CEN-505 ENVIRONMENTAL HYDRAULICS

Dr. Pramod Kumar

Associate Professor
Civil Engineering Department
Indian Institute of Technology Roorkee
E-mail. prmdkfce@iitr.ac.in; drprmdkmr@gmail.com
Syllabus: Credits: 4. CWS-25, MTE-25, ETE-50. Total-100. Exam Duration: 3h
• Introduction and scope, review of basic principles of engineering fluid mechanics,
continuity, momentum, and energy equations, steady flow through pipes- hydraulic
gradient and total energy line, basics of open channel flow; Ground water, well hydraulics,
well design and constructions [7 hours]
• Parallel, compound and equivalent pipes, head losses in pipes, design of pressurized
conduits [2 hours]
• Mixing for environmental engineers, various forms of mixing in the environment [6 hours]
• Modeling the mixing: advection diffusion equation, various forms of advection diffusion
equation and its solution. Dispersion [10 hours]
• Special cases of mixing, density stratified flow, tide, literature review [4 hours]
• Mass transfer in gas-liquid and liquid-liquid systems with special emphasis on aeration,
literature review [10 hours]
• Project presentation [3 hours]

Suggested Books

• Chadwick, A., Morfett, J., Borthwick, M. "Hydraulics In Civil and Environmental

Engineering", 5th Edition, CRC Press. 1998
• Lee, C. C., Lin, S.D. "Handbook of Environmental Engineering Calculations", McGraw Hill.
2004
• Roberson, J.A., Cassidy, J.J., Chaudhry, M.H. "Hydraulic Engineering", 2 nd Edition, Wiley.
1996
• Schnoor, J.L., Environmental Modeling: Fate of Chemicals in Water, Air and Soil, John
Wiley & Sons, New York. 2007
Environmental Hydraulics

• Motion and mixing of fluids in the environment

• Transportation of fluid and calculations of substances and properties transported by the
fluid
• Examples include junction of two streams, contaminant spill into ocean or river,
smokestack discharge into atmosphere, mixing of warm and cold water in lakes, mixing
of salt and fresh water in estuaries and wastewater discharge into stream, estuary, or
ocean.

Basic Definitions

• Continuum Mechanics: Continuum is a state of matter where its properties are more or
less constant within the boundaries of space/domain and may be considerably different
from those outside the boundaries. Engineering behavior of materials is modeled as a
continuous mass rather than as discrete particles, e.g. solid mechanics, fluid mechanics,
aerodynamics.
• Fluid Mechanics: Rheology, i.e. science of deformation and flow. Deformation depends
on shear rate or velocity gradient, G, that is the relative velocity of two adjacent fluid
layers divided by distance between the layers. Typically, G=10-6 to 10-3 s-1 in sedimentation
process and 10-1000 s-1 in mixing process. G has units m/s/m, i.e. s-1.
• Viscosity: Resistance to deformation = shear stress/shear rate (i.e. μ=τ/G). Water viscosity
= 1 cP = 10-3 Pa.s at 20 0C [1 Poise, P =0.1 Pa.s]. It decreases by 3% for each 0C
temperature rise. Air viscosity = 1.825x10-5 Pa.s at 200C. It increases by 0.26% for each 0C
temperature rise. Coefficient of viscosity of Non-Newtonian fluids is by convention called
apparent viscosity, μa.
Herschel-Bulkley Law

Generalized relationship between τ and G, τ = τ0+KGn, where τ0 is yield stress, K is consistency

coefficient and n is flow behavior index. When τ0 =0, this relationship is called Power law of
viscosity, viz. τ = KGn . In addition, when n=1, it is called Newton’s Law of viscosity. Water obeys
Newton’s law and is also called Newtonian Fluid.

𝐸𝐴
For a Newtonian Fluid, 𝜇 = 𝛽𝐴 𝑒 𝑅𝑇 ,, where 𝛽𝐴 is called Arrhenius Coefficient, 𝐸𝐴 is activation energy
(kJ/mol) = 15.6 kJ/mol for water, R is universal gas constant = 8.3 kJ/kg.K and T is absolute
temperature (K).

In the following description, plastics have tendency to retain deformation even when the stress
is removed; thixotropic fluids, literally sensitive to touch like touch me not plants, show reduction
in viscosity under stress and regain its original value when stress is removed; rheopectic and
dilatant liquids do not have tendency to flow and their resistance to flow, i.e. apparent viscosity,
increases with increase in shear rate; while rheopectic liquids require that shear rate should be
sustained for longer durations to increase their μa, dilatant liquids immediately increase their μa
on increasing shear rate.

Non-Newtonian Fluids

𝜌𝑢2−𝑛 𝐷 𝑛 (4𝑛+2)(5𝑛+3)
Effects on Flow Characteristics: = 3𝑛+1 𝑛 𝑛−1
; 𝐶𝑟𝑖𝑡𝑖𝑐𝑎𝑙 𝑅𝑒 = 2100
𝐾( ) 8 3(3𝑛+1)2
4𝑛

Velocity profiles in Pressure driven flow through cylindrical pipe, viz. Poiseuille Flow:
𝑟2
Newtonian Laminar (n = 1): u = 2𝑢̅; 𝑢 = 2𝑢̅ (1 − 𝑅2 )
max
𝑟2
Newtonian Turbulent (n = 1): u = 1.2𝑢̅; 𝑢 = 2𝑢̅ (1 − 𝑅2 )
max

𝑛+1
n > 1: umax > 2𝑢̅ 3𝑛+1 𝑟 𝑛
Non- Newtonian Laminar Flows: } 𝑢 = 𝑢̅ 𝑛+1 [1 − (𝑅) ]
n < 1: umax < 2𝑢̅
 1000 Non-linear Plastic (Herschel-
Bulkley) e.g. Minced meat,
mashed potatoes
900
Bingham Plastic: e.g. Tooth
paste
800

700

600
Shear Stress, τ

500
Pseudo plastic, Shear thinning,
Thixotropic: n<1; e.g. Sauce,
Cream, Yogurt, Mayonnaise, etc.
400
Here G↑=> μa↓and at constant
G, τ↑ as t↑
τ
300
Newtonian: n=1; e.g. water, fruit
juices, milk, honey, vegetable
oils
200
Slope = μa
Rheopectic, Dilatant, Shear
100
thickening: n>1; e.g. Starch
solution, Lubricants, Printers’
ink; G↑ => μa↑ and at constant
0
G, τ↓ as t↑
0 1 2 3 G4 5
Shear Rate, G

Fig. Herschel-Bulkley Law

Transport

Water is an excellent solvent; transports substances or physico-chemical or biological properties;

for example, Heat, Turbidity, Color, Suspended Solids, Salinity, Dissolved oxygen, Dissolved solids,
Metals, Pesticides, BOD, pH, Fish eggs, Protozoa, Bacteria, Viruses
• Advection: Transport by an imposed flow as in a river or coastal waters.
• Convection: Vertical transport induced by flow instability such as the flow over a heated
plate, or below a chilled water surface in a lake.
• Molecular Diffusion: The scattering of particles by random motions of molecules due to
their kinetic energy in the stagnant fluid. This motion is called Brownian motion. It is called
turbulent diffusion when fluid particles have motion due to external forcing function.
• Dispersion or mixing: The scattering of particles or a cloud of contaminants by the
combined effects of shear and transverse diffusion. Mixing occurs due to any process
which causes one parcel of water to be mingled with or diluted by another. For example
turbulent diffusion in buoyant jets and plumes.

Bernoulli’s Equation
𝑝1 𝑣12 𝑝2 𝑣22
+ 𝛼1 + 𝑧1 + ℎ𝑝 = + 𝛼2 + 𝑧2 + ℎ𝑡 + ℎ𝐿
𝛾 2𝑔 𝛾 2𝑔
hp = head supplied by a pump
ht = head given to a turbine
hL = mechanical energy converted to thermal
1 𝑉
α = kinetic energy correction factor= 𝐴 ∫(𝑉̅)3 𝑑𝐴
𝑉𝑚𝑎𝑥
= 1+ε2 where ε = ̅
-1
𝑉

= 2 for laminar flow

= 1.04 – 1.11 for turbulent flow.

α is assumed 1 for turbulent flow, without much error. Flow is invariably turbulent in most cases
in environmental hydraulics.
 EGL (or TEL) and HGL

𝑝1 𝑣2
EGL= + 𝛼1 2𝑔1 + 𝑧1
𝛾 𝑝1
HGL= + 𝑧1
𝛾

Pressure head (w.r.t. Velocity Position head (w.r.t. datum) Energy Grade
reference pressure) head Line (EGL)
Velocity head, Hydraulic Grade
𝑣2 Line (HGL)
𝛼1 2𝑔1

Static head, z Pressure head,

𝑝1
𝛾

Position head,
z
Pump

z=0 Datum/Reference

Head-loss
Major Head-loss: Straight length of pipe
2∆𝑝 2𝑔ℎ𝐿
Based on Dimensional Analysis, 𝐶𝑝 = − 𝜌𝑉 2 ; 𝛾ℎ𝐿 = −2∆𝑝 = −∆(𝑝 + 𝛾ℎ) ⇒ 𝐶𝑝 = ;
𝑉2

𝐷 2𝑔ℎ𝐿 𝐷 𝐿 𝑉2
𝑓 = 𝐶𝑝 𝐿 = ⇒ ℎ𝐿 = 𝑓 𝐷 2𝑔 : Darcy-Weisbach equation.
𝑉2 𝐿
64
Laminar Flow: Hazen-Poiseuille: 𝑓 = 𝑅𝑒

1 𝜀 2.51
Transition Flow: Colebrook-White: = −2𝑙𝑛 (3.7𝐷 + )
√𝑓 𝑅𝑒√𝑓

0.25
Or, Swamee-Jain explicit (ε/D<0.02, Re>3000): 𝑓 = 𝜀 5.74 2
(𝑙𝑛{ + })
3.7𝐷 𝑅𝑒0.9

1 𝐷
Turbulent Flow: Nikuradse: = 1.14 + 2𝑙𝑛 ( 𝜀 ) ; 𝑓 = 0.02 when 𝑅𝑒 > 2 × 105
√𝑓
 Pipe Material Pipe Roughness ε (mm)
Glass, Drawn Brass, Copper 0.0015
Commercial Steel, Wrought Iron 0.045
Asphalted Cast Iron, PVC 0.12
Galvanized Iron 0.15
Cast Iron 0.26
Concrete 0.18-0.6
Rivetted Steel 0.9-9.0
Corrugated Metal 45

Minor Head-Loss: Pipe fittings

Most minor losses cannot be obtained analytically, so they must be measured. They are often
expressed as a loss coefficient, K, times the velocity head. In the absence of detailed analysis,
they are assumed as 30 percent of major losses.

2∆𝑝 2𝑔ℎ𝐿 𝑉2
𝐶𝑝 = − 𝜌𝑉 2 ⇒ 𝐶𝑝 = ⇒ ℎ𝐿 = 𝐾 2𝑔;
𝑉2

K = 0.5 Entrance;
= 1.0 Exit into atmosphere
𝐴 2
= (1 − 𝐴1 ) for sudden expansion
2

Hazen-William’s Expression
It considers effects of aging of pipes on increase in their surface roughness. Bureau of Indian
Standards codes of practice also recommend using this expression for pipe design/analysis.

10.675𝐿 𝑄 1.852
ℎ𝐿 = (𝐶 ) in SI units where C = Hazen-William’s coefficient
𝐷 4.8704
 C Condition
150 PVC, Copper, Brass
140 Extremely smooth, straight pipes; asbestos cement
130 Very smooth pipes; concrete; new cast iron
120 Wood stave; new welded steel
110 Vitrified clay; new riveted steel
100 Cast iron after years of use (Default for metal pipes)
95 Riveted steel after years of use
60-80 Old pipes in bad condition

Entrance Length: Flow characteristics are undefined within entrance length. Le/D = 0.06 Re for
Laminar flow and 4.40 Re1/6 for Turbulent flow
Multiport Diffusers
 Port, Vp

EGL

Riser
Pipe, Vr

HGL

Diffuser
Pipe, Vd

Using Bernoulli’s equation between sections just before the riser in the diffuser pipe and port
outlet:

𝑝𝑑 𝑣𝑑2 𝑝𝑝 𝑣𝑝2 𝑝 𝑣𝑑2 𝑣𝑝2

+ + 𝑧𝑑 = + + 𝑧𝑝 + ℎ𝐿 ; 𝐻 = + 𝑧 ⇒ 𝐻𝑑 + = 𝐻𝑝 + + ℎ𝐿
𝛾 2𝑔 𝛾 2𝑔 𝛾 2𝑔 2𝑔
𝑣𝑝2
𝑉𝑑 as well as 𝐻𝑝 are negligible in comparison to 𝑉𝑝 and 𝐻𝑑 respectively ⇒ 𝐻𝑑 = 2𝑔 + ℎ𝐿

ℎ𝐿 = ℎ𝑒𝑛𝑡𝑟𝑦 + ℎ𝑟𝑖𝑠𝑒𝑟 + ℎ𝑒𝑙𝑏𝑜𝑤

𝐷𝑝 2
𝑣𝑟 𝐷𝑟2 = 𝑣𝑝 𝐷𝑝2 ⇒ 𝑣𝑟 = 𝑣𝑝 ( )
𝐷𝑟
 𝑣𝑝2 2𝑔𝐻𝑑
𝐻𝑑 = 𝐾𝑟 ⇒ 𝑣𝑝 = √
2𝑔 𝐾𝑟

𝐿𝑟 𝐷𝑝 4
𝐾𝑟 = 1 + (𝐾𝑒𝑛 + 𝑓 + 𝐾𝑒𝑙 ) ( )
𝐷𝑟 𝐷𝑟
𝜋𝐷𝑝2 2𝑔𝐻𝑑
Therefore 𝑄𝑝 = √ . This is called orifice equation. 𝐻𝑑 can be calculated by 𝐻𝑑 =
4 𝐾𝑟

2
𝐾𝑟 4𝑄𝑝
( )
2𝑔 𝜋𝐷 𝑝

𝜋𝐷𝑝2 2𝑔𝐻𝑑
𝑄𝑝 = √
4 𝐾𝑟

Flow separation (≈ 5Dd length)

𝑣𝑖 𝑄𝑝𝑖
𝑣𝑖+1 = 𝑣𝑖 −
𝐴𝑑
𝑄𝑖 ℎ𝐿 = ℎ𝑒𝑛𝑡𝑟𝑦 + ℎ𝑟𝑖𝑠𝑒𝑟 + ℎ𝑒𝑙𝑏𝑜𝑤
𝑣𝑖+1 𝑄𝑝𝑖
𝐻𝑑𝑖 𝐻𝑑𝑖+1 = 𝐻𝑑𝑖 + 𝛥𝐻𝑖 + 𝛥𝐻𝑝𝑖𝑝𝑒 𝑖 ; 𝛥𝐻𝑖 =
𝑔𝐴𝑑

Flow separation causes uniformly distributed pressure across entire diffuser area just before riser
and momentum transferred through reduced flow area, just after riser.

