Problem 1.

As part of a water quality study, you have been asked to assess the diffusion of a new
fluorescent dye. To accomplish this, you do a dye study in a laboratory tank (h = 40 cm).
You release 100 g of the dye at a depth of 20 cm (spread evenly over the area of the
tank) and monitor its development over time. Vertical profiles of dye concentration in
the tank are shown in the figure below; the x-axis represents depth and the y-axis
represents the reading on your fluorometer.
Problem 1.4

Note:
(a)the densities of air and of water are 1.23 and 1000 kg/m3 respectively
(b)the diffusion coefficient of gaseous carbon dioxide is 0.14 cm2/s in air and
1.71x10-5 g/cm3 in water
Problem 2.1 Nutrient Removal Wetland

Wetlands can be used to remove nutrients, such as Phosphorus or Nitrogen, from

agricultural run-off. The largest such project is the Everglades Nutrient Removal
Program . Consider a fixed basin size for the wetland and a fixed flow rate, such that
TRis pre-determined. Which circulation type would provide the greatest phosphorus
removal? Plug-Flow or Stirred Reactor? Consider two cases.

1. Phosphorus is removed at constant rate, with all removal completed in time TR.

2. Phosphorus is removed in a first-order reaction with removal rate, k = 1/TR.

Problem 3.1

Two airplanes are conducting tests to estimate the coefficient of diffusion in a stagnant
atmosphere. The first airplane, flying 1-km above the earth's surface, instantaneously
releases 1-kg mass of trace gas. A second airplane flies through the cloud 60 minutes
later and measures a maximum concentration of 0.03 mg/l within the cloud.

Estimate the coefficient of diffusivitywithin this region of atmosphere. Hint 1 Hint 2 Hint
3
Estimate the coefficient of diffusivitywithin this region of atmosphere.

Hint 1- Can you assume uniform concentration in any direction?

The gas is effectively released into an unbounded domain, so that one cannot expect it to

mix rapidly to uniform conditions in any direction, i.e. ∂C/∂y ≠0, ∂C/∂z ≠0, ∂C/∂y ≠0.

All three dimensions must be retained in the governing equation.

Hint 2 -What assumption can you make about the air currents?

The atmosphere is stagnant, which implies that there are no ambient air currents. You

may assume that u=v=w=0. The passage of each plane will create air movement. This

movement may locally and temporarily enhance the dilution of the cloud. However, this

effect dies out within a few minutes, so that over an hour time frame, the effect is

negligible.

Hint 3 -What assumption must you make about the coefficient of diffusion?

Since you are only given one value of concentration, you can solve uniquely for only one

value of diffusion. Thus, to make the problem tractable, you must assume that the

diffusion is isotropic.

With the above assumptions, the transport equation reduces to,

Problem 3.2

A friend of yours is coming to meet you at youroffice, which is located mid-way along a
very long (100-m) hallway. In preparation for a small prank, he is carrying a vial
containing 10-g of a noxious smelling gas. He trips 20-m before reaching your office
door, and the vial breaks. The gas rapidly mixes vertically and horizontally within the
hallway, which is 2-m wide and 3-m high. The human nose will detect the gas at
concentrations greater than 10-µg/l. Assume an isotropic diffusion, D = 0.05 m2/s.

What governing equationdescribes the evolution of the gas concentration in the

hall?
At what timeafter the spill do you smell the gas?
When does the smell, as perceived by humans, disappear from the hallway?
What governing equationdescribes the evolution of the gas concentration in the
hall?
For isotropic diffusion, the governing equation is:

Align the coordinate x to the length of the hallwayand define the y-z plane as the cross-
section. Since the gas "mixes rapidly" in the vertical and horizontal, we assume ∂C/∂y =
∂C/∂z = 0. The problem statement gives no information about air currents in the hallway,
so we assume they are negligible, u = 0. With these assumptions, the governing equation
is reduced to,

If the spill location is taken as x = 0, the initial condition is then, M = δ(x)

At what timeafter the spill do you smell the gas?
The governing equation and initial condition above describe an instantaneous, point
release
diffusing in one-dimension. The concentration field is described by Equation 10in chapter
3

Use 3.10 to find the time at which the concentration at your door, C(x=20m, t) is 10 µgl-1
. As time appears in both the exponential and leading terms, it is simpler to use a
graphical solution.
Whendoes the smell, as perceived by humans, disappear from the hallway?
To answer this question we need information on the end conditions of the hallway. Lets
first
consider that the hallway is open at both ends, so that odor can diffuse beyond the end of
the
hall. Then, we need only consider the evolution of the maximum concentration, located at
the
site of the spill, x = 0. We seek the time for which

At this time, the length of the cloud will be 4ζ=4 2Dt =266m , indicating that the cloud
has diffused beyond the length of the hallway. The above time scale is correct, only if
the hallway is open at both ends. If the hallway is shut off by fire doors at both ends, then
in reality the cloud cannot diffuse beyond the length of the hall. Under these conditions,
the final concentration in the hallway set the maximum possible dilution, which is
determined by distributing the total mass released over the total volume ofthe hallway
(2m x 3m x 100m).

Since the final concentration is above the detection limit, the smell will not disappear
until the fire doors are opened.
a. While both clouds are fully visible (C > 10-g l-1), which cloud will appear larger,
and by how much?

Since the size of the cloud increases in proportionto the diffusion coefficient, the blue
cloud will grow more rapidly, and thus appear bigger, than the red cloud. Specifically,
the length scale of each cloud, as defined in chapter 3, equation 26, is

b. At what time and at what location will the two dye clouds first appear to
touch?
Hint 2:Simplify governing equation with assumptions.
Begin with the full transport equation that governs the evolution of both dye drops.
Note that the exponential is written to be one at the center of the red cloud, i.e. at
(x=10,y=0), with the position given in units of centimeters.

