ChemText for CE 150/150L

(Fall 2023)

A supplementary text for introductory classes in environmental engineering at the

Department of Civil Engineering, California State University Sacramento
by John Johnston, Ph.D., P.E.

1
 Table of Contents

1. Introduction 3
2. Environmental Engineering Calculations 4
Significant figures, chemical calculations, statistical analysis
3. Particles in Water 27
Physical categories, engineering categories and measurements
4. Equilibrium Processes 41
Chemical equilibrium
5. Adsorption 48
Equilibrium isotherms, applications
6. Gas Solubility and Transfer Kinetics 60
Henry’s Law, gas transfer kinetics
7. Acids, Bases, and pH 71
Dissociation of water and pH, normality, monoprotic acids,
diprotic acids, carbonate buffering (alkalinity, titration
curves), acids, bases, and their salts
8. Mineral Solubility 93
Precipitation and dissolution equilibria
9. Coagulation/Flocculation 98
Coagulation, flocculation and mixing
10. Biochemical Oxygen Demand 106
11. Reactor Analysis 117
Introduction to tracer tests
12. Dissolved Oxygen Modeling (DO Sag Curve) 123
Derivation of the DO sag (Streeter-Phelps) equation.
Applications in simple environments. Using the equation in
more complex environments. Limitations of the equation.

2023
John R. Johnston
All Rights reserved
This book is not to be used outside CSUS classes or similar venues
covered by fair use copyright rules.

2
1. Introduction

Knowledge of basic chemistry is needed to understand and appreciate the field of

environmental engineering and to perform the calculations expected for the
Fundamentals of Engineering exam. If you took Chem 1E at Sac State, you have
at least been introduced to the topics we’ll cover in CE 150. (Hint: You will
probably find your Chem 1E notes to be helpful.) If you took chemistry
elsewhere, you may have missed some topics because they usually appear in
Chem 1B. Not to worry, your instructors are not assuming that you are chemistry
experts, only that you are willing to learn material that is normally taught in Space for
freshman chemistry. In other words, you can do this. your notes.

This monograph was written to supplement the CE 150 textbook in these topics
and to provide background material for the lab exercises used in CE 150L. The
topics covered are listed below.

Chemical calculations Particles and their measurement in water

Equilibrium Coagulation/Flocculation
Adsorption Biochemical oxygen demand (BOD)
Gas solubility Reactor Analysis
Acid base chemistry Dissolved oxygen model (DO Sag)
Mineral solubility

This manual is not designed to stand alone. Throughout the text, references will
be made to the second edition of Mihelcic and Zimmerman (MZ), the authors of
the current textbook for CE 150 A (see the yellow boxes below). You will find
that studying the referenced material in the assigned textbook will be valuable
and will be expected by your instructor. In addition,
you may find it helpful to read about these topics in Look for text boxes like these for
references to similar material in your text
your basic chemistry textbook (assuming that you (MZ, 2nd edition). The notation Sec 2.4*
still have it) or online (e.g., Khan Academy and means 2.4 plus included subsections.
similar websites). Seeing alternate explanations for
various topics can often be helpful.

References

Current text (MZ): Mihelcic, James and Julie Beth Zimmerman, Environmental
Engineering, Fundamentals, Sustainability, Design, 2ed, Wiley, 2014 (ISBN
9781118741498)

Previous text: Mines, R.O. and L.W. Lackey, Introduction to Environmental

Engineering, Prentice Hall, 2009 (ISBN 0-13-234747-4)

3
2. Environmental Engineering Calculations

More than other sub-disciplines in civil engineering, environmental engineering

uses both International (SI) and U.S. Customary (so-called “English”) units. This
is because of the central role that chemistry (which is almost all SI units) plays in
the field. Beyond this, though, engineering is becoming increasingly globalized
and being able to operate in both systems will be useful in your career. Many
conversion factors and tables are available such as the one in your
text. You may have these programmed into your calculator. MZ
Conversion programs are also readily available online. Sec 2.1*, 2.4, 2.5*
Nevertheless, knowing some of the common conversions by heart
will save you time and effort, and help you check numbers in your head. For
instance, if a drawing calls out a 20-mm diameter pipe, and you know that there
are about 25 mm per inch, then you know that this pipe is about ¾ inches. Now
you can make a mental check – is a pipe the diameter of a garden hose about the
right size for this situation or not? Memorizing numbers is a drag, no doubt
about it. The author has found that memorizing the relatively few conversions
that are starred (*) below allows him to derive a large majority of the unit
conversions needed in this class without resorting to tables. He recommends that
practice to you.

Conversion Factors of Particular Usefulness in this Course

Length Mass/Weight/Concentration
2.54 cm/inch * 2.205 lb/kg *
3.28 ft/m * 454 g/lb
5280 ft/mi * 62.4 lb/ft3 for H2O *
Area 2000 lb/ton (U.S.) *
43,560 ft2/ac * 1000 kg/mton (metric ton) *
104 m2/ha (hectare) 1 mg/L = 1 g/m3 *
(= 100 m × 100 m) *
Volume Force/Pressure
3.785 L/gal * 1 N = 1 kg·m/s2 *
7.48 gal/ft3 * 1 Pa = 1 N/m2 *
1000 L/m3 * 6.9 kPa/psi
Energy/Power 14.7 psia/atm *
1 W = 1 N·m/s * 101.3 kPa/atm
0.746 kw/hp; 1.34 hp/kw Gas constant values, R
Miscellaneous 0.08205 atm.L / mol.K
g = 9.81 m/s2 = 32.2 ft/s2 * 8.314 Pa·m3/mol.K
273°K = 0°C

4
SIGNIFICANT FIGURES

In addition to units, understanding and properly expressing valid engineering

answers is important. You do this by using significant figures properly. The key
issue to remember is that you cannot claim credit for more than one “doubtful
digit” and that any operation performed with a doubtful digit, results in a
doubtful digit. In the example below, the doubtful digits are shown ion red, bold
numbers. You can verify the doubtful digits in the answers by doing the
calculation by hand. Watch how the doubtful digits propagate through the
calculation. Any number multiplied by a doubtful digit is itself doubtful and any
number added to a doubtful digit is doubtful.

15.42
× 13.6
9.252
46.26
154.2
-------
209.712

Since we keep only one doubtful digit in the final answer, round off accordingly.
So, the answer is 210 after expressing the correct number of significant figures.
Notice that the number of significant figures in the answer is limited by the
multiplicand with the least significant figures. In other words, the answer has 3
significant figures because 13.6 has three significant figures. This is true for
division as well.

Addition and subtraction are a little trickier. Again, any operation with a
doubtful digit will result in a doubtful digit answer. Try this:

15.243
– 14.67
------
0.573

In this case, the 3 and the 7 in the operands are doubtful, so any number they
interact with are doubtful. Also, the zero that we mentally add to the 14.67 is
also doubtful. So, in this case, the answer is 0.57, which is two significant
figures even though the other numbers had three and four. Watch out for this
situation in your lab calculations. Your calculator and Excel will not round
numbers based on significant figures.

5
A few more tips:

• Zeros in front of a number (leading zeros) are not significant. Scientific notation
Accordingly, 0.00234 has three significant figures. Think of Proper form is 2.34×10-3
Calculator/Excel shorthand is 2.34E-3
the number written in scientific notation like so: 2.34×10-3. Use proper form in your hand calculations.
It’s easier to see the significant figures in this form.

• Trailing zeros, those at the end of a number, are significant if they

express an actual measurement or calculation. For instance, if your
analytical balance tells you one sample weighs 1.2345 g and a second
sample weighs 1.2340 g, the zero is significant because, as shown in the
first sample, the instrument can actually measure ten-thousandths of a
gram. Don’t throw away significant figures. Don’t report 1.234 g if the
instrument can read 1.2340 g. In spreadsheets Excel will round 1.2340
to 1.234. You should manually adjust Excel displays to reflect the
correct number of significant figures in your lab reports.

• Integers have an infinite number of significant figures. Don’t limit your

answer’s significant figures because of integers.

• Successive rounding at each step in a long calculation can introduce

error. A good practice is to keep a few extra digits throughout a
calculation, then round to the allowable number of significant figures at
the end.

• Because all engineering calculations rely on a measurement at some

point, be careful not to report answers that are more precise than the
measurements allow. Having said this, normally in engineering, we
report a minimum of 2 significant figures even if the strict application of
the rules of significant figures don’t technically justify them.

For homework problems, because the sources and accuracy of the numbers aren’t
known, report your answers with a reasonable number (2 to 4) of significant
figures. In lab calculations on the other hand, you know a lot more about the
data, so report your answers in the number of significant figures justified by the
data.

CHEMICAL CALCULATIONS

The following discussion is a quick review of some of the material found in MZ,
Chapter 2. We will solve the example problems during the first lab.
6
Mass and Concentration

Mass concentration is the mass of material dissolved or suspended in a given

volume of water. In environmental engineering, the volume is usually 1.0 liter
(L) and the mass is usually milligrams (mg) so the concentration is mg/L. It
might also be micrograms per liter (µg/L).

Mass = (Concentration) × (Volume) = CV

Mass/time = (Concentration) × (Volume/time) = CQ

where Q = volumetric flow.

For the practice problems, numerical answers
Mass flow is often referred to as “load” in environmental are shown in square brackets. Hints are
engineering. provided in the Appendix. Try doing the
problems first without looking at the hints.

Practice Problems

2-1. How many grams of K3PO4 are in 75 L of water if the concentration is 50

mg/L? [3.75 g in 75 L]

2-2. If you had 5.0 g of potassium phosphate (K3PO4), how much water would
you need to make a 150 mg/L solution? [33.3 L]

2-3. Suppose you had 2 L of a 150 mg/L K3PO4 solution and you want to dilute
it down to 120 mg/L. How much water would you need to add, assuming that the
water had no K3PO4? [Add 0.5 L]

2-4. Suppose you want to make up 2 L of a 30 mg/L solution of NaCl using clean
water and a stock solution containing 500 mg/L NaCl. How much water and
how much stock solution do you need? [120 mL of stock solution and 1880 mL
of water.]

In some environmental situations, concentration is the important factor; in others

it is mass or the mass flow per time. For example, fish (and people) need a
minimum concentration of dissolved oxygen to live. On the other hand, mercury
and lead poisoning depend on the mass accumulated in the body. Some sensitive
algae are harmed at low concentrations of copper. But algae growth in a lake
depends on the mass of nutrients (N and P) that have been deposited in the water.

7
Practice Problems

2-5. Dirty Creek flows into Muddy Lake. If the creek’s water contains 200 mg/L
of total suspended solids (TSS) and the flow is 3 m3/s, what is the daily TSS
mass flow (daily load) to the lake? [51,800 kg/d. Check the units]

2-6. Turbid Creek flows through Scummy Pond. As it does, 50 mg/L of total
suspended solids (TSS) in the creek water settles out in the pond. If the creek
flow is 4000 m3/d, many tons of solids accumulate in the pond over a 50-year
period? Express your answer in both SI and US units. [3650 MT, 4020 US tons]

2-7. Suppose your drinking water contains 8 µg/L of lead (Pb) and you drink 2 L
per day for 30 years. Assuming that all the Pb you drink stays in your body and
that your mass is 70 kg, how many grams per kg of body mass would you
accumulate? [2.5x10-3 g/kg body mass]

Large concentrations may be expressed as mass fractions (or mass percentages).

For dilute aqueous solutions (the ones of concern in this class), 1% = 10,000
mg/L. Mass percentages are also used in expressing concentrations in solid
phase such as the mass of pollutant per mass of soil.
MZ
There’s a periodic
mass A mass A table on the back
mass A (in %) = = × 100 cover.
total mass mass A + mass B

where A is the constituent of interest and B is the “diluting” medium (e.g., body
mass or mass of soil). For example, if a soil is 12% sand, then each 100 pounds
of soil would contain 12 lb of sand and 88 lb of other components (e.g., sand, silt,
and clay).