BE in diffuser pipe between just before and just after ith riser, sudden expansion
2 2
𝑣2 𝑣𝑖+1 𝑣2 𝑣𝑖+1
𝐻𝑖 + 2𝑔𝑖 = 𝐻𝑖+1 + + ℎ𝐿𝑖 ⇒ 𝛥𝐻𝑖 = 𝐻𝑖+1 − 𝐻𝑖 = 2𝑔𝑖 − − ℎ𝐿𝑖
2𝑔 2𝑔

𝑣𝑖2 𝑣𝑖+1
2
(𝑣𝑖 − 𝑣𝑖+1 )2 𝑣𝑖+1 (𝑣𝑖 − 𝑣𝑖+1 )
𝛥𝐻𝑖 = 𝐻𝑖+1 − 𝐻𝑖 = − − =
2𝑔 2𝑔 2𝑔 𝑔
𝑄𝑝𝑖 𝑄𝑝𝑖 𝑣𝑖+1 𝑄𝑝𝑖
𝑄𝑖 = 𝑄𝑝𝑖 + 𝑄𝑖+1 ; 𝑣𝑖 − 𝑣𝑖+1 = ⇒ 𝑣𝑖+1 = 𝑣𝑖 − ⇒ 𝛥𝐻𝑖 =
𝐴𝑑 𝐴𝑑 𝑔𝐴𝑑

Tℎ𝑒 𝑝𝑖𝑒𝑧𝑜𝑚𝑒𝑡𝑟𝑖𝑐 ℎ𝑒𝑎𝑑 𝑐𝑎𝑛 𝑏𝑒 𝑐𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑒𝑑 𝑏𝑦: 𝐻𝑑𝑖+1 = 𝐻𝑑𝑖 + 𝛥𝐻𝑖 + 𝛥𝐻𝑝𝑖𝑝𝑒 𝑖
Example.
Flow 2.6 m3/s Diffuser dia 1m; Port spacing 4m; Number of ports 20; Riser diameter 0.5m; Port
diameter 0.23m. Therefore Kr= 1.042, Hd=0.478m
PORT X Vi Hi ΔHi Hdi
1 4 3.310 5.31E-02 4.02E-02 0.478 0.995
2 8 3.144 5.03E-02 3.61E-02 0.491 0.957
3 12 2.979 4.75E-02 3.22E-02 0.505 0.924
4 16 2.813 4.47E-02 2.85E-02 0.520 0.894
5 20 2.648 4.19E-02 2.51E-02 0.536 0.867
6 24 2.482 3.91E-02 2.18E-02 0.553 0.844
7 28 2.316 3.63E-02 1.88E-02 0.571 0.824
8 32 2.151 3.35E-02 1.60E-02 0.588 0.806
9 36 1.985 3.07E-02 1.35E-02 0.606 0.792
10 40 1.820 2.79E-02 1.11E-02 0.623 0.779
11 44 1.654 2.51E-02 9.02E-03 0.640 0.769
12 48 1.488 2.23E-02 7.12E-03 0.656 0.760
13 52 1.323 1.95E-02 5.45E-03 0.671 0.753
14 56 1.157 1.67E-02 4.00E-03 0.685 0.748
15 60 0.992 1.39E-02 2.78E-03 0.698 0.744
16 64 0.826 1.11E-02 1.77E-03 0.709 0.740
17 68 0.660 8.35E-03 9.96E-04 0.718 0.738
18 72 0.495 5.56E-03 4.41E-04 0.726 0.736
19 76 0.329 2.76E-03 1.09E-04 0.731 0.735
20 80 0.164 1.69E-06 4.07E-11 0.733 0.733
Ei Notes.
 Multiport diffusers typically
inject the fluid into water
bodies horizontally. The ports
are placed at least 1m above
the bottom of water bodies to
avoid sinking of fluids due to
Cowanda effect (suction is
created near the bottom of
water body if fluids injected
move near the bed). The design
of these systems can be done
for constant pressure in the
diffuser or for constant
discharge through all ports.
 The sum of all port areas must
be less than the diffuser pipe
area. The best area ratio (port
area/diffuser area) is usually
between 1/3 and 2/3. In this
range, the effects of pipe
friction and pressure recovery
at expansion will tend to cancel
when Dd=fLd/3, where Ld is a
the total length of the diffuser
pipe and the friction factor, f, is
obtained by iteration since it is
a function of the pipe
diameter.
 1.2

1.0
EGL

0.8
Head (m)

0.6

0.4 HGL

Exact Shape of HGL

0.2

0.0
0 16 32 48 64 80
Distance along diffuser, x (m)

Pipe Networks

10.675𝐿 𝑄 1.852
Darcy-Weisbach Equation: ℎ𝐿 = (𝐶 ) ⇒ ℎ𝐿 = 𝐾𝑄 𝑛 ;
𝐷 4.8704

10.675𝐿 𝑄 1.852
Hazen-Williams Equation: ℎ𝐿 = (𝐶 ) ⇒ ℎ𝐿 = 𝐾𝑄 𝑛
𝐷 4.8704
Consider a pipe network having P pipes and N nodes.

Node 1 Pipe 1 Node 2

Pipe 4 Pipe 2

Pipe 3
Node 3
Node 4

• Continuity or node equations: The algebraic sum of the flows at any node must be zero,
i.e. flows into and away from each node must balance. ΣQij + Si = 0 for any node i. Qij = flow from
node i to node j. Si is the external supply (+)/demand (-) at node i. Maximum number of
independent node equations is (N-1).
• Loop equations: The integrated head loss around any loop must be zero. Σhij= 0 for each
loop hij= Hi - Hj = Kij.Qijm. Maximum number of independent loop equations is P-(N-1). Total
number of independent equations =P
• Sign Convention: While traversing a loop, clockwise positive

Boundary conditions:
• Magnitude of supplies and demands are known; one nodal pressure is known; or
• Magnitudes of pressures at supply nodes and magnitudes of demands are known.

Direct solution of systems of non-linear simultaneous equations is not feasible, hence it is

necessary to use iterative solution methods. Commonly used methods are as below.

• Loop by loop flow correction (Hardy Cross) Method

𝑑ℎ
F= Σhij = 0 => Σhij = Σ (Hi- Hj) = Σ Kij.Qijn = 0; −ℎ𝑖𝑗 = 𝑑𝑄𝑖𝑗 ∆𝑄𝑖𝑗
𝑖𝑗

𝑛
ℎ 𝛴𝐾𝑖𝑗 𝑄𝑖𝑗 𝛴ℎ𝑖𝑗
⟹ ∆𝑞𝑖𝑗 = − 𝑑ℎ𝑖𝑗 𝑖𝑗 ; ∀𝑖, 𝑖 ∈ 𝐿 ⟹ ∆𝑞 = − 𝑛𝛴𝐾 𝑛−1 =- ℎ
⁄𝑑𝑞 𝑖𝑗 𝑄𝑖𝑗 𝑛𝛴( )𝑖𝑗
𝑖𝑗 𝑄
Flow corrections are repeated on a loop by loop basis until the maximum out-of-balance loop
head is reduced to a specified tolerance value.

• Simultaneous loop flow correction

𝜕ℎ
For each loop, −ℎ𝐿 = ∑𝑖 𝜕𝑄 𝐿 𝛥𝑄𝑖𝑗 𝑖 = 1,2, … , 𝑁𝑜. 𝑜𝑓 𝐿𝑜𝑜𝑝𝑠
𝑖𝑗

𝜕ℎ
[𝐶][∆𝑄] = [−ℎ] Where [𝑐(𝑖, 𝑗)] = 𝜕𝑄𝑖𝑗
𝑖𝑗

• Linearization of Network Equations

̅𝑖𝑗 𝑄𝑖𝑗 = 0
Σhij = 0 => Σhij = ΣKij.Qijn = 0; and ΣKij.Qijn-1 Qij= 0 Or 𝛴𝐾
̅ pipe
After each solution of the full set of linearized network equations, the new set of 𝐾
coefficients is calculated for use in the next iteration.

Convergence of all iterative methods is aided by judicious selection of the initial flow distribution.
The Hardy Cross method provides the least rapid convergence. This is because each loop flow
correction is made in isolation from the rest of the network, whereas the other two methods
incorporate full network interaction in each iteration.

Example
A water supply distribution system is shown in the figure below. All pipes are cast iron with
lengths and diameters as provided in table below. Perform pipe network analysis and calculate
water flow in all branches. ε=0.26mm, HLoss = K Q² ; K = 8fL/ π²g D 5
F(I) : K1.Q1²+K3.Q3²-K8.Q8²-K4.Q4²-K2.Q2²=0
F(II) : K5.Q5² + K7.Q7² - K6.Q6² - K3.Q3² = 0
F(III): K6.Q6²+K10.Q10²-K9.Q9²+K8.Q8²= 0

The Jacobian matrix, viz. matrix of derivatives, JL, is such that JL.ΔQ = - F(Qm-1) =>JL = δF / δ(ΔQ)
where Qm-1 is vector of pipe flow, ΔQ is vector of loop flow corrections and F(Qm-1) is the vector
of re siduals of loop conservation of energy equations evaluated at Qm -1. JL is first derivatives of
the loop equations evaluated at Qm-1. pipe flows are updated by the loop corrections:
Qm = Qm-1 +/- ΔQ.

Pipe No. L, m D, m Q, m3/s K hL nhL/Q

1 300 0.30 0.2 199 7.95 79.49
2 250 0.25 0.1 434 4.34 86.79
3 125 0.20 0.12 690 9.94 165.60
4 300 0.20 0.10 1662 16.62 332.34
5 350 0.20 0.08 1948 12.47 311.74
6 350 0.20 0.07 1955 9.58 273.72
7 125 0.20 0.03 722 0.65 43.29
8 125 0.15 0.05 3148 7.87 314.81
9 350 0.20 0.05 1976 4.94 197.58
10 125 0.15 0.10 3111 31.11 622.13

𝜕𝐹1 𝜕𝐹1 𝜕𝐹1

𝜕(∆𝑄1 ) 𝜕(∆𝑄2 ) 𝜕(∆𝑄3 )
𝜕𝐹2 𝜕𝐹2 𝜕𝐹2
𝐽𝐿 =
𝜕(∆𝑄1 ) 𝜕(∆𝑄2 ) 𝜕(∆𝑄3 )
𝜕𝐹3 𝜕𝐹3 𝜕𝐹3
[𝜕(∆𝑄1 ) 𝜕(∆𝑄2 ) 𝜕(∆𝑄3 )]
Derivative for single pipe: δ(K1.Q1²)/δ(ΔQI) = 2.K1.Q1 = 2.HL1/Q
Diagonal terms: Add derivatives of all pipes in a loop [always positive]: δFI/δ(ΔQI) = n(|HL1/Q1|
+ |HL3/Q3| +|HL8/Q8|+ |HL4/Q4| + |HL2/Q2|)=979.03
Off-diagonal terms: The gradients for pipes that appear in different loops [always negative]:
For example, δFI/δ(ΔQII) = -n(|HL3/Q3|) = δFII/δ(ΔQI) = -165.6

979.03 -165.60 -314.81

𝐽𝐿 = [-165.60 794.35 -273.72 ]
-314.81 -273.72 1408.23
1.19 × 10−3 3.65 × 10−4 3.37 × 10−4
JL-1= [3.65 × 10−4 1.46 × 10−3 3.65 × 10−4 ]
3.37 × 10−4 3.65 × 10−4 8.57 × 10−4

F(Qm-1) is evaluated based on head loss across each loop: FI = HL1 + HL3 - HL8 - HL4 - HL2=-10.84
JL.ΔQ = - F; ΔQ = [ 0.00066 -0.00261 -0.03133]
Adjust the flows, e.g. Q3 = Q30 + ΔQI – ΔQII = 0.12 + (0.00066) - (-0.00261) = 0.123
Repeat this until ΔQ ≈ 0

Notes: If a pipe’s flow direction changes from the assumed value, the signs for that pipe head
loss terms are switched for all loops containing the pipe during the next iteration in loop
equations. JL matrix signs will remain same as above.

Network Flow Controls

• Non-return valves
If the computational procedure reveals a backward flow through a non-return valve, a correction
can be made by introducing an equal and opposite loop flow or alternatively the network can be
re-analyzed with the pipe containing the non-return valve omitted.

• Flow-regulating valves
Such devices cause a step-change in head H, which can usually be related to the flow Q by an
equation of the form: ℎ𝑣 = 𝐾𝑣𝑄2 where Kv is a variable coefficient whose value is a function of the
extent of valve opening. The flow correction ∆Q for a loop containing a valve is found by adding
the appropriate valve terms to equation
∑ ℎ𝑖𝑗 + ℎ𝑣
∆𝑄 =
ℎ
𝑚 ∑ (𝑄 ) + 2𝐾𝑣 𝑄
𝑖𝑗

• Pumps
Pumps cause a step-increase in head in the direction of flow, which can generally be
determined from pump characteristic equation of the form ℎ𝑝 = 𝐴0 + 𝐴1𝑄 + 𝐴2𝑄2, where A0 is
the ‘shut-off’ head, and A1 and A2 are constant coefficients
∑ ℎ𝑖𝑗 + ℎ𝑝
∆𝑄 = −
ℎ
𝑚 ∑ (𝑄 ) + 2𝐴2 𝑄 + 𝐴1
𝑖𝑗

• Points of Fixed Heads

∑ ℎ𝑖𝑗 + (𝐻𝑥 − 𝐻𝑦 )
∆𝑄 = −
ℎ
𝑚 ∑ (𝑄 )
𝑖𝑗

Example

L1

61m A
0.11
C
B

L2
0.09
E H I
L4
2
D H=72-35Q
0.14 0.17
L3

0.14
F G
 Pipe L D Q hL nhL/Q
AB 300 0.75 0.6 1.00 3.08
BC 600 0.6 0.2 0.77 7.16
CE 450 0.6 0.15 0.34 4.20
BD 600 0.4 0.4 20.12 93.16
DE 670 0.15 0.06 79.50 2453.77
DF 300 0.4 0.2 2.79 25.81
EG 450 0.3 0.12 6.59 101.70
FG 600 0.2 0.06 17.54 541.26
CH 900 0.15 0.06 106.79 3296.11
GH 900 0.15 0.04 50.40 2333.31
HI 0.19 2X35Q=13.3

JL: 3309.2 -7.2 0.0 -3296.1 F: -114.75

-7.2 2558.3 -2453.8 -4.2 98.50

0.0 -2453.8 3122.5 -101.7 -65.76

-3296.1 -4.2 -101.7 5735.3 164.11

JL-1: ΔQ: -1.23E-02

7.1E-04 6.3E-05 6.3E-05 4.1E-04
6.3E-05 1.6E-03 1.3E-03 5.9E-05 7.72E-02

6.3E-05 1.3E-03 1.3E-03 6.0E-05 4.03E-02

4.1E-04 5.9E-05 6.0E-05 4.1E-04 2.23E-02

Open Channel Flow
1 2⁄ 1⁄
Manning’s Equation: 𝑉 = 𝑛 𝑅 3𝑆 2

V= Average Velocity (m/s)