Hint 3: Write mathematical criteria for condition when clouds first touch

From geometry, the two clouds willfirst touch along the line y = 0. We wish to find the
time, t, and location, x, for which CB(x, y=0, t) = CR(x, y=0, t) = 10 g/l. This could be
tackled analyticaly by first finding the position xBR(t) at which CB=CR. Then solve for t
using this position to constrain x, e.g. solve CB(x = xBR, y=0, t). However, a simpler
approach is to graph Cba nd CRin an interactive graphing package, such as Excel, and
then vary time until the intersection of the two concentration curves lies at C = 10 g/l
Solution:- At t = 20500s, the intersection of the blue and red concentration curves
corresponds to C = 10 g/l and is located at x = 3.1 cm. The clouds will first appear to
touch and x = 3.1 cm.

Make a rough estimate of the locationusing your result from part a?

Based on the definition used in a. and the definition sketch shown above, the two clouds
first appear to touch when, (LB/2)+(LR/2) = 5. Additionally, LR= (LB/2), so the edge of
the blue cloud will be at x = (LB/2) = 5/1.5=3.3cm, when it first touches the red cloud.
c. At what time will the line connecting the release points be completely purple?
Hint 4:- Define a mathematical criteria for this to occur?
This condition requires that CBand CR> 10 g/l in the region x = 0 to 10 cm. Use the
spreadsheet created in part b to interrogate the concentration field over a range of time.
Solution- Graph CB and CRin an interactive graphing package, such as Excel and vary
time until the above criteria is met. You will find thatthe criteria is never met. Between
0<x<5 cm, when CR ≥10g/l then CB< 10 g/l, and when CB≥10g/l, CR< 10 g/l.

Problem 4.1 -A slug of mass, M, is released instantaneously into the corner of a large,
shallow box. The full width and length of the box are LX = LY = 100L, and the height of
the box is LZ= 0.01L. Every wall of the box is a no-flux boundary. The mass is released
a distance L from two adjacent walls, and mid-way between the top and bottom
boundary. Assume isotropic diffusion withinthe box, represented by diffusivity, D.
Describe the concentration field inside the box from t = 0 to t = L2/D. Hint 1 Hint 2 Hint
3 Hint 4 Solution

Describe the concentration field inside the box from t = 0 to t = L2/D.

Hint 1 - When will the mass be mixed uniformly in the vertical?

From equation 4.24 a mass released mid-way between two parallel boundaries a distance
LZ apart will be mixed to a uniform concentration between those boundaries in time t =
LZ2/4D, where LZis the distance between the boundaries. The time for the concentration
to become well-mixed in the vertical is then, t = (0.01L)2/D = 0.0001 L2/D.
Hint 2 - Estimate when the mass will reach each vertical wall in the box.

Hint 3 - How will each boundary impact the solution in the time t = 0 to L 2/D?
Use the time scales determined in hint 1 and 2. Because the concentration field becomes
uniform in the vertical very rapidly, within 1/10,000th of the time of interest, we will
assume that the concentration is instantly uniform in z, i.e. ∂C/∂z = 0 for all time. In the
time of interest, the cloud will never reach the far walls (545L2/D >> L2/D), so these
boundaries do not impact the solution. The cloud will reach the near walls (L2/18D <
L2/D), and a correction must be made to satisfy the no flux condition at these walls.
Hint 4 - Place images sources to satisfy the no-flux boundary condition.
Image sources are needed at the following locations. Image 2 balances the loss of mass
from Image 3 across the x-axis and the loss of mass from Image 1 across the y-axis.
Solution- The concentration in the box is described by a superposition of two-
dimensional, instantaneous, slug releases (equation 3.23) at each of the above sources.

Problem 4.2
You own a house and dock along a boat canal, which ends 25 m upstream from you. One
day, your neighbor has a small (1 kg) fuel spill. Due to the boat traffic, the diffusivity
inthe canal is quite high, D = 0.01 m2/s. The current in the canal is negligible, such that
the fuel is transported to your house (x = -50 m) by diffusion only. Assume the fuel
mixes rapidly across the width and depth, and that there is no flux through the canal
walls.

(a) What is the concentration at your house10 hrs after the spill ?
(b) What is the maximum concentration at your house, and when does it occur ?
(c) Suppose the safety limit is 0.2 g/m3. At what timeafter the spill is this concentration
reached?
(d) Repeata, b & c assuming that the boundary at x = -75 m is totally absorbing.
(a)An image source is needed at x = -150 m to account for the no-flux boundary at the
end of the canal. The concentration of fuel in the canal is described by the superposition
of these two instantaneous, 1-D sources.
a) Estimate the time scale, T, at which the dye becomes uniformly distributed in x.
The diffusion of dye released at the end wall (top figure) will be similar to the diffusion
of dye released mid-way between end walls placed twice as far apart (bottom figure).
This is because the diffusion in both systems proceeds as a function of exp(-x2/4Dt),
which is symmetric about x = 0. From this similarity we expect that the dye will be well
mixed in both systems in the time-scale already established for the bottom system.
Specifically, from eq. 4.24applied to the bottom system, T =(2L)2/4D= L2/D= 107sec.

b) Confirm your estimate by plotting C(x) at the times t = T/10, T/4, T/2, T.
Theoretically an infinite number of images is needed to satisfy a no-flux boundary
condition at two parallel boundaries. For simplicity we initially consider only one image
for each boundary. The boundary at x=0 requires image I1 at x=0, i.e. co-located with
the source. The boundary at x=L requires image I2 at x=2L. The concentration field is,
The solution indicates that two images are not sufficient, because mass is not conserved
in the time of interest. The final concentration should be 1mg/(10cm3) = 0.1 mg cm-3, but
the above solution yields 0.66 mg cm-3 at t =L2/D. In addition, the gradient of
concentration should be zero at the boundaries, ∂C/∂x = 0, to satisfy the no-flux
condition. This condition is not met at x = 10cm. For comparison, now consider a
solution with six images, each denoted by I.