Practice Problem

2-8. Commercial bleach is 5.5% sodium hypochlorite (NaOCl).

a. Express this concentration in mg/L. [55,000 mg/L]
b. What is the weight of a liter of bleach at room temperature, assuming the
chlorine molecules occupy a negligible volume, and the water density is 998 g/L?
[1.053 kg]

8
Moles and Molar Concentrations

In chemistry, molar concentrations are often used instead of mass concentrations.

A mole (symbol: mol) contains an Avogadro’s Number of molecules (6.02×1023).
The number of grams per mole is equal to the molecular weight (MW) in grams.
In chemical calculations, it is useful to assign pseudo-units of “g/mol” to MW.
Carrying this unit in your calculations will help you avoid errors.

A solution that contains 1.0 mol/L is called a “one molar solution” and is written
as 1.0 M where M is the molarity.

mol (mass / MW )
M = molarity = =
L L

Practice Problems

2-9. Calculate the molecular weight of potassium phosphate (K3PO4). [212.3

atomic mass units or g/mol]

2-10. What mass of potassium phosphate (K3PO4) must you add to 0.25 m3 of
water to make a 0.1 M solution? [5.31 kg]

2-11. Suppose the molar concentration of potassium phosphate (K3PO4) is 0.56

mM (millimolar).
a. What is the mass concentration of K3PO4 (in mg/L)? [118.9 mg/L K3PO4]
b. What is the mass concentration of K alone? [65.7 mg/L K. We would say that
the solution contained “65.7 mg/L as K”.]

2-12. The nitrate (NO3-) standard in drinking water is 10 mg/L as N. What is the
nitrate concentration? [44.3 mg/L]

Mole Fraction

In some chemical calculations, concentrations are expressed as mole fraction (χ)

where:

( mol of particular chemical specie )

χ=
∑( mol of all species in solution)

9
“All species” includes water, and for most dilute solutions (i.e., concentrations in
the mg/L range) the moles of water dominate the summation.

( mol of particular chemical specie )

χ≅
∑( mol of water )

This will be true for natural fresh water (rivers and lakes) and even wastewater
(sewage). For high-strength industrial waters or seawater, you ought to check
this assumption by doing an exact calculation.

Practice problems

2-13. How many moles of water are in a liter of pure water at 10°C? [55.56
mol/L]

2-14. What’s the mole fraction of oxygen in a solution containing 10 mg/L

oxygen (as O2) and 50 mg/L of NaCl? (Don’t forget the water.) [5.6x10-6, no
units because it is a fraction]

Properties of Water
Reference: Crowe, C.T., D.F. Elger, and J.A. Roberson, Engineering Fluid Mechanics, 8th ed., Wiley,
2005, Appendix A. A similar table is in the inside front cover of your text.

Specific Dynamic Kinematic

Temp Density (ρ)
weight (γ) viscosity (μ) viscosity (ν)
C kg/m3 N/m3 N-s/m2 m2/s
5 1000 9810 1.510 x10-3 1.510 x10-6
10 1000 9810 1.310 x10-3 1.310 x10-6
15 999 9800 1.140 x10-3 1.140 x10-6
20 998 9790 1.000 x10-3 1.000 x10-6
25 997 9781 8.910 x10-4 8.940 x10-7
30 996 9771 7.970 x10-4 8.000 x10-7
F slug/ft3 lb/ft3 lb-s/ft2 ft2/s
40 1.94 62.43 3.230 x10-5 1.660 x10-5
50 1.94 62.40 2.730 x10-5 1.410 x10-5
60 1.94 62.37 2.360 x10-5 1.220 x10-5
70 1.94 62.30 2.050 x10-5 1.060 x10-5
80 1.93 62.22 1.800 x10-5 9.300 x10-6

10
Balancing Chemical Reactions

Conservation of mass requires that the number of elemental atoms on

one side of a reaction must equal the number on the other because MZ
elements are neither created nor destroyed in chemical reactions. Section 3.3

Also, the net charge on one side of a reaction must equal the net
charge on the other.

Practice problem

2-15. Balance the following reaction by providing the stoichiometric coefficients

in the spaces indicated. The coefficients should be whole numbers. [a=3, b=1,
c=1, d=3]
a KOH + b H3PO4 → c K3PO4 + d H2O

Stoichiometric Calculations

Reactions show the molar ratios between reactants and products. These can be
converted into mass ratios, which can then be used to calculate the mass (or
weight) of one reactant needed to react with another, or to calculate the mass of
product produced from a given mass of reactant.

Practice problem

2-16. For the reaction in the previous problem …

a. How much phosphoric acid (H3PO4) is needed to react completely with 250 g
of KOH in the reaction shown above? [145 g H3PO4]
b. How much potassium phosphate is produced? [315 g K3PO4 produced].

11
DEALNG WITH DATA

Environmental engineering data sets are variable. Rarely are two measurements
the same. These variations are partially due to random measurement errors in the
instruments and measurement protocols, and partially due to environmental
factors. These latter variations are more pronounced in environmental
engineering because we are measuring natural systems which are complicated
and dynamic (e.g., a river as opposed to a manufactured steel bar). When we
look at data, we want to consider (1) what is the best estimate of the true value of
the thing we are measuring, and (2) how confident are we that the estimate
represents the true value?

Accuracy

Sometimes we do know the true value of a parameter. For instance, we can buy a
reference or standard mass whose accuracy is well documented. If our balance is
accurate, when we weigh the standard, the balance readout should match what we
know the standard value to be. Likewise, we can make a standard solution using
chemicals whose weight and composition are known or can be very accurately
measured.

The accuracy of an instrument or a procedure is judged by how close the

measurement or experimental result is to a standard. We judge the “closeness”
by calculating the percent difference. The mathematical expression used is the
percent difference (PD).

X − Xref
PD = � � × 100
X ref

In this equation, X is a measured number that we want to compare against some

reference, Xref. The reference could be a theoretical value, a standard, or just
another number of interest (e.g., the concentration before and after a treatment
system). In this class, we usually want to know if the measured value is higher or
lower than the reference. Therefore, in this class, do not take the absolute value
of the difference.

Precision

The precision of an instrument or method is the reproducibility of the result.

Suppose you analyze duplicate samples multiple times. In theory you should get
the same answer each time, but of course, you won’t. Small variations in the
duplicate data set indicate a very precise method. Large variations indicate an
imprecise method.

12
The way we measure the variability by the sample standard deviation (s).

1/ 2
 ∑ ( xi − x) 2 
s= 
 n −1 
 

The standard deviation is a measure of the variation of the data about the mean.
It indicates the width or spread of the data set and can be used to determine the
fraction of values contained in given intervals. For instance, if the data follow a
normal distribution, 68.27% of the data points will be within ±1s from the mean
and 95.45% will be within ± 2s. (Note that not all data sets follow a normal
distribution.)

Be careful to use the sample standard deviation and not the population standard
deviation you learned about in statistics class. The difference is that the
population standard deviation (σ) has an n in the denominator instead of n-1.

The standard deviation by itself may not be the best indicator. Imagine you want
to test the precisions of two scales. On Scale 1 you weigh identical 10-lb
concrete blocks many times and find that standard deviation in the recorded
measurements is 2.5 lb. On Scale 2 you do the same thing with 100-lb blocks
and find the same 2.5-lb standard deviation. Are the two scales equally precise?
A spread of 2.5 out of 10 is much greater than 2.5 out of 100. So even though the
standard deviations are the same, you would rate Scale 2 as more precise. We
can account for this by using the Coefficient of Variation (CV).

𝑠𝑠
𝐶𝐶𝐶𝐶 = × 100 (𝑡𝑡𝑡𝑡 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑎𝑎𝑎𝑎 𝑎𝑎 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝)
𝑥𝑥̅

The CV (sometimes called COV) normalizes the variation by the mean. A data
set with a low CV is more tightly clustered than one with a high CV, so a
measurement procedure with a low CV is more precise than one with the high
CV. In the previous example, Scale 2 had a lower CV (2.5% vs. 25%), even
though the standard deviations were equal.

In figure below, five data sets are plotted vertically. The data points are circles;
the mean values are red lines. Look at each, and based on the descriptions above,
judge their accuracy and precision. The true value is 5.

13
 Figure 2-1. Accuracy and precision exercise (mean = 5).

Data set 1 is precise (the data are tightly clustered) but not accurate (the mean is
not near 5). Data set 2 is also precise but not accurate. Data set 3 is neither
precise nor accurate. Data set 4 is accurate but not precise. Data set 5 is both
precise and accurate.

Measurement of Central Tendency

In considering a data set containing multiple values, we are most often interested
in a single number that describes what is going on, a measure of the central
tendency of the data. In this class, we’ll most often calculate the arithmetic
mean.

∑ 𝑥𝑥
𝑥𝑥̅ =
𝑛𝑛

The calculated mean isn’t the whole story, however. We have limited resources
and can analyze only a finite number of samples. We could take five
measurements, say, and average them. Often, we assume that the calculated
mean (𝑥𝑥̅ ) is the same as the true mean (µ). This, however, may not be (usually
isn’t) true. A calculated mean is the true mean only if the calculation includes
every member of a population. If we take only 5 samples out of an infinite
population, for example, we cannot say that the calculated mean is the same as
the true mean. Even though we don’t know the value of the true mean, we can
use statistics to estimate the interval (the so-called confidence interval) in which
the true mean is likely to reside.

14
Based on the t-distribution, the confidence interval for the true mean (µ) of the
population can be calculated from:

𝑡𝑡𝑡𝑡 𝑡𝑡𝑡𝑡
�𝑥𝑥̅ − � < 𝜇𝜇 < �𝑥𝑥̅ + �
√𝑛𝑛 √𝑛𝑛

where 𝑥𝑥̅ is the calculated mean of the data set, s is the sample standard deviation
(calculated), n is the number of samples, and t is chosen from the t-distribution
table at the end of this chapter. The table requires you choose the desired level of
confidence (90%, 95%, 99%) and the number of degrees of freedom needed for
the t-distribution. The number of degrees of freedom (df) for a single variable
data set is n-1. For a data set with 6 values, the t-statistic for a 99% confidence
interval is 4.032. (Go look.)

A 95% confidence interval (C.I.) means that there is a 95% probability that the
true population mean (µ) lies inside this interval. (Note that this is NOT the
same as assuming a normal distribution and saying that 95% of the data values lie
within ±1.96s of the mean. Here we are discussing the C.I. of the mean, not the
of the data set.)

The C.I. is affected by both the variation (indicated by s) and the number of
samples. In the equation above, as sample size (n) increases, the interval narrows
because we are dividing by the square root of n. The more samples we analyze,
the larger the fraction of the whole population we are involving in our
calculations, and the closer our calculated mean is to the true mean. If we
sampled infinite times (n=∞), the calculated mean would be exactly equal to the
population mean. The term ts/n0.5 goes to zero when n goes to infinity. Also,
note that as n increases, the number of degrees of freedom increase and the t-
statistic drops.

The effect of increasing the number of data points is shown in Figure 2-2. The
three red lines are the upper limit (UL) of the C.I., the mean, and lower limit
(LL). Even though the mean is the same for the three data sets, the C.I.’s get
more and more narrow as the number of data points increases.

15
Figure 2-2. Effect of increasing n from 6 to 10 to 19 points. The CI shrinks as n
increases.

Here are the numbers used to calculate the UL and LL. See if you can duplicate
the C.I. results shown in the graph (mean = 3.0).

Data set Standard deviation n

1 1.10 6
2 1.01 10
3 0.88 19

The other effect you should be aware of is that higher levels of confidence (say
from 90 to 95 to 99%), result in wider confidence intervals. The more certain
you want to be that the true mean is within your C.I., the wider your interval
must be to assure that you are actually capturing the true mean. If you want to be
100% confident that you have captured the true mean, your confidence interval
would be have to be infinitely wide.

This effect is illustrated in the following figure. In this example, all the data
points are the same, meaning the mean and standard deviation are also the same
(s = 1.01, n=10). What is different is that the C.I. for data set 1 is 90%; the C.I.
for data set 2 is 95%; and the C.I. for data set 3 is 99%. Can you verify the
values shown?

16
Figure 2-3. Increasing the degree of confidence from 90% to 95% to 99% (data
sets 1, 2, and 3 respectively). Notice how the CI increases.