R= Hydraulic Radius (m)
S = Energy Gradient (m/m)
n = Manning’s Coefficient

′
𝑛 𝑑 𝑑 2 𝑑 3 𝑑 4
𝐶𝑎𝑚𝑝 𝑠 𝐸𝑥𝑝𝑟𝑒𝑠𝑠𝑖𝑜𝑛: = 𝑓(𝑑) = 1.04 + 2.3 ( ) − 6.86 ( ) + 7.76 ( ) − 3.27 ( )
𝑁 𝐷 𝐷 𝐷 𝐷
d/D n/N
0.0 ≤ d/D ≤ 0.03 1+ d/(0.3D)
0.03 ≤ d/D ≤ 0.1 1.1 + (d/D – 0.03)(12/7)
0.1 ≤ d/D ≤ 0.2 1.22 + 0.6 (d/D – 0.1)
0.2 ≤ d/D ≤ 0.3 1.29
0.3 ≤ d/D ≤ 0.5 1.29 – 0.2(d/D – 0.3)
0.5 ≤ d/D ≤ 1.0 1.25 – 0.5 (d/D – 0.5)

Sewers
Crown level
Parts of sewers:
Soffit level

Spring line

Invert level

Barrel level
Sewers are designed to flow only partially full, say 80-85% =>d ≈ 0.8D

d D

Hydraulic Elements: θ = 2cos-1[ (D - 2h)/D ]; R = A/P; P = πD – Dθ/2 ; A = πD2/4 – D2(θ – sinθ)/8

Typical flows: Maximum Flow = 2x Average Flow; Minimum Flow = Average Flow/3
Typical velocities: Minimum velocity = 0.6 m/s; Maximum velocity = 3 m/s

Additional Flows:
Infiltration Minimum Maximum
L/ha.d 5000 50000
L/km.d 500 50000
L/d.MH 250 500
Slopes:
Sewer Dia, mm Min. Slope, % (1 in n)
150 0.60 (170)
200 0.40 (250)
250 0.28 (360)
300 0.22 (450)
375 0.15 (670)
450 0.12 (830)
≥525 0.10 (1000)
Nomenclature:

[Note: The Hydraulic elements may also be written as for example v/V or q/Q in place of V/V f or
Q/Qf respectively.]
Example. Design the trunk sewer for Roorkee city with the following information: Population 2
lac; Water supply rate = 150 lpcd; Wastewater generated = 80% of water supplied; peak factor =
3; minimum flow = one-third of average flow. Design requirements are: minimum velocity = 0.6
m/s; maximum velocity = 3 m/s; maximum depth of flow = 80% full depth.

Solution:

As observed from the nomogram, q/Q ≈1 at d/D ≈0.8, where a/A ≈ 0.88. Average flow =
0.8x150x200000=2,40,00,000 L/d= 24 MLD = 0.278 m3/s. Minimum flow 0.278/3= 0.093 m3/s and
peak flow = 3x0.278 = 0.833 m3/s and. X-sect area required for peak flow, a = 0.833/3= 0.278 m2.
Since a/A=0.88, A=0.278/0.88=0.316 m2. Diameter of pipe required = (4x0.316/π)1/2= 0.634m.
Nearest available diameter = 700mm that has x-sect area = 0.385 m2. For this pipe, a/A=
0.278/0.385 = 0.72 at which d/D=0.67 and q/Q=0.83. Therefore it can carry 0.833/0.83 = 1 m3/s
flow. Velocity at full depth flow= 1/0.385=2.6 m/s. Using n=0.013 for CI pipe,
V=2.6=(1/0.013)(0.7/4)2/3S1/2=> S=0.0117=> Provide 1 in 80 slope.

Check for average flow: q/Q=0.278 whence d/D=0.39 and v/V = 0.85 =>v=0.85x2.6 = 2.2 m/s OK

Check for minimum flow: q/Q=0.093, whence d/D =0.23 and v/V = 0.6 =>v=0.6x2.6=1.56 m/s. OK.

You may like to do these calculations using the curves for variable Mannings’ coefficient.
Bernoulli’s Equation: Unsteady Flow

Consider a tank conveying water through a long pipe of length L as shown below. Assuming
inviscid flow and h constant in the tank, we can write unsteady Bernoulli’s equation, basically
Euler’s equation, as below.

h
L

a b 2

𝜕𝑣𝑠 𝜕𝑣𝑠 𝜕𝑃
𝜌 + 𝜌𝑣𝑠 = − 𝜕𝑠 + 𝜌𝑔𝑠
𝜕𝑡 𝜕𝑠
2 𝜕𝑣𝑠 2 𝜕𝑣𝑠
⇒ ∫1 𝜌 𝑑𝑠 + ∫1 𝜌𝑣𝑠 𝑑𝑠 = −(𝑃2 − 𝑃1 ) − 𝜌𝑔 (𝑧2 − 𝑧1 )
𝜕𝑡 𝜕𝑡
2 𝜕𝑣𝑠 1 1
⇒ ∫1 𝜌 𝑑𝑠 + (𝑃 + 2 𝜌𝑣𝑠2 + 𝜌𝑔𝑧) − (𝑃 + 2 𝜌𝑣𝑠2 + 𝜌𝑔𝑧) = 0
𝜕𝑡 2 1
2 𝜕𝑣𝑠 1 1
⇒ ∫1 𝜌 𝑑𝑠 + (𝑃𝑎 + 2 𝜌𝑣22 + 𝜌𝑔(0)) − (𝑃𝑎 + 2 𝜌(0)2 + 𝜌𝑔ℎ) = 0
𝜕𝑡
2 𝜕𝑣𝑠 𝑎 𝜕𝑣𝑠 𝑏 𝜕𝑣𝑠 2 𝜕𝑣𝑠 2 𝜕𝑣𝑠 𝜕𝑣2
∫1 𝜌 𝜕𝑡
𝑑𝑠 = ∫1 𝜌 𝜕𝑡
𝑑𝑠 + ∫𝑎 𝜌 𝜕𝑡
𝑑𝑠 + ∫𝑏 𝜌 𝜕𝑡
𝑑𝑠 ≈ ∫𝑏 𝜌 𝜕𝑡
𝑑𝑠 = 𝜌 𝜕𝑡
𝐿

𝜕𝑣2 1 2𝐿
⇒𝜌 𝐿 + 𝜌𝑣22 = 𝜌𝑔ℎ ⇒ 𝑑𝑣2 = 𝑑𝑡
𝜕𝑡 2 𝑣22⁄
2𝑔ℎ − 2
𝑣2 𝑡 𝑣2 𝑡
𝑑𝑣2 𝑑𝑡 1 𝑑𝑣2 𝑑𝑡
⇒∫ 2 = ∫ ⇒ ∫ 2 = ∫
0 𝑣
2𝑔ℎ − 2⁄2 0 2𝐿 2√2𝑔ℎ 0 2𝑔ℎ − 𝑣2⁄ 0 2𝐿
2
𝑣2 𝑡
1 1 1 𝑑𝑡 √2𝑔ℎ
⇒ ∫ ( + ) 𝑑𝑣2 = ∫ ⇒ 𝑣2 = √2𝑔ℎ 𝑡𝑎𝑛ℎ ( 𝑡)
2√2𝑔ℎ 0 √2𝑔ℎ + 𝑣2 √2𝑔ℎ − 𝑣2 0 2𝐿 𝐿
The system has Torricelli velocity, 𝑉2 = √2𝑔ℎ and characteristic time to flow L length, 𝑇 =
𝐿/𝑉2 . For example, for h=10m and L=2km, 𝑉2 = 14 𝑚/𝑠 and T=142.8 s. Using this notation,
𝑣2 𝑡
= tanh (𝑇).
𝑉2

1.2

0.8
v/V

0.6

0.4

0.2

0
0 2 4 6 8 10
t/T

Consider the same situation again. After the velocity through the pipe reaches steady state, the
valve at the end of pipe, at point 2, is closed gradually. In this case the pressure developed just
upstream of point 2 can be calculated as below when the cross-section area of flow through the
valve is linearly reduced to zero.
2 𝜕𝑣𝑠 1 1
∫1 𝜌 𝑑𝑠 + (𝑃 + 2 𝜌𝑣𝑠2 + 𝜌𝑔𝑧) − (𝑃 + 2 𝜌𝑣𝑠2 + 𝜌𝑔𝑧) = 0
𝜕𝑡 2 1
2 𝜕𝑣𝑠 1 1
∫1 𝜌 𝑑𝑠 + (𝑃2 + 2 𝜌𝑣22 + 𝜌𝑔(0)) − (𝑃𝑎 + 2 𝜌(0)2 + 𝜌𝑔ℎ) = 0
𝜕𝑡
2 2
𝜕𝑣𝑠 𝜕𝑣𝑠 𝜕𝑣2
∫ 𝜌 𝑑𝑠 ≈ ∫ 𝜌 𝑑𝑠 = 𝜌 𝐿 as before and 𝑃2 − 𝑃𝑎 = 𝑃
1 𝜕𝑡 𝑏 𝜕𝑡 𝜕𝑡
𝜕𝑣2 1 𝑄 𝑡
⇒𝜌 𝐿 + 𝜌𝑣22 + 𝑃 − 𝜌𝑔ℎ = 0; v2 = ; Let 𝐴(𝑡) = 𝐴 (1 − )
𝜕𝑡 2 𝐴 𝑇
𝑄2 𝐿 𝑑𝑄 𝑄2 𝑡
Therefore 𝑃 = 𝜌 [𝑔ℎ − 2 − ] = 𝜌 [𝑔ℎ − 2 − 𝐿𝑄 (1 − )]
2𝐴 𝐴 𝑑𝑡 2𝐴 𝑇
Consider the same situation yet again. An application requires the discharge to be varied only
linearly with time. It can be done via variation in the area of valve opening at the end of the pipe.
𝑡
Let 𝑄(𝑡) = 𝑄 (1 − 𝑇) ;
𝑄2 𝐿 𝑑𝑄
𝑃 = 𝜌 [𝑔ℎ − 2𝐴2 − 𝐴 𝑑𝑡 ] as before
1 1
Consider a point 3 just outside the valve: 𝑃2 − 𝑃3 = 2 𝜌𝑣32 − 2 𝜌𝑣22

𝑄2 𝐿 𝑑𝑄 1 𝑄2 𝑄2 𝐿 𝑑𝑄 𝑄2
⟹ 𝜌 [𝑔ℎ − − ] = 𝜌 [ − ] whence, 𝑔ℎ − =
2𝐴2 𝐴 𝑑𝑡 2 𝐴2 𝐴20 𝐴 𝑑𝑡 2𝐴2
𝑡
𝐿 1 1 𝑡 2 𝐴0 (1− )
⟹ 𝑔ℎ − 𝐴 [− 𝜏 𝐴0 √2𝑔ℎ] = 𝜌𝑔ℎ𝐴20 (1 − 𝑇) ⟹ 𝐴 = 𝑇
0 2𝐴2 𝐿
√(1+ )√2⁄𝑔ℎ
𝑇

Reynolds Transport Theorem

Control Mass (CM) or System Control Volume (CV): X-Sect Area = Control Surface (CS)

Eulerian
Lagrangian Frame
Frame At time t=0
At time
t =0

CM CV CM and CV CV CM
At time t-Δt At time t At time t+Δt

Mass Conservation:
𝑑𝑀
𝑀 = ∫ 𝜌 𝑑∀; | = 0;
𝑑𝑡 𝐶𝑀
Mass conservation in CV:
At time t,
𝜕
∫ 𝜌 𝑑∀ + ∫ 𝜌𝑉⃗ . 𝑑𝐴 = 0
𝜕𝑡 𝐶𝑉 𝐶𝑆

This is called Reynolds Transport Theorem or Green’s Theorem, which is fundamental to all
calculations where CM and CV coincide with each other. The direction of area vector points
perpendicularly outward from the area. The inflows are therefore considered negative and
outflows, positive.

Example
Oxygen is supplied from a pressurized tank 30 cm in dia. and 2m high. A valve is used to maintain
the oxygen flowing out at a constant velocity of 2 m/s. The valve has an opening area of 10 -4 m2
and the temperature inside the tank is maintained at 25°C. If the initial pressure inside the tank
is 10 MPa gauge, calculate the duration to reduce the pressure in tank to 90% of its initial value.
CV CS
2 m/s

Solution.
Control Volume: Tank; Control Surface: Valve Opening
P
MCV = ρ∀ and ρ = RT for ideal gas; R = 260 J/kg.K for O2
𝜕 ∀ dP dP
Since T is a constant, 𝜕𝑡
∫ ρ d∀ = RT d𝑡 = 18.2 × 10−7 d𝑡

P
⃗ . 𝑑𝐴 = −ρ𝐴𝑉 = − × 10−4 × 2
Mass flow from tank = ∫𝐶𝑆 𝜌𝑉 RT
−9 −7
dP dP
= −2.58 × 10 𝑃 = 18.2 × 10 ⇒ = −0.00142 𝑑𝑡
d𝑡 𝑃
9 𝑑𝑃 9
Integrating, ∫10 𝑃 = − ∫ 0.00142 𝑑𝑡 ⇒ 𝑙𝑛 (10) = −0.00142 𝑡
Simplifying, t = 73.94 s, That is, it takes approximately 74 seconds.
Boundary Layer
The theory of boundary layer was introduced by Prandtl in 1904.

𝑢∗ 𝑦
𝑢̅(𝑦) = ln(𝑦 )
𝜅 0

5𝜈
𝛿𝑠 = ; 𝑢̅(𝑦) = 𝑢∗2 𝑦/𝜈
𝑢∗

̅
𝜕𝑢 ̅
𝜕𝑢 𝜏
𝑢𝑟𝑚𝑠 ~ 𝜕𝑥 ; i.e. u scales on 𝜕𝑥 . u’  u = shear velocity =√ 𝜌0 = √𝑔𝑑𝑆 ∼ 0.1𝑢̅
rms *
5.2𝑥 0.3𝑥
𝛿 = (𝑅𝑒 0.5 i. e. ∝ 𝑥 0.5 for laminar and δ = (𝑅𝑒 0.2 i. e. ∝ 𝑥 0.8 for turbulent flow.
𝑥) 𝑥)
𝑧 𝜐
Viscous Sublayer: z<z0 where z0 = z at which Re=1; 𝑢̅(𝑦) = 𝑢∗ 𝑓 (𝑧 ) ; 𝑧0 = 𝑢∗
0

Turbulence

It is a state of the instability (oscillatory movements) of fluid particles. The properties of the flow
are characterized by intensity of the velocity fluctuations (random variations in magnitude and
direction) and size of the fluctuations (viz. length scale)
 𝑢

𝑢′
𝑢̅

Reynolds Process

In the process of induction of turbulence in the flow, the velocity fluctuations are described
as 𝑢 = 𝑢̅ + 𝑢′ , that is, Instantaneous Velocity = Mean Velocity + Velocity Fluctuation.

∑ni=1(ui − u̅)2
′
u = urms =√
n−1
Boussinesq Assumption

Large scale circulations called eddies are set-forth due to turbulence in the flow. The size of these
eddies typically corresponds to smallest dimension of flow. These large eddies feed energy to
small eddies and eventually the energy is dissipated as heat at molecular level by viscosity.
Boussinesq assumed that the momentum transfer caused by turbulent eddies can be modeled
with an eddy viscosity.