With six images the concentration is correct at T = L2/D. At longer times mass will
eventually be lost due to unbalanced images and the concentration will decline. However,
longer times hold no interest, because once the system is well mixed in x (∂C/∂x = 0 for
all x), the solution may be ended asno further evolution of the true profile will occur.
Note, the shape of the concentration profiles are similar with 2 and 6 images, and both
solutions indicate a uniform distribution in x is achieved by T = L2/D.
Problem 4.4:
Consider the two systems shown below. System 1 is enclosed by no-flux walls which
define a domain of dimensions 1m x 1m x 0.1m. System 2 is defined by parallel,
horizontal (x-y plane), no-flux boundaries at z = ± 0.1 m, but is otherwise unconstrained.
Both systems have an isotropic diffusivity of D = 2 cm2s-1 . At t = 0 a mass, M = 100g, is
released into both systems at x=0, y=0, z=0. A concentration probe (A and A‘) is located
in each system at the position (x = -0.5 m, y = 0, z = 0). The detection limit of these
probes is 10 ppm (gm-3).
a) Estimate the timeat which the concentration measuredat A and A‘ begin to
diverge?
b) What is the final concentration measuredin each system, and when is this
concentration achieved?
c) Confirm your estimatesfrom a) and b) by plotting the concentrations at A and A‘.

b) What is the final concentration measured at A (A‘), and when is this

concentration achieved?
The final concentration in System 1will be C = (100g)/ (1m x 1m x 0.1m) = 1000 ppm. It
is achieved when the mass is fully mixed across the domain. Using the largest dimension
to estimate this time-scale, t = (L)2/(4 D) = 1250 s. As noted in the chapter text, this is a
conservative estimate, and t = (L)2/(8 D) = 625 s, is also a reasonable estimate. Because
System 2 is unbounded, infinite dilution is possible and the final concentration will be C
= 0 ppm. Theoretically, this will take infinite time. Because the probe has a detection
limit of 10 ppm, zero concentration will be recorded for any concentration C < 10 ppm,
which occurs in a finite time.
c) Confirm your estimate from a) and b) by plotting concentrations at A and A‘.
In both systems the concentration is uniform in z at t = (10cm)2/(4x2 cm2s-1)=12.5 sec
after the release. Because thisis short relative to other timescales of interest (1250 sec),
we can neglect the three-dimensional phaseof the cloud and use a two-dimensional
solution. For System 1, an infinite number ofimage sources are needed to satisfy the no
flux boundaries. Following equation 4.19,

In practice an infinite number of images isnot needed. For time less than required to reach
a well-mixed condition between the boundaries (t < L2/4D), three images per boundary is
sufficient to approximate the full solution with infinite images. Beyond this time, the
concentration is steady and uniform, and the detailed solution above is no longer needed.
System 2 is described by a simple two-dimensional slug-release (eq. 3.23),

FromC1 and C2 the concentration at A (A‘) is plotted (see corresponding animation).

Problem 5.1. One gram of dye is injected into a long, thin tube and instantly mixes across
the cross-section, A = 1 cm2. Water is pumped through the tube at Q = 0.1 cm3s-1 .
Assume that the flow is uniform across A. If the flow is laminar, assume a molecular
diffusion coefficient, DM= 10-5cm2s-1. Otherwise, assume D = 100 DM. The
concentration is measured at x = L = 100 cm downstream from the injection point.
Estimate the length of the dye patch as it passes x = L and the maximum concentration
experienced at this point.
Problem 5.2. A 1-kg mass of dye is dumped into a 10-m2channel at a location designated
as x = 0. Downstream, at x = L = 100 m, the concentration of the dye is measured. The
test is repeated for three different flow rates, u = 0.001, 0.1, 1 ms-1. For simplicity, the
diffusion coefficient is D = 1 m2s-1 for each condition.
a) For each case estimate the advection and diffusion time-scales, and the Peclet Number.
b) The curves below represent the three cases. Label each curve with the appropriate
Peclet number.
c) For which curves will the peak concentration observed at L occur at the advection time
scale? Calculate the peak concentration for these cases.
d) The injection of mass required 10 seconds. Considering the other relevant time scales,
can this injection be considered instantaneous?
a) and b) For all cases the length scale is L = 100 m and the diffusion rate D = 1m2s-1.

c) For Fickian diffusion, the peak concentration occurs at the center of mass which
arrives at the advection time scale, i.e. at time TU the peak concentration in the cloud
will be located at L. However, the peak concentration observed at L will not
necessarily occur at TU. This is because the magnitude of the concentration is
changing as the cloud passes the measurement point. If the cloud is passing very
slowly (Pe << 1), the concentration may decline considerably as the cloud is passing,
such that the peak concentration seen at L will occur before the center of mass arrives.
If the concentration changes very little as the cloud passes (advection is faster than
diffusion, Pe >>1) the peak concentration observed at L will correspond to the
advection time scale. We expect this to occur for the Red and possibly the Green
systems.
To find the concentration we must first determine whether the cloud is three-, two-, or
one-dimensional as it reaches the measurement position. This will depend on the time
required to mix the dye across the channel area, A = 10 m2. If we assume a square
cross-section, then the width and depth of the channel is, Ly= Lz= √10 m. The mixing
time-scale (see chap 4, eq 24), is ti= Li2/ 4Di= 10m2/(4 m2s-1) = 2.5 s. For each curve
both TD and TU>> 2.5 s, so we can safely assume that the dye is well-mixed across the
channel when it arrives at L. Thus, we can use a one-dimensional solution to estimate
the concentration observed at L at t = L/U.

This is LESS than the concentration observed at L at t = 4000s (C = 0.25 gm -3). For
this case the peak concentration observed at L occurs before TU, i.e. before the center
of mass passes L. This is because the concentration in the cloud is dropping off faster
(via diffusion) than the center of mass can travel (via advection), consistent with Pe
<< 1. The spatial distribution, C(x), for this condition is shown below.
d) We are interested in the concentration observed at L = 100m. In all cases the transport
time scale for this location is 100 s or more, which is much longer than the injection time
scale. For this reason the injection can be assumed to be instantaneous.