The most commonly used probability is 95%, but it isn’t the only one you can
choose. The criterion depends on the problem you are addressing. For example,
if you are testing whether a chemical is a toxin, you might want to be very sure of
your answer and raise the criterion to 99%. If you don’t need to be quite so sure,
you could go to 90% or some lower value.

Comparing of the Means of Two Data Sets

In the previous section, we discussed how mean values have a certain amount of
uncertainty associated with them because of the variability of their underlying
data sets. In this section, we want to incorporate this understanding into
comparing means.

Imagine a company marketing a new weight-loss product, a lotion that works

when you rub it on your belly. The company’s advertisements claim that it has
scientific proof that the product works. This proof comes from a study
comparing the weight loss of two groups -- one using the new lotion and one
using a simple moisturizer cream (the control). The mean weight loss was 8
pounds in the experimental group getting the lotion and 4 pounds in the control
group. The data ranges in the two groups were 4 to 12 lb. in the experimental
group and 0 to 8 lb. in the control group. Do you think the reported difference in
means is proof that the lotion works? Or is it possible that the difference in
means is just a random result, given the high variability in the data?
17
In environmental engineering, a similar application might be a new water
treatment process that promises to remove a contaminant much more cheaply
than standard methods. We can test its efficacy by testing influent and effluent
concentrations and comparing the means. In this case, we hope that the means
are different indicating that the system works. Another application would be
comparing the concentrations of a contaminant in a river at two locations --
upstream and downstream of a wastewater discharge pipe. In this case, we hope
that the means are not different because if they are, then state regulators will be
on our case for polluting the river. In both cases, we need a method to decide
whether an observed difference in means is due to chance or due to a real
difference between the data sets.

Statisticians approach the problem this way. Imagine that you had two identical
populations of data with random errors. If they are identical, then their means and
variances are the same (i.e., (µ1 = µ2 and σ1 = σ2.). Now choose a random subset
of data from each population. Calculate the means of the two subsets and
subtract them. If you repeat this process many times, the calculated differences
in means will be normally distributed (see Figure 2-4). The horizontal axis in
Figure 2-4 is z. You might remember that for a single data set, z is the mean
divided by the standard deviation. In this case the difference in means is divided
by a weighted average standard deviation based on the variances of the two data
sets. The center point is zero because µ1 - µ2 = 0. The variation around zero
arises from the fact that you are choosing random subsets of data to compare.
Because these subsets may not be exactly representative of the whole data set,
which introduces a random error.

Let’s understand what this curve is saying. For any observed difference in means
from two data sets, the curve tells you the probability that the two data sets are
subsets of identical populations. If |z| is small, the probability is high that the
populations are identical. If |z| is large, then the chance that the two data subsets
drawn from data sets are drawn from identical populations is small. For a
Normal distribution, 95% of the area under the curve lies between ± 1.96
standard deviations. Therefore, there is a 95% probability that two data subsets
drawn from identical populations will lie in the ± 1.96z region. Only 5% of the
subset comparisons (2.5% on each side) will lie outside these bounds.

18
 Rejection region Rejection region

Figure 2-4. Normal distribution showing rejection regions for α=0.05.

Here’s the basic logic of the comparison test. If the probability is very low that
your experimental difference in means would occur if the data sets were the
same, then we would conclude that the observed difference is due to a real
difference in the experimental data sets. The terminology is that the difference is
statistically significant. On the other hand, if your observed difference has a
good chance of occurring if the data sets were identical, then we would say that
the difference is not significant. In other words, you cannot claim that the
difference in means represents a true difference between the underlying data sets.

If you have seen hypothesis testing in a previous statistics course, that’s what we
are doing here. The null hypothesis is that the two data sets are equal (µ1 - µ2 =
0). We reject the null hypothesis if the experimental results are unacceptably
improbable (i.e., the probability is too low). Visually, we define a rejection
region on the fringes of the probability curve (see Figure 2-4). If your
experimental difference in means lands in that region, the difference is
considered statistically significant.

The size of the rejection region represents the probability that you are mistakenly
rejecting the null hypothesis when it is true (i.e., deciding that the difference is
real when it isn’t). This is called a Type I error and the probability is denoted by
α. An α of 0.05 means that we have a 5% chance of making this error. In other
words, there is only a 5% chance that two identical populations would produce a
difference in means this high or low. The area in the rejection region is 5% of
19
the total area under the curve (remember that the whole area under the curve
represents a cumulative probability of 1.0). In a two-tailed test, α is split between
the two tails (2.5% on each side if α is 0.05). In a one-tailed test, α is assigned to
only one side or the other.

Who determines α? You do, just like you did with confidence intervals. If you
want to minimize the chances that you are mistakenly deciding that the difference
is significant when it in fact it isn’t, choose a small α. A commonly used value
for α is 0.05 but you can choose values that are larger or smaller. The smaller the
α you choose, the smaller the rejection region, and the harder it is to demonstrate
a statistical difference.

Here are the mechanics. Because we are usually dealing with relatively small
samples sizes (small n’s), we will use the t-distribution rather than the Normal
distribution and calculate a t-statistic rather than a z-value.

1. Calculate an experimental t-statistic from your data using the following

formula. 𝑥𝑥̅ is the mean; n is the number of points in each data set; and s
is the sample standard deviation of each data set. The denominator is
called the Standard Error (SE).

(𝑥𝑥̅1 − 𝑥𝑥̅2 ) (𝑥𝑥̅1 − 𝑥𝑥̅2 )

𝑡𝑡 = 0.5 =
𝑠𝑠 2 𝑠𝑠22 𝑆𝑆𝑆𝑆
�𝑛𝑛1 + 𝑛𝑛 �
1 2

2. Calculate the degrees of freedom for your t-distribution from the

following equation. The variables are as defined above. The df formula
will likely give you a fractional answer. Round down before going to the
next step.

2
𝑠𝑠 2 𝑠𝑠 2
�𝑛𝑛1 + 𝑛𝑛2 �
1 2
𝑑𝑑𝑑𝑑 = 2 2
𝑠𝑠 2 𝑠𝑠 2
� 1� 𝑛𝑛1 � � 2 �𝑛𝑛2 �
⎛ ⎞+⎛ ⎞
𝑛𝑛1 − 1 𝑛𝑛2 − 1
⎝ ⎠ ⎝ ⎠

3. Go into the tabulated t-statistics and choose the one associated with your
rounded df and α value. Compare your calculated t-value with the
chosen t-statistic. If the absolute value of your calculated t-value is
larger than the t-statistic from the table, you can conclude that difference
between means is statistically significant with α degree of certainty. If

20
 not, you cannot claim that the observed difference indicates a true
difference between the data sets.

Example problem (with solution)

Operators at a treatment plant that serves a factory measure the TSS of the plant
influent (INF) at several times on days 1 and 2. On day 2 they also measure the
effluent (EFF) values. The data are shown below (all in mg/L). Do two analyses
of the difference in means – one to decide whether the influent changed from day
to day and a second to decide whether the treatment plant was being effective on
day 2 (i.e., whether the effluent is different from the influent). Use α=0.05.

Calculations were performed using the equations described above. For the first
test, the two data sets were the INF values from day 1 and day 2. For the second
test, the data sets were the day 2 INF and day 2 EFF values. The results are
shown in a table on the next page. See if you can duplicate these results. For
both comparisons, the df turned out to be 4 and the t-statistic from the table for
α=0.05 is 2.776. You don’t need to generate this graph for your analysis but for
illustration, a plot of the t-distribution for df=4 is shown. The rejection region is
the part of the graph below -2.776 and above +2.776.

The t-values from the data were 0.33 for the INF vs. INF comparison (large
round dot on the plot) and 5.61 for the INF vs. EFF (triangle marker). The
observed difference in means for the two INF data sets is not statistically
significant because the t-statistic lies in the acceptance region (i.e., 0.33 < 2.776).
In contrast, the t-statistic from comparing the INF2 and EFF2 data sets lies in the
rejection region (i.e., 5.61 > 2.776). Therefore, we would conclude that the
difference between influent and effluent is statistically significant.

Statistics are great but you should always check them against physical reality. In
this case, the statistical results for INF2 vs. EFF2 make sense because the
treatment plant is designed to improve the quality of the water. If the influent
and effluent means were not statistically different, we’d suspect that the plant
was not working. For the INF1 vs. INF2 comparison, if the conditions in the
factory haven’t changed, it is not surprising that influent values are not
statistically different from day to day.

21
Example Problem -- Comparison of Means
Data
INF - day 1 INF - day 2 EFF - day 2
130 120 14
165 125 23
140 130 25
150 150 24
120 160 19
160 15
n 6 5 6
Mean 144.2 137.0 20.0
Std Dev 17.4 46.4 4.7

INF1 vs. INF2 INF2 vs. EFF2

Observed difference 7.2 117.0
s12 / n1 50.69 431.15
s2 / n2
2
431.15 3.73
Standard Error 21.95 20.85
|t-value| 0.33 5.61

df calc 4.94 4.07

df rounded down 4 4
t-statistic from table 2.776 2.776

Significant? No Yes

Rejection region

22
A practical application of this analysis arose in early Caltrans studies of the
effectiveness of stormwater treatment devices. Caltrans researchers set up
various treatment devices (filters, screens, etc.) on the side of freeways and
measured the influent and effluent TSS concentrations for different storms.
Stormwater data sets collected in the field have a lot of variation (high s) and low
numbers of samples (small n). The observed differences in means were not
statistically significant at the α=0.05 level. The influent and effluent TSS means
were different, just not different enough to pass the statistics test. Yet,
researchers could open the devices and see that see that pollutant solids had been
trapped. After some discussion, it was decided that α=0.05 was too stringent a
test given the high variability of the data. When an α=0.10 was applied, most
devices passed the test. Using α=0.10 meant that there was a 10% chance of
mistakenly deciding that a stormwater device worked when it didn’t. Caltrans
managers were comfortable with this level of error in this setting. However, that
would probably not be the case if this were a crucial traffic safety device, or
structural element in a bridge. In those cases, smaller α values would be favored.

23
Practice problems – Use your calculator for these.

2-17. Here is a set of total suspended solids data (in mg/L) collected from a river
over a 24-hour period: 23, 27, 16, 24, 22, 20, 21, 18
Learn how to calculate means
a. Calculate the mean. [21.4 mg/L] and standard deviations on
your calculator. You won’t
have a computer in exams.
b. Calculate the sample standard deviation. [3.5 mg/L]

c. Calculate the 95% confidence interval. [18-24 mg/L]

d. If the maximum legal allowance for the mean is 25 mg/L, are you 95% sure
that the river water is meeting the standard? Are you 99% sure?
[We are 95% sure but not 99% sure because 25 mg/L is outside the 95% CI but
not the 99% CI. See hints for CI values.]

2-18. Here is another set of data from samples collected downstream from the
sampling location in the previous problem: 20, 25, 19, 15, 16, 14. Is this data set
statistically different from the previous one (i.e., did the river characteristics
change) at the 95% confidence level? [No. The calculated t-statistic is less than
the tabulated one. See hints for values.]