The energy dissipation (or energy transfer) in fluid flow is a cascading process, facilitated with
the help of eddies:
 Size of eddy is turbulent length scale that vary considerably. However, the smallest one is
larger than molecular length scale.
 Eddies overlap in space with large eddies carrying small eddies.
 Eddies convect with flow and therefore turbulence is not a local phenomenon; it depends
on history of the eddy.
 Cascading process transfers the turbulent kinetic energy from larger eddies to smaller
ones where energy is dissipated as heat energy through molecular viscosity.

This figure shows eddies at the interface of two fluids mixing together. These structures represent
the outcome of a phenomenon called Kelvin-Helmholz instability formation (typical wave pattern
observed at the interface of two fluids having relative motion. The cascading process of energy
dissipation through eddies (or whirls) has been expressed in a poetic form as below.
“Big whirls have little whirls, Which feed on their velocity;
And little whirls have lesser whirls, And so on to viscosity”. – Richardson
While boundary layer development over a flat surface is conventionally a good example for
demonstration, flow pattern developed in a cigarette smoke shown above also behaves exactly
the same way if the flow is inviscid and the thickness of flat plate is negligible. Shown below is
another good example of K-H instability in clouds.

In any fluid flow, the amount of kinetic energy 𝑢′ produced equals energy dissipation for
maintaining equilibrium. The characteristic dimension at which viscous forces dissipate energy is
called Kolmogorov length scale, Lk, defined as
 3
ν ⁄4
Lk = 1 ,
ε ⁄4

Where, ν is kinematic viscosity and ε is rate of energy dissipation, i.e. KE dissipated per unit
time. This is the smallest size of eddies produced in the energy dissipation process, while the
largest size corresponds to the smallest dimension of the flow.

The time series is well correlated at shorter durations and the velocity components become
completely random at a characteristic time, called Kolmogorov time scale or integral time scale,
tI, defined as

ν
𝑡𝐼 = √ε.

The Reynolds process is valid only for t>>tI. tI is also interpreted as average turbulent diffusion
time. Average turbulent diffusion dimension or length scale = L k. The ration of Lk and 𝑡𝐼 is called
Kolmogorov velocity or average turbulent diffusion velocity. Time averaged quantities of
turbulent flows can be calculated using Kolmogorov scales. For example

1 𝑡+𝑡𝐼
𝑢̅ = 𝑡 ∫𝑡 𝑢𝑑𝜏 .
𝐼

In case of homogeneous turbulence,

̅ = 1 ∫𝑡+𝑡𝐼 𝑢′ 𝑑𝜏 = 0.
𝑢′ 𝑡 𝑡 𝐼
Mass Transport

Mass transport through a fluid may occur by the following means:

• Advection: Movement caused by shearing that causes fluid flow and/or convection, i.e.
buoyancy due to temperature (or density) difference between the mass and its surroundings.
• Diffusion: Movement caused by random motion of particles of fluid, e.g. Brownian motion. It
is also known as Molecular Diffusion.
• Dispersion: Movement caused by molecular diffusion as well as non-uniformity of the velocity
field causing advection (e.g. stretching).

Mass Transport via Advection

M
 Advective Transport = 𝑢𝐶. 𝑑𝐴 [ T ]

 Flow characteristics (Laminar, Transition or Turbulent flow) = f (relative importance of

fluid friction (viscosity) and flow inertia).
 Reynolds number, Re = UL/ν = the ratio of inertial to viscous forces; U = Characteristic
velocity scale, and L= length scale for a system, ν is the kinematic viscosity of the fluid.
• Surface water systems: Characteristic length scale is the basin- scale (1 m to 10
km) => most surface water systems are turbulent.
• Groundwater Systems: Characteristic length scale is the pore scale (< 1 mm)
=>groundwater flow is nearly always laminar.
• Open/Closed Channels: Characteristic length-scale for a channel of is the hydraulic
radius, R = A/P where A is flow area and P is the wetted perimeter.

Characteristic scales of some systems have been tabulated below.

Convective Transport
It is a property of thermally stratified flows caused by either heating from below or cooling
from above. Flows where their characteristics or properties change in layers are called
stratified flows. Transport is governed by convective velocity, w*, similar to shear velocity,
1⁄
𝛼𝑔ℎ𝑄 3
𝑤∗ = [ ]
𝜌𝐶𝑝

Where, α=thermal expansion coefficient;

ρ=fluid density,
h=height,
Q=heat flux (W/m2) and
Cp = heat capacity at constant pressure.
It characterizes the velocity scale of rising or sinking thermals. The height h at which
convective velocity = shear velocity is called Monin-Obukhov (MO) Length, L.
𝜌𝐶𝑝 𝑢∗3 𝐿 𝑢∗3
𝐿= ⟹ = ∗3
𝛼𝑔𝑄 ℎ 𝑤

The following table shows the differences in mixing by wind and that by convection:
Mixing by Winds Mixing by Convection
Large L, Small h Small L, Large h
Strong u*, Weak w* Weak u*, Strong w*
High Shear Turbulence High Convective Turbulence
Convection near ground Shear Turbulence near ground

Stability of Thermals: Richardson Number

Fluid parcels having different temperatures than surroundings are called thermals.

Consider two parcels in a stratified fluid as shown. These

parcels are oscillating due to their instabilities.
Density ρ+∆ρ/2
Volume V As the parcels interchange their positions, the potential
Velocity U energy of the system increases by
Position z+H ∆𝜌
(𝜌 + ∆𝜌/2)𝑔𝑉𝐻 − (𝜌 − ) 𝑔𝑉𝐻 = ∆𝜌𝑔𝑉𝐻
2

It can occur at the expense of available kinetic energy of the

1
system, viz. 2 × 2 𝜌𝑉𝑈 2 = 𝜌𝑉𝑈 2

The ratio between net gain in potential energy to kinetic

energy available is called Richardson Number, Ri.
∆𝜌𝑔𝐻
𝑅𝑖 =
𝜌𝑈 2
Density ρ-∆ρ/2
• Ri <<1 =>vertical motions are only weakly affected by
Volume V
the stratification.
Velocity U
• Ri ≈ 1 => vertical motions are significantly affected by
Position z
the stratification.
• Ri >> 1 => vertical motions are extremely affected by
the stratification.
Mass Transport: Fick’s First Law
It is an empirical relationship that describes the mass transport. According to this law, mass flux
is proportional to gradient of mass concentration, viz.

𝑑𝐶 𝑑𝐶
𝐽𝑚 ∝ − 𝑑𝑥 𝑜𝑟 𝐽𝑚 = −𝐷𝑚 𝑑𝑥

Jm = Mass flux
C = Concentration (e.g. mg/liter)
Dm= Coefficient of molecular diffusion = f(solvent, solute, temperature)

Flux is a quantity or mass moving across an area, perpendicular to it. The causative agent or
forcing function is not actually the concentration gradient; it is the kinetic energy of molecules.
Solvent Solute Dm (cm2/s)
Water Oxygen 2.4x10-5
Water Table salt 1.545x10-5
Water Glucose 0.673x10-5
Air Hydrogen gas 0.634
Air Oxygen gas 0.178
Air Carbon dioxide 0.139

Heat/Thermal Energy Transport: Fourier’s Law of Heat Transport

𝑑𝑇
𝑞𝐻 = −𝜌𝐶𝑝 𝐾
𝑑𝑥
K = coefficient of conductivity or heat diffusivity

Momentum Transport: Newton’s Law of viscosity

du d(ρu) d(ρu)
τ = −ρν or τ = −ρν ; or τ = −(ν + η)
dy dy dy

ν = (kinematic) viscosity or momentum diffusivity

𝑑𝑢
η = Turbulent Diffusivity=𝜌𝑙 2 𝑑𝑧 ; 𝑙 = 𝜅𝑧; κ=von Karman Constant (≈0.4)
Reynolds Analogy
In turbulent processes, all properties (momentum, heat and mass) are exchanged at the same
rate. It is because fluid transport dominates over molecular transport.

Fick’s Second Law: Diffusive Transport

C Only diffusion, no advection

dy

 J 
A Jm A  Jm  m dx  A x
dz  x 

dx

Mass Balance: Change in mass in control volume = Mass in – Mass out

  Jm  C Jm  dC C  2C
(CAdx)  JmA   Jm  dx A     (Dm )   Dm 2
t  x  t x x dx t x
  x2  12
exp 4D t 
M
C(x , t) 
A 4 Dm t  m  
10 1s
Cmax (t) 
M

1 
Concentration (g/mL)

A 4 Dm t t 8

 x 2 6
C(x , t)  Cmax (t) exp 4D t  
 m  4 10 s
For Dm = 0.673 x 10-5 cm2/s 
2 100 s
M=1g 

A = 1 cm2 -0.1 -0.05 0 0.05 0.1
Distance (cm)
Turbulent Diffusion

• Mechanism of turbulent diffusion is different than the mechanism of molecular

diffusion, but the effect is mixing only.
• Scale of the motion generating turbulent diffusion is much larger than for molecular
diffusion.

C  2C
 Dt 2
t x Dt  Dm

Advection-Diffusion Equation: 1-D ADE

Advective Flux, 𝐽𝐴 = 𝑢𝐶
𝑑𝐶
Diffusive Flux, 𝐽𝐷 = −𝐷
𝑑𝑥
𝑑𝐶
Total Flux, 𝐽𝑇 = 𝐽𝐴 + 𝐽𝐷 = 𝑢𝐶 − 𝐷
𝑑𝑥

Δy

𝜕 𝐽𝑇
𝐽𝑇 . 𝐴 ( 𝐽𝑇 + ∆𝑥) . 𝐴 x
𝜕𝑥
Δz A

Δx

Per unit time, change in mass in control volume= Mass in – Mass out

𝑑𝑀 𝑑(𝐶𝑉) 𝑑𝑉 𝑑𝐶 𝑑𝐶 𝑑𝐶 𝜕𝐽𝑇
= =𝐶 +𝑉 =𝑉 = ∆𝑥𝐴 = 𝐽𝑇 . 𝐴 − ( 𝐽𝑇 + ∆𝑥) . 𝐴
𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡 𝜕𝑥
𝜕 𝑑𝐶 𝜕(𝑢𝐶) 𝜕 𝑑𝐶 𝑑𝐶 𝑑(𝑢𝐶) 𝑑2𝐶
= − (𝑢𝐶 − 𝐷 ) ∆𝑥𝐴 = − ∆𝑥𝐴 + (𝐷 ) ∆𝑥𝐴 ⟹ + =𝐷 2
𝜕𝑥 𝑑𝑥 𝜕𝑥 𝜕𝑥 𝑑𝑥 𝑑𝑡 𝑑𝑥 𝑑𝑥
 dC dC d2 C
Whence the ADE is + u dx = D dx2
dt

𝑑𝐶 𝑑2𝐶
Transforming the coordinates system by 𝜂 = 𝑥 − (𝑥0 + 𝑢𝑡) and 𝜏 = 𝑡, =𝐷 2
𝑑𝜏 𝑑𝜂

𝑀 (𝜂)2
It has the solution as before: 𝐶(𝜂, 𝑡) = 𝐴√4𝜋𝜏𝐷 𝑒𝑥𝑝 (− 4𝜏𝐷 ) ;
𝑥 𝑥

Substituting for η and τ and let x0 = 0 for simplification, we obtain:

𝑀 (𝑥 − 𝑢𝑡)2
𝐶(𝑥, 𝑡) = 𝑒𝑥𝑝 (− )
𝐴√4𝜋𝑡𝐷𝑥 4𝑡𝐷𝑥

Dispersion in rivers

The following expressions are commonly used.

̅̅̅̅̅̅̅ u∗ z
Mixing: u t (z) = u
̅+ k
(1 + ln d); k=0.4

Vertical: Dt,z = 0.067u∗ d; (±25% Error)

Transverse: Dt,𝑦 = Dt,𝑥 = 0.15u∗ d; (±50% Error)
Dt,𝑦 = Dt,𝑥 = (3.4)u∗ d (Strong meander)
Dt,𝑦 = Dt,𝑥 = (0.2)u∗ d (Straight Chennel)
Dt,𝑦 = Dt,𝑥 = (0.6)u∗ d = (0.4 − 0.8)u∗ d (Gentle meander)

Average lateral diffusivity:

1 d ′ z 1 z ′ 1 W ′ y
1 y
DL (z) = − ∫ u ∫ ∫ u dz dz dz = − ∫ u d ∫ ∫ u′ d dy dy dy
d 0 0 Dz 0 A 0 0 Dz d 0
Dispersion in Groundwater

AdvectionFlux, JA  nuC
C
DiffusionFlux, JD  nD0
x
C
Mechanical DispersionFlux, JM  nDM
x

C
D0  DM  DH  J  JA  JD  JM  nuC  DH
x

D0 is molecular diffusivity (m2/s) and DH is hydrodynamic diffusivity, n is porosity, C is

concentration. In the absence of detailed information, DM = αLU where αL is longitudinal
dispersivity ≈ 0.1* travel distance; the transverse dispersivity in the horizontal direction, αTy is of
the order of 0.1 αL and that in the vertical direction, αTz is of the order of 0.01 αL

τ is tortuosity i.e. ratio of straight length between two points in flow to total zig-zag path length
followed by water molecules between those two points as shown below. Typically τ = 0.7 for silts,
less for clay and more for sand. Typical values of diffusivities in groundwater have been listed
1 1
below. Tortuosity has also been correlated with the porosity: 𝜏 = 𝜃 −3 𝑡𝑜 𝜃 −2 .
In case of groundwater, a parameter called retardation coefficient (Rd) is a parameter that is used
to determine the transport velocity of centroid of the contaminant mass. If f mass fraction of
hydrophobic contaminant is adsorbed to the stationary media in the aquifer, i.e. attains velocity
=0 and remaining 1-f fraction moves with velocity, Vf, of fluid in a porous media, then effective
velocity of contaminant mass, Vc, can be obtained using momentum balance,

1xVc = fx0 + (1-f)Vf => Vc = (1-f) Vf => Vf / Vc = Rd = 1/(1-f).

Cation D0 (10-6 cm2/s) Anion D0 (10-6 cm2/s)

H+ 93.1 OH- 52.7
Na+ 13.3 F- 14.6
K+ 19.6 Cl- 20.3
Rb+ 20.6 Br- 20.1
Cs+ 20.7 HS- 17.3
Mg2+ 7.05 HCO3- 11.8
Ca2+ 7.93 CO32- 9.55
Sr2+ 7.94 SO42- 10.7
Ba2+ 8.48
Ra2+ 8.89
Mn2+ 6.88
Fe2+ 7.19
Cr3+ 5.94
Fe3+ 6.07

𝜕𝑚 𝜕𝐽 𝜕𝜌𝑏 𝐶𝑠 𝜕𝐶𝑤 𝜕 𝜕𝐶 𝜕𝐶
=− + 𝑆𝑜 − 𝑆𝑖 ⇒ + = − (𝑛𝑢𝐶 − 𝑛𝐷𝐻 ) + 𝑆𝑜 − 𝑆𝑖
𝜕𝑡 𝜕𝑥 𝜕𝑡 𝜕𝑡 𝜕𝑥 𝜕𝑥 𝜕𝑡
𝑢 𝜕𝐶 𝐷𝐻 𝜕 2𝐶 𝑢 𝜕𝐶 𝐷𝐻 𝜕 2 𝐶
=− + = − +
𝜌𝑏 𝐾𝑑 𝜕𝑥 𝜌𝑏 𝐾𝑑 𝜕𝑥 2 𝑅𝑑 𝜕𝑥 𝑅𝑑 𝜕𝑥 2
1+ 𝑛 1+ 𝑛
 𝜕𝐶 𝑢 𝜕𝐶 𝐷𝐻 𝜕 2 𝐶
⟹ + =
𝜕𝑡 𝑅𝑑 𝜕𝑥 𝑅𝑑 𝜕𝑥 2

Where,
1+ ρb Kd/n = Rd or retardation coefficient
So and Si are sources and sinks that have been assumed to be absent
Subscript w has been dropped for consistency with literature on the subject.

