You are on page 1of 14

COMSOL Conference 2015 Boston

Session: Computational Fluid Dynamics

Session Chair: William Vetterling, Zink Imaging, Inc.


October 8, 2015 1:00 PM – 2:30 PM

Hydrodynamic Modeling of a Rotating Cone


Pump Using COMSOL Multiphysics™

Mikhail Vasilev
Anoop Uchagawkar
Patrick L. Mills*

Department of Chemical and Natural Gas Engineering


Texas A & M University-Kingsville
Kingsville-TX 78363-8202 USA
*Patrick.Mills@tamuk.edu
Comparison of Several Different Pump Types

Pumps

Conventional Centrifugal pumps Rotating Cone Pump Viscous Micro Pumps

Fig.2 Eccentric cylinder pump[2]

Fig. 1 Common centrifugal


pump design[1]

• Reynolds number > 1000 • Reynolds number 10-100 • Reynolds number < 10
• High fluid throughput • High fluid throughput • Moderate fluid throughput
• High pump head (meters) • Low pump head (millimeters) • Low pump head (millimeters)
Model Equations: Continuity, Momentum Transport and Turbulence Model

Turbulence Model Equation of Continuity


𝜕𝜕𝜕𝜕
k-ε Model Equations + 𝛻𝛻 � 𝜌𝜌𝒖𝒖 = 0
𝜕𝜕𝜕𝜕
• Turbulent kinetic energy, k, is given by:
𝜕𝜕𝜕𝜕 𝜇𝜇 𝑇𝑇
𝜌𝜌 + 𝜌𝜌 𝒖𝒖 � 𝛻𝛻 𝑘𝑘 = 𝛻𝛻 � 𝜇𝜇 + 𝛻𝛻𝛻𝛻 + 𝑃𝑃𝑘𝑘 − 𝜌𝜌𝜌𝜌
𝜕𝜕𝜕𝜕 𝜎𝜎𝑘𝑘

• Turbulent dissipation, ε, is given by:


𝜕𝜕𝜕𝜕 𝜇𝜇 𝑇𝑇 𝜖𝜖 𝜖𝜖 2
𝜌𝜌 + 𝜌𝜌 𝒖𝒖 � 𝛻𝛻 𝜖𝜖 = 𝛻𝛻 � 𝜇𝜇 + 𝛻𝛻𝛻𝛻 + 𝐶𝐶𝑐𝑐𝑐 𝑃𝑃𝑘𝑘 − 𝐶𝐶𝑐𝑐𝑐 𝜌𝜌 Momentum Transport Equation
𝜕𝜕𝜕𝜕 𝜎𝜎𝜖𝜖 𝑘𝑘 𝑘𝑘
𝜕𝜕𝜕𝜕
𝜌𝜌 + 𝜌𝜌 𝑢𝑢 � 𝛻𝛻 𝑢𝑢 =
• Turbulent viscosity is modelled by: 𝜕𝜕𝜕𝜕
𝑇𝑇
2
𝑘𝑘 2 𝛻𝛻 � −𝑝𝑝𝑝𝑝 + 𝜇𝜇 + 𝜇𝜇 𝑇𝑇 𝛻𝛻𝛻𝛻 + 𝛻𝛻𝛻𝛻 − 𝜌𝜌𝜌𝜌𝜌𝜌 + 𝐹𝐹
𝜇𝜇 𝑇𝑇 = 𝜌𝜌𝐶𝐶𝜇𝜇 3
𝜖𝜖
• Production of turbulent kinetic energy defined as:
𝑇𝑇 2
𝑃𝑃𝑘𝑘 = 𝜇𝜇 𝑇𝑇 [𝛻𝛻𝒖𝒖: 𝛻𝛻𝒖𝒖 + 𝛻𝛻𝒖𝒖 − 𝜌𝜌𝜌𝜌𝜌𝜌 � 𝒖𝒖
3
Equations for 1-D Momentum Transport Model

Approximation of an annular region flow as a narrow slit flow.