24
 Values of the t-Statistic for Confidence Intervals and Deciding
Significance

For single variable data sets, the degrees of freedom, df = n-1 where n =
number of data points

t-statistic for two-sided distribution

df 90% 95% 99%
(α=0.10) (α=0.05) (α=0.01)
1 6.314 12.706 63.657
2 2.920 4.303 9.925
3 2.353 3.182 5.841
4 2.132 2.776 4.604
5 2.015 2.571 4.032
6 1.943 2.447 3.707
7 1.895 2.365 3.499
8 1.860 2.306 3.355
9 1.833 2.262 3.250
10 1.812 2.228 3.169
11 1.796 2.201 3.106
12 1.782 2.179 3.055
13 1.771 2.160 3.012
14 1.761 2.145 2.977
15 1.753 2.131 2.947
16 1.746 2.120 2.921
17 1.740 2.110 2.898
18 1.734 2.101 2.878
19 1.729 2.093 2.861
20 1.725 2.086 2.845
21 1.721 2.080 2.831
22 1.717 2.074 2.819
23 1.714 2.069 2.807
24 1.711 2.064 2.797
25 1.708 2.060 2.787
30 1.697 2.042 2.756
∞ 1.645 1.960 2.576

Reference: Walpole, R.E., R.H. Myers, and S.L. Myers, Probability and Statistics for
Engineers and Scientists, 6th ed., Prentice Hall, 1998

25
 Practice Problem Hints

Chap 2 Calculations
1 Hint: m = CV
2 m m
Hint: C = ; V =
V C
3 Hint: Mass is the same in both solutions. C150 V = m =
C120 (V + ∆V)
4 Hint: The mass in the desired solution comes from the stock
solution. (30)(2000) = (500)VS where VS is the volume of stock
solution. Of course, the final solution must equal 2000 mL.
5 Hint: If we call the mass flow �m� then �m� = QC
t t
6 Hint: m = �m� t = CQt
t
7 Hint: m = �m� t = CQt then divide by 70 kg
t
8 Hint: If the chlorine volume is negligible, the mass per liter add.-
9 --
10 Hint: Molar concentration = mol = mass
V
11 Hint (a): Follow the units. Mg = �5.6 × 10−3 mol� �MW g �
L L mol
Hint (b): There are 3 K’s per K3PO4 so the molarity is three times
larger, but the molecular weight (g/mol) is smaller.
12 Hint: 10 mg/L as N means the N part of the total weighs 10 mg/L.
gN
14
molNO 3
The fraction of N in NO− 3 is gNO3
62
molNO3
13 --
14 --
15 --
16 Hint (a): First figure out the mass ratio of H3PO4 to KOH from the
reaction.
g
1 mol H3PO4 98 H3PO4 gH3PO4
� �� mol
g �=
3 mol KOH 56 KOH g KOH
mol
Hint (b): Figure out the mass ratio of K3PO4 to KOH from the
reaction.
17 Hint: Be sure to calculate the sample standard deviation (s, n-1 in
denominator) and not the population standard deviation (σ.)
We are 95% sure that the mean is between 18 and 24 mg/L, so it
does meet the standard. However, the 99% CI is 19-26, so we are
not 99% sure that the mean is less than 25 mg/L; it could be 26.
18 Hint: The observed difference in the means is 3.2 mg/L. The
calculated df is 9.8 which is rounded to 9. The calculated t-value is
1.55 and the tabulated t-statistic is 2.26, so the two data sets are not
statistically distinguishable.

26
3. Particles in Water and their Measurement

One way of looking at water quality is to consider all suspensions and solutions
as mixtures of water molecules and other particles. Particles come in all sizes,
from molecular diameters to cobbles. A summary of the different particles and
their respective sizes is shown in Figure 3-1.

Figure 3-1. Particles in water and their sizes

In looking at Figure 3-1, you should note first the scale (the black line in the
middle). It is a logarithmic scale in units of mm. On the right is 100, which is 1
mm. In the middle is 10-3 mm which is 10-6 m, also called a micrometer (μm) or
a micron. For reference, the width of a human hair is about 100 µm (0.1 mm)
and a human skin cell is about 30 µm across. Microbes are smaller. A cholera
bacterium is 4-5 µm long and an Influenza A virus is about 0.1 µm in diameter.

PHYSICAL CATEGORIES

If you talk to a physicist or a chemist, he/she will define three different types of
particles based on their size and behavior – suspended, colloidal, and dissolved.

27
Suspended Solids

Suspended solids are physically distinct from the surrounding liquid. In other
words, there is a phase boundary, as shown in Figure 3-2. Strictly speaking, all
particles that show this phase boundary would be suspended particles and their
sizes would range down to 10-6 mm (bottom line in Figure 3-1). In practice,
however, there is a distinction made between particles large enough to settle and
those that aren’t. The boundary between the two is nominally about 10-2 to 10-3
mm, or 1 to 10 μm.

Phase boundary

Figure 3-2. Particulate solids have a distinctive phase boundary

Suspended particles can be removed by physical means, that is, processes that
don’t require a chemical reaction. They can be settled, screened, or filtered.
Sands and silts are suspended particles, as are algae, protozoa, and the larger
bacteria (see Figure 3-1).

Colloidal Solids

Colloids also exhibit a phase boundary, i.e. they are physically distinct from the
liquid. They are so small however, that thermal currents in water can keep them
from settling. (See the “non- sink range in Figure 3-1.) Also, colloids are small
enough that their surface charge can significantly affect the way they act.
Because they are physically distinct from water, colloids can be removed by
physical means – screening, filtering, and centrifugation (settling under very high
gravity). Clays, small bacteria, and viruses are colloids (see Figure 3-1).

Dissolved Solids

Dissolved solids include common molecules in water such as ions from dissolved
salts (e.g., Na+, Cl-, Mg2+, Ca2+, K+, CO32-, SO42-, NO3-, NH4+), dissolved gases
(O2, N2, CO2, H2S, NH3), and organic molecules of many kinds. What
distinguishes dissolved particles is that they are so small that they cannot be
physically distinguished from the liquid. They are intimately intermixed with the
water molecules and have no phase boundary. Here we are in the size range of
molecules, smaller than 10-3 μm or 0.000001 mm (see Figure 3-1). Dissolved
28
substances are so small that we usually must change them into another phase in
order to remove them from the water.

There are several kinds of phase changes used to remove truly dissolved
substances:
a. Changing into a solid by a chemical reaction (precipitation) or by
evaporating off all the water (distillation).
b. Attaching to a solid surface (adsorption).
c. Changing into a gas (volatilization).
d. Dissolving into another liquid (extraction).

Extraction is a way of removing hydrophobic substances from water.

Hydrophobic (“water-fearing”) substances such as fats and oils are not very
soluble in water and are more soluble in organic solvents. If you add the organic
solvent to the water sample and shake or mix the two, hydrophobic substances
will preferentially move to the organic solvent. Removing the solvent removes
the hydrophobic substance. This process is used in laboratory and industrial
settings.

Mixture of hydrophilic Add organic solvent Let the solvent and water
(hearts) and hydrophobic and shake. separate. The hydrophilic
molecules (triangles) molecules stay in the water.

Figure 3-3. Extraction with an organic solvent

Hydrophilic (“water-loving”) substances are very soluble in water and cannot be

extracted in this manner.

29
ENGINEERING CATEGORIES AND MEASUREMENTS

In engineering, there are several ways of measuring particles in water. These

measurements are used to judge water quality on natural bodies of water and to
monitor the effectiveness of treatment systems.

Settleable Solids

Settleable solids include the upper range of the suspended solids in Figure 3-1.
These are particles that settle quickly. They are measured in an Imhoff cone (see
Figure 3-4).

One liter of a water sample is placed in the cone and allowed to settle MZ
Sec 2.5.4
for 30 minutes. The volume of material that settles is read from
graduations on the side of the cone. Thus, the units of settleable solids
are mL/L (mL of solids per L of original sample).

1L

Volume
gradations

Settleable solids

Figure 3-4. Imhoff cone (Johnston)

The other engineering categories are summarized in Figure 3-5. Basically, if the
particle is captured on a filter, it is classified as suspended; if it goes through it is
“dissolved” (including colloids because they act as if they are dissolved; they
don’t settle). If the captured solids can be combusted to CO2 they are organic
(also called “volatile” because they are turned into gas in the high temperature
oven). Although not shown in the figure, colloids and molecules can be volatile
30
or nonvolatile also. You may want to re-visit this figure after reading through the
next several pages.

Burns away
Volatile (volatilizes)
(organic) at 550°C
Total Suspended
Solids (TSS)
Fixed
(non-
Total volatile)
Solids

Glass fiber
Total Dissolved filter
Solids (TDS) Colloids and
Molecules

Figure 3-5. Summary of engineering solids categories (after Parkhurst)

Total Solids

Total solids are measured by evaporating all of the water away (see Figure 3-6).
This can be done using a steam bath followed by baking or by direct flame as
shown in the figure.

Evaporating water

Liquid Sample

Flame (or steam bath)

Figure 3-6. Total solids test procedure

The pan is weighed empty and then again after the evaporation is complete. The
difference is the mass of solids in the water sample. By dividing by the original
31
volume of sample, a measure of the mass per unit volume is generated. Thus,
typical units are mg/L. Note that total solids (TS) include everything except
dissolved gases, i.e., suspended solids, colloids, and dissolved molecules.

Total Suspended Solids (TSS)

TSS is a very common measure of solids in water. In this test, a water sample is
poured through a glass fiber or polycarbonate filter. Photomicrographs of the
two kinds of filters are shown in Figure 3-7. In this lab, we will use glass fiber
filters.

Figure 3-7. Photomicrographs of filters

(Tchobanoglous and Schroeder, Water Quality, 1986)

The procedure for measuring TSS is to weigh a clean filter, pass a known volume
of water through it, dry it, and reweigh it (see Figure 3-8). The TSS are particles
caught on the filter. Their mass can be determined by the difference in the
masses of the filter before and after the sample has been processed through it.
TSS are reported in mg/L.

Commonly, the solids that pass thorough this filter are called “total dissolved
solids” (TDS). Note that this isn’t strictly true. Because the nominal pore size is
about 1 µm, colloids can pass through these filters (see Figure 3-1). We know
this is true because particles in the filtrate (water passing through the filter)

32
reflect and scatter light, which is measured as turbidity (see later discussion).
Truly dissolved particles don’t cause turbidity. Sometimes these small particles
are called “soluble” solids because, even though they may not all be dissolved,
they generally act as if they are (i.e., they don’t settle out).

Sample

TSS caught
on filter

Filter

“TDS” passes through

Figure 3-8. TSS test procedure (Johnston)

Commonly, the solids that pass thorough this filter are called “total dissolved
solids” (TDS). Although this isn’t strictly accurate, it is a common term.

Photomicrographs of suspended solids collected on filters and photographed with

an electron microscope are shown in Figure 3-9.

Total Dissolved Solids (TDS) and Electroconductivity (EC)

Total dissolved solids (TDS) are measured by evaporating the water from a
known volume of filtrate (i.e., the water passing through the filter). They are
reported in mg/L.

In an earlier discussion we noted that a large component of dissolved particles

are ions. The greater the number of ions in a water sample, the easier it conducts
electricity. In many natural waters, an empirical correlation can be made between
the TDS or salinity and the electroconductivity (EC). This is illustrated in Figure
3-10. These correlations are not universal because the EC depends on the
particular mix of ions in the water sample. Nevertheless, a correlation for a
particular river or lake or industrial water might be applicable over a wide range
of TDS.

33
 Figure 3-9. Suspended solids in water
(Tchobanoglous and Schroeder, Water Quality, 1986)

34
 Electrodes in sample
Conductivity meter

TDS

EC

Figure 3-10. Electroconductivity (EC) and total dissolved solids (TDS)

Volatile Solids (VS) and Volatile Suspended Solids (VSS)

It is often valuable to know how much of the TS, TSS, or even TDS is composed
of organic compounds. As we will see later in the semester, organic compounds
can be used as food by microorganisms whose metabolic processes remove
oxygen from the water, which in turn can lead to fish kills. One way of judging
whether a material is organic is to see if it burns. The MZ
oven temperature used in the TSS test (105 °C) is low Another measure of organics is
Biochemical Oxygen Demand,
enough to dry organic materials without combusting them.
which is in MZ Sec 5.4 and in a
If we raise the temperature to 500 °C, the organic chapter in this document.
compounds will burn and be converted to CO2 (see Figure
3-11). The loss of a gas is called volatilization, and so the
organic compounds are called “volatile” solids. The materials left on the filter
(or pan) after combustion are called “fixed” (i.e., nonvolatile) solids. The mass
of volatile solids is the mass of the filter (or pan) before combustion minus the
mass after combustion. Dividing by the original sample volume gives a mass per
unit volume unit. Typically, VS and VSS are reported in mg/L.

35
 High Temperature Oven (Muffle furnace)

Organics burn off (volatilize)

Filter
Fixed (inorganic) solids remain

Figure 3-11. VSS test procedure

A couple of useful relationships between the various gravimetric (mass-based)

measures follow:

The volatile fraction can’t be greater than the total.