The above expression can also be directly inferred because Dm is proportional to flow velocity,
both flow velocity as well as Dm need to be divided by Rd and therefore the ADE is

C u C DH  2C
 
t Rd x Rd x 2

dC dC d2 C
Or, + u dx = D dx2
dt

Analytical Solutions

Three-Dimensional Forms

Continuous Source:

C 1 L  v t  1  v L  L  v t 

 Erfc x
 
 4D t  2 
E xp
x

E rfc
 4D t 
x

C0 2  x   D x   x 

Instantaneous Source:
𝑀 (𝑥 − 𝑢𝑡)2 (𝑦)2 (𝑧)2
 𝐶(𝑥, 𝑦, 𝑧, 𝑡) = 𝑒𝑥𝑝 − ( + + )
3⁄
(4𝜋𝑡𝐷) 2 4𝑡𝐷𝑥 4𝑡𝐷𝑦 4𝑡𝐷𝑧

M = C0V
D = (DxDyDz)1/3
Two-Dimensional Forms

Instantaneous Source:
𝑀 (𝑥 − 𝑢𝑡)2 (𝑦)2
𝐶= 𝑒𝑥𝑝 − ( + )
4𝜋𝑡 √𝐷𝑥 𝐷𝑦 4𝑡𝐷𝑥 4𝑡𝐷𝑦

Continuous Source:
𝑀̇ 𝑥 − 𝑢𝑡
𝐶= 𝑒𝑟𝑓𝑐 (− )
4𝜋𝑡 √𝜋𝑢𝑡𝐷𝑦 4𝑡𝐷𝑥

One-Dimensional Forms

Instantaneous Source:
M (x − ut)2
C= exp (− )
√4πtDx 4tDx

Continuous Source:
𝑀̇ 𝑥 − 𝑢𝑡
𝐶= 𝑒𝑟𝑓𝑐 (− )
2𝑢 4𝑡𝐷𝑥

Erf = Eror Function and Erfc = Complementary Error Function, given and approximated by,

𝑥
1 2
erf(𝑥) = ∫ 𝑒 −𝑡 𝑑𝑡 𝑎𝑛𝑑 erfc(𝑥) = 1 − erf(𝑥) ≈ 1.3693exp[−0.8072(𝑥 + 0.6388)2 ]
√𝜋 −𝑥
Dispersion in Atmosphere

AdvectionFlux, JA  uC
C
DiffusionFlux, JD  D0
x
C
Turbulent Transport Flux, JT  D T
x

(a) Puffs or Volume elements

𝜕𝐶 𝜕𝐶 𝜕 2𝐶 𝜕 2𝐶 𝜕 2𝐶
𝐾 = 𝐷0 + 𝐷𝑇 ⇒ +𝑢 = 𝐾𝑥 2 + 𝐾𝑦 2 + 𝐾𝑧 2
𝜕𝑡 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑧
𝑄∆𝑡 1 (𝑥 − 𝑢𝑡)2 𝑦 2 (𝑧 − 𝑧0 )2
𝐶(𝑥, 𝑦, 𝑧) = 1/2
𝑒𝑥𝑝 {− ( + + )}
8(𝜋𝑡)3/2 (𝐾𝑥 𝐾𝑦 𝐾𝑧 ) 4𝑡 𝐾𝑥 𝐾𝑦 𝐾𝑧

𝑄∆𝑡 (𝑥 − 𝑢𝑡)2 𝑦2 (𝑧 − 𝑧0 )2
𝜎𝑖2 = 2𝐾𝑖 𝑡 ⇒ 𝐶(𝑥, 𝑦, 𝑧) = 𝑒𝑥𝑝 {− ( + + )}
(2𝜋)3/2 𝜎𝑥 𝜎𝑦 𝜎𝑧 2𝜎𝑥2 2𝜎𝑦2 2𝜎𝑧2
(b) Plumes or Continuous Emissions

∞ (x − ut)2
Q y2 (z − z0 )2
C(x, y, z) = ∫ exp {− ( + 2+ )} dt
0 (2π)
3/2 σ σ σ
x y z 2σ2x 2σy 2σ2z

This equation can be integrated using conditions that σ’s are proportional to time up to
few kilometer, about 10km, from source and then after some more simplifications,

Q y2 (z − z0 )2
C(x, y, z) = exp {− ( 2 + )}
2πu σy σz 2σy 2σ2z

This is a popular equation, called Gaussian model equation of plume. The equation is
modified to include the effects of plume buoyancy due to temperature and its momentum
due to velocity of gases released from the source. The net effect of these entities is
gradual increase in the centerline of plume, which is called plume rise and which is
𝑉𝑠 𝐷 𝑇𝑠 −𝑇𝑎
modeled in the Indian practice using Holland’s formula: ∆𝐻 = {1.5 + 2.68𝑃𝐷 }
𝑢 𝑇𝑠

where ∆H= Plume rise (m), Vs=exit velocity (m/s), D=Source Diameter (m), u = wind
velocity at source height (m/s), P = atmospheric pressure (bar), Ts=Temperature of
emissions at source (K) and Ta=Ambient temperature (K).

∆H is multiplied by 0.8 and 1.1 for stable and unstable atmosphere (see table on next
page). The effect of plume rise is modeled by substituting z0+∆H for z0 in the equation.

The plume rise is also affected by wake developed in flow field on the leeward side of the
𝑉
source. This effect is empirically modeled as ∆𝐻 ′ = 2𝐷 𝑢𝑠 − 1.5 𝑓𝑜𝑟 𝑉𝑠 < 1.5𝑢 and 0

Otherwise. This quantity is deducted from ∆H.

Gaussian Plume Model

• The wind velocity in this expression is at the centerline of the plume, called effective
height (He), viz. z0+∆H. Wind velocity at z height (uz) can be calculated using power law:
𝑢𝑧 𝑧 𝑝
= (𝑧 ) where ur is the wind velocity at reference height zr, usually 2 or 10m.
𝑢𝑟 𝑟

• All parameters in this equation depend on lapse rate = - temperature gradient that
determines the mixing behavior of atmosphere, called stability. Commonly used values
for these parameters for dispersion on flat terrains have been tabulated below. Values
for other types of terrains are available in literature.

Source Characterization: Phase Partitioning

Water
Air

Solid

≈ 0.2 m in sand to
≈ 2 m in clays

Mass of constituents per bulk volume of bulk concentration, m = θwCw+ θaCa+ ρbCs
Ca=Conc. in air (mg/L), Cw = Conc. in water mg/L, Cs= Conc. in solids mg/kg, and θ’s are
porosities or fraction of void spaces occupied by various phases.
NAPL -Non Aqueous Phase Liquids or OIL -Organic Immiscible Liquids are hydrophobic
compounds, typically petroleum products or organic solvents or their mixtures. In addition to
other phases, these compounds also partition into their own phase, i.e. OIL.

Thus Ca= K’H Cw; Cs= Kd Cw and Co=KoCw where K’H =Henry’s constant and Kd= Soil partition
coefficient and Ko is partitioning coefficient for OIL phase.

Mass of constituents per bulk volume of bulk concentration: m= θoCk

m = θwCw+ θaCa+ θoCo+ρbCs where ρb=(1-n) ρs

= θwCw+ θa KH’ Cw + θo KoCw +ρb Kd Cw
= (θw+ θa KH’ + θo Ko +ρb Kd) Cw = BwCw

Where, K’H = KH/RT. Bw is called Bulk Water Partition Coefficient. Similar expressions can be
written for other phases; i.e. m=BaCa, m=BoCo and m=BsCs; whence Cw=m/Bw;
Ca=m/Ba=K’Hm/Bw; Cs=m/Bs =kdm/Bw and Co=m/Bo=Kom/Bw

Kd = Koc.foc where Koc is organic carbon partition coefficient and foc is fraction of organic carbon
in the soil matrix. Koc can be estimated from Kow, the Octanol-Water Partition Coefficient.

Ko can be determined by Raoult’s law, i.e. aqueous phase concentration of a volatile compound
equals aqueous phase solubility of the constituent in equilibrium with pure constituent
multiplied by mole fraction of constituent in OIL phase. That is, Cwk=Sk.Xk where Sk is solubility
of kth constituent and
Ck / MWk
Xk 
 Ck / MWk
Example
A gasoline mixture has the following composition:
Constituent Concentration (g/L) Molecular Weight (g/M)
Benzene 8.2 78
Toluene 43.6 92
Xylene 71.8 106
1-hexane 15.1 84
Cyclohexane 2.1 84
n-hexane 20.4 86
Other aromatics 74 106
Other paraffins (C4-C8) 336.7 97.2
Heavy ends (C9 and above) 145.1 128

The mixture spills on a soil whose bulk density is 1.6 kg/L and organic carbon fraction 0.01. Its
porosity 0.4 is divided among water, NAPL and air phases as 0.15, 0.05 and 0.2 respectively.
Determine the partitioning of BTX compounds among soil, water, NAPL and air phases.

Partitioning characteristics of BTX compounds are as below.

Compound MW Density Solubility Vapor Pr. KH (atm. Koc (L/kg)
(g/m3) (mg/L) (atm) m3/M)
Benzene 78.1 0.87 1750 0.125 5.59E-3 83
Toluene 92.1 0.86 535 3.7E-2 6.37E-3 300
Xylene 106.2 0.88 175 9.0E-3 5.1E-3 830
Molar concentration of gasoline mixture =
9
Ci
 MW
i1 i

= 7 mol/L
Bulk conc. of each constituent mi=θCi
Whence mB=0.05x8.2 = 0.41 g/L; Similarly mT=2.18 g/L and mX=3.59 g/L
K’H=KH/RT where R=Universal Gas Constant = 8.2E-5 atm.m3/mol-K and T=298 K
The following partitioning constants are obtained using these partitioning relationships:
Constituent KH Kd (L/kg) Ko BW
Benzene 0.23 0.83 310 17
Toluene 0.26 3 1200 65
Xylene 0.21 8.3 4200 220

BTX concentrations in different phases can be calculated by following expressions:

Cw=m/Bw
Ca=m/Ba=KHm/Bw
Cs=m/Bs =kdm/Bw
Co=m/Bo=Kom/Bw

As percentages
In Water =100 θw/Bw %
In Air =100 θaKH/Bw %
In OIL=100 θoKo/Bw %
In Solids = 100 ρbkd/Bw %
Compound Water Air Soil NAPL
mg/L (%) mg/L (%) mg/kg (%) g/L (%)
Benzene 24 (0.88) 5.5 (0.27) 20 (7.8) 7.5 (91)
Toluene 34 (0.23) 8.7 (0.08) 100 (7.4) 40 (92)
Xylene 16 (0.07) 3.4 (0.02) 130 (5.9) 67 (94)
Integral Plume Rise

𝑤
z+dz

ρ 𝑢 ρ 𝑢
a p e
e

𝑤
z
ρ
r

Boussinesq Assumption: ρ ≈ ρ ≈ ρ
r a p

Buoyancy Force: B = g(ρa − ρp )

ρa −ρp
Reduced Gravity: g ′ = g ρr

g dρ
Buoyancy Frequency: N = √− ρ (Brunt Väisälä Frequency)
a dz

Entrainment Hypothesis: ue =∝ w; ∝=0.1 for water, and 0.3 for air

Top hat profile w(r, z) = wz
r2 r2
Self Similarity: w(r, z) = wc (z)exp(R2 ); and g ′ (r, z) = g ′c (z)exp(R2 )
When a parcel of air moves vertically in a stratified flow, it starts vibrating with Brunt Väisälä
Frequency, typically at 10-2 s-1 above surface layer in atmosphere. However, if the fluid is linearly
stratified, this frequency = 1 s-1 .

Mass Conservation: [Mass flux exiting control volume] = [Mass entering control volume]+[Mass
entraining from side]
(πR2 wρp )z+dz = (πR2 wρp )z + (2πRue ρa )dz
(πR2 wρp )z+dz - (πR2 wρp )z
Simplifying, = (2πRαwρa )
dz
d(πR2 wρp )
Or, = (2πRαwρa )
dz
d(R2 w)
Using Boussinesq Assumption: = 2Rαw
dz

Momentum Conservation:
[Upward buoyancy] =[weight of displaced fluid]-[actual weight of plume segment]
= (πR2 ρa g)dz − (πR2 ρp g)dz = πR2 g(ρa − ρp )dz = (πR2 ρr g′)dz
[Momentum flux exiting control volume]=[Momentum entering control volume] + [Momentum
entraining from side] + [Upward buoyancy]
(πR2 w 2 ρp )z+dz = (πR2 w 2 ρp )z + 0 + (πR2 ρr g′)dz
(πR2 w 2 ρp )z+dz = (πR2 w 2 ρp )z
= (πR2 ρr g′)
dz
d(πR2 w2 ρp ) d(R2 w2 )
= (πR2 wρr g′) ⟹ = (R2 g′) on using Boussinesq Assumption.
dz dz

Buoyancy conservation:
[Buoyancy flux exiting control volume] =[Buoyancy flux entering control volume] +[Influx of
buoyancy entraining from side]
[πR2 wg(ρr −ρp )]z+dz = [πR2 wg(ρr −ρp )]z + [2πRue g(ρr −ρa )]dz
[πR2 wg(ρr −ρp )]z+dz − [πR2 wg(ρr −ρp )]z
= 2Rαw(ρr −ρa )
dz
d(πR2 w) d[R2 w(ρr −ρa )] d(ρr −ρa )
= (ρr −ρa ) = − (R2 w)
dz dz dz
 d[R2 w(ρr −ρa )] d(ρr )
⟹ = (R2 w)
dz dz
Multiplying both sides by g/ρr
d[R2 wg′]
= − N2 R2 w
dz

Boussinesq assumption is used only for inertia related terms, e.g. continuity, momentum and not
the buoyancy related terms as small differences in densities matter for buoyancy driven
transport.

dW
Let W = wR2 , V = wR and F = wR2 g ′ ⟹ = 2αV
dz
dV4 dV2 FW dV2 1 dV4
= 2FW { dz = ; V2 =2 }
dz V2 dz dz

dF
= −N2 W
dz

The set of equations is solved simultaneously using 4th order Ranga-Kutta method:
dy
= f(x, y); y(0) = y0 ; yi+1 = yi + (a1 k1 + a2 k 2 + a3 k 3 + a4 k 4 )h
dx
h k1 h
h = xi+1 − xi ; k1 = f(xi , yi ); k 2 = f (xi + , yi + );
2 2
h k2 h h
k 3 = f (xi + 2 , yi + ) ; k 4 = f(xi + 2 , yi + k 4 h)
2

Boussinesq assumption is not necessary to formulate and numerically solve these equations.
Brunt Väisälä Frequency = 1 for linearly stratified atmosphere further simplifies this model.