Problem 5.3.
You are the safety officer for a new factory with a permit to discharge a treated waste
stream into the center of a river at a flow rate of 2 m3s-1 . The river has a square cross
section with an area of 20m2. Under standard operating procedures the discharge is fully
treated before release. The discharge is monitored, and if the chemical concentration in
the waste stream becomes elevated above a safety threshold of 100 mgl-1the factory will
shut-down. However, the response time for the shut down is 100 seconds. You are asked
to prepare an environmental impact report for a fish hatchery located 1-km downstream
of the plant. The hatchery draws water from the river to flush the hatchery tanks. Based
on studies of chronic exposure to juvenile fish the safety standard at the hatchery is set at
C = 0.5 mgl-1.
Write an expression describing the concentration experienced at the hatchery. Then, for
each scenario below find the peak concentration and the duration of exposure at the
intake to the fish hatchery. Then describethe exposure threat - considering duration and
level - for each flow regime and state whether you would issue a water quality warning.

Start by defining the system dynamics using time scales

We are interested in predicting the concentration at L =1000 m. Because 1) the channel is
square; 2) Dy= Dz, and 3) the discharge is at the center of the channel, the time to mix
across the cross-section is, Tmix= 20m2/(4Dy). All time scales are given in seconds.

In both cases Tmix<< TDand TU, so we can assume that the chemical is uniformly
distributed in y and z long before it reachesthe hatchery. Thus, we can use a 1-D solution
to describe the concentration experienced at L = 1000 m. In addition, the release time
[100 s] is also very short compared to either transport time scale, which justifies the use
of an instantaneous release model. Finally, since concentration is uniform in y and z, no
images are needed to correct for the no-flux boundaries, i.e. the condition ∂C/∂y = 0 and
∂C/dz = 0 is met at the boundaries becauseit is true everywhere after t = Tmix. With these
approximations we represent the concentration at the hatchery as
Problem 5.4
A chemical plant releases exhaust gas at a flow rate of 10 m3/s through a smokestack of
height H = 20 m. A failure in the filter system allows a pulse of sulfur dioxide, SO 2, to be
released with the exhaust. The concentration of SO2 in the exhaust is normally zero, but
during the failure increases to 10 mg/m3. This release occurs for TR= 5 minutes. Assume
an isotropic diffusivity, D = 1 m2/s, and uniform wind speed, U = 2 m/s.

a) Using the coordinate system shown in the sketch above, and assuming the release
occurs as an instantaneous slug of mass, M, write an expression for concentration of
chemical XX downstream. Assume that the ground is a no-flux boundary.
b) Calculate the Pe# for x = 500 m and explain what the value indicates about transport.
c) At what distance downstream will the release appear as an instantaneous release?
d) Your home is 500 m downwind of the plant. What is the maximum concentration you
would expect to observe at ground level at your house?

a) Since the system is unbounded in each coordinate direction (x, y, z), the concentration
can never become uniform in any direction and a three-dimensional solution will apply.
Assuming the release occurs at t = 0 s, the trajectory of the chemical cloud's center of
mass is (x = Ut, y = H, z = 0). To satisfy the no-flux boundary at y = 0, we must add an
image source at y = -H. The center of mass of the image source follows the trajectory
(x = Ut, y = -H, z = 0). With these conditions, and isotropic diffusivity the concentration
solution is

Problem 5.5.
Consider a long, narrow channel connected to a harbor that receives drainage water from
the city at its landward end (x = 0) at an average rate, Q = 0.005m3/s. The channel is 75-
m long, 2-m deep, and 10-m wide. One morning a boat spills a tank of soluble degreasing
fluid at x = 25 m. Use the following questions to assess the impact of this spill on the
harbor. Assume an isotropic diffusivity, D = 0.1 m2/s.
a. Estimate the Peclet No., Pe =UL/D, for the system and explain its meaning.
b. What time do I have to execute preventive measures to keep the contaminant out
of the harbor?
c. Estimate the time scale for the contaminant to become well-mixed in the lateral
and vertical.
d. To describe the concentration of contaminant observed at the harbor (x = 75 m),
should I use the 1-D, 2-D, or 3-D evolution equation? Justify your answer.
e. Using symbolic language [U, D, x, t, M, etc] write a mathematical expression that
describes the evolution of the spill in time and space, after the concentration has
become well-mixed across the width and depth of the channel. Assume that the
contaminant cannot pass through the back wall of the inlet at x = 0.
f. State two limitations of the model you have chosen.

a.Assume the flow fills the channel uniformly, then U =Q/A = 2.5 x 10-4ms-1. Using L =
75, Pe = 0.19. Alternatively, for L = (75-25) = 50 m, Pe = 0.13. In either case the Peclet
number indicates that the system is dominated by diffusion.
b.According to the Peclet number, transport is dictated by diffusion. The time-scale for
the contaminant to reach the harbor can be estimated as the time-scale for diffusive
transport over the 50-m between the spill and the harbor entrance. TD= (50m)2/(8 x
0.1m2s-1) = 3125 s ≈1hr. So, I have one hour to put up a contaminant-absorbing barrier
and protect the harbor
c.In the vertical, tmix=(2 m)2/(4 x 0.1 m2s-1) = 10 s
In the lateral, tmix=(10 m)2/(4 x 0.1 m2s-1) = 250 s
d.Since the mixing time scales in both the lateral and vertical are much shorter than the
time required for the contaminant to reach the harbor, we can assume the contaminant is
mixed across the channel area when it reaches x = 75 m. If concentration is uniform
(well-mixed) in the lateral and vertical, we can drop these two dimensions, and use a one-
dimensional solution.

The image source is needed to satisfy the no-flux boundary at x = 0. Since Pe<<1, we
could also neglect U entirely and still get a good representation of C(x,t).
f. U and D are assumed to be uniform, i.e. not functions of (x, y, z), but in fact the no-slip
condition at the channel boundaries will make U = f(y,z). We assume that D is isotropic.
In fact, turbulent diffusion in the longtudinal direction will be much more rapid than in
the vertical or horizontal. This is discussed further in Chapter 9. We neglect losses to the
atmosphere, when in fact the gasoline is volatile. We assume that no gasoline absorbs to
the sediment. We assume that the flow and thus velocity are not functions of time.