VS ≤ TS
VSS ≤ TSS
VDS ≤ TDS

The total is the sum of the suspended and the soluble.

TS = TSS + TDS
VS = VSS + VDS

Turbidity

Another way to detect small solids is turbidity. Turbidity can be thought of as

the cloudiness of a water sample, which depends indirectly on the number of
particles. As shown in Figure 3-12, turbidity is measured with light. A light
beam is projected into a sample. The particles reflect and scatter the light. A
photodetector measures the intensity of the light that is scattered at a 90° angle to
the incoming beam. This technique is called “nephelometry” and the units are
Nephelometric Turbidity Units (NTU). NTUs are relative units, based on a
standard made with an agreed-upon recipe that produces a suspension with 1000
NTU.

36
 Scattered
Light

Photo detector and meter (or digital readout)

Input Light

Figure 3-12. Measurement of turbidity by nephelometry (90° light scatter)

Turbidity is a very quick and easy measurement compared to TSS. There are
even handheld battery units for the field. It is, however, an aggregate measure,
meaning it measures a combination of parameters. Light scattering is a function
of the number, shapes, sizes, and optical properties of particles. Therefore,
turbidity does not indicate anything about particle mass, but it does tell you
something about the relative abundance of particles in the sample. Where
turbidity is particularly useful is in clean water applications
because it is much more sensitive than TSS. The difficulty of MZ
Oddly, turbidity is not
measuring very small masses effectively sets a lower bound on discussed in the text
accurate TSS measurements. So, turbidity used instead of TSS although it is a common
for defining and monitoring the quality of drinking water and measurement.
other clean process waters. The turbidity standard for municipal
drinking water is 0.5 NTU.

Secchi Depth – A Field Measure of Clarity

One other measure that is used in the field is Secchi depth. Secchi was a priest in
the Papal Navy who was interested in knowing how far down into the water a
lookout could see obstructions. The test is very simple. A standard Secchi disk
is lowered into the water until the disk disappears (see Figure 3-13). This is the
Secchi depth. Cleaner waters produce larger Secchi depths. There is, of course,
a certain imprecision to this measurement caused by different lighting conditions
and operators. Nevertheless, it is a useful semi-quantitative indicator of clarity.

Locally, the most famous use of Secchi depth has been at Lake Tahoe. In the
1960’s, one could see a Secchi disk down to a depth of 100 ft. In fact, Tahoe is

37
one of the clearest large lakes in the world. Over the last 40 years, though, the
Secchi depth has steady decreased to about 65 ft today. The loss of clarity was
originally attributed to algae growth in the lake stimulated by nutrients from
sewage discharges. In the 1970’s and 80’s the sewage collection and treatment
systems were completely rebuilt, but even though sewage is now pumped out of
the Tahoe Basin, clarity has not improved. It is now thought that the majority of
the clarity loss is due to very small particles (<20 µm) that wash into the lake
from stormwater runoff and are blown into the lake by the wind. Looking back
at Figure 3-1, you can see that these particles are at the low end of the suspended
solids range, and so don’t settle readily.

Secchi
depth

Secchi Disk

8 inch

Figure 3-13. Measuring clarity using a Secchi disk

38
Practice Problems

3-1. (ML 8.4 edited by JJ). Solids analysis is one of the most widely used
parameters for assessing water quality. The following tests were performed on a
water sample. The sample volume was 150 mL in each test (filtered or
evaporated). Calculate the parameters requested using the following data.

Mass, g
1 Tare mass of evaporating dish 24.3520
2 Mass of evaporating dish plus residue after evaporation 24.3970
@ 105°C
3 Mass of evaporating dish plus residue after ignition @ 24.3850
550°C
4 Mass of Whatman filter and tare 1.5103
5 Mass of Whatman filter and tare after drying @ 105°C 1.5439
6 Mass of Whatman filter and tare after ignition @ 550°C 1.5399

a. Total solids. [TS = 300 mg/L]

b. Volatile solids (total). [VS = 80 mg/L]

c. Total suspended solids. [TSS = 224 mg/L]

d. Volatile suspended solids. [VSS = 27 mg/L]

e. Total dissolved solids. [76 mg/L]

Notes:
Here are some checks that you can apply to your calculations.
TSS<TS
VS<TS
VSS < TSS
TS = TSS + TDS
TSS = VSS + FSS

39
 Practice Problem Hints

Chap 3 Solids
1 Hint (a): (Row 2 – Row 1) ÷ volume

Hint (b): (Row 2 – Row 3) ÷ volume. The volatile solids are those
that burned off (volatilized), not those left behind.
(Row 3 – Row 1) ÷ volume gives you the solids left on the dish.
These are called the “fixed” (inorganic) solids (FS).

Hint (c): (Row 5 – Row 4) ÷ volume

Hint (d): (Row 5– Row 6) ÷ volume. Same logic as in b. The fixed

suspended solids are called FSS.

Hint (e): TDS = TS – TSS. In the lab this could be measured by

capturing the filtrate and evaporating it. We don’t have these
numbers in the problem statement.

40
4. Equilibrium Processes
In nature there are reversible and irreversible reactions. An irreversible reaction
goes one way only. When you combust hydrogen and oxygen you produce
water. The water, though, does not spontaneously revert to hydrogen and oxygen
gas. In contrast, reversible reactions do move in either forward or reverse
directions. Table salt, for instance, will dissolve into sodium and chloride ions
and under the right circumstances these ions will recombine to create solid salt (a
process called “precipitation”).

In this class we will examine four equilibrium processes or reactions of

environmental importance. They are:

1. Adsorption
2. Gas solubility
3. Acid/base chemistry
4. Mineral solubility.

It is important to remember that the word “equilibrium” in this context does not
mean the same as it does in structures. Structures are in equilibrium when the
sum of the forces on them is zero. In other words, the forces are balanced. This
is often called “static” equilibrium. In contrast, chemical equilibrium is
“dynamic” because it is based on an equality of rates.

Imagine two tanks of water with two pumps as shown below. The system is set
up so that the rate at which the pumps operate depends on the volume of water in
the tank -- the more water, the higher the pump rate. Mathematically, Q=kV
where Q is the flow rate, V is the volume, and k is the control constant.

41
 Q = kV

Q1 forward
V1
V2
Q2 reverse

Figure 4-1. Fluid equilibrium example

No imagine a situation where V1 is 2000 gallons and V2 is 0. Assume k1 is 0.006

gpm/gal and k2 is 0.008 gpm/gal (that is, gallons per minute flow per gallon of
water stored in the tank). Initially, Q1 is 2000×0.006 = 12 gpm out of tank 1 and
into tank 2. Q2 is 0 initially, but as water pours in from tank one, V2 increases
from 0, causing Q2 to increase as well. Over in tank 1, the outflow of water is
reducing V1 which causes Q1 to decrease as well. You can see where this goes in
the following graph:

Figure 4-2. How flow and volume move toward equilibrium values in the fluid
equilibrium example

42
As the volumes change, the flowrates change until the flowrates are equal.
At this point, the volumes do not change any Computer Link
further. Notice that the volumes aren’t equal; You can try different experiments on the
Chemical Equilibrium demonstration
they just aren’t changing. In dynamic spreadsheet on Canvas.
equilibrium the rates are equal and the volumes
are constant.

Le Chatelier's Principle or the Law of Mass Action

If a system at equilibrium is subjected to a change in conditions that moves it

away from equilibrium, then the rates adjust so that a new equilibrium state is
reached. The system adjusts in a way that offsets, or
opposes, the change of condition that disrupted the Henri Louis
equilibrium. Le Chatelier
(1850 - 1936) was an
A mechanical analogy would be a ball in a U-shaped track influential French
(see below). The equilibrium position of the ball is the chemist. (Wikipedia)
bottom. If the equilibrium is disturbed by pushing the ball
up one side, the ball tries to return, that is, it opposes the change of condition.

System at System being System opposing the

equilibrium displaced away from displacement and returning
equilibrium to equilibrium

Figure 4-3. Reacting to a disturbance

In chemical equilibrium, though, the new equilibrium state is not the same as the
original. Let’s return to the water tank system. Suppose at t=200 minutes, we
instantaneously drop 1000 gallons of water into tank 1. Q1 would increase
because V1 is greater but Q2 would both change at first. The system would be out
of equilibrium and would work to adjust itself as shown in the graphs below.

43
Figure 4-4. Disturbing and returning to equilibrium in the fluid system example

You can see that at t=500 the rates are once again equal and the volumes are
constant. Notice though, that neither the rates nor the volumes are the same as
they were before. This is a different equilibrium state than the original. In this
example we increased V1. The system “opposed” our action by coming to a new
equilibrium with a smaller V1. If we had taken water out of tank 1, the system
would “oppose” our action and add water back in during the process of attaining
a new equilibrium state.

CHEMICAL EQUILIBRIUM

Reversible chemical reactions go in two directions:

A + B  C + D (forward), or

A + B  C + D (reverse)

We express this on paper as: A + B  C + D

or in shorthand: A + B = C + D

44
For example,

2SO2(g) + O2(g)  2SO3(g).

In this reversible reaction some SO3(g) decomposes, like so:

2SO2(g) + O2(g)  2SO3(g)

When the rate of the forward (left to right) reaction equals the rate of the reverse
reaction, the system is in "dynamic equilibrium". The word "equilibrium" means
the relative concentrations of reactants and products do not change with time.
"Dynamic" reminds us that the reactions are still happening. The concentrations
are steady because the reactions are proceeding at the same rate, not because the
reactions have stopped. Keep in mind that just because the reaction rates are
equal, doesn't mean the concentrations are equal. In the example above, the
concentrations in one equilibrium state are:

[SO2] = 1.65 M (M = molarity = mol/L)

[SO3] = 12.43 M

Although there are many equilibrium states, we have equations from

thermodynamics that tell us the relationships among the concentrations when a
system is at equilibrium.

In general, for a system where [Reactants]  [Products]

[ Products]
K eq =
[Reactants]

For the chemical reaction: aA + bB = cC + dD

[C] c[ D] d
Keq =
[ A] a[ B] b
where the [ ] nomenclature means molar concentration.

Keq depends on the relative reaction rates (forward and reverse) which in turn
depend on the concentrations of reactants and products. Think of it this way. In
a chemical reaction, the reacting substances must interact with each other. The
greater the concentration or reactants, the higher the probability that the
individual molecules will interact and the faster the reaction rate will be. (That’s
one reason the concentrations are expressed in mol/L. Moles represent the
number of molecules rather than their mass.) In practice, Keq is usually

45
determined by experiment, and is affected by various environmental factors, most
notably, temperature.

Here’s how Le Chatelier's Principle would be applied to a chemical system.

Assume the following system of four reactants.

A + B = C + D

[C ][ D ]
The equilibrium equation is: K eq =
[ A][ B ]

If we add more "D" from some outside source (say, by dissolving in a

hypothetical salt called E2D), the increased concentration of D increases the
opportunities for C and D to react. The additional D will increase the rate of the
reverse reaction (A + B ← C + D). This causes the concentrations of C and D to
decrease and the concentrations of A and B to increase. The disappearance of D
“opposes” our action of artificially increasing the D concentration in the solution.
This process continues until the equilibrium equation is satisfied again.

If you wish, you can consider this from a purely mathematical point of view. If
we artificially increase D, then for the Keq equation to hold true, C must decline,
and A and B must increase. This change in concentration is accomplished by the
chemical reactions described above.

As in the water tank system, the chemical system doesn’t return to the same
equilibrium point; it comes to a new equilibrium. In other words, the
concentrations of A, B, C, and D after the system equilibrates will not be the
same as they were before the disturbance. They will, however, satisfy the
equilibrium constant equation, and the rates of forward and reverse reactions will
be equal.

46
Practice problems

4-1. Consider the following equilibrium reaction.

AB2  A+ + 2B-
If you add A+ to the solution, will you get more AB2 or less? [More]
Hint: Increasing the A+ concentration initially increases the reverse reaction
rate. Because it is larger than the forward rate, there will be an accumulation of
AB2. As the AB2 concentration increases, the forward reaction rate increases
and as the A+ and B- concentrations decrease, the reverse reaction rate declines
until the two rates are equal again.