Stratified Atmosphere

Similarity theory characterizes the advection and diffusion parameters of the stratified
atmosphere. The theory was developed in 1950s by Russian scientists Monin and Obukhov. This
theory is therefore also known as Monin-Obukhov (or simply MO) similarity theory. The theory
is based on dimensional analysis (Buckinghan pi theorem) of parameters that characterize the
turbulence in the atmosphere. Dimensionless parameters used for characterizing fluxes of heat
and momentum (thermal and mechanical turbulence) have the same functional forms or similar
profiles everywhere, hence it is called similarity theory.

The theory essentially describes all fluxes as universal functions (i.e. applicable everywhere) of
similarity parameter, ξ (=z/L), where z is height above ground surface and L is MO length, i.e. z at
which both thermal and mechanical turbulences are equal in magnitude. Stratification in fluids
means variation in density with z, due to any reason, e.g. temperature (in atmosphere, oceans,
lakes), suspended particulate matter (in atmosphere, oceans and lakes) or dissolved salts (in
oceans, lakes).

There are is a similarity functions, Φ(ξ) for each flux of momentum, heat and water vapor (i.e.
Φm, Φh, and Φq, respectively) that can be integrated to obtain profiles or concerned quantities
(i.e. wind velocity, temperature and or moisture content respectively). A function Ψ(ξ) represents
the influence of diabatic processes on idealized adiabatic conditions and is defined in terms of ξ
and Φ(ξ).
The MO similarity theory was originally proposed for characterizing the flux profiles of near
neutrally stratified atmosphere in the surface layer (about 100m bottom layer in troposphere). It
has now been extended to other stratifications (stable, unstable and convective) as well as
beyond surface layers. A variant of this theory, called local similarity theory uses z/Ʌ where Ʌ =
Obukhov length, defined in a way similar to L and using local values of parameters. This theory
relies on the fact that above the surface layer, the local values of heat and momentum fluxes
govern the turbulence and not their magnitude at ground level. In convective mixed layers, the
profiles are functions of z/zi where zi is height of boundary layer or mixing depth as the similarity
parameter.

Boundary Layer Depth

T 14400 12 cos ϕ cos δ π π F(1−A)

h(t) = √P0 21Γ ×√ [sin (12 t R ) − sin (12 t)] + (t − t R )sinϕsinδ × √ κ
0 0 −14Γ1 π 1+√ a
κg

2π(day of year−80) 12
Where, sinδ = sinϵ [ ]; t R = 12 − cos−1 (−tanϕtanδ)
365.2422 π

T0 = Surface temperature; P0= Surface pressure; Γ0 = Adiabatic lapse rate; Γ1 = Environmental lapse
rate, typically of inversion or negative; ϵ= Angle of tilt of Earth’s axis=23.5 degrees; t=Time of day
 W
in hours; t R =Time of sunrise in hour of day; F=Integrated solar irradiance (m2 )over all

wavelengths at the surface; A = Surface albedo; κa =Thermal diffusivity of air; κg =Thermal

diffusivity of ground.

Using the characteristic assumption that any non-dimensional quantity is a function of

parameter, ξ (=z/L), the following expressions are used:
̅ kz
∂u z u∗ z z ξ 1−∅m (ξ)
= ∅m (L) ⇒ u̅(z)= [ln z − Ψm (L)] ; Ψm (ξ) = Ψh (ξ) = ∫ξ dξ
∂z u∗ k 0 0 ξ
z z
∅m ( ) = ∅m (0) = 1 for neutral atmosphere, | | ⟶ 0
L L
z z z
∅m (L) = 1 + 4.7 L for stable atmosphere, L > 0; ∅h = ∅m
1
z z −4 z
∅m (L) = (1 − 15 L) for unstable atmosphere, L < 0; ∅h = ∅2m

While Ψm (ξ) can be directly integrated for stable and neutral conditions, the integration for
unstable conditions is as below.
1
z 1+x 1+x2 −1 π z 4
Ψm ( ) = 2 ln + ln − 2tan x + ; Where, x = (1 − 15 )
L 2 2 2 L

Similar expressions can be written for ∅ℎ (= ∅𝑞 ) to obtain fluxes of heat and humidity. k is von
Karman constant ≈ 0.4. The power law exponent of wind profile =f(ξ).

ku∗ z
The vertical diffusivity of matter, K z = ∅ and horizontal diffusivity may be taken as 2, 5 and 6
h (ξ)

times of vertical diffusivity for unstable, neutral and stable atmosphere, respectively.
Stratified Lakes

Wind
Winter Summer
0 4 10 20
0
Temperature 0C Epilimnion

Thermocline Metalimnion
Depth

10 (Inflection Point)

20
Hypolimnion

30

Deep lakes typically turn over during autumn and spring when the water temperature in
epilimnion also becomes 40C. Thermal stratification breaks in this condition and surface winds
set forth circulation in the water mass.
Source: Boehrer and Schultze, Reviews of Geophysics, 46 (2008) DOI: 10.1029/2006RG000210

Notes.
Chemocline: Inflection point of salinity profile
Monimolimnion: Stable, high salinity bottom layer
Meromictic Lakes: Permanently stratified lakes, that never turn-over

TS Diagrams

TS diagrams represent variation of temperature with salinity as shown below. Pair of

temperature and salinity data obtained at various depths are plotted. The contours of TS diagram
represent the equal-density lines.
PSU or psu or Sp or 0⁄∞ (obsolete) = Practical Salinity Unit = parts per thousand, ppt or (= g/L,
actually g/kg of seawater). The reference value of salinity, SR is 35.

The approximate ionic composition of ocean water is Sodium 30.6%, Magnesium 3.7%, Calcium
1.2%, Potassium 1.1%, sulfates 7.7% and chlorides 55%.

Completely Mixed Analysis

Consider a lake with BOD discharged to it.

Volume=V
S Conc = C Q
Qin Cin BOD rate C
constant = k
S=Source, kg/s
3
Qin = inflow, m /s
3
Cin = Influent conc., kg/m

dm dCV dV dC dC
= =C +V = V  when volume is constant
dt dt dt dt dt

dC ∑S 𝑆 + 𝑄𝐶𝑖𝑛
V = ∑ 𝑆 − QC − kCV = 0 ⇒ C∞ = =
dt Q + kV Q + kV

dC Q dC
= − (k + ) (C − 𝐶∞ ) analogous to  = −k'C
dt V dt
Q
C(t) = C∞ + (C0 − C∞ )exp [− (k + ) 𝑡]
V

For example,
if V=1.0x107m3 ; k= 0.2/d Qin=5m3/s, Cin = 10 mg/L;
Plus another source Qw=0.5m3/s, Cw=100 mg/L =>C∞=3.5mg/L

On stopping second source at t =100 d: C∞=1.8mg/L; e.g. C(7) = 2.1 mg/L.

 4

3.5

2.5
Conc. ppm

2
Both Sources
1.5 Stopping Source 2

0.5

0
0 50 100 150 200
Time, d

Similarly if the pollutant added settles in the lake, e.g. phosphates,

dC Q𝐶𝑖𝑛 + 𝑆
V = S + 𝑄Cin − 𝑄𝐶 − 𝑣𝑠 𝐴C = 0 ⇒ C∞ =
dt Q + 𝑣𝑠 𝐴
𝐶(𝑡) = 𝐶∞ + (𝐶0 − 𝐶∞ ) exp[ − (𝑄 + 𝑣𝑠 𝐴)/𝑉)𝑡]

For example, A=80x106 m2; Qin=15m3/s Cin=0.01 mg/L; S = 1g/s; vs=10m/y =>C∞=0.028mg/L
For C∞=0.01mg/L; S=0.25 g/s

g ρp −ρw 2
Settling velocity can be calculated using Stokes law: vs = 18 d
μ

If particles follow a distribution: vs= Σfivsi ,where vsi=settling velocity of particles having mass
fraction, fi.
Condition for completely mixed analysis: Potential Energy Anomaly

Potential energy per unit volume: ψ = ρ ⋅ g ⋅ H

0
Stratified Fluid: ψ = g ∫−H ρ(z) ⋅ z ⋅ dz

0
Mixed water column: ψm = g ∫−H ρ ⋅ z ⋅ dz

Energy difference between a mixed and a stratified water column

0

ϕ = ψm − ψ = g ∫(ρ − ρ) ⋅ z ⋅ dz
−H
dϕ
= Power from (mixing by wind + cooling/heating + precipitation)
dt
dϕ
= τu∗
dt
2
τ ρa cD u10 ρa cD
u∗ = √ = √ =√ u = ϑ ⋅ u10
ρ ρ ρ 10

dϕu 3
= ϑτu10 = ϑρa cD u10
dt
d wind at 10m
  Effective fraction of ~ 0.00116
dt
a  1.2 kg/m 3 ;   1020 kg/m 3 ; and c D  0.00114
dϕu 3
= 1.3 × 10−6 ρa u10  W/m2
dt

dϕW 3 3
Alternately,  = δk s ρa u10 = 1.4 × 10−6 ρa u10   W/m2
dt
Where, δ is a mixing efficiency coefficient = 0.023; ks is a drag factor that equals 6.4x10-5
Heating/Cooling
𝑑𝜙𝐶 𝛼𝑔𝐻𝑄
=
𝑑𝑡 2𝑐𝑝
α is the thermal expansion coefficient of lakewater ~ 1.6x10-4 °C-1
cp is the specific heat of seawater ~ 4x103 J/(kg °C)
Q is the cooling/heating rate (W/m2)

Precipitation
𝑑𝜙𝑃 𝑔𝐻𝛥𝜌
= 𝑃𝑟
𝑑𝑡 2
Δρ is the density difference between fresh water and lake water and Pr is the precipitation rate
(m/s)

Stratified Estuaries

Ozmidov scale:

Maximum size of eddies before gravity arrests them.

It is the largest scale at which there is enough energy in the turbulence to have overturns, viz.
Tharpe Overturn Scale. At larger scales, there are no overturns and internal waves are generated.

Characterization of Turbulence

𝛥𝜌 𝑔 𝜕𝜌 1
𝑔′ = 𝑔; 𝑁 2 = − 𝜌 𝜕𝑧 ; 𝑇𝐸 = 2 𝜌𝑢2 + 𝜌𝑔𝑧 + 𝑝
𝜌
ℎ1 +𝛥𝑧
1 1 𝜕𝜌 𝑢𝑇
𝜌 𝑢 2 = 𝜌𝑖 𝑔𝛥𝑧 + 𝑔 ∫ 𝜌(𝑧)𝑑𝑧 = 𝑔 (𝛥𝑧)2 ⇒ 𝑢𝑖 = 𝑁𝛥𝑧𝑖𝑓  𝑢𝑇 = 𝑢1 ⇒ ℓ𝑜 =
2 𝑖 𝑖 2 𝜕𝑧 𝑁
ℎ1
wind unstratified flow:
Only boundary-layer turbulence
Wind-driven
velocity turbulence eddy viscosity/
Diffusity ℓ

shear-driven stress 𝐾𝑚 = 𝜅𝑢∗ 𝑧(1 − 𝑧/ℎ)

turbulence ≈ 𝑢𝑇 ℓ𝑢𝑇 ≈ 𝑢∗ ℓ
≈ 𝜅𝑧(1 − 𝑧/ℎ)ℓmax

u Boundary-layer
turbulence

2 2
Turbulent velocity scale Bottom stress τB /ρ= Cdub =u
uT ~u ~ 0.05 ub *
*

eddy viscosity
𝛥𝑧 = ℓ𝑜 (Ozmidov scale)
stratification 𝐿 𝑇 = ℓ𝑜
decreases Ozmidov scaling: 𝜀 1/2
𝜌𝑖
turbulence z LT=uT/N = ( )
𝑁3
near pycnocline Boundary layer Scaling:
LT ~ kz;
Max LT =0.2h at h/2π

stress
𝑧 1/2
𝐿𝑇 = 𝐿𝐵𝐿 = 𝜅𝑧 (1 − )
ℎ
Mixing in Oceans

Seawater flows along the horizontal plane and in the vertical. Typical speeds of the horizontal
flow or currents are 0.01-1.0 m/s; vertical speeds within the stratified ocean are much small,
close to 0.001 m/s. Two forces produce the non-tidal ocean currents: the wind exerting a stress
on the sea surface and by buoyancy (heat and freshwater) fluxes between the ocean and
atmosphere that alter the density of the surface water. The former induces what we call the wind
driven ocean circulation, the latter the thermohaline circulation. The wind driven circulation is
by far the more energetic but for the most part resides in the upper kilometer. The sluggish
thermohaline circulation reaches in some regions to the sea floor, and is associated with ocean
overturning linked the formation and spreading of the major water masses of the global ocean,
such as North Atlantic Deep Water and Antarctic Bottom Water.

Wind induced upwelling: The wind stress acting on the surface layer of the ocean induces
movement of that water. This is called Ekman Layer transport, which extends to the surface 50
to 200 meters. The Ekman transport is directed at 90° to the direction of the wind, to the right of
the wind in the northern hemisphere, left of the wind in the southern hemisphere. As the wind
varies from place to place, Ekman transport can produce divergence (upwelling or suction) or
convergence (sinking or pumping) of surface water.

Geostrophic Currents: The surface layer is less dense (more buoyant) than the deeper layers,
therefore a spatially variable Ekman transport field acts to redistribute the buoyant surface
water: thinning the buoyancy surface layer in divergence regions, thickening the buoyant surface
layer in convergence regions. As the ocean is in hydrostatic equilibrium, the redistribution of the
buoyant surface layer induces sea level depressions in divergent regions and upheavals in
convergence regions. While these variations amount to only a 1.5 meter in amplitude, they are
sufficient to induce horizontal pressure gradients which initiate the wind driven circulation
following the geostrophic balance concept. The ocean currents are for the most part geostrophic,
meaning that the Coriolis Force balances the horizontal pressure gradients. The wind driven
circulation is characterized by large clock-wise and counter clock-wise flowing circulations or
gyres, such as the subtropical and sub polar gyres.

Thermohaline Circulation: As surface water is made denser through the removal of heat or
freshwater, the surface layer descends to deeper depths. If the stratification is weak and the
buoyancy removal sufficient, the descent would reach the deep sea floor. Such deep reaching
convention occurs in the northern North Atlantic (North Atlantic Deep Water) and around
Antarctica (Antarctic Bottom Water). The thermohaline circulation engages the full volume of the
ocean into the climate system, by allowing all of the ocean water to interact directly with the
atmosphere (on a time scale of 100-1000 years).

Typical composition of sea water is shown below.

