• A general wall friction model for a non-Newtonian fluid has been developed and compared successfully to experimental data.
• The model computes the velocity profiles, for given rheology, velocity, pipe diameter and wall roughness, and results in a generic friction factor which can be applied to fluids with non-Newtonian behavior. The transition between laminar and turbulent flow is included.
• In the presentation we compare model and experimental pressure drops for various rheologies.
• A case of cleanup of a well containing water based mud is discussed
Improved fluid control by proper non-Newtonian flow modeling
Stein Tore Johansen, SINTEF Sjur Mo, SINTEF
• Non-Newtonian flows are present in many situations
• All oils where precipitation is thermodynamically possible
• Precipitated particles may cause non-Newtonian rheology
• Precipitates, such as asphaltenes, wax, naphthenic acids, inorganic salts • Emulsions formed by the flow and where fluid interfaces are modified and
stabilized by surfactants or particles
• Drilling operations. The drilling fluids contain different types of particles. The fluid base may be some stabile emulsion.
• Completion fluids
Typical flow assurance situations where non-Newtonian effects must be
respected
• Flows with precipitated wax, hydrates, as well as emulsions • Well clean-up: Leftover drilling fluids and completion fluids
• Assumption: the magnitude of the stress in the fluid is a pure function of strain rate (velocity gradient)
• Can be obtained from rheometer tests
Rheology
0( )
S
KS
n;
S
U
1 0( ) / S
S
/
S
KS
n
• The steady state momentum equation for flow in a pipe gives us:
• The rheology of the fluid is Yield Power Law (Herschel-Bulkley):
A generic model for pipeflow
(1
)
( )
w ty
U
R
S
y
S
0( )
S
KS
n;
S
U
y
( )
(1
)
w ty
S
U
R
S
y
The turbulent viscosity
Ref: ASHRAFIAN, A. & JOHANSEN, S. T. 2007. Wall boundary conditions for rough walls. Progress in
Computational Fluid Dynamics, 7, 230-236
2
0
Generic transition function Wall
viscosity
High Re dimensionless kinematic turbulent viscosity
max( , ) ; max( , ) 51.98 ( ) ( , , , ) 11.4 max( , ) ; max( , ) 51.98 w t w w S s y s y S g R S s y s y
1 ( ) (-0.0285*max(60, )+1.61) 0 0 Transition function Non-newtonian low Re correction( , , ) ( , , , ) 1 where , R , , R, , R, R w w wall wall u S S f R g R e U y s y s y s y ( ) wall U y ( ) s f
• By guessing the wall shear stress we find • The velocity profile
• The averaged velocity
• If the required velocity is not met correct wall shear stress until convergence • Compute friction factor versus Re
• ReS: Use |∂U/ ∂y|Wall based on laminar Newtonian flow
• ReW: Use actual |∂U/ ∂y|Wall
Numerical solution
(1
)
( )
w ty
U
R
S
y
S
2 4 4 4 8 ReS S U U R R U U D U D U DS D U 1/ 0 2 2 Re n w W w w U R U R K IMPLICIT !Predicted effect on wall roughness of friction factors for
Newtonian flows
Roughness transition: s+<70 2 8 Re Re 8 S U D U Predicted effect on wall roughness of friction factors for
non-Newtonian flow
2 8 Re Re 8 S U D U 0.806 3 (S) 2.827 Pa+0.047 S 1045 kg/m Model validation (data from Chilton & Stainsby)
CHILTON, R. & STAINSBY, R. 1998. Pressure Loss Equations for Laminar and Turbulent Non-Newtonian Pipe Flow. Journal of Hydraulic Engineering, 124, 522-529
0.613 3 (S) 1.268 Pa+0.214 S 1024 kg/m 0.664 3 (S) 0.727 Pa+0.069 S 1011 kg/m
Model validation (data from Chilton & Stainsby)
CHILTON, R. & STAINSBY, R. 1998. Pressure Loss Equations for Laminar and Turbulent Non-Newtonian Pipe Flow. Journal of Hydraulic Engineering, 124, 522-529
0.806 3 (S) 2.827 Pa+0.047 S 1013 kg/m 0.594 3 (S) 1.273 Pa+0.189 S 1016 kg/m
Model validation (data from El-Nahhas et al., 2005)
EL-NAHHAS, K., GAD EL-HAK, N., ABOU RAYAN, M. & EL-SAWAF, I. FLOW BEHAVIOUR OF NON-NEWTONIAN CLAY SLURRIES. Ninth International Water Technology Conference, IWTC9 2005, 2005 Sharm El-Sheikh,
0.610 3 (S) 1.950 Pa+0.090 S 1170 kg/m 0.505 3 (S) 6.0 0.325 1223 kg/m Pa S
Model validation (data from El-Nahhas et al., 2005)
EL-NAHHAS, K., GAD EL-HAK, N., ABOU RAYAN, M. & EL-SAWAF, I. FLOW BEHAVIOUR OF NON-NEWTONIAN CLAY SLURRIES. Ninth International Water Technology Conference, IWTC9 2005, 2005 Sharm El-Sheikh,
0.465 3 (S) 14.0 0.755 1283 kg/m Pa S Model inaccuracy:
Laminar – turbulent transition model issue !
Velocity profiles
• High yield stress velocity profile becomes more "laminar" at high Re • Log region is modified
• Velocity profile controls local viscosity and sedimentation / separation of
0.806 3 (S) 2.827 Pa+0.047 S 1013 kg/m
• Reservoir pressure : 91 bar • Exit pressure: 16 bar
• Well PI = 0.5 kg/(bar sec)
• Initially well is filled with stagnant water, oil and gas
Simulations of 2.3 km long well
Restart scenarios (pure brine or water based mud in well)
Restart scenarios (pure brine or Water Based Mud in well)
Water based mud in well: Restart 0.80
(S) 2.5Pa 0.40 S
Pressure distribution some time after restart
Water based mud in well: Pressure after 17331 sec Brine in well: Pressure after 12348 sec
Profiles of mass fractions of water based mud in water
zone
Only brine in well
Very small inflow after restart with water based mud in
well
• A numerical model is established for friction in non-Newtonian flows • The model can represent available experimental data well
• The model gives for the first time a method to include wall roughness into non-Newtonian wall friction predictions
• The model can be applied directly as boundary condition in CFD (Computational Fluid Dynamics) simulations
• The model can easily be extended to represent heat transfer (mildly temperature dependent rheologies)
• The model is quite computationally intensive: i) Huge potential for speed optimization
ii) Easy to generate pre-calculated friction factor curves for any fluid rheology – allows superfast friction calculations
•
A non-Newtonian model has been implemented into LedaFlow
•
Here we investigated the "water phase" as a mixture of brine and WBM.
•
Water based mud sitting in a well is very challenging to clean out due to
•
Yield stress of mud
•
High effective viscosity of mud
•
A combined effect of the two above is that initial velocity during clean-up
may be too low, such that only trapped gas can escape, making the situation
worse
•
Mud may only partially be cleaned out, restricting wellbore area and limiting
production
• The sponsors contributing to this development are gratefully acknowledged: • The NFR sponsored project Advanced Wellbore Transport Modelling,
with partners Statoil, GDF SUEZ E&P Norge, IRIS, UiS, NTNU and SINTEF
• LedaFlow Technologies DA