The Finite Volume Method

4. Pressure- Velocity Coupling

 Introduction

The convection of a scalar variable 𝜙 depends on the value and direction of the local velocity field
Hypothèse : le champ de vitesses est connu

The velocity field is not known and it must be part of the overall solution of the flow variables

Convection term no linear 𝜕 𝜌𝑢𝑢 𝜕 𝜌𝑣𝑢 𝜕 𝜕𝑢 𝜕 𝜕𝑢 𝜕𝑝

+ = 𝜇 + 𝜇 − + 𝑆𝑥
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑥
Pressure field not known
𝜕 𝜌𝑢𝑣 𝜕 𝜌𝑣𝑣 𝜕 𝜕𝑣 𝜕 𝜕𝑣 𝜕𝑝
+ = 𝜇 + 𝜇 − + 𝑆𝑦
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑦

𝜕 𝜌𝑢 𝜕 𝜌𝑣
+ =0
2 new difficulties 𝜕𝑥 𝜕𝑦

𝑢, 𝑣 appear in 3 equations
3 intrinsically coupled equations No equation for p
𝑃 appears in 2 equations
The pressure is indirectly specified by the continuity equation
where Δ indicates a centred difference (“right minus left”).
Suppose pressure and velocity are co-located (stored at the same positions) and
that advective velocities (cell-face velocities in the mass fluxes) are calculated by linear interpolation.

Hence, the discretised momentum equation has the form:

 𝑃𝐸 + 𝑃𝑃 𝑃𝑃 + 𝑃𝑊
𝜕𝑃 𝑃𝑒 − 𝑃𝑤 − 𝑃𝐸 − 𝑃𝑊
2 2
= = =
𝜕𝑥 𝛿𝑥 𝜕𝑥 2𝜕𝑥

𝑃𝑁 + 𝑃𝑃 𝑃𝑆 + 𝑃𝑊
𝜕𝑃 𝑃𝑛 − 𝑃𝑠 2 − 2 𝑃𝑁 − 𝑃𝑆
= = =
𝜕𝑦 𝛿𝑦 𝜕𝑦 2𝜕𝑦

𝑃𝑃 does not appear despite the fact that the pressure field exhibits a spatial oscillation in 2 directions

𝜕𝑃 𝜕𝑃
= =0
𝜕𝑥 𝜕𝑥 A uniform pressure field Physically unrealistic behavior
 In the staggered-grid arrangement, velocity components are
stored half-way between the pressure nodes that drive them.

This leads to different sets of control volumes: one set

for pressure (and other scalars) and others for the
different velocity components.
 In the momentum equation

In the continuity equation

No interpolation is required for cell-face values

and there is a strong linkage between successive
pressure nodes, avoiding odd-even decoupling.
The alternative approach uses co-located pressure and velocity
but employs a different, pressure-dependent, interpolation for
advective velocities (cell-face velocities in the mass fluxes).
Correcting Pressure to Enforce Mass Conservation
Momentum Equation

𝑃𝐼,𝐽 − 𝑃𝐼−1,𝐽
𝑎𝑖,𝐽 𝑢𝑖,𝐽 = ෍ 𝑎𝑛𝑏 𝑢𝑛𝑏 − ҧ 𝑢
∆𝑉𝑢 + 𝑆∆𝑉 𝑎𝑖,𝐽 𝑢𝑖,𝐽 = σ 𝑎𝑛𝑏 𝑢𝑛𝑏 + (𝑃𝐼−1,𝐽 − 𝑃𝐼,𝐽 ) 𝐴𝑖,𝐽 + 𝑏𝑖,𝐽
𝛿𝑥𝑢
Momentum Equation

𝑃𝐼,𝐽 − 𝑃𝐼−1,𝐽
𝑎𝑖,𝐽 𝑢𝑖,𝐽 = ෍ 𝑎𝑛𝑏 𝑢𝑛𝑏 − ҧ 𝑢
∆𝑉𝑢 + 𝑆∆𝑉 𝑎𝑖,𝐽 𝑢𝑖,𝐽 = σ 𝑎𝑛𝑏 𝑢𝑛𝑏 + (𝑃𝐼−1,𝐽 − 𝑃𝐼,𝐽 ) 𝐴𝑖,𝐽 + 𝑏𝑖,𝐽
𝛿𝑥𝑢
Momentum Equations