4-2. Given the equilibrium reaction shown below, if you add more D and let the
system come back into equilibrium, what are the new concentrations compared to
the originals?
A + B  C + D
[A is higher, B is higher, C is lower, D is higher.]

4-3. Given the equilibrium reaction shown below, if you want to reduce the
concentration of A, should you add B, C, or D?
A + B  C + D
[Add more B.]

Practice Problem Hints – none for these problems

Chap 4 Equilibrium
1 --
2 --
3 --

47
5. Adsorption

In adsorption, molecules dissolved in a fluid preferentially accumulate at a solid

surface. The dissolved substance is called the adsorbate; the solid is called the
adsorbent. The fluid can be either liquid or gas. The cause of the preferential
accumulation is thought to be weak physical and chemical bonds between the
adsorbate and the adsorbent. Although the exact nature of these bonds isn’t fully
understood, the end result is that adsorbate molecules are in a lower
energy state on the surface than they are in the fluid, which is why MZ
they accumulate at the surface. Sec 3.10 and 8.11

EQUILIBRIUM ISOTHERMS

Adsorption is a reversible process. If “B” represents some chemical substance,

the adsorption process can be written as a pseudo-chemical reaction like so:

B in fluid  B on solid

As some molecules of B attach themselves to the surface, others are kicked off
back into the fluid. At equilibrium, the rate of movement of molecules onto the
surface (adsorption) and the rate movement off the surface (desorption) are equal.
Both rates are dependent on concentrations.

We can write an equilibrium expression for the pseudo-reaction like so:

C solid
=k
C fluid

The concentration in the fluid is usually denoted by Ceq (the equilibrium

concentration) and is expressed in concentration units like mg/L. The
concentration on the surface is denoted by x/m where x is the mass of adsorbate
adsorbed onto mass m of solid. For example, if B is the dissolved adsorbate and
D is the solid adsorbent, the units for x/m might be (mg of B) ÷ (g of D). Using
this terminology, the equation above can be written as:

(𝑥𝑥⁄𝑚𝑚)
𝐾𝐾𝑝𝑝 =
𝐶𝐶𝑒𝑒𝑒𝑒

48
where Kp is an empirical partitioning coefficient (that’s where the “P” comes
from). Normally, though, the equations are written in this form:

𝑥𝑥 Note, in many texts x/m

= 𝐾𝐾𝑝𝑝 𝐶𝐶𝑒𝑒𝑒𝑒
𝑚𝑚 is given the symbol q.

Equations that describe the concentrations at equilibrium are called isotherms.

(The word “isotherm” literally means “constant temperature”, which reflects the
fact that the isotherm constants are temperature dependent. The equation above
is the Linear isotherm, which is often applicable to situations in the natural
environment where concentrations are typically low (e.g., groundwater or surface
water bodies).

Many other isotherms have been derived. The Freundlich (pronounced froynd-
lick) isotherm is a general empirical equation developed by Dr. Herbert
Freundlich in 1909 (not that long ago compared to other engineering equations).

𝑥𝑥
  =  𝐾𝐾𝐶𝐶𝑒𝑒𝑒𝑒 1/𝑛𝑛
𝑚𝑚

where K and 1/n are constants that can be determined from experimental data.
The linear isotherm can be seen to be a special case of the Freundlich when 1/n =
1.

Unlike the Freundlich isotherm, the Langmuir isotherm was derived from
theoretical considerations and confirmed by experiments in situations where
adsorbate molecules occupy a large fraction of all the sites on a surface that could
potentially be occupied. The Langmuir isotherm equation looks like:

𝑥𝑥 𝑘𝑘𝐶𝐶𝑒𝑒𝑒𝑒 𝑥𝑥
 = � �� �
𝑚𝑚 1 + 𝑘𝑘𝐶𝐶𝑒𝑒𝑒𝑒 𝑚𝑚 𝑚𝑚𝑚𝑚𝑚𝑚

where k is an empirical equilibrium coefficient and (x/m)max is the amount of

adsorbate (x) per unit mass of adsorbent (m) contained in a mono-molecular layer
on the solid surface. In theory, (x/m)max is the maximum mass of adsorbate that
could be retained

In the graph below, all three isotherms are plotted together. The coefficients
used are listed in the table.

49
 100
90
80
70
x/m (mg/mg)

60
50
40
30
Linear
20
Freundlich
10 Langmuir
0
0 5 10 15 20 25 30
Ce (mg/L)

Langmuir k = 0.5 (x/m)max = 40

Freundlich K = 15 1/n = 0.5
Linear Kp = 10

Figure 5-1. Plots of linear, Freundlich, and Langmuir isotherms

As can be seen in this example, all three equations describe the region below
Ceq=4 equally well. In this region, you could substitute a linear isotherm for
either of the other two. However, when Ceq increases (say to about 5 in this
example), the Freundlich and the Langmuir isotherms depart from the linear. For
larger Ceq values, the denominator of the Langmuir isotherm (1+kCeq) approaches
kCeq. At the limit, the top and bottom kCeq terms in the equation cancel and
(x/m) = (x/m)max. You can see the Langmuir plot in the graph lean over toward
(x/m)max. In contrast, neither the Freundlich nor Linear isotherms have an upper
boundary. Looking at the graph, the Freundlich isotherm looks approximately
linear in the 15-30 mg/L range. A linear equation in that range, however, would
not go through zero. Like any empirically-derived relationships, these
isotherms should only be used in the Ce ranges for which they were calibrated.

The coefficients of the Freundlich isotherm can be determined from a fairly

simple set of batch experiments. In these experiments, a known amount of
adsorbent is added to a beaker containing adsorbate with a known volume and
concentration. After equilibrium conditions are established, which may take
some time, the concentration in the liquid is Ceq and x/m can be calculated from
the change in adsorbate concentration. With a set of Ceq and x/m data pairs, the
linearized form of the Freundlich equation can be plotted.
50
 x/m = KCeq1/n

ln(x/m) = ln(K) + (1/n) ln(Ceq)

Linear regression gives the slope of the line (= 1/n) and the y-intercept (= lnK).
For the Linear isotherm, the math to determine Kp is even easier. Plot x/m vs. Ceq
and force the regression through 0. The slope is K.

The degree to which an adsorbate is removed from the water (as expressed by
Keq) is a function of two factors:

1. Chemical affinity between the adsorbent and the adsorbate

2. Surface area of the adsorbent

The first factor means that each adsorbate will react differently with a given
adsorbent. In nature, the chemical characteristics of the adsorbent depend on its
composition. For engineered adsorbents, the affinity factor is
controlled by proprietary steps in the manufacturing process.
Regardless of the affinity, surface area is an important factor because
adsorption occurs on the solid surface. For equal affinities and
concentrations, small particles will adsorb more per unit mass than
large particles because of the difference in surface area.

APPLICATIONS

Adsorption is a common phenomenon. In the natural environment, it is a

mechanism by which dissolved compounds are immobilized. Here are some
examples.

• Nutrients (nitrogen and phosphorus compounds) adsorb to soil particles, and

thus become available to plant roots. Without adsorption, many of these
compounds would leach away from surface soils and end up in groundwater.

• In natural water bodies, compounds will adsorb to suspended particles. The

adsorbed compounds then move with the particles to which they are attached.
If the particles are large and heavy and settle easily, adsorbed pollutants can
accumulate in sediments on the bottoms of lakes, rivers, and bays. By this
mechanism, some pollutants are separated from swimming organisms. On
the other hand, bottom-dwelling organisms receive a larger dose. Filter
feeders (animals like shellfish whose diet consists of small particles filtered
from the water) are put at particular risk.

51
• In aquifers, pollutants adsorb and desorb from soil particles, complicating
efforts to track and remove groundwater pollution.

• Adsorption is also used in engineered systems as a treatment process to

remove dissolved contaminants. Examples include organic pesticides,
solvents, and compounds that cause tastes and odors. You may have seen
home versions of these systems in the form of filters that fit onto water
faucets. Engineered systems are also used to adsorb contaminants from air.

In engineered systems, the adsorbent of choice is very often activated carbon.

Activated carbon is charcoal that has been specially-treated to increase its surface
area and its affinity for target compounds. The activation process is a closely-
held trade secret in most cases. As described below, there are two major methods
of applying activated carbon in treatment systems.

Adsorption by Direct Addition of Adsorbent

When an adsorbent is added to a solution, adsorbate moves from the

dissolved phase to the solid surface. The concentration in the solution PAC
decreases until it reaches Ceq. Adding more carbon causes the concentration
in solution to decrease further. In a treatment plant, a common adsorbent is
powdered activated carbon (PAC), which is used because it has a very large
surface area per unit mass. PAC is added toward the beginning of the
treatment train. After giving it some time to come to equilibrium, a
downstream process, such as settling or filtration, removes it from the water.

When treating a water stream to meet a specified limit, operators just keep
adding PAC until that desired target concentration (Ctarget) is achieved. At the
end, the PAC is in equilibrium with the desired target concentration. This fact
can be used to calculate the amount of PAC needed to treat a given volume (or
flow) of water.

Mass of adsorbate to be removed from a given volume, X = (Cinf – Ctarget)V

x/m = KCe1/n where Ce is set at the target concentration, Ctarget

PAC needed = X / (x/m)

Another name for the “target” concentration (Ctarget) is the “effluent”

concentration (Ceff) because the target concentration is the desired concentration
in the treatment plant effluent.

52
Adsorption in Columns

An alternative method of treating a water (or air) stream is by passing the fluid
through a stationary bed of adsorbent. In both water and air, a common
adsorbent is granular activated carbon (GAC) because its particles are larger
than PAC particles. Larger particles are chosen to reduce the fluid head loss in
the column to a reasonable value (bigger particles, bigger pores). The most
common configuration is a multi-stage column.

Imagine the top section of the column as a

beaker full of solid GAC (the star shapes in the
figure) to which you add adsorbate with a given
influent concentration (Cinf = 9.5) Follow
along with the graph on the next page. Fill the
beaker with influent and let it sit there for a
while. As the GAC and the water contact each
other, dissolved constituent adsorbs to the
carbon particles to establish equilibrium. In this
example, let’s assume V = 1, m = 1, and K =
0.5. At equilibrium, the mass of adsorbate
removed from the water (Ci-Ceq multiplied by
V) must equal the mass adsorbed onto the
adsorbent (x). Substituting this into the isotherm equation, we get …

�𝐶𝐶𝑖𝑖 − 𝐶𝐶𝑒𝑒𝑒𝑒 �𝑉𝑉

= 𝐾𝐾𝐶𝐶𝑒𝑒𝑒𝑒
𝑚𝑚

If you solve for Ceq, you get 6.33 (remember V = 1, m = 1, and K = 0.5). In the
graph (volume 1), the dissolved concentration initially drops to 6.33 and x/m for
the GAC increases from 0 to 3.17.

𝑥𝑥 (9.5 − 6.33𝑚𝑚𝑚𝑚/𝐿𝐿)(1𝐿𝐿)
= = 3.17𝑚𝑚𝑚𝑚/𝑔𝑔
𝑚𝑚 (1𝑔𝑔)

(Note that the units are mg of adsorbate per g of adsorbent. Don’t try to cancel
the grams.)

Now imagine pouring off the liquid while keeping the adsorbent and then
refilling the beaker with new water at the original concentration (Cinf = 9.5).
Again, adsorption takes place, but because the GAC already has some adsorbate
attached to it, less is removed from the liquid phase and Ceq is higher than it was
before. Let’s check whether this makes sense. The x/m value after the second
53
equilibrium is the value after the first equilibrium plus what was adsorbed from
the second volume of water put into the beaker.
The equation looks like this:

𝑥𝑥 (9.5 − 𝐶𝐶𝐶𝐶𝐶𝐶 )𝑉𝑉

� � = �3.17 + � = 𝐾𝐾𝐾𝐾𝐾𝐾𝐾𝐾
𝑚𝑚 2 𝑚𝑚

Again, assuming V=1 and m=1 for simplicity, if you solve for Ceq you get 8.44.
The liquid concentration drops from 9.5 to 8.44 and the adsorbed mass (x)
increases from 3.17 to 4.22 as shown below.

