Sie sind auf Seite 1von 4

Chemical Engineering Science, 1958, Vol. 9, pp. 74 to 77. Pergamon Press Ltd.

A note on mean residence-times in steady flows of arbitrary


Imperial College of Science and Technology, London

(Receioed 24 January 1958)

Abstract-The mean residence-time of a tracer injected into a steady incompressible flow of

arbitrary complexity is shown to be equal to the vessel volume enclosed between inlet and outlet
planes divided by the volumetric flow-rate, provided that the tracer is prevented from diffusing
upstream of the injection plane. Velocities and effective diffusion coefficients may vary with
position, but not with concentration or time.

If the tracer is not prevented from diffusing upstream of the injector plane, the mean residence-
time exceeds the above by an amount equal to the effective longitudinal diffusion coefficient in
the inlet pipe, divided by the square of the velocity in that pipe.

R&sum&L’auteur montre que le temps de contact moyen d’un traceur introduit dans un
fluide incompressible de complexite arbitraire, lors de l’ecoulement permanent, est &gal au
rapport du volume du recipient compris entre les plans d’entree et de sortie et de la vitesse
d’ecoulement volumique, ir condition d’empecher la diffusion en amont du plan d’injection.
Les vitesses et coefficients effect& de diffusion peuvent varier avec la position, mais ne varient
pas en for&ion de la concentration ou du temps.

Si le traceur diffuse en amont du plan d’injection, la dur6e de contact est superieure B celle
d&rite precedemment dune quantite &gale au coefficient effectif de diffusion longitudinale dans
la tubulure d’entree divise par Is car& de la vitesse dans ce tube.

Zusammenfassung-Die mittlere Verweilzeit eines Indikatom, der in eine stationlre, inkom-

pressible StrGmung belibiger Anordnung injiziert wird, ist, wie gezeigt wird, gleich dem Behiil-
tervolumen zwischen Eintritts - und Austrittsebene geteilt durch den Volumenstrom, voraus-
gesetzt, dass der Indikator nicht von der Injektionsebene stromaufwiirts diffundiert. Gesch-
windigkeiten und effektive Diffusionskoeffizienten k&men von Ort zu Ort, nicht aber mit der
Konzentration oder der Zeit variieren.

Wenn der Indikator von der Injektionsstelle stromaufwiirts diffundieren kann, wird die
mittlere Verweilzeit gr6sser als die oben angegebene, und zwar urn einen Betrag gleich dem
effektiven Liingsdiffusionskoeffizienten im Eintrittsrohr, geteilt durch das Quadrat der Gesch-
windigkeit in diesem Rohr.

1. PURPOSE OF NOTE em = v/v (1)

If a pulse of tracer material is injected uniformly where v is the volumetric rate of fluid flow.
across the inlet duct to a vessel of volume V, This note examines the conditions under
through which incompressible fluid flows steadily, which this formula is valid. It will be shown
and the concentration of tracer is measured that it holds if the tracer material is prevented
vs. time at the outlet section, a simple formula from diffusing upstream at the injector plane,
[l] for the mean residence-time 0, (time from provided that the effective diffusion coefficients
instant of injection to instant of appearance at and velocities are independent of concentration,
exit of an element of tracer, averaged over all even for flows of arbitrary complexity. If
tracer elements) is upstream diffusion at the injector plane can occur,

A note on mean residence-times in steady flows of arbitrarycomplexity

the formula underestimates the mean residence- 3. ANALYSIS

time. Equation of conseroatirm

2. STATEMENTOFPROBLEMFORINJECTORPLANE It will be supposed that the tracer travels by

IMPERMEABLETO TRACER convection and diffusion. Then its concentration
c at any point obeys the equation
Consider first the vessel shown in Fig. 1. At
time 8 = 0 a small tracer quantity S is suddenly k/38 = v (D v C)- v t4 q (3)
injected uniformly into the fluid crossing the inlet
where Q is the local velocity vector, and D is the
plane 1. A “ sieve ” prevents the tracer from
local effective (laminar and turbulent) diffusivity.
diffusing upstream. An analysis device measures
Both q and D vary with position, but not with
the concentration of tracer cg at the outlet plane 2.
concentration or time.
The volume of vessel and duct between the two
planes is F. The flow velocities can be regarded
as uniform across each of the planes 1 and 2. Theorem 1

Integrating the equation from 8 = o to
V - - ,2 0 = co, (3) becomes
m m
c m -co=V(DV c&?)-V(q cd@ (4)
& s s
0 0

Now at all points, cm = 0 ; for at long enough

FIG. 1. The flowsystemconsidered.
times no tracer material remains. Also, for points
downstream of the injector plane, co = o ;
for the tracer takes a finite time to reach such
The analysis. device gives a result such as is points in finite concentrations. The L.H.S.
illustrated in Fig. 2. We have to find the value of (4) is therefore zero.
For convenience, put
1~ de (5)
Then (4) becomes

V(DVI) - V(q1) =.o (3)

This will ,be recognised as the steady-state
diffusion equation in the absence of sources.
Without formal analysis (which could of course
FIG. 2. Typical concentration time diagram at outlet. be given), it is easy to see that, since gradients
of c, and therefore of I, are zero at the vessel
walls, the only possible solution is
of the mean residence-time which, in accordance I = const. everywhere
with the above definition, is given by = I for inlet stream
= s/v (7)

c c2 0 de Equation (7) represents the first theorem,

which is obvious enough. It means among other
em= :, (2) things that, if a steady stream (train of equal

c2 de pulses) of tracer is injected, the concentration is
uniform throughout the vessel.