𝑃𝐼,𝐽 − 𝑃𝐼,𝐽−1 𝑎𝐼,𝑗 𝑣𝐼,𝑗 = σ 𝑎𝑛𝑏 𝑢𝑛𝑏 + (𝑃𝐼,𝐽−1 − 𝑃𝐼,𝐽 ) 𝐴𝐼,𝑗 + 𝑏𝐼,𝑗
𝑎𝐼,𝑗 𝑣𝐼,𝑗 = ෍ 𝑎𝑛𝑏 𝑢𝑛𝑏 − ҧ 𝑣
∆𝑉𝑣 + 𝑆∆𝑉
𝛿𝑦𝑣
Momentum Equations

𝑃𝐼,𝐽 − 𝑃𝐼,𝐽−1 𝑎𝐼,𝑗 𝑣𝐼,𝑗 = σ 𝑎𝑛𝑏 𝑢𝑛𝑏 + (𝑃𝐼,𝐽−1 − 𝑃𝐼,𝐽 ) 𝐴𝐼,𝑗 + 𝑏𝐼,𝑗
𝑎𝐼,𝑗 𝑣𝐼,𝑗 = ෍ 𝑎𝑛𝑏 𝑢𝑛𝑏 − ҧ 𝑣
∆𝑉𝑣 + 𝑆∆𝑉
𝛿𝑦𝑣
 SIMPLE Algorithm

To initialize the SIMPLE calculation, a pressure field 𝑃∗ is assumed

∗ ∗
The discretized equations are solved using the pressure field 𝑃∗ to obtain a velocity field 𝑢𝑖,𝐽 and 𝑣𝐼,𝑗

∗ ∗ ∗ ∗
𝑎𝑖,𝐽 𝑢𝑖,𝐽 = σ 𝑎𝑛𝑏 𝑢𝑛𝑏 + (𝑃𝐼−1,𝐽 − 𝑃𝐼,𝐽 ) 𝐴𝑖,𝐽 + 𝑏𝑖,𝐽
Obtain 𝑃∗ 𝐺𝑢𝑒𝑠𝑠
∗ ∗ ∗ ∗
𝑎𝐼,𝑗 𝑣𝐼,𝑗 = σ 𝑎𝑛𝑏 𝑣𝑛𝑏 + (𝑃𝐼,𝐽−1 − 𝑃𝐼,𝐽 ) 𝐴𝐼,𝑗 + 𝑏𝐼,𝑗

𝑃 = 𝑃∗ + 𝑃,
correct 𝑢 = 𝑢∗ + 𝑢, correction

𝑣 = 𝑣∗ + 𝑣,
guess

𝑎𝑖,𝐽 𝑢𝑖,𝐽 − 𝑢∗ 𝑖,𝐽 = ෍ 𝑎𝑛𝑏 𝑢𝑛𝑏 − 𝑢∗ 𝑛𝑏 + 𝑃𝐼−1,𝐽 − 𝑃∗ 𝐼−1,𝐽 − 𝑃𝐼,𝐽 − 𝑃∗ 𝐼,𝐽 𝐴𝑖,𝐽

𝑎𝐼,𝑗 𝑣𝐼,𝑗 − 𝑣 ∗ 𝐼,𝑗 = ෍ 𝑎𝑛𝑏 𝑣𝑛𝑏 − 𝑣 ∗ 𝑛𝑏 + 𝑃𝐼,𝐽−1 − 𝑃∗ 𝐼,𝐽−1 − 𝑃𝐼,𝐽 − 𝑃∗ 𝐼,𝐽 𝐴𝑖,𝐽
 𝑎𝑖,𝐽 𝑢, 𝑖,𝐽 = ෍ 𝑎𝑛𝑏 𝑢, 𝑛𝑏 + 𝑃, 𝐼−1,𝐽 − 𝑃, 𝐼,𝐽 𝐴𝑖,𝐽