Ion g/kg
Sodium 10.781
Magnesium 1.284
Calcium 0.4119
Potassium 0.399
Strontium 0.00794
Chloride 19.353
Sulfate 2.712
Bicarbonate 0.126
Bromide 0.067
Borate 0.0257
Fluoride 0.00130
Totals 35.169
Ekman Transport
Jets in Oceans

Example: Multiport diffusers in oceans for disposal of wastewater

Empirical Relationships
uo
Densimetric Froude number = 𝐹𝑟 = Δρ
= Inertia/ buoyancy force per unit mass
√ρ 0 gd
a

For plumes Fr ≈ 1; for jets Fr → ∞. For most outfalls the Fr ranges from 4 to 18 (max. about 26)
and the flow can be classified as Buoyant jets.

The velocity and concentration distributions are nearly Gaussian. The initial volume and specific
momentum fluxes for a circular jet are
𝜋 𝜋
𝑄𝑜 = 𝐷2 𝑈 and 𝑀0 = 𝜌0 𝐷2 𝑈
4 4
1
𝑄
Average dilution for a round jet 𝑆̄ = 𝑄 = 0.29 𝑀𝑜2 𝑥 𝑄𝑜−1
𝑜
1 5
Average dilution for a round plume 𝑆̄ = 0.163 𝐹𝑜3 𝑧 3 𝑄𝑜−1
𝜋 𝛥𝜌
Where the initial buoyancy flux is given by 𝐹𝑜 = 𝑑 2 𝑢𝑜 𝜌 𝑔
4 𝑎

Analytical Descriptions: Forcing Functions

Mixing affected by inertia of turbulent eddies

• Jets: momentum (continuous source)
• Puffs: momentum (instantaneous source)
Mixing affected by inertia produced by buoyant force
• Plumes: buoyancy (continuous source)
• Thermals: buoyancy (instantaneous source)
Mixing affected by inertia of turbulent eddies as well as that produced by buoyant force
• Buoyant jets: momentum+ buoyancy (continuous source)
• Buoyant puffs: momentum+ buoyancy (instantaneous source)
All turbulent round jets have the same opening angle (11.8 0) in same density homogeneous
fluids, irrespective of the fluid characteristics (air, water, other), orifice diameter (d) and injection
speed (U).

Self-Similarity: Repetition of same patterns

Determine a proper scale to measure characteristic patterns of the self-similar profiles. This
would require a change in the frame of reference; the new one will measure entities with new
scale. Self-similarity zone starts where the ambient fluid penetrates to centerline in a jet. The
figure below shows the self-similar profiles of some geometries. A similarity solution is
theoretically the one in which the number of independent variables is reduced by at least one,
usually by a coordinate transformation. Normally, coordinates are collapsed into dimensionless
groups e.g. that scale the velocities. One can take advantage of self-similar (self-preserving)
solutions to achieve useful simplifications in solving problems or comparing data from other
experiments. Usually the self-similar profile is assumed to be Gaussian, whence
𝑟2
Velocity: 𝑢(𝑟, 𝑥) = 𝑢𝑚𝑎𝑥 𝑒𝑥𝑝 (− 2𝜎2 ) ; σ is the standard deviation

Radius, R(x)=x/5 for jet; x/6 for plumes; σ is defined as R/2

50𝑟 2 50𝑟 2
Velocity: 𝑢(𝑟, 𝑥) = 𝑢𝑚𝑎𝑥 𝑒𝑥𝑝 (− ); Concentration: 𝐶(𝑟, 𝑥) = 𝐶𝑚𝑎𝑥 𝑒𝑥𝑝 (− )
𝑥2 𝑥2
r

Self-Similarity Zone
20d from source
d,U,ρ0

11.8 x

5d/2
Development Zone

Near Field: Dominated by jets; Short time and length scales: Seconds to minutes and less than a
meter to few meters
Far Field: Dominated by ocean processes. Long time and length scales: Hours to days; Hundreds
of meters to kilometers

Some works (Integral Theory) define self-similar profile as, for example 𝑢(𝑟, 𝑥) =
𝑟2 25𝑟 2 25𝑟 2
𝑢𝑚𝑎𝑥 𝑒𝑥𝑝 (− 𝑅2 ). In this case, 𝑢(𝑟, 𝑥) = 𝑢𝑚𝑎𝑥 𝑒𝑥𝑝 (− ); and 𝐶(𝑟, 𝑥) = 𝐶𝑚𝑎𝑥 𝑒𝑥𝑝 (− )
𝑥2 𝑥2

−2
𝑢 𝜂2 𝑦 𝑢 𝑦
Görtler Theory: Round Jets: 𝑢 ≈ (1 + ) ; 𝜂 ≈ 15.2 𝑥 Plane Jets: 𝑢 ≈ 𝑠𝑒𝑐ℎ2 (10.5 𝑥 )
𝑚𝑎𝑥 4 𝑚𝑎𝑥

𝑟50% = radial distance

where u= umax/2
Coanda Effect

The jet starts sinking if it is released too close to the ocean floor. Coanda effect occurs only in
turbulent jets and not in laminar flows.

Semi-empirical analysis: Minimum height above sea-floor to avoid Coanda attachment is 1m or

𝜋 0.25
0.12 (4 ) 𝑑. Fr , whichever is more.

[Source. Shao and Law (2009), Jr. Turb. DOI:10.1080/14685240903426505].

Conservation laws
• Mass is not conserved because ambient fluid entrains into jet.
2𝜋 ∞
𝑚̇(𝑥) = ∫0 ∫−∞ 𝜌𝑢(𝑥, 𝑟, 𝜃)𝑟𝑑𝑟𝑑𝜃 ≠ 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡
𝜕𝑃
• Momentum is conserved because = 0, i.e. no pressure gradient
𝜕𝑥
2𝜋 ∞ 2
𝑀(𝑥) = ∫0 ∫−∞ 𝜌𝑢 (𝑥, 𝑟, 𝜃)𝑟𝑑𝑟𝑑𝜃 = 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡
• Kinetic Energy is not conserved because ambient fluid entrains into jet and energy
transfer takes place.
2𝜋 ∞ 1 2
𝐾𝐸(𝑥) = ∫0 ∫−∞ 2 𝜌𝑢(𝑥, 𝑟, 𝜃)𝑢 (𝑥, 𝑟, 𝜃)𝑟𝑑𝑟𝑑𝜃 ≠ 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡
∞ 𝜋 𝜋
Mass Balance: 𝑚̇ = 𝜌𝑄 = ∫0 𝜌𝑢. 2𝜋𝑟 𝑑𝑟 = 50 𝜌𝑢𝑚𝑎𝑥 𝑥 2 = 10 𝜌𝑈𝑑𝑥
𝑑𝑄 𝜋
Therefore, entrainment of ambient water into profile, u𝑒 = = 10 𝜌𝑈
𝑑𝑥
∞ 𝜋𝑑2 5𝑑
Momentum Conservation: ∫0 𝜌𝑢. 𝑢. 2𝜋𝑟𝑑𝑟 = 𝜌0 𝑈. 𝑈 4
⟹ 𝑢𝑚𝑎𝑥 = 𝑥
𝑈
 1 ∞ 5𝑑𝑢 𝑢𝑚𝑎𝑥
Therefore Average Velocity at any distance 𝑢 = 𝜋𝑅2 ∫0 𝑢. 2𝜋𝑟𝑑𝑟 = =
2𝑥 2
∞ 𝜋𝑑2 5𝑑 5𝑑 50𝑟 2
Contaminant Flux, ∫0 𝐶𝑢. 2𝜋𝑟 𝑑𝑟 = 𝐶0 𝑈 ⟹ 𝐶𝑚𝑎𝑥 = 𝐶0 ⟹ 𝐶(𝑟, 𝑥) = 𝐶0 𝑒𝑥𝑝 (− )
4 𝑥 𝑥 𝑥2

Let by any standards, the reference value of contaminant concentration is 𝐶𝑟𝑒𝑓 [: 𝐶𝑟𝑒𝑓 = 𝐶(𝑟, 𝑥)].
𝑥2 𝐶0 5𝑑
Therefore, for the farthest extent of concentration exceeding reference level, 𝑟 2 = 50 𝑙𝑛 [𝐶 ]
𝑟𝑒𝑓 𝑥

𝐶0 5𝑑 5𝑑𝐶0
Zone of excess concentration condition, 𝑥𝑚𝑎𝑥 : 𝑟 2 > 0 𝑖𝑠 𝑔𝑖𝑣𝑒𝑛 𝑏𝑦 = 1 ⟹ 𝑥𝑚𝑎𝑥 =
𝐶𝑟𝑒𝑓 𝑥 𝐶𝑟𝑒𝑓

𝜕𝐶 𝜕2 𝐶 𝜕2 𝐶
For finding effective diffusivity, steady state advection diffusion equation: 𝑢 𝜕𝑥 = 𝐷 [𝜕𝑦 2 + 𝜕𝑧 2 ]
𝑆⁄
𝑢 𝑦 2 +𝑧 2 𝑆 5𝑑𝑈𝑟 2 5𝑑𝑢
⟹ 𝐶(𝑥, 𝑦, 𝑧) = 2 𝑒𝑥𝑝 (− 4𝐷 𝑥 ) = 4𝜋𝐷𝑥 𝑒𝑥𝑝 (− ); Since 𝑦 2 + 𝑧 2 = 𝑟 2 ; 𝑢 =
⁄𝑢 8𝐷𝑥 2 2𝑥
(√4𝜋𝐷 𝑥⁄ )
𝑢

50𝑟 2 5𝑑𝑈𝑟 2 5𝑑𝑈 𝑈𝑥 80𝑥

Comparing exponential terms, = ⟹𝐷= ⟹ 𝑃𝑒 = = ≥ 200 𝑎𝑡 𝑥 = 2.5𝑑
𝑥2 8𝐷𝑥 2 80 𝐷 𝐷

Puffs
V = mR3 ; m = 3 in place of 4/3 π (= 4.19)
d 3 3
R30 u0
(ρVu) = 0 ⟹ R u = R 0 u0 ⟹ u = 3
dt R
1
d d dR R30 u0 4au0 t 4
(ρV) = (ρmR3 ) = aR2 uρ ⟹ 3mR2 = aR2 3 ⟹ R = R 0 ( )
dt dt dt R 3mR 0
3 3
4au0 t −4 dx t
4au0 t −4 1
u = u0 ( ) = ⟹ x = ∫ u dt = 4u0 ( ) t 4 or x = nR; n = 4 [1.7 ≤ n ≤ 6.3]
3mR 0 dt 0 3mR 0
1 Sverdrup (Sv) = 1 Mm3/s

Basic Design Considerations

• Neglect mass transport due to waves, tides and gravity waves.
• Surface velocity in Ekman Layer: About 3 cm/s (Neglect Advection).
• Vertical Diffusivity in Oceans: About 2x10-6 to 0.05 m2/s (Neglect Vertical Diffusion).
• Horizontal Diffusivity in Oceans: About 2 m2/s (Use two-third of this value in both directions)
(𝛼𝑡)2
or, 𝐾ℎ = 𝐾𝑧 {1 + 12
}; α = 0.01 s-1
Numerical Solution

Finite difference methods (FDMs) are practically suited to solve the advection diffusion equation.
Other methods include finite element method, finite volume method, particle in cell etc.

FDM uses Taylor series approximation:

′ (x)
𝑑𝑥 2
𝑓(𝑥 ± 𝑑𝑥) = 𝑓(𝑥) ± dx f ± 𝑓"(𝑥) + ⋯
2!
𝜕𝑓 𝑓(𝑥 + 𝑑𝑥) − 𝑓(𝑥)
Forward Difference: = lim + O(dx)
𝜕𝑥 𝑑𝑥→0 𝑑𝑥
𝜕𝑓 𝑓(𝑥) − 𝑓(𝑥 − 𝑑𝑥)
Backward Difference: = lim + O(dx)
𝜕𝑥 𝑑𝑥→0 𝑑𝑥
𝜕𝑓 𝑓(𝑥 + 𝑑𝑥) − 𝑓(𝑥 − 𝑑𝑥)
Central Difference: = lim + O(dx 2 ).
𝜕𝑥 𝑑𝑥→0 2𝑑𝑥

𝜕2 𝑓 𝑓(𝑥+𝑑𝑥)−2𝑓(𝑥)+𝑓(𝑥−𝑑𝑥)
Second order derivative: 𝜕𝑥 2 = lim + O(dx 2 )
𝑑𝑥→0 𝑑𝑥 2

O(∆𝑥) (pronounced as “big Oh delta x”) is a notation for reporting errors. O(∆𝑥) means “of the
order of ∆𝑥”.

1-D Advection Diffusion Equation using Einsteinian Notation:

𝜕C u 𝜕C DH 𝜕 2 C
+ =
𝜕t R d 𝜕x R d 𝜕x 2

Or
𝜕C 𝜕C 𝜕 2C
+𝑈 =D 2
𝜕t 𝜕x 𝜕x

A combination, forward in time and centered in space (FTCS) is frequently used.

Error = O(∆𝑡, ∆𝑥 2 ).
𝐶𝑗𝑛+1 −𝐶𝑗𝑛 𝑛
𝐶𝑗+1 𝑛
−𝐶𝑗−1 𝑛
𝐶𝑗+1 −2𝐶𝑗𝑛 +𝐶𝑗−1
𝑛
+𝑈 =𝐷
∆𝑡 2∆𝑥 (∆𝑥)2
𝑈∆𝑡 𝐷∆𝑡
𝐶𝑗𝑛+1 = 𝐶𝑗𝑛 − 2∆𝑥 (𝐶𝑗+1
𝑛 𝑛
− 𝐶𝑗−1 𝑛
) + (∆𝑥)2 (𝐶𝑗+1 − 2𝐶𝑗𝑛 +𝐶𝑗−1
𝑛
)

Implicit FDM: Concentration is calculated after un-predefined ∆t

Explicit FDM: Concentration is calculated after predefined ∆t. It is commonly practiced method.

Errors in FDM
1. Round off errors: Due to computer representation of digits
2. Residual errors: Errors that remain when convergence criteria are fulfilled.
3. Truncation Errors: Due to higher order terms neglected in Taylor series expansion.
4. Oscillations: Due to coarse digitization in the time and space in the region of large
concentration gradients.
5. Instability: Errors grow at each iteration or time step so that the solution at the end
becomes infinity, viz. NaN, i.e. Not a Number.

U∆t
Courant No. = ≤ 1 : Controls the time step, viz. ∆t
∆x
U∆t
Peclet No. = ≤ 1 : Controls numerical dispersion
2𝐷
D∆t 1
Neumann criteria: (∆𝑥)2 ≤ 2 : Controls grid size
U∆𝑥
Cell Reynolds No. = ≤ 2 : Controls oscillations
𝐷

In multidimensional applications, these upper bounds are for sum of parameters in all directions:
𝑈𝑥 ∆t 𝑈𝑦 ∆t 𝑈𝑧 ∆t
e.g. Courant No. = + + ≤ 1. Courant criteria is also known as Courant-Friedrich-
∆x ∆y ∆z

Lewy (CFL) Criteria.