Theorem 2

s s
00 co

To approach 8,, we first multiply equation V c2 e de = v c2 de (14)

(3) by 8, and then integrate from 0 = 0 to 0 0

8 = co, obtaining Introducing 8, from (2), we have therefore

@=m proved that
- v(4s”ede)
(8) em= v/v (15)
8=0 0 0 which is our “ simple formula ” (1).
This is Theorem 2. A somewhat different
The L.H.S. may be written as
proof has previously been given by DANCKWERTS
[l]. It will be noted that the theorem holds


a9dc = [6’ c]; - c dB = 0 - I. (9) irrespective of the flow pattern in the vessel.
B=O 0 There may even be regions of recirculation in
the vessel where the streamlines are closed.
N.B. (In order that the symbols within the
Changes in flow pattern will introduce changes
square bracket shall represent zero at the upper
in the shape of the curve of Fig. 2, but we have
limit, c must fall off more rapidly than l/8. It
not had to make any assumption about this shape,
always does so ; for otherwise I would not be
being concerned only with the position of the
centroid of the area beneath it.
Putting, for convenience,
J E c&‘dB, (10) PLANE
0 We now carry out the corresponding analysis
for the case in which the “sieve ” is absent
equation (8) now becomes
from the inlet duct, so that tracer material is
V(DVJ) - V(qJ) +I = 0 (11) free to diffuse upstream.

Equation (11) will be recognised as the steady-

Distribution of rc d6’
state diffusion equation with a distributed source
term 1, which Theorem 1 has shown to have
constant magnitude. The solution to the equation No change in the analysis is needed in the
can again immediately be written down without derivation of Theorem 1 for the region down-
formal analysis (which of course could again stream of plane 1. There 1 = S/v still holds. We
easily be given). Thinking of J as representing now have to consider the distribution of I in
enthalpy and I as a heat source, and noting that the inlet duct.
wall gradients of c, and so of J, are zero, it is Let the area of the duct be A,, so that the
evident that the problem is akin to that of fluid flow velocity is v/A,. Let the diffusion coefficient
flow through a heater. Since the total “ heat in the duct have the uniform value D,. Then
input rate ” is IV, it is clear that the difference equation (6) for this region runs
between J at outlet and J at inlet is given by
D d_“f
_ -- ’ dI = 0
v(J, - J1) =I V (12) ’ dx2 Al dx

Now J1, if we take for clarity a plane just where x = distance along the inlet duct in the
upstream of 1, is evidently zero. Therefore stream direction.
The solution of this equation, with the boundary
vJ,=IV (13) conditions I = S/v at x = xi (inlet plane), and
I = 0 at x = - co (far upstream), is
Reverting to the earlier notation, this can be
written as x1 < x : I = (S/v) exp [v (8 - q)/D,A,] (17)
A note on mean residence-times in steady flows of arbitrary complexity

This means that I falls off exponentially in an This is Theorem 3. The second term may be
upstream direction. interpreted as a correction to the “ simple
formula,” necessary because some of the tracer
material lengthens its residence-time by diffusing
Determination of Q, : Theorem 3 upstream before being swept downstream.
The previous analyses leading to (11) is still Alternatively D,A12/v could be treated as a
valid; the solution for J, will be seen to differ correction which must be added to the vessel
however. For ease of physical understanding, we volume V to account for the permeability of the
will continue to regard I as a heat input term ; inlet plane.
no rigour is lost thereby. 5. DISCUSSION
The total “ heat input rate ” now contains
The result contained in equation (20) may be
two terms, one for the vessel as before, namely
regarded as a generalisation to flows of arbitrary
SV/v, and another for the inlet duct. The latter
PI complexity of a result reported recently by
must be written as I A, dx. Introducing LEVENSPIEL and SMITH [2], for the case of
I constant-area flow with a uniform diffusion
equation (17), valid inmthis region, we have : coefficient.
Xl However, the suggestion of these authors
that, in complex flows, an average value of
I A, dx = (SAX/v) exp [V(8 - xl)/D,A,) dx
diffusion coefficient should be used in the
s s
-CO -a’ residence-time correction, appears to be ill-
= S D,A,2/v2 founded : it is the diffusion coefficient in the
inlet pipe which matters. This will often be
The solution of (11) is therefore much smaller than the average.
It also appears that the time-correction
z,J, = S V/-J + SD, At/v2 (19) increases as the square of the intake area. This
means that the effect of an impermeable injector
Introducing the original notation as before,
plane can always be obtained by reducing the
a new expression for the mean residence-time is
cross-section .of the inlet duct in proportion to
obtained, namely
the other dimensions of the vessel. In many
V D,A: practical cases therefore the correction will be
em=-+7 entirely negligible.

[l] DANCKWERTS P. V. Chem. Engng. Sci. 1957 2 1.
[2] LEVENSPIEL 0. and SMITH W. K. Chem. Engng. Sei. 1957 6 227;


Das könnte Ihnen auch gefallen