 ∂uz   ∂ 2uz ∂ 2uz ∂ 2uz   ∂ux ∂uy ∂uz  ∂p
ρ  − µ  + +  + ρ ux + uy + uz
 + ∂z = 0
 ∂t   ∂ x 2
∂y 2
∂z 2
  ∂x ∂y ∂z 

• Steady-state;
𝐻𝐻 � cos(𝛼𝛼) • 𝑢𝑢𝑥𝑥 , 𝑢𝑢𝑦𝑦 = 0;
• 𝑢𝑢𝑧𝑧 = 𝑓𝑓(𝑥𝑥);
 ∂ 2uz  ∂p
µ 2  + =0
 ∂ x  ∂z
Separation of variables gives:
2𝜋𝜋𝜋𝜋 Δ𝑃𝑃 � 𝐵𝐵2 𝑥𝑥 2
𝑢𝑢𝑧𝑧 = 1−
2𝜇𝜇𝜇𝜇 𝐵𝐵

Integration over cross section gives:


4𝜋𝜋𝐵𝐵3 𝜌𝜌 sin𝛽𝛽 1
𝑤𝑤 = Δ𝑃𝑃 + ( 𝜌𝜌Ω2 𝑠𝑠𝑠𝑠𝑠𝑠2 𝛽𝛽) � (𝐿𝐿22 − 𝐿𝐿21 )
𝐿𝐿 8
3𝜇𝜇 ln( 2�𝐿𝐿 �
1
COMSOL Model Setup
COMSOL Model Description

Assumed Geometry

R0

α
H
H - Height of the cone
κ - Ratio of inner to outer cylinder radii
Pin – Inlet pressure
Ω - Angular velocity
α - Semi-angle of the cone
Meshing

Tetrahedral Mesh Mesh Independent Solution


(48,000 elements)

• Tetrahedral mesh with boundary 101500

Output pressure, Pa
layers has been used
• Number of elements ~48000 101000
• Average quality of the mesh - 0.75

100500
30000 70000 110000 150000
Number of elements in grid
Velocity and Pressure Profiles: Effect of Semi angle

Pa Pa
m/s m/s

Velocity and pressure profiles for a semi angle of 45°, Velocity and pressure profiles for a semi angle of 12°,
Q = 1 ml/s Q = 1 ml/s
The pressure profiles show that the hydraulic head of the pump is a
weak function of the cone semi angle and height of the rotating cone.
Pump Head vs Flow Rate: Effect of Rotational Speed

Head Curve for a Semi Angle of π/4


• Outlet pressure decreases almost
160
140
linearly with increasing volumetric
120 flowrate.
Pump head, Pa

100
Ω=6000 • Pump head is proportional to the
80 Ω=12000
60 Ω=3000 square of rotational speed.
40
20 • Pump outlet pressure does not exceed
0
135 Pa even for highest values of
0 0.5 1
Flow rate, ml/s angular velocity.
Pump Head vs Flow Rate: Approximate Solution vs CFD Solution

Comparison of Head Curve Acquired from Approximate solution for rotating cone pump problem is
Approximation and CFD Results available in open literature. (Bird et al., 2007)

500 Key assumptions


Ω=12000 (CFD)
400 Approximation • Laminar flow
Pump head, Pa

300 • Curvature and entrance effects are neglected.


y = -33.38x + 139.61
200 4𝜋𝜋𝐵𝐵3 𝜌𝜌 sin𝛽𝛽 1
R² = 0.9847 𝑤𝑤 = (𝑝𝑝1 −𝑝𝑝2 ) + ( 𝜌𝜌Ω2 𝑠𝑠𝑠𝑠𝑠𝑠2 𝛽𝛽) � (𝐿𝐿22 − 𝐿𝐿21 )
𝐿𝐿 8
100 3𝜇𝜇 l n( 2�𝐿𝐿 �
1

0 where
0 0.5 1
Flow rate, ml/s L1, L2 are heights corresponding to pressures p1 & p2
respectively.
Pump Head vs Flow Rate: Effect of Fluid Density and Viscosity

Comparison of head curve


Fluid Density, kg/m3 Viscosity, cP
for water and diethyl ether

160
Water 999.66 1.01
140
120
Diethyl ether 713.58 0.24
100
80
60 • Rotating pump head increases with increasing
40
Water viscosity and density.
20
Diethyl ether
0
0 0.2 0.4 0.6 0.8 1 1.2 • Rotating pump head behavior follows the same
trend as that for a centrifugal pump.
COMSOL Application for Rotating Cone Pump
Conclusions

• The rotating cone pump is a simple design compared to other pump


configurations that has applications where a low pump head is sufficient,
such as in microprocess systems.
• Analysis of the rotating cone pump performance can be facilitated by
using the COMSOL Multiphysics CFD Module.
• The CFD model predicts the correct trends in pump head performance
for various model parameters.
• The approximate formula, which is based upon a 1-D fluid mechanics
model, over predicts the pump head performance by about a factor of 2.
• Optimization of the cone pump design, which might include modification
to the cone head surface, e.g., addition of spiral fins, would be facilitated
by COMSOL CFD module simulations vs using empirical approaches.
Thank you for your attention

You might also like