Lax-Richtmyer Theorem. For a consistent finite difference method for a well-posed (i.e. there
exists a unique solution) linear initial value problem, the method converges (i.e. numerical
method yields the same result as by analytical method) if and only if it is stable. This is a
fundamental principle in solving partial differential equations by FDM.
𝜕C 𝜕C 𝜕 2C 𝜕C 𝜕 2C
+𝑈 =D 2⇒𝑈 =D 2
𝜕t 𝜕x 𝜕x 𝜕x 𝜕x

𝑛 𝑛
𝐶𝑗+1 −𝐶𝑗−1 𝑛
𝑈 (∆𝑥)2 = 𝐶𝑗+1 − 2𝐶𝑗𝑛 +𝐶𝑗−1
𝑛
2∆𝑥𝐷

𝑛 𝑛 𝑛
𝐶𝑗+1 −𝐶𝑗−1 𝐶𝑗+1 −2𝐶𝑗𝑛 +𝐶𝑗−1
𝑛
𝑈 =𝐷
2∆𝑥 (∆𝑥)2

𝑈𝐿
Exact Solution: 𝑅𝐿 = 𝐷
𝑥 C=0 U C=1
𝑒𝑥𝑝 (𝑅𝐿 𝐿 ) − 1
𝐶=
𝑒𝑥𝑝(𝑅𝐿 ) − 1
L
𝑛 𝑛
U∆𝑥 𝐶𝑗+1 −𝐶𝑗−1 𝑛
𝑅= ⟹𝑅 = 𝐶𝑗+1 − 2𝐶𝑗𝑛 +𝐶𝑗−1
𝑛 𝑛
⟹ (𝑅 − 2) 𝐶𝑗+1 + 4𝐶𝑗𝑛 − (𝑅 + 2) 𝐶𝑗−1
𝑛
=0
𝐷 2

Try 𝐶𝑗𝑛 = 𝑞 𝑗
(𝑅 − 2) 𝑞 𝑗+1 + 4𝑞 𝑗 − (𝑅 + 2)𝑞 𝑗−1 = 0
2+𝑅
⟹ (𝑅 − 2) 𝑞 2 + 4𝑞1 − (𝑅 + 2) = 0 ⟹ q = 1 𝑎𝑛𝑑
2−𝑅

𝑗 𝑗 2+𝑅 𝑗
𝐶𝑗𝑛 = 𝐴1 𝑞1 + 𝐴2 𝑞2 ⟹ 𝐶𝑗𝑛 = 𝐴1 + 𝐴2 ( )
2−𝑅
2+𝑅 𝑁
𝐶0𝑛 = 𝐴1 + 𝐴2 = 0 and 𝐶𝑁𝑛 = 𝐴1 + 𝐴2 (2−𝑅) = 1
2+𝑅 𝑗
( ) −1
⟹ 𝐶𝑗𝑛 = 2−𝑅
2+𝑅 𝑁+1
( ) −1
2−𝑅
 Solution
1
0.8
0.6
0.4
0.2
0
-0.2 0 2 4 6 8 10

-0.4
-0.6
-0.8
-1
Exact Solution R=1 R=3

Stability Criteria:
𝑈∆𝑡 𝑈∆𝑡 𝐷∆𝑡 1 𝑈∆𝑥
(a) ≤ 1 (b) ≤ 1 (c) (∆𝑥)2 ≤ 2 (d) ≤2
2𝐷 ∆𝑥 𝐷

FTCS Discretization:
𝑈∆𝑡 𝐷∆𝑡
𝐶𝑗𝑛+1 = 𝐶𝑗𝑛 − 2∆𝑥 (𝐶𝑗+1
𝑛 𝑛
− 𝐶𝑗−1 𝑛
) + (∆𝑥)2 (𝐶𝑗+1 − 2𝐶𝑗𝑛 +𝐶𝑗−1
𝑛
)

Case 1. U and D Constant:

Calculate ∆𝑡 from (a) and then ∆𝑥 from (b) and satisfy (c) and (d)
Case 2. U and/or D Variable:

Finite difference methods are basically not meant for space and/or time varying coefficients of
the advection diffusion equations. Even then good estimates can be obtained if minimum and
maximum values are available for the domain of application and then use minimum grid size
and minimum time step in such a way that the four conditions of stability criteria could be
satisfied.
 ↑
Time

n+1

n
J-1 J X→
Case 1. U and D are Constants
1. Specify source to occupy the adequate number of cells. These cells are assigned the
concentration at the source. The concentration may be calculated by total quantity emitted/
total source extent as calculated by no. of cells in 1, 2 or 3D times cell dimensions.
2. Assign all cells a value that represents the background concentration or nil.
3. For a fixed n, iterate for j, viz. j-1, j, j+1 …., etc. including all dimensions.
4. For a fixed j, iterate for n, viz, n, n+1, n+2, …
5. Add the background concentration in all these values in each cell.
Case 2. U and/or D Variable(s)
1. Keep track of dimensions of cells while doing the iteration. The space dimensions of the cells
will now vary and number of steps in the domain need to be dynamically decided based on
dimensions of the domain.
2. The basis of finite difference approximation, that the entire cell is occupied at once leads to
numerical errors called artificial diffusion. Though some smoothening may be provided, yet
it is a rather crude methods and more error prone. Finite element methods are practiced via
various solvers available for the purpose.
Interphase Mass Transfer

Gas-Liquid Interface: Henry’s Law

Form H for O2 in water, at 250C
CL = H. PG 769.23 L.atm/molG
CG = Hc.CL 3.18x10-2
PG = He. CL 1.3x10-3 molG/L.atm
PG = Ha. X 4.259x104 atm. molL/molG
Note. Reported values of H vary in literature

C d ln(H) ∆h
Effect of temperature on Henry’s coefficient: H = P L ; 1 = ; ∆h=Heat of absorption (-ve)
G d( ) R
T

1 1
For oxygen in water, H(T) = H(T0 )exp {−1700 ( − )} ; T0 = 298 K
T T0
Conversion of forms of Henry’s coefficient: PV=nRT; R = 0.0821 atm-L/mol-oK
n Hc.PG Hc n 1
PG = He. CL ; CG = V = Hc. CL = ⟹ = V.P = RT = 40.87
H H G

Theories of Interphase Mass Transfer:

• Nernst (1904) Film Theory: J α D
• Higbie (1935) Penetration Theory: J α √D
• Danckwerts (1951) Surface Renewal Theory: J α √D
These theories do not work because in reality as observed flux: J α D(0.5 to 1)

Two film theory

It is a modification of Nernst Film Theory proposed by Lewis and Whitman (1923). It uses three
hypothetical steps of mass transfer of gas to liquid or vice-versa:
• Transfer of gas from the bulk gas to the gas film.
• Transfer of gas across the gas-liquid films’ interface.
• Transfer of gas from the liquid film to the bulk water.
In reality there is no film and practically it is the top layer of liquid and thickness of the
imaginary film is inversely proportional to kinetic energy of molecules on either side of
interface.

The theory states that transfer rate can be expressed in terms of an overall transfer coefficient
and resistances on either side of the interface. Gas film governs the mass transfer of highly
soluble gases while liquid film controls mass transfer of less soluble gases in liquid.

Steady-state flux balance through each film

DL,G
J = k L (CL − CLi ) = k G (CGi − CG ); k L,G = δL,G

Where, subscripts: G-Gas, L-Liquid and i=Interface

Overall mass transfer coefficients for these phases
CG
J = K L (CL − CLe ) = K G (CGe − CG ); CLe = andCGe = HC CG
HC
Incorporating interface liquid concentration in liquid phase overall flux,
J = K L [(CL − CLi ) + (CL − CLe )] = K L [(CL − CLi ) + (CGi − CG )/HC ]
J J 1 1 1 1 HC 1
Therefore, J = K L [k + k ]⟹K =k +k and K = +k
L G HC L L G HC G kL G

Corrections to overall mass transfer coefficient:

Temperature: 𝜃: K L,T = K L,20 (𝜃 = 1.024)𝑇−20
KL 𝑊𝑎𝑠𝑡𝑒𝑤𝑎𝑡𝑒𝑟 CLe 𝑊𝑎𝑠𝑡𝑒𝑤𝑎𝑡𝑒𝑟
Surfactants:𝛼 = = 0.2 − 1; TDS:𝛽 = = 0.85 − 1;
KL 𝑃𝑢𝑟𝑒𝑤𝑎𝑡𝑒𝑟 CLe 𝑃𝑢𝑟𝑒𝑤𝑎𝑡𝑒𝑟
Example
Consider an oxygen bubble 0.1cm diameter injected to an oxygen free stirred tank at 25 0C.
Solution. Oxygen is less soluble in water and therefore the liquid film governs the mass transfer.
If the bubble diameter reduces to 0.054 cm after 10 minutes,
dCG V
= JA;
dt
4 dr 3
J = k L (CL − CLi ) ⟹ πCG = 4πr 2 (CL − CLi ); CL = 0.
3 dt
dr kL (CL −CLi ) kL (CL −CLi ) CG (r1 −r0 )
⟹ = ⟹ r1 − r0 = t ⟹ kL
dt CG CG (CL −CLi )𝑡

po2 1 atm mol

If po2 = 1 atm, then CG = by ideal gas law = = 0.041
RT 0.08206 × 298 L
Using Henry’s coefficient for air water interface at 250C, viz 43800 atm, i.e. P=HaX
1 atm 997𝑔 𝐻2 𝑂/𝐿
CLi = ( mol O2 /𝑚𝑜𝑙 𝐻2 𝑂) × = 1.26 × 10−3 mol O2 /L
43800 atm 18 𝑔/𝑚𝑜𝑙
(0.05 − 0.027)(0.041)
⟹ Mass Transfer Coefficient = k L = = 1.25 × 10−3 𝑐𝑚/𝑠
1.26 × 10−3 × 600

Aeration Practice
Note. Conventionally the aeration system of wastewater treatment plants are designed using
α=0.4 for fine bubble diffusers, 0.8 for coarse bubble diffusers and 0.85 for surface aerators.

Semi-empirical Mass Transfer Coefficients

μ Momentum Diffusivity
Schmidt Number: Sc = ρD = Mass Diffusivity

(ρL −ρG )ρL gd3b Buoyancy Force

Grashof Number: Gr = =
μ2L Viscous Force

KL L Convective Mass Transport

Sherwood Number: Sh = =
D Diffusive Mass Transport
1 1
Large Bubbles (>2.5mm dia, db): Sh = 0.42Gr 3 Sc 2
1 1
Small Bubbles (<0.6mm dia, db): Sh = 2 + 0.31Gr 3 Sc 3

When there are large number of bubbles in the system and it is not feasible to determine the
actual area across which mass transfer occurs, the mass transfer coefficient, K L is multiplied by
P 0.7
a=(A/V) of interface and correlations like K L a = 0.0020 (V) vs0.2 are used. This expression is

for non-coalescing bubbles, P/V is in Watt/m3 [500≤P/V ≤ 10000] and vs is the superficial gas
velocity (m/s) = volumetric flow rate/ Surface area; KLa is in s-1. Typical range of superficial gas
velocities in plants with diffused aeration: 0.2 to 0.3 cm/s.

Example
Consider 3 mm dia air bubbles stirred in water at 37 0C. To estimate the mass transfer
coefficient,
1 1
𝑆ℎ = 0.42𝐺𝑟 3 𝑆𝑐 2
1/3
𝐾𝐿 𝐿 (𝜌𝐿 − 𝜌𝐺 )𝜌𝐿 𝑔𝑑𝑏3 𝜇𝐿 1/2
= 0.42 { } { }
𝐷 𝜇𝐿2 𝜌𝐿 𝐷
1/3
(𝜌𝐿 − 𝜌𝐺 )𝜌𝐿 𝑔𝑑𝑏3 𝜇𝐿 1/2 𝐷
𝐾𝐿 = 0.42 { } { }
𝜇𝐿2 𝜌𝐿 𝐷 𝐿
Substituting at 37 0C: 𝜌𝐿 =0.994 g/cc; 𝜌𝐺 =0.00113 g/cc; g=981 g/cc; 𝜇𝐿 =6.95x10-3 g/cm.s;
D=3.3x10-5 cm2/s; db=L=0.3cm in this expression, 𝐾𝐿 =0.055 cm/s
Now consider 0.3 mm dia bubbles,
1 1
𝑆ℎ = 2 + 0.31𝐺𝑟 3 𝑆𝑐 3 or
1/3
𝐾𝐿 𝐿 (𝜌𝐿 − 𝜌𝐺 )𝜌𝐿 𝑔𝑑𝑏3 𝜇𝐿 1/3
= 2 + 0.31 { } { }
𝐷 𝜇𝐿2 𝜌𝐿 𝐷
1/3 1/3 𝐷
(𝜌𝐿 −𝜌𝐺 )𝜌𝐿 𝑔𝑑𝑏3 𝜇
Substituting these values, 𝐾𝐿 = [2 + 0.31 { } {𝜌 𝐿𝐷} ] 𝐿 = 0.019 𝑐𝑚/𝑠
𝜇𝐿2 𝐿

ASCE Method
dC
Surface Aerators: dt = K L a(C∞ − Ca ) ⟹ ln(C∞ − Ca ) = ln(C∞ − C0 ) − K L a(t − t 0 )

Where
dC
= Oxygen Transfer Rate, OTR
dt

𝐾𝐿 𝑎 = volumetric mass transfer coefficient (1/T),

𝐶∞ = average DO concentration at infinite time (mg/L)
𝐶𝑎 = effective average DO concentration in the liquid phase.
 (P+ρgh)
Submerged Aerators: Csat = C∞ = Cs at top; and Csat = Cs at h depth in the tank.
P
h
∫0 He Y (P+ρgz)dz
Average saturation concentration, Csat = = Csat at middepth
h

[Barometric Equation: = 𝑃0 exp(−1.1856 × 10−4 𝐴𝑙𝑡) ≈ 𝑃0 (1 − 1.05 × 10−4 Alt)]

Where
Alt = Altitude in m
P = Atmospheric pressure
ρ = Weight density of water
He = Henry’s law coefficient
𝑃0 = Atmospheric pressure at msl
Y = Mole fraction of oxygen in gas phase (0.2095)
Cs = Tabulated value of saturation concentration.
Csat = Estimated average saturation concentration
h= Aerator submergence depth, m, usually approximated by liquid depth above the
aerator level

Example

Consider a 10x10x4m aeration tank at Roorkee, Alt=269m. 𝑃0 = 101.325 kPa. T=25 0C. If coarse
bubble diffuser is to be used then a=A/V=(10x10/(10x10x4)=0.25.
P ≈ P0 (1 − 1.05 × 10−4 Alt) = 101.325 × (1 − 1.05 × 10−4 × 269) = 98.463 = 0.972 atm.
He = 1.3x10-3 M/atm ⟹ 𝐶𝑒 = 1.3x10-3 × 0.2095 × 0.972 𝑀 = 0.0002647𝑀 =
0.0002647x32x1000 = 8.47 mg/L.

Average Ce may be calculated from that at mid-depth, viz. 2m: P + ρgz =

98.463+ 0.997 x9.81 x2 = 118.024 kPa = 1.165atm ⟹ Ce = 10.15 mg/L
dC
= K L a. 𝛼(𝛽C∞ − Ca ); Typically 𝛼=0.8 and 𝛽=0.99. Let Ca =0 and 𝐾𝐿 = 5.5 10−4 m/s as
dt
dC
calculated in the previous example. Substituting the values, maximum OTR = = 5.5 × 10−4 ×
dt

0.25 × 0.8(0.99 × 10.15 − 0) = 11.05 × 10−4 mg/L.s