Problem 6.1
A small stream was found to be contaminated with Lindane, a pesticide known to cause
convulsions and liver damage. Groundwater wells in the same region have also been
found to contain Lindane, and so you suspect that river contamination is due to
groundwater inflow. To test your theory you conduct a dye study. Based on the
information given below, estimate the ground water volume flux, QGW, and the
concentration of Lindane, CLGW, in the groundwater.
Station 1: A 50-mg/l solution of tracer is injected at the rate of Qi= 100cm3/s.
Station 2: Located 100-m downstream of Station 1. Dye concentration, Cdye = 10µg/l.
Lindane concentration, CL2= 0.5 µg/l
Station 3: Located 200-m downstream of Station 1. Dye concentration, Cdye = 8 µg/l
Lindane concentration, CL3= 0.9 µg/l

Problem 6.2
A clumsy professor trips and spills 1 g ofchemical 10 m down the hall from your door.
The hallway is 2 m wide and 2 m high. The chemical evaporates in25 seconds. Assume
isotropic turbulent diffusion D = 0.05 m2/s.
Case 1.The safety doors are closed so that there is no air current in the hall. Write an
expression for the concentration at your door, justifying all assumptions. Plot the
exposure concentration at your door versustime, indicating the maximum and final
concentration.
Case 2. The safety doors have been propped open (bad, bad, bad), so that there is a 4 m/s
breeze blowing down the hall from the spill towards your door. Calculate the magnitude
and duration of the maximum concentration at your door. Sketch the concentration, c(t),
clearly indicating the maximum and final concentration.
Case 1. The safety doors are closed so that there is no air current in the hall.
Use time scales to determine if, for the exposure at your office, the release is
instantaneous or continuous. The distance to your door is dx=10-m. With no advection
(zero current), the transport time scale is Tdiffusion= dx2/8D, based on the arrival of the
2ζ contour. Thus, Tdiffusion= (10-m)2/(8 x 0.05m2s-1) = 250 s. This is much longer than
the release time scale (25 s), so the concentration observed at your office will appear as if
from an instantaneous release.
Similarly, use a comparison of time-scales to determine if the concentration at your door
appears as if from a 1-D, 2-D, or 3-D source. If we assume the release occurs at mid-
width, the time-scale to mix across the hallway is, tmix,y= (2m)2/(4 x 0.05 m2s-1) = 20 s.
For the vertical mixing time-scale, note that the spill is at the floor boundary. A release at
the boundary must diffuse across twice the distance, as compared to a release at mid-
height, so the length-scale must be doubled. The time-scale to mix over the hall height is
thus, tmix,z= (2x2m)2/(4x0.05m2s-1)= 80s. Since tdiffusion> > tmix, the concentration at
your door appears as if released as a one-dimensional source, i.e. distributed evenly over
y and z.
Finally, with the closed doors at x = ±20 are no-flux boundaries. To account for these
boundaries we put in an infinite set of images located at x = 40n, where n = ±1,±2, etc.
The real source is located at n=0. The concentration observed at your door then follows,
 The concentration observed at your door (x= 10 m) is plotted above. Note the
concentration at your door at the diffusion time scale, 250 s, is not negligible. This time
scale is based on the arrival of the 2ζ contour, not on a specific value of concentration.
The time-scale based on cloud size (2ζ ) is not necessarily a good indicator of exposure
risk. Also, if the absolute concentration at 2ζ is below the detection limit, you would not
detect it, even though you predict it has arrived. The final concentration is reached at
around 1600 s, which is shorter than tmix,x= LX2/4D = 8000 s, with LX= 40m (defined in
chap. 4). This again emphasizes the conservative nature of the standard mixing-time
estimator, L2/4D.
Case 2. Safety doors are open, u= 4 m/s from the spill towards your door.
The advection time scale is, Tadvection= dx/u = 10-m/4ms-1= 2.5 s << Tdiffusionand <<
Trelease. So, the transport is dominated by advection, and the chemical arrives at your door
before the release ends. So, during most of the release the concentration at your door
appears as if released from a continuous source. In addition, the mixing time scales
derived above still hold, so we can write, Tmix,y and Tmix,z> Tadvection, such that the
chemical cloud is not mixed over y and z at the door. The cloud is thus fully three-
dimensional at your door. For simplicity we assign the origin of the release to be (x = 0, y
= 0, z = 0). Image sources are need to accountfor the no-flux boundaries at y = ± 1-m and
z = 0 and 2m. Consider the fourclosest images, (y,z)=[ (0,0); (-2,0); (2,0); (0,4)]. During
most of the release,the concentration at your door will be,
One can see from the values in the last expression that only the image coincident with the
real source contributes significantly to the solution. The above concentration is observed
at the door only during the duration of the release. More specifically, the concentration is
observed starting approximately at t = 2.5 s (the advection time scale for the front) and
ending at (25 +2.5) = 27.5 seconds, the duration of the release plus the advection time
scale. The maximum exposure concentration at your door is depicted below. Because of
longitudinal diffusion, the front and tail of the passing cloud are not sharp, butwill have a
transition region of length ≈4ζ ,where ζ is estimated at the advection time-scale, 2.5s.
Thus, the transition region at the front and tail of the exposure cloud will 4 √2(0.05m2s-
1
)(2.5s) = 2 m long, which translates into an transient period of (2-m)/(4m/s) = 0.5 s.
Note that the final concentration is zero, everything eventually flushes out of the hall by
the air current.
Problem 6.3
Indeck Corporation has proposed to construct a new 350 MW gas-fired power plant in
Smithfield, Rhode Island. Environmental reports submitted for the plant indicate it
would discharge 150 metric tons per year of carbon monoxide from a 57.9-m high stack.
Your house is located 1-km downwind of the proposed stack, so you are very interested
in the potential carbon monoxide exposure. Assume a steady wind of U = 2 m/s, a
vertical turbulent diffusion coefficient of Dz= 1 m2/s, and a horizontal turbulent diffusion
coefficient of Dx= Dy= 2 m2/s.
a. Write the equation for the carbon monoxideconcentration downwind of the stack.
b. Plot the vertical profile of concentration, C(z), at your house. If your house is 10 m
high, what is the maximum concentration you might experience?