𝑎𝐼,𝑗 𝑣 , 𝐼,𝑗 = ෍ 𝑎𝑛𝑏 𝑣 , 𝑛𝑏 + 𝑃, 𝐼,𝐽−1 − 𝑃, 𝐼,𝐽 𝐴𝑖,𝐽

For simplification, the nb terms are neglected, this is the main approximation in SIMPLE
𝑢, 𝑖,𝐽 = 𝑑𝑖,𝐽 𝑃, 𝐼−1,𝐽 − 𝑃, 𝐼,𝐽
𝐴𝑖,𝐽 𝐴𝐼,𝑗
𝑑𝑖,𝐽 = 𝑑𝐼,𝑗 =
𝑎𝑖,𝐽 𝑎𝐼,𝑗
𝑣 , 𝐼,𝑗 = 𝑑𝐼,𝑗 𝑃, 𝐼,𝐽−1 − 𝑃, 𝐼,𝐽

𝑢𝑖,𝐽 = 𝑢∗ 𝑖,𝐽 + 𝑑𝑖,𝐽 𝑃, 𝐼−1,𝐽 − 𝑃, 𝐼,𝐽

𝑣𝐼,𝑗 = 𝑣 ∗ 𝐼,𝑗 + 𝑑𝐼,𝑗 𝑃, 𝐼,𝐽−1 − 𝑃, 𝐼,𝐽

Same expression for

𝑢𝑖+1,𝐽 = 𝑢∗ 𝑖+1,𝐽 + 𝑑𝑖+1,𝐽 𝑃, 𝐼,𝐽 − 𝑃, 𝐼+1,𝐽 𝐴𝑖+1,𝐽 𝐴𝐼,𝑗+1
𝑑𝑖+1,𝐽 = 𝑑𝐼,𝑗+1 =
𝑎𝑖+1,𝐽 𝑎𝐼,𝑗+1
𝑣𝐼,𝑗+1 = 𝑣 ∗ 𝐼,𝑗+1 + 𝑑𝐼,𝑗+1 𝑃, 𝐼,𝐽−1 − 𝑃, 𝐼,𝐽+1
The velocity field is subject to a constraint, it must satisfy the continuity equation

𝜌𝑢𝐴 𝑖+1,𝐽 − 𝜌𝑢𝐴 𝑖,𝐽 + 𝜌𝑣𝐴 𝑖,𝐽+1 − 𝜌𝑣𝐴 𝑖,𝐽 =0

𝜌𝑖+1,𝐽 𝐴𝑖+1,𝐽 𝑢∗ 𝑖+1,𝐽 + 𝑑𝑖+1,𝐽 𝑃, 𝐼,𝐽 − 𝑃, 𝐼+1,𝐽 − 𝜌𝑖,𝐽 𝐴𝑖,𝐽 𝑢 ∗ 𝑖,𝐽 + 𝑑𝑖,𝐽 𝑃, 𝐼−1,𝐽 − 𝑃, 𝐼,𝐽
+
𝜌𝑖,𝐽+1 𝐴𝑖,𝐽+1 𝑣 ∗ 𝐼,𝑗+1 + 𝑑𝐼,𝐽+1 𝑃, 𝐼,𝐽 − 𝑃, 𝐼,𝐽+1 −𝜌𝐼,𝑗 𝐴𝐼,𝑗 𝑣 ∗ 𝐼,𝑗 + 𝑑𝐼,𝑗 𝑃, 𝐼,𝐽−1 − 𝑃, 𝐼,𝐽

𝜌𝑑𝐴 𝑖+1,𝐽 + 𝜌𝑑𝐴 𝑖,𝐽 + 𝜌𝑑𝐴 𝐼,𝐽+1 + 𝜌𝑑𝐴 𝐼,𝐽 𝑃, 𝐼,𝐽

− 𝜌𝑑𝐴 𝑖+1,𝐽 𝑃, 𝐼+1,𝐽 + 𝜌𝑑𝐴 𝑖,𝑗 𝑃, 𝐼−1,𝐽 + 𝜌𝑑𝐴 𝐼,𝑗+1 𝑃, 𝐼,𝐽+1 + 𝜌𝑑𝐴 𝐼,𝑗 𝑃, 𝐼,𝐽−1

+ 𝜌𝑢 ∗ 𝐴 𝑖,𝐽 − 𝜌𝑢∗ 𝐴 𝑖+1,𝐽 + 𝜌𝑣 ∗ 𝐴 𝐼,𝑗 − 𝜌𝑣 ∗ 𝐴 𝑖,𝐽+1

𝑎𝐼,𝐽 𝑃, 𝐼,𝐽 = 𝑎𝐼+1,𝐽 𝑃, 𝐼+1,𝐽 + 𝑎𝐼−1,𝐽 𝑃, 𝐼−1,𝐽 + 𝑎𝐼,𝐽+1 𝑃, 𝐼,𝐽+1 + 𝑎𝐼,𝐽−1 𝑃, 𝐼,𝐽−1 + 𝑏 , 𝐼,𝐽

Pressure correction equation 𝑃,

Resulting from incorrect velocity field 𝑢∗ 𝑣 ∗
By solving this equation, the corrected pressure field 𝑃, is obtained in all nodes
The pressure, 𝑃 the speeds 𝑢𝑖,𝐽 , 𝑣𝑖,𝐽 are then obtained

The omission of termes σ 𝑎𝑛𝑏 𝑢, 𝑛𝑏 does affect the solution , since corrections 𝑃, 𝑢, 𝑣 , 0 when solution converge
𝑃 = 𝑃∗, 𝑢 = 𝑢∗ , 𝑣 = 𝑣∗

For fear of divergence of the pressure correction equation, under-relaxation is desirable during the iterative process:

𝑃𝑛𝑒𝑤 = 𝑃∗ + 𝛼𝑝 𝑃, 0 < 𝛼𝑝 < 1

too large; speeds up the solution, too small ensures a more stable solution

𝑢𝑛𝑒𝑤 = 𝛼𝑢 𝑢 + 1 − 𝛼𝑢 𝑢(𝑛−1)

𝑣 𝑛𝑒𝑤 = 𝛼𝑣 𝑣 + 1 − 𝛼𝑣 𝑣 (𝑛−1)

without relaxation
Discretized equations with under relaxation :

𝑎𝑖,𝐽 𝑎𝑖;𝐽 (𝑛−1)

𝑢𝑖,𝐽 = σ 𝑎𝑛𝑏 𝑢𝑛𝑏 + 𝑃𝐼−1,𝐽 − 𝑃𝐼,𝐽 𝐴𝑖,𝐽 + 𝑏𝑖,𝐽 + 1−∝𝑢 𝑢𝑖,𝐽
∝𝑢 ∝𝑢

𝐴𝑖,𝐽 ∝𝑢 𝐴𝑖+1,𝐽 ∝𝑢
𝑑𝑖,𝐽 = 𝑑𝑖+1,𝐽 =
𝑎𝑖,𝐽 𝑎𝑖+1,𝐽

𝑎𝐼,𝑗 𝑎𝐼;𝑗 (𝑛−1)

𝑣𝐼,𝑗 = σ 𝑎𝑛𝑏 𝑣𝑛𝑏 + 𝑃𝐼,𝐽−1 − 𝑃𝐼,𝐽 𝐴𝐼,𝑗 + 𝑏𝐼,𝑗 + 1−∝𝑣 𝑣𝐼,𝑗
∝𝑣 ∝𝑣

𝐴𝐼,𝑗 ∝𝑢 𝐴𝐼,𝑗+1 ∝𝑢
𝑑𝐼,𝑗 = 𝑑𝐼,𝑗+1 =
𝑎𝐼,𝑗 𝑎𝐼,𝑗+1