has been a lecturer for several courses on fluid mechanics at both the University of Bologna and Eni Corporate University. He earned a first-class (5 years) degree in physics (1989) at University of Bologna.

Matteo Paglialunga ([email protected]) is a flow assurance lead at Saipem SpA. He has been involved in a number of different theoretical and practical scientific fields: non-Newtonian fluid flow, solid-liquid suspensions flow, three-phase gas-oil-water flow, modeling of fittings (valves, orifices, area changes), thermal transmission problems (natural and forced convection, conduction through pipe walls), friction analysis, and gas thermodynamics. Paglialunga earned a first-class degree (2007) in chemical engineering at University of Bologna.

Daniela Zenobi ([email protected]) is project manager for engineering of offshore pipeline projects within Saipem's engineering and construction business unit. She joined Saipem in 1989 and also serves as project manager in contract with Nord Stream AG for engineering support to Nord Stream operation and maintenance. Zenobi was also project manager for engineering during Nord Stream's construction follow-on phase and technical manager responsible detailed engineering of the Nord Stream pipelines. She previously served as technical manager for the longest trunkline in the North Sea, Langeled. Zenobi holds the first-class degree in civil engineering from the Polytechnic University of Marche, in Ancona, Italy.

Marco Casirati ([email protected]) works for Nord Stream as project manager for precommissioning, commissioning, gas-in, and shore approach activities at the Russian landfall. He has overall responsibility for development of the concept, design, and performance of operations and has held this role since October 2008. Before his assignment with Nord Stream, Casirati worked for Saipem, where his last position was branch manager and project director for the EPCI Project Dolphin Sealines and Export Pipelines in Abu Dhabi. He has been working in the offshore oil and gas construction industry since 1995, including managing large subsea trunkline projects, such as the Libya Gas Transmission System (GreenStream) and the offshore section of Blue Stream. Casirati holds a PhD in aeronautical engineering from the Polytechnic University of Milan, Italy.

Jarleiv Maribu ([email protected]) works with Nord Stream as an advisor on engineering, precommissioning, and commissioning. He has been in this role since October 2008. Before this, Maribu spent 30 years with Statoil, managing the precommissioning department for many years, responsible for precommissioning concepts and planning and execution of all the company pipeline systems. Projects included Europipe, Zeepipe, Franpipe, and Langeled. Maribu holds an MS in mechanical engineering from the Norwegian Technical University, Trondheim.

Vladimir Borovik ([email protected]) works for Nord Stream AG as deputy technical director for operation-dispatching of the Nord Stream Pipeline System, including landfalls (in Russia and Germany), subsea pipelines, and the control center in Zug, Switzerland. He has participated in Nord Stream concept development, design, and operations since September 2008. Borovik has worked in the gas industry since 1974. Before Nord Stream he worked for Gazprom, Russia, most recently as technical director of Blue Stream Pipeline Co. running from Russia to Turkey. He holds a BS in mechanical engineering from Ufa (Russia) Oil & Gas Academy for design and operation of oil and gas trunklines and underground gas storage.