The stack will run continuously, so use a continuous release solution. Let H = 57.9-m be
the stack height. With z = 0 at the ground, the source is at (x=0, y=0, z = H). One image
source is needed at (x = 0 , y = 0, z = -H) to account for the no-flux boundary at the
ground.
Problem 6.4
Pickles are made in large wooden vats filled with brine of 10 g L-1 salt concentration. The
base of each vat is square, 5-m to a side. At the Dill Pickle factory one of the vats begins
leaking brine directly into a shallow aquifer at a rate of 0.1 m3week-1. The leaking is
distributed evenly over the vat's base area. The aquifer is 2-m deep and sits atop an
impermeable clay layer. The density of the brine allows it to quickly disperse over the
depth of the aquifer. The Pickle Company is not concerned about losing this small
amount of brine, but the community using the aquifer for drinking water is concerned.
The community well is located 500-m from the vats in the direction of groundwater flow.
Assess the possible impacton drinking water quality by estimating the maximum salt
concentration at the well. How long after the leak begins will this concentration be
reached? The transport velocity (pore velocity) in the aquifer is 10-6 m/s. The molecular
diffusion of brine is 10-9m2 s-1. The longitudinal dispersion coefficient is 10-5m2s-1. The
lateral dispersion coefficient is 10-6m2 s-1.

Answer 6.4
The release is described as continuous, and we are told that it rapidly distributes over the
aquifer depth. No lateral boundaries are indicated. As a first approximation, we believe
the concentration field will resemble that from a continuous release in two dimensions,
(x,y), with conditions uniform in the vertical coordinate, z. No image sources are needed,
because boundaries only restrict flux in the z-direction, and the concentration is already
assumed to be constant in z, such that the no-flux condition, ∂C/∂z = 0, is automatically
satisfied. However, the source is not a point, but rather is distributed over a 5-m x 5-m
square. We can treat it as a point if thisarea is small compared to the extent of the plume
at the well (see B below), as the concentration observed at the well will then appear as if
it came from a point source. Weestimate the lateral dimension of the plume as B = 4 ζy=
4 √(2KyL/U), where L = 500m is the distance to the well, and the lateral dispersivity is
Ky= 10-6m2s-1. Then, B = 126m. Since 5m << 126m, the concentration at the well
appears as if it originated from a point source. Finally, we write the expression for brine
concentration downstream of the leak.

This salinity is not detectable by the human palate. For reference, salt water is 30 ppt.
The solution above assumes that steady state has been reached at the well. The
timerequired for this to occur depends both on advection and longitudinal dispersion. The
advection time-scale, TU = 500m/10-6m/s = 5x108s ≈15 years, defines the arrival of the
mid-front which will have the concentration Cmax/2. The time interval ∆t between the
arrival of the mid-front and the end ofthe transient period is approximately ∆t = (2ζx)/u
Finally, note that molecular diffusion, while present, is negligible compared to the lateral
and longitudinal dispersion, and so is not considered in the plume transport calculations.

Problem 6.5
Tracer is released at 10mgs-1 at the sidewall of a rectangular irrigation channel. The
concentration is measured mid-channel at three downstream locations, x = 20m, 40m, and
60m. The injection starts at t = 0. Estimate the flow in the channel, the mean cross-
sectional area, and the coefficient of longitudinal dispersion.
Problem 7.1
Velocity profiles were measured at two locations downstream of a jet exiting a nozzle of
diameter D. The locations are: X = 20 D and 40 D.
1) Sketch the profile, vrms(y), at each station. Explain the shape of your plot.
2) What is the length scale of the eddies which dominate mixing within the jet? Sketch
this length scale and the corresponding eddy below.

1) Profiles of vrms are sketched below, with x=20D shown as a solid line and x = 40D
shown as a dashed line. The strongest turbulent fluctuations occur at the points of
maximum mean shear. Because the maximum shear is weaker at the downstream station,
the fluctuations are weaker here as well.
2) First, let B(x) be the width of the jet atposition x. While many eddy scales coexist in a
turbulent jet, it is the biggest scalesthat dominate mixing. The biggest eddies have the
scale of the flow, which for a jet is B. We find the sign of rotation of these eddies by
imaging the motion of a rigid stick placed in the flow. Because the sign of the shear (and
thus vortex rotation) reverses at mid-jet, a single eddy cannot grow larger than the half-
width of the jet.Thus, we can say more specifically that the largest eddies in the jet will
bethe scale of the half-width, B/2.

Problem 7.2
The velocity profile shown below was measured above a sandy bed for which the mean
sand grain diameter was d = 1mm. Use the plot to answer the following questions.
a) Use the logarithmic section of the profile to find the friction velocity, u*
b) Using the answer to a), what is the bed stress, ηW?
c) What is the thickness of the laminar sub-layer?
d) Using a velocity measured within the laminar sub-layer estimate the bed stress.
e) Is this flow rough or smooth turbulent?
f) Imagine that you introduce two sources of dye at elevations, y = 3 mm, and y = 3 cm.
Sketch the dye streak-line and describe the differences in flow at these two points.
a)The logarithmic region is fitted with a red line. From this line we estimate the value of
u*.