𝑥𝑥 (9.5 − 8.44)(1)
� � = �3.17 + � = 4.23
𝑚𝑚 2 (1)

𝑥𝑥
� � = 𝐾𝐾𝐶𝐶𝑒𝑒 = (0.5)(8.44) = 4.22
𝑚𝑚 2

(The difference between 4.23 and 4.22 is rounding.)

As you repeat this process over and over, x/m increases with each cycle and the
dissolved concentration changes less and less. Eventually, the dissolved
concentration does not change at all and x/m is in equilibrium with the influent
concentration (Cinf).

The results shown in Figure 5-2 represent only the top section of
the column. The water leaving the top section travels downward
through the column. Water passing through the different sections
is analogous to being poured from one beaker to the next. The
concentration changes in beaker (section of the column). Over
time though, each beaker (column section) removes less and less
adsorbate (gets darker and darker in the figure). Eventually, the
liquid in each beaker is at the influent concentration as predicted in
Figure 5-2.

54
Figure 5-2. How dissolved and adsorbed concentrations vary as a system moves
toward equilibrium in a column configuration

When this happens, there is no change in concentration as the water moves

through the column and the x/m value for the adsorbent everywhere is 4.75. (We
sometimes say that the column is “saturated” at this point.) When the GAC
cannot remove any more pollutant, it is replaced with fresh carbon.

In practice, these
columns are often strung
together as shown here.
The effluent from the
1 2 3
first column (Point A)
would be monitored to
determine when the
column is no longer
adsorbing. At that point, A
column #1 would be
removed, refilled with new GAC, and then placed in the #3 position, so that the
cleanest GAC was always the last in line and “dirtiest” at the front. (This would
be done by rerouting the flow using valves rather than physically moving the
column.)

55
The amount of carbon used to treat a given volume (or flow) is calculated in the
same way as shown above before except that in the isotherm equation Ceq is set
equal to Cinf rather than some target Ceff .

The mass of adsorbate that is captured by the adsorbent at equilibrium is

calculated from the isotherm, assuming the adsorbent is in equilibrium with the
influent concentration (that is, when the column is saturated).

x/m = KCe or KCe1/n

where Ce is set at the influent (incoming) concentration, Cinf.

𝑥𝑥 𝑋𝑋
𝑎𝑎𝑎𝑎 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 =
𝑚𝑚 𝑀𝑀

where X is the mass of adsorbate captured from the volume of water and M is the
mass of adsorbent doing the capturing.

Note: This discussion is a great simplification of adsorption systems.

Equilibrium calculations only predict endpoints, not the speed of getting there.
In real design, you have to think about the flowrate because adsorption is not
instantaneous; it takes some time to come to equilibrium. The concentration
decreases as the water moves down through a column and from column to
column. Don’t think that all the adsorption happens in the first column only.
The design goal is to have enough adsorbent in the water’s pathway that the
concentration in the effluent from the system is below the target value. If you
push the water through the columns quickly, you’ll need a greater length than if
you do it slowly. The point of this discussion is to show that the capture of
adsorbate by adsorption depends on geometry and engineering design, as well as
on the adsorbent material.

Comparison of Direct Addition and Column Systems

Assuming equilibrium occurs, x/m will be proportional to Ce. Therefore, x/m for
a column is larger than the x/m resulting from directly adding the carbon to the
water because Cinf is larger than Ceff based on a standard. Larger x/m values are
desired because less carbon is needed to treat a given volume of water.

In practice, engineering considerations dictate which mode is used. The

economic advantage of columns in terms of carbon use is lessened somewhat by
the fact that the K value for PAC is larger than that for GAC because of PAC's

56
larger surface area per unit mass. Also, using columns usually requires lifting the
water, which involves expending energy (potentially a lot because water is
heavy). On the other hand, you cannot use PAC unless you have facilities in
place to remove the PAC after it has done its job in the water. So, in surface
water treatment plants, which do include settling tanks and filters, direct addition
is often the method of choice. Where you don’t have those facilities, such as
when you are treating water from well and want to send it directly to the
customer, columns are preferred.

Figure 5-3. GAC columns on a City of Davis well (Johnston)

57
Practice problems

5-1. Suppose you are adding activated carbon (PAC) to a water stream to
remove a contaminant. The contaminant is adsorbed according to a linear
isotherm with K = 0.5 L/g. The influent concentration (Ci) is 5 mg/L. The target
effluent concentration is 1.5 mg/L.

a. What is the solid phase concentration (x/m) when the activated carbon is at
equilibrium with the target concentration? [7.5x10-4 g/g]

b. What mass of contaminant must be removed from 1.0 million liters of water
to reach the target concentration? [3500 g]

c. How many kg of activated carbon would be needed to treat 1.0 million liters
of contaminated water down to the target concentration? [4667 kg]

5-2. Repeat Problem 1 (a, b,and c) for a column configuration assuming the
column removes the contaminant to the same target concentration. [2.5x10-3 g/g,
3500 g and 1400 kg]

5-3. Repeat Problem 1 (a, b, and c) assuming a Freundlich isotherm with K =

0.2, 1/n = 0.5, and dissolved concentration units in g/L. [7.75x10-3 g/g, 3500 g
and 452 kg]

5-4. Repeat Problem 4-3 for a column configuration assuming the column
removes the contaminant to the same target concentration. [1.41x10-2 g/g, 3500 g
and 247 kg]

Note: In these problems the contaminant mass removed is always the same,
based on the influent and effluent concentrations. The column always requires
less adsorbent that the direct addition for the same isotherm because of the
higher concentration in the isotherm equation.

58
 Practice Problem Hints

Chap 5 Adsorption
1 Hint (a): Because it is a direct addition scenario,
the equilibrium concentration is the effluent
X
(target) value. = KCe
m
Hint (b): The mass removed is the change in
concentration in the volume treated.
Hint (c): You know x/m from the isotherm and x
from part b; calculate m.
2 Hint: The only change is that the equilibrium concentration is the
influent value because it is a column scenario.
3 Hint: Because it is a direct addition scenario, the equilibrium
concentration is the target value. You need to adhere to the
specified units for the Freundlich isotherm equation. Because the
units are specified as g/L, the number for your equation is 1.5x10-3
g/L.
4 Hint: Set the equilibrium value to the influent concentration.

59
6. Gas Solubility and Transfer Kinetics

When a liquid and a gas come into contact, molecules can move across the phase
boundary. Molecules moving into the liquid are said to be dissolving; molecules
leaving the liquid are said to be volatilizing. If we write this as a pseudo-
reaction, it looks like:

Gas phase  Liquid phase

Dissolution is the forward rate (to the right) rate; volatilization is the reverse (to
the left) rate. Notice that we called this a pseudo-reaction because the molecules
of gas in the gas phase and the liquid phase are not reacting. The molecules that
dissolve or volatilize are not changed; they’re just moving from one phase to the
other. For example, oxygen molecules in water are in the same for as in the air,
i.e.,O2 molecules.
William Henry
HENRY’S LAW published his
results in
1803.
At equilibrium …
(Wikipedia)
(rate of dissolution) = (rate of volatilization)

The concentrations are not equal but they are constant, and because of this, their
ratio is a constant. This is Henry’s Law. This can be expressed mathematically
as:
MZ
𝑃𝑃 𝐶𝐶 Sec 2.3, 3.6*
𝐾𝐾 = � � 𝑜𝑜𝑜𝑜 � �
𝐶𝐶 𝑃𝑃 Several different
expressions of Henry’s
Law are shown in Table
In the equation above, P is the gas concentration and
3.3. The units of the
C is the liquid concentration. The value and the constant tell you what
units of the constant, k, depend on the gas, the ratio the constant applies
to
temperature, and the measures of concentration.

Liquid concentrations can be expressed is several ways – mass concentration

(mg/L), molar concentration (mol/L), and mole fraction (χ). Mole fraction was
defined earlier in this monograph.

Gas concentration is often expressed by the partial pressure of the gas as

calculated from Dalton’s Law. Dalton imagined that in a mixture of gases, each

60
gas contributes a “partial” pressure and that the sum of all the partial pressures
(Pi) is the total pressure (PT) inside a vessel of volume V.

PT = ∑ Pi for a constant V

P1
If we apply the ideal gas law, we can see that the ratio of the
partial pressure to the total pressure is equal to the mole
P2
fraction.
P3
PiV n RT
= i
PTV nT RT
Pi n
= i = χi
PT nT
Pi = χ i PT

In these equations n is the number of moles and R is the gas constant. It turns out
that the mole fraction is also the volumetric fraction according to Amagat’s Law.
Amagat imagined that in a mixture of gases with pressure P, each gas contributes
a partial volume and that the sum of all the partial volumes is the total volume.

VT = ∑Vi at constant P
V1 V2
If we apply the ideal gas law again, we can see that the mole
fraction is also the volumetric ratio.
V3
PVi n RT
= i
PVT nT RT
Vi n
= i = χi
VT nT

The most common expression for partial pressure combines these results.

V 
Pi =  i  PT
 VT 

PT is the total pressure, (Vi/VT) is the volumetric fraction of gas(i) and Pi is the
partial pressure of gas(i). Volumetric fraction is unitless. It can be calculated
from percent divided by 100 (as in oxygen is 20.9% of the atmosphere or 0.209)
or in parts per million by volume divided by one million (as in the CO2

61
concentration is 400 ppmv or 4x10-4). Partial pressure has the same units as the
total pressure, so it can be expressed in psi, atmospheres (atm), pascals (Pa),
inches of mercury, or other units. When the total pressure goes up or down, the
partial pressure does the same. The partial pressure of oxygen at the top of Mt.
Everest is less than the partial pressure at sea level even though the volumetric
fraction is the same at both places. This is why it is hard to breathe at the top of
Mt. Everest or why you are short of breath when jogging at Lake Tahoe
compared to jogging in Sacramento.

In the table below are some values of Henry’s Law constants. Notice that they
vary by temperature.

Henry’s Law Constants

Temp °C H2S N2 O2 CH4 Volatile organic compounds,

KH in units of 10 atm*
4 Units of atm/(mol/L), all at 25°C
0 0.027 5.29 2.55 2.24 Chloroform 4.1
10 0.037 6.68 3.27 2.97 Trichloroethene 11.6
20 0.048 8.04 4.01 3.76 Tetrachloroethene 26.9
30 0.061 9.24 4.75 4.49 Benzene 5.6
KH in units of atm/(mol/L) Toluene 6.4
10 6.67 1204 589 535 Pi =0 in the atmosphere for these
20 8.65 1449 723 677 compounds.
30 10.99 1665 856 808
* Read this way: for oxygen at 10°C, KH = 3.27×104 atm

Reference: Davis, M. and S. Masten, Principles of Environmental Engineering and Science,

McGraw-Hill, 2004 and extended by the author.

What’s more important to notice in the table above is that the K units are
different. The units on a Henry’s Law constant tell you the form of the
equation. The Henry’s Law equations for the two constants shown below are as
follows:

𝑝𝑝 𝑃𝑃 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑖𝑖𝑖𝑖 𝑎𝑎𝑎𝑎𝑎𝑎

𝐾𝐾(𝑎𝑎𝑎𝑎𝑎𝑎) = = =
𝑐𝑐 𝜒𝜒 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 (𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢)

𝑎𝑎𝑎𝑎𝑎𝑎 𝐿𝐿 𝑃𝑃 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑖𝑖𝑖𝑖 𝑎𝑎𝑎𝑎𝑎𝑎

𝐾𝐾 � �= =
𝑚𝑚𝑚𝑚𝑚𝑚 𝑀𝑀 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 (𝑚𝑚𝑚𝑚𝑚𝑚 ⁄𝐿𝐿)

62
In the lab procedures is another form:

𝑚𝑚𝑚𝑚𝑚𝑚 𝑀𝑀 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 (𝑚𝑚𝑚𝑚𝑚𝑚 ⁄𝐿𝐿)

𝐾𝐾 � �= =
𝐿𝐿 𝑎𝑎𝑎𝑎𝑎𝑎 𝑃𝑃 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑖𝑖𝑖𝑖 𝑎𝑎𝑎𝑎𝑎𝑎

You should be able to figure out the form of the Henry’s Law equation for any
combination of units.

