(c)1998 by Prof. M.

Kostic
Filling-up and Overflow in a Tank Example http://www.ceet.niu.edu/faculty/kostic/
NOTE: This is a typical, general type problem, for which no specific prerequisites are necessary. An
intelligent man may solve it without using calculus, see the last methods (c) and (d) on page 3. This is a simple
problem solved as an example to show how several different methods may be used in general.

Problem: A tank providing water to a pump is shown in

Figure. Water enters the tank trough a 1-in diameter supply
pipe at a constant mass flow rate of 6.8 lb/s (min), and exits
to the pump through a 1-in diameter (de) pipe. The diameter
of the tank is 18 in. (dT), and the top of the 2-in diameter
(dof) overflow pipe is 2 ft (z top) from the base of the tank.
The velocity Ve, in ft/s, of the water exiting to the pump
varies with the height z of the water surface from the
bottom, in ft, according to: Ve(z)= 8.16z1/2. Determine the
time (T) , in seconds, for the control volume enclosing the
initially empty tank to reach steady state (when overflow
begins, z=ztop). At steady state, what is the mass flow rate,
in lb/s, that water spills out through the overflow pipe?

Setting up the problem:

1 18 2
Given: d e dT d of z top 2 V e( z ) 8.16 . z , also: ρ 62.34
12 12 12
2 π 2 π ... exit-pipe and tank-base areas,
de . A e = 5.454 .10 d of .
3 2
Ae AT dT A T = 1.745
4 4 respectively

m in 6.8 ... constant in-flow mass rate m e( z ) ρ .A e .V e( z ) .. variable exit-flow mass rate
Conservation of mass: "small" mass (dm) and height (dz) increases
dm ρ .A T .dz m in m e( z ) .dt
for a small time period (dt), see the darker shaded area on the Figure
above:
dz m in m e( z ) m in Ae
...then, D C 1 C 2 . z where: C 1 C 1 = 0.0625 C 2 .8.16 C = 0.0255
2
dt ρ .A T ρ .A T AT

a) Solving the above differential equation for "dt" by separation of variables and integrating:
...MathCAD numerical integration ...MathCAD symbolic/analitic integration
z top = 2
z top z top
1 1
Ta dz T a = 54.479 dz 54.4789751206557729
C1 C 2. z C1 C 2. z
0 0
...the solution, i.e. the time needed to fill-up the empty tank till its overflow level.

NOTE: Different subscripts may be used for the same physical variables to indicate
the corresponding method, for example (Ta) or (Tb) or (TI ), or to differentiate results
obtained using different methods.

1
Filling-up and Overflow in a Tank ( Cont'd ) Page 2

C 2 .z
2 2 2
ln C 1 z C1 z ...the analytical indefinite
I( z ) .C 2. 2. .atan C2 .
2
2 1 integral of the above
C2 C2
C 2. C2 . C1
2 2 2
C1 expression

Also, T I I z top I( 0 ) T I = 54.479 ...the solution, i.e. the time needed to fill-up the empty tank till its
overflow level.
As we often see, the analytical integration is usually lengthy or impossible, so we, engineers, resort to
numerical solutions. That is no longer difficult, if we utilize available (and now powerful and inexpensive) PC
computational hardware and application software, see below.

b) Solving the above differential equation (see page 1) by Runge-Kutta numerical method (with rkfixed( ),
a built-in MathCAD function - have to be used if the separation of variables is not possible):

z0 0 D( t , z ) C1 C 2 . z0 ...initial value and derivative of the function z(t)

t0 0 tL 100 N1 500 ...initial- and end-values of the independent variable t

and number of points over that range
<1 >
Z rkfixed z , t 0 , t L, N 1 1, D N rows( Z ) N = 500 M cols( Z ) M=2 Z N 1
= 3.001256

<0 > <1 >

tb Z zb Z ...the solution of the above differential equation (see page 1) over the range t 0<t<tL.

4 The Runge-Kutta method (see also MathCAD Help, or press F1

54.48
key while cursor is on "rkfixed" function) is a very effective to solve
any first-order ordinary differential equation (ODE), or a system of
the first order ODEs (thus higher-order DE) for a given initial
2
value(s). However, to evaluate the independent variable, t in this
zb 2 case, for a given value of dependent variable z=ztop=2, we have to
"look" in the table of solution values, or on the diagram, or to try
different t-values until we obtain the given z=2 value, in this case.
Luckily, MathCAD has the built-in "Given-Find" solver to do
almost any iterative "search" for us, see next (This procedure
0 may be also used to solve the so-called boundary-value
0 50 100 150
tb problems using the initial value problem methods):

Define the solution as <1 >

Z b t top rkfixed z , t 0 , t top , N 1, D z Top t top Z b t top
function of ttop: N 1

Guess: t top 40 ...and use the "Given-Find" solver below to find new Tb=ttop to satisfy the required condition

Given
z Top t top 2 ... given condition

Tb Find t top T b = 54.479 ...the solution, i.e. the time needed to fill-up the empty tank till its
overflow level.
Check the z-value for the calculated Tb value, i.e. that the condition above is z Top T b = 2
satisfied:

2
Filling-up and Overflow in a Tank ( Cont'd ) Page 3
c) Solving the above differential equation (see page 1) by dividing the function z-range in N divisions
and using the finite difference approximation (this is a "common sense" method):

z top
N 100 z top 2 ∆z i 0 .. N 1 t0 0
N
i
i . zi zi 1 ∆z
zi z top zN z top z av ∆t ti 1
∆t
N 2 i k
C 2 . z av
i
C1 k= 0
i
K if( N < 12 , N , 12 )
K = 12 Tc ∆t T c = 54.48 also, tN = 54.48

k 0 .. K ...the solution, i.e. the time needed to fill-up the empty tank till its overflow level.
tk = zk =
0 0 2
2
0.334 0.02 Tc
0.678 0.04
1.03 0.06
1.389 0.08 z 1
1.753 0.1
2.124 0.12
2.499 0.14
2.879 0.16 0
0 10 20 30 40 50 60
3.264 0.18 t

3.653 0.2
4.046 0.22
...first K-values tabulated on the right and all values plotted above.
4.444 0.24

d) Solving the above problem (see page 1) by approximating variable exit flow rate with a constant
flow rate based on average water height during the fill-up
NOTE: This is a "common sense" simplified method which may or may not
be appropriate for a given case. Be very cautious when using it.

0 z top ρ .A T .z top Td
z avg m e_av m e z avg Td T d = 54.057 = 99.2 %
2 m in m e_av TI

The Td compares pretty well with the above results (Ta, TI, Tb, Tc), since the relation between z and t is
"almost" linear (see above diagram). However, we have to be very cautious in general when using such
approximations.

The steady-state solution (at z = ztop= 2 = constant):

m in = 6.8 ... in-flow mass rate
m e_top m e z top m e_top = 3.924 ... exit-flow mass rate

m overflow m in m e_top m overflow = 2.876 ... over-flow mass rate

(c)1998 by Prof. M. Kostic
http://www.ceet.niu.edu/faculty/kostic/