d) The blue point (u = 0.2 cm/s, y = 0.2 cm) is within the laminar sub-layer. Because
turbulence is excluded from the laminar sub-layer, the wall stress can be found from the
molecular viscosity and the shear, ηW= µ∂u/∂y = µu/y = (0.001 kgm-1s-1)(0.2cms-1)/(0.2
cm) = 0.001 Pa, consistent with the estimate based on u*
e) yo= 0.011 cm from the graph. For smooth turbulent flow, yowould be ν/9u*= 0.011
cm. Since this agrees with the observed yo, the flow must be smooth turbulent.
Alternatively, for a sand grain the Nikuradse roughness is the sand diameter, ε= d =
0.1cm. Thus, the roughness Reynolds number is εu */ν= (0.1 cm)(0.1 cms -1)/(0.01 cm2s-1)
= 1 < 5, also indicating that the boundary is Smooth Turbulent.
f) Dye released at y = 3 mm will bewithin the laminar sub-layer. Its trace will be a
straight line, as the dye will tracer a single streamline without deviation. Dye released at
y = 3 cm will be above the laminar sub-layer, and so will feel the effects of turbulence.
The dye tracer will be squiggly, as the dye is distorted by the action of many tiny eddies.
Problem 7.3.Nikuradse used uniform sand grains of diameter εto characterize the impact
of surface texture on velocity profiles.Based on his experiments the length scale ε, now
called the equivalent sand grain diameter, has become a standard for describing
roughness of any shape. The ”equivalency‘ isinterpreted as providing the same drag. The
equivalent roughness is estimated by measuring the velocity profile and fitting it to a
logarithmic profile to determine yo. If the flow is Rough Turbulent, then yo= ε/30. This
tells us that the roughness provides the same drag to the flow as uniform sand grains of
diameter 30 yo.
Consider the roughness elements depicted below. Although the shapes have the same
physical scale, d, they will not necessarily offer the same roughness to the flow, that is
they will not necessarily produce the same yo.Consider two cases, d << δs, and d >> δs,
and order the three cases from highest to lowest yo.
If d << δs, every shape will be buried within the laminar sub-layer. The outer flow will
not feel any of the shapes and the flow will be Smooth Turbulent. Under these conditions
all cases have the same yo= ν/9u*.
If d >> δsthe flow is Rough Turbulent and the flow feels each shape. The shape that
disturbs the flow the most, i.e. creates the most separation and flow distortion, will create
the greatest momentum sink and will feel the roughest. This is shape C. Shape A is the
most streamlined, and thus produces the least disturbance and separation. Based on this
we expect yoC> yoB> yoA.

Problem 7.4
Below is a top view of a channel of width b. The cross section of the channel is constant,
so that the velocity, U, is alsoconstant along the channel. Part of the channel is filled with
vegetation whose morphology is uniform over depth, and emerges through the water
surface. The mean stem diameter is d, and the mean spacing between stems is ∆S.
Consider the model, Dt,y~ v’ly, to describe the lateral diffusivity.
a) How will the lateral turbulent diffusivity change as the flow enters the vegetation?
b) Compare the diffusivity in the vegetated zone for Ud/ν=1 versus Ud/ν= 1000?
c) Suppose the flow is unconfined, i.e. no side-walls, but the lateral extent of the
vegetated zone is unchanged, how will the turbulence scales, turbulence intensity, and
diffusivity differ in the vegetated and unvegetated zones?

Answer 7.4
a) In the unvegetated zone the lateral eddy scales are set by the channel width, ly ~ b.
But, in the vegetated zone eddies of scale b are destroyed by the physical obstruction of
the plant stems. In this zone the lateral eddy scale is set by the vegetation spacing and
diameter, such that ly ~ d or ∆S. Because of the confining walls, the mean channel
velocity is approximately the same in both regions. Because the eddy scale is so much
reduced in the vegetated zone (∆S, d << b) we expect the lateral diffusivity to be reduced
in this region, relative to the unvegetated zone.
b) If Ud/ν= 1, the wakes behind each stem will be laminar and will not contribute
additional eddies to the flow. If Ud/ν= 1000, the stems contribute stem-scale eddies to the
flow through separation and the generation of Kelvin-Helmholtz vortices. So, for the later
case we expect that v' will be augmented by the wake-vortices, and that Dt.y will be
greater with the higher Reynolds number.
c) If the flow is unconfined the vegetative drag will lead to a reduction of velocity
within the vegetated zone as flow is redirected around the patch of vegetation. Then, the
velocity in the vegetated zone, U V, will be less than the velocity in the free stream, U. If
UV< U and the stem wakes are not contributing additional turbulence (UVd/ν<≈100) we
would expect that the turbulence intensity in the vegetated zone is also less, v' V< v'. In
addition, the turbulence scales will be smaller in the vegetated zone, as discussed in a).
With both v' and ly smaller in the vegetated zone, then Dt,y will also be smaller in the
vegetated zone. Even if the stem Reynolds' number, U Vd/ν, is sufficiently high that the
stems contribute additional eddies, their scale (ly ~ d) is too small compared to those in
the free stream, ly > b, and still the diffusivity in the vegetated zone will remain small
compared to the unvegetated zone.

Problem 7.5
The velocity profile shown below was measured in a stream above a gravel bed for which
the mean gravel size was 1 cm.
a) Find the friction velocity, u*.
b) What is the thickness of the laminar sub-layer?
c) Does the gravel‘s roughness contribute to the resistance at the bed?
d) What diameter of spherical sand grain would produce the same flow resistance?
Problem 8.4
A small slug of tracer is released at mid-channel, shown below. The channel is very deep,
such that vertical shear and vertical domain limits can be neglected. The channel width is
B, and isotropic diffusivity D. Describe the evolution of this cloud at t1 << B2/4D and t2
>> 0.4 B2/D. For each time period, describe the shape of the cloud and the rate at which
its length increases.

Answer 8.4
Time t1<< B2/4Dy: When the slug is initially released, it is very small compared to the
width of the channel. Released in the center of the channel, the variation in velocity
(shear) across the patch is negligible, and the entire patch advects at the same speed.
Because the patch is not experiencing differential advection, the spreading of the cloud in
the longitudinal direction is due to longitudinal diffusion only. The cloud's longitudinal
length scale is 4(2Dt)1/2 . More specifically, letting the release pointbe (x, y, z) = (0, 0, 0),
theconcentration field evolves as,

Time t2>> 0.4 B2/Dy.By this time the patch has grown to uniformly fill the lateral
dimension of the channel. In addition, sufficient time has passed for the
longitudinaldispersion due to the lateral shear to reach Fickian behavior. The longitudinal
length-scale of the patch is now 4(2Kxt)1/2 , i.e. the patch growth rate is dictated by the
dispersion coefficient, KX. In the vertical direction the cloud continues to grow via
vertical diffusion, such that the concentration field evolves as,