Practice problems

6-1. Calculate the saturation concentration for an air bubbling experiment using
both kinds of Henry’s Law constants from the table presented earlier in this
chapter. Assume a temperature of 10°C. [11.4 mg/L]

6-2. Calculate the saturation concentration for a bubbling experiment using pure
oxygen rather than air. Assume a temperature of 11°C. [52.9 mg/L]

Hints at the end of the chapter.

63
GAS TRANSFER KINETICS

Henry's Law describes the concentration relationship at equilibrium. What if you

are not at equilibrium? Le Chatelier’s Principle tells you how the system will act
to establish or re-establish equilibrium, but it
MZ
doesn’t tell you how fast this will happen. The Gas transfer kinetics is not explicitly
speed of a process or reaction is described by covered in MZ. Later, you’ll realize
kinetic equations. that it is simply another example of
first-order kinetics, which is
covered in Sec 3.11*.
Suppose we have a jar of water containing only a
little oxygen in solution. (A jar is the classic batch
reactor discussed in lecture.) Now suppose we place the jar on a countertop and
open it to the atmosphere. If we were to monitor the concentration of oxygen in
the water over time, the results might look like:

Figure 6-1. Theoretical increase in dissolved oxygen as a system approaches

equilibrium (C = saturation value, Cs)

As can be seen, the concentration asymptotically approaches some maximum

concentration at which equilibrium conditions are re-established. Because the
concentration doesn't go any higher than this value, it is said that the water is
saturated with oxygen and the maximum concentration is called the saturation
concentration (Cs). CS is the equilibrium concentration predicted by Henry's
Law.

Experiments like that described above have shown that the rate of change of
dissolved gas concentration can be expressed mathematically by:

dC
= k (C s − C )
dt
64
where
C = the concentration of dissolved gas (mg/L or mol/L)
CS = the concentration of gas under saturated (equilibrium) conditions
k = gas transfer coefficient and t = time

Cs is the Henry’s Law value based on equilibrium calculations.

Let's see if this equation makes sense by combining Le Chatelier’s Principle with
the mathematics.

• When C = CS, the system is at equilibrium, the rate of dissolution (rd)

is equal to the rate of volatilization (rv), and dC/dt = 0 (see above).

• When C < CS, rv < rd and the net movement of gas is from the air into the
water. Mathematically, when C < CS then dC/dt > 0, indicating that the
change of concentration in the water over time is positive (because more
gas is dissolving into the water).

• When C > CS, rv > rd and the net movement of gas is out of the water.
Mathematically, CS - C < 0, indicating that dC/dt < 0 and the
concentration in the water is declining over time (because the gas is
leaving the water).

If we integrate this equation and apply the appropriate boundary condition (at t =
0, C = C0), the resulting equation is:
(Cs − C ) = (Cs − C0 )e − kt

(Do you remember how to do this?)

In the environment, C is usually less than CS. How far the concentration is below
the saturation concentration (CS - C) is called the deficit, D. Rewriting the
equation above in terms of deficit yields:

D = D0 e − kt

Ah ha, this is an exponential decay function. Can you see this in the graph of
experimental results shown above? The distance between CS and C (the deficit)
is getting smaller with time as C asymptotically approaches CS.

It is helpful to think of the deficit (or CS - C) as a "driving force" for net gas
transfer. Looking at the differential equation dC/dt = k(CS - C), the slope of the C
65
curve is steepest when C is farthest from CS (i.e., when the deficit is largest). As
C approaches CS, the slope (dC/dt) gets smaller and smaller. In other words, the
driving force for net gas transfer diminishes as the system approaches
equilibrium. Le Chatelier's Principle says that when a system's equilibrium is
disturbed, it attempts to re-establish that equilibrium. What we see here is that
the larger the disturbance, the "harder" the system "works" to re-establish the
equilibrium.

Besides the deficit, what other factors do you think might affect the rate of
gas transfer? Recall our hypothetical experiment. We left the jar sitting
quietly on a countertop and watched the oxygen climb to the saturation
value. What if we mixed the jar? What if we bubbled air through the jar?
What if we poured the water back and forth between our jar and a beaker?
Would the degree of mixing or the size and number of bubbles, or the
thickness of the water stream between the jar and the beaker affect the
results? Yes, they would.

Physical and chemical factors change the gas transfer rate by changing the
transfer coefficient, k. These factors include surface area, temperature, and the
chemical characteristics of the liquid solution. You should see immediately why
surface area is perhaps the most important factor. Gas transfer occurs at the
interface between the two phases, so the greater the surface area, the more
opportunities the molecules have to transfer between phases. Surface area can be
increased by putting bubbles into the liquid. The liquid can also be spread out in
the air by making it a flat sheet or a collection of small droplets.

Transfer coefficients can be determined by a relatively simple experiment. We

linearize the integrated deficit equation as shown below:

D = D0 e-kt
ln(D) = ln(D0) – kt = ln(D0) –+ (-k)t

This equation is in the form of y = b + mx where ln(D) = y, t = x, m = -k and b =

intercept = ln(D0).

If we perform an experiment in which we measure the concentration of the gas in

the liquid as a function of time, we can plot that data as shown below:

66
 time (min)

0 10 20 30 40
2.5

y = -0.0301x + 1.7272
2.0 2
R = 0.9617

1.5
ln (D)

1.0

0.5

0.0

Figure 6-2. Plot of natural log of deficit vs. time in an aeration experiment

After doing the linear regression, the original equation can be written with the
best-fit linear regression values.

m = -k = -0.0301/min
k = 0.0301/min
ln(D0) = 1.7272, so D0 = 5.62 mg/L

D = 5.62e −0.03t
(C S − C ) = 5.62e −0.3t
C = C S − 5.62e −0.03t

When this equation is plotted with the experimental data, the graph shown below
is generated. Note that the curve from the derived equation (the red line) fits the
data pretty well except for the first value. (More on the first value below.)

67
 8

7
Concentration (mg/L)

0
0 10 20 30 40
Time (min)

Figure 6-3. Plot of measured concentrations (triangles) and two

mathematical models, one using the intercept of the line (solid red) and
one using the measured initial deficit (dashed blue)

An alternate form of the linearized equation is:

 D
ln  = − kt
 D 0 

Mathematically, this is equivalent to the version suggested above. D0 is Cs - C0,

so it doesn’t seem hard to use either. However, in this form a heavy burden is
placed on the C0 measurement. Because every other value is divided by D0, any
error in the C0 measurement is propagated throughout the whole analysis and
affects the regression. For example, in the graph above, C0 looks inconsistent
with the other values in the series. If the equation above is used to derive k, the
blue dotted line is generated. You can see that it doesn’t fit the data nearly as
well. So in your experiment you will be asked to calculate a hypothetical C0
from the intercept of the ln(D) vs. t graph.

68
Importance of Kinetics in Civil Engineering

Process kinetics is not an esoteric concept of interest only to chemists. The faster
a process goes, the less time is needed to go from a given starting point to a
target. In a municipal wastewater treatment plant, for example, the starting point
is the concentration of a pollutant entering the plant from the sewer, and the
target is the concentration of treated water that can be safely discharged to a
river. The time available to accomplish this treatment is the hydraulic detention
time.
V
θH =
Q
θH is the hydraulic detention time; V is the volume; and Q is the flowrate. For a
given flowrate, the hydraulic detention time is established by the volume. A fast
treatment process (i.e.,a high k value) requires a short time (a small θH) to
accomplish the required treatment, which in turn requires a small tank (V). If the
process is slow, θH is high and V is large. Large and small tanks are a civil
engineering issue.

Practice problems

6-3. You are setting up a new tank for your pet fish, Spot. You know that Spot
needs a minimum dissolved oxygen concentration of 5.0 mg/L, but your tap
water contains only 2.1 mg/L. So before putting Spot into the new tank, you turn
on the air pump and bubble air through the water. If the oxygen transfer rate
coefficient is 1.4/h, and the water temperature is 20°C, what is the minimum time
that you must bubble air through the water before you can put Spot into the tank?
[22.3 min]

6-4. The Johnstown, CA sewage treatment plant produces an effluent with a

dissolved oxygen (DO) concentration of 1.0 mg/L. This flow is to
be discharged into Dry Creek, a stream that contains a sensitive Dry Creek
species of fish, the Golden Hornet trout. The plant’s discharge Treatment
permit requires that where the wastewater enters the creek, the plant
wastewater oxygen content shall be not less than 85% of the
saturation value of the creek water. To raise the DO, it is proposed Aeration channel
to run the flow through a long, narrow rectangular channel with a
bubbler aeration system built into its bottom. The hydraulic detention time in
this tank is 6 minutes. If the creek temperature in the summer is 9 °C in the
summer, what transfer coefficient (k) is needed to meet the regulatory
requirement? [0.31/min]

69
 Practice Problem Hints

Chap 6 Gas Solubility and Transfer

1 Hint: For KH = 3.27×104 atm, and Pi = 0.209 atm for air, χ =
6.39×10-6. The molar concentration is 3.55×10-4M. For KH = 589
atm/(mol/L), the molar concentration is the same.
2 Hint: Use linear interpolation to get KH at 11°C (either form
works). The partial pressure of oxygen is 1.0 atm because it is a
pure gas stream.
3 Hint: Using Henry’s Law as in the previous two problems, CS is
9.25 mg/L. Now for the integrated gas transfer equation, C0 is 2.1
mg/L, and C is 5 mg/L. Watch your units on t.
4 Hint: The target concentration after aeration (C) is 85% of the
saturation value for the creek. CS for the creek is 11.6 mg/L
(interpolate the values in the KH table at the beginning of the
chapter). The wastewater initial concentration (C0) is 1.0 mg/L.
The aeration time is the hydraulic detention time in the channel.

70
7. Acids, Bases, and pH
For our purposes, the Bronsted-Lowry definition of acids and bases is adequate.

• Acid: a substance that donates a hydrogen ion (H+).

• Base: a substance that reacts with a hydrogen ion. Many common bases
perform this action by donating a hydroxide ion (OH-) which in turn
reacts with the hydrogen ion to form water.

Acids and bases influence the hydrogen ion concentration of aqueous solutions
which is measured by pH (defined below). You may recall from high school
chemistry that low pH values indicate acidic conditions and high pH values
indicate basic or alkaline conditions.

Acids, bases, and pH conditions have multiple important applications in the study
of environmental engineering. Here are some examples:

• Aquatic habitat – All aquatic organisms have pH ranges in which they

can live. Move out of these ranges and the organisms die. Generally
speaking, most organisms are safe in a pH range of 6-9. Rainbow trout
fry (baby fish), however, will die at pH values below 6.5. A study
showing this contributed to Caltrans abandoning chemical treatment of
stormwater runoff to reduce its pollution load on Lake Tahoe. In the
northeastern U.S, eastern Canada, and Scandinavia, acidic rain caused by
air pollution has lowered the pH of lakes to a point where almost nothing
can live in them. They look pristine and beautiful – clear, clear water –
but that is because they are nearly sterile.

• Weathering and erosion – The natural pH of rain is about 5.6, which

qualifies it as a dilute acid. This acid slowly dissolves rocks in a process
known as weathering. It’s the starting point of the natural erosional
cycle of the earth. This can happen to manmade structures as well. In
Europe, where for many decades the rain was made acidic by air
pollution, limestone and marble structures are being dissolved. There are
100-year old photographs of sculptures with intricate detail that are now
gone because they have been dissolved away.

• Metals – Acidic water dissolves metals and concrete. In mines, rain

water percolating through the mine will react with sulfide minerals and
create sulfuric acid which can then dissolve metals in rock. When this
happens in a mine acid mine drainage is created. Acid mine drain water
71
 is both very acidic, and carries a large load of dissolved metals, both of
which are bad for aquatic organisms. Locally, the Iron Mountain Mine
near Lake Shasta is an example of this phenomenon. This can happen