Appendix B
Derivation of the Taylor-Aris Dispersion

B.1 Derivation of the Taylor-Aris Dispersion

It is considered that the flow inside a straight cylindrical pipe is steady, driven by a constant pressure gradient (i.e, Poiseuille flow). The average velocity over the pipe cross-section can be given as:

u(r) = 2u¯ (1 r2 R2 ) , (B.1)

where:

u¯ = 1 πR202πd𝜃0Rrudr. (B.2)

In these equations u¯ denotes the average quantity of the velocity flowing through the pipe. If it is assumed that an axisymmetric distribution of material c(r, z, t) is released into the flow, the evolution of the propagation is described by the ADE in cylindrical form.

c t + u(r)c z = Dm ( 2c z2 + 1 r r (rc r )). (B.3)

Since no particle can leave the system, the boundary conditions are must satisfy cr = 0 at r = R. By separating c using Reynold’s decomposition method,c is separated into its cross-sectional average and r variable parts.

c(r,z,t) = c¯(z,t) + c(r,z,t), (B.4)

where:

c¯ = 2 R20Rrcdr. (B.5)

since the average of deviation is zero (c¯ = 0) the equation can be written as:

c¯ t + c t + u(r) (c¯ z + c z ) = Dm ( 2c¯ z2 + 2c z2 + 1 r r (rc r )). (B.6)

Taking the cross-sectional average of Eq. (B.6) yields the following simplification taking into account that ct = 0 on r = R.

c¯ t + u(r)¯c¯ z + u(r)c z ¯ = Dm 2c¯ z2. (B.7)

The the mean concentration c¯ depends on the average advection of the r-varying part of c (i.e., c (r, z,t)), which is calculated by subtracting Eq. (B.7) from Eq. (B.6) reveals the r-varying component of Eq. (B.6),

c t + (u(r) u¯) c¯ z + uc z uc z ¯ = Dm2c. (B.8)

Based on this equation, an approximation is made whereby after a time of in the order t = R2 Dm the radial diffusion to have almost smoothed out variation in the r-axis. Thus for t O (R2 Dm), it is expected for c¯ c . In addition, the gradients in the r-direction are greater than those in the z-direction. Therefore the primary balance is:

(u(r) u¯ ) c¯ z Dm r r (rc r ), (B.9)

Introducing Eq. (B.1) into (B.9), the following expression is derived.

r (rc r ) = u¯ Dm c¯ z (r 2r3 R2 ) . (B.10)

As shown in the Reynold’s decomposition of c in Eq. (B.4), c¯ is independent from r, so Eq. (B.10) can be integrated twice over,

c = u¯ Dm c¯ z (r2 4 r4 8R2 + A + Blnr). (B.11)

Since c is regular at r = 0 B can be declared the value of 0. Furthermore, c has zero average. This yields:

0Rrudr = 0, (B.12)

This equation give A the value of:

A = R2 12 , (B.13)

c = u¯Rr2 24Dm c¯ z (6Rr2 3R r4 2)whereR r = r R. (B.14)

Equation (4) requires the term u(r)cz¯, which is

u(r)c z ¯ = R2u¯ 48Dm 2c¯ z2. (B.15)

Substituting this result into (B.6), ADE for the mean concentration c¯(z, t) is derived.

c¯ t + u¯c¯ z = (Dm + R2u¯2 48Dm ) 2c¯ z2 = Deff 2c¯ z2. (B.16)