Problem 9.1
A smokestack of height H = 20 m releases two gases, dichlorodifluoromethane (Freon
12) and trichloroethene (TCE), each at a rate of 5 kg/min. Freon 12 is conservative. TCE
undergoes first-order degradation in the atmosphere at a rate of kTCE= 0.1 day-1,
producing the highly toxic chemical phosgene (C(=O)Cl2). Assume that the wind
blowssteadily and uniformly at 5 m/s in the positive x direction. The atmospheric
turbulence is homogeneous but anisotropic, with the vertical diffusivity, DZ= 0.1 m2s-1,
smaller than the horizontal diffusivities, DX= DY= 1 m2s-1 . For both gases the ground
acts as a no-flux boundary. Find the maximum concentration of Freon
and TCE 10-km downwind of the stack.
Answer 9.1
For a continuously operating smokestack and steady climatic conditions, we can assume
a steady concentration field, i.e. ∂C/∂t = 0. The wind is given as u = 5 m/s, implying v =
w = 0. We are told to assume a uniform wind, i.e. no shear, so we neglect shear-
dispersion. For the length-scale of interest, LX= 10,000m, the Peclet number is
(5m/s)(10,000m)/(1m2/s) = 50,000 >>> 1. With this high value of Pe, the longitudinal
diffusion term is negligible relative to longitudinal advection, and we drop it. With the
above assumptions, the transport equation
In any transverse dimension for which the plume is unbounded (here the y-direction), the
maximum concentration is at the centerline of the plume (here y = 0). The vertical
coordinate, however, is bounded by a no-flux boundary at the ground, z = 0. Once the
plume reaches the ground, concentration will build up at the no-flux boundary. Because
the upper edge of the plume is not bounded, the vertical concentration field will
eventually become asymmetric with the maximum concentration at the ground. We
estimate the distance at which this will occur using the time-scale for the edge of the
plume (the 2ζcontour) to reach the ground,
(e) T2ζ= H2/(8DZ) = (20m)2/ (8 x 0.1 m2s-1) = 500 s.
Thus, for x >> u T2ζ= 2500 m, which includes the point ofinterest, we expect the
maximum concentration to be at the ground. Therefore,the maximum concentration at x =
10,000 m will be Cmax= C (x =10000 m , y =0, z = 0). Evaluating (d) for Freon and TCE
we find,

Problem 9.2
A small channel is h = 5 cm deep and b = 10 cm wide. It carries flow at U = 10 cms -1.
The stream-wise coordinate is x. The vertical coordinate is z, with z = 0 at the bed and
positive upward. A continuous source of dye is injected at a rate of m& =1 gs -1 at mid-
depth and mid-width, and at x = 0. Assume that the channel has no dye upstream of the
injection point. The bed of the channel is a perfect absorber for the dye,such that the
concentration of dye in equilibrium with the bed is zero, and thus C(z=0) = 0. The
molecular diffusivity for the dye is D = 10-5cm2s-1 . What is the maximum concentration
in the channel 20 m downstream of the source?
Answer 9.2
A perfectly absorbing boundary can be treated like a dissolving boundary with C eq= 0.
The boundary is a sink rather than a source, otherwise the process of exchange between
the bed and the water column is the same. Here, dye is injected as a continuous point
source and the evolving plume experiences a sink at the absorbing boundary. If the
system has fast mixing, then we can assume that ∂C/∂z = ∂C/∂y = 0 and use the
fastmixing model for bed-exchange. Then the effects of the boundary sink are modeled as
a distributed sink S. For steady-state conditions and Pe >> 1, the transport equation

However, if the dye mixes slowly overthe cross-section, we cannot assume ∂C/∂z =
∂C/∂y = 0 in the channel. Under these conditions we would use the solution for a 3-D,
steady, continuous release (equation 8, chapter 6), with positive image sources to account
for the no-flux side-boundaries, and a negative image source to account for the perfectly
absorbing bed. Before proceeding with (1), we must check all the assumptions. First, we
will determine if the flow is turbulent, and if it is we will estimate turbulent diffusivities.
The hydraulic radius is (5cm x 10cm)/ (10cm + (2 x 5 cm)) = 2.5 cm. The Reynolds
number based on hydraulic radius is, ReH= (10cms-1x2.5cm)/(0.01 cm2s-1) = 2500,
which indicates the flow is likely to be turbulent. Next to each boundary there is a
laminar sub-layer with thickness, δs= 5 v/u*. The friction velocity is estimated as u*
≈0.1U = 1 cms-1. This gives δs= 0.05 cm.
Now, we estimate the coefficients of turbulent diffusion using the empirical relations for
a straight channel given in Table 1 of Chapter 7.
Dt,x= 0.45 u*h = 2.3 cm2s-1
Dt,y= 0.15 u*h = 0.75 cm2s-1
Dt,z= 0.067 u*h = 0.34 cm2s-1
Confirm Fast-Mixing Bed Exchange Model
To determine if the system will follow a fast-mixing or slow-mixing model of bed-
exchange, we compare the time scale required for the channel to mix vertically with the
time scale for diffusive flux to cross the laminar sub-layer.

To be very confident that the fast-mixing model is appropriate, we require that T δs is an

order of magnitude greater than TL. Here the time scales only differ by a factor of three.
However, the system is closer to the fast-mixing model then the slow-mixing model, so
we proceed with that assumption.
Confirm well-mixed conditions (∂C/∂y = ∂C/∂z = 0) for plume evolution.
To use (1) to describe plume evolution, wemust confirm that the plume rapidly mixes
over the channel cross-section. We need to find the distance from the source at which the
plume is uniform in y and z. These distances are,

This indicates that for distances greater than 333 cm from the source, the plume will be
uniform in y and z. We are interested in the position x = 2000 cm, so we can model the
concentration as if it originated from a one-dimensional source at x = 0. That is, we can
assume ∂C/∂y = ∂C/∂z = 0.
Confirm assumption of Pe >>1
If Pe = ULx/Kx>> 1, we can neglect longitudinal dispersion relative to longitudinal
advection. The relevant length-scale is the distance at which we want to predict the
concentration, L = 2000 cm. The longitudinal dispersion is KX= 5.9u*h = 30 cm2s-1 .
Then, Pe = (10 cms-1x 2000 cm)/(30 cm2s-1) = 666>> 1. So, this assumption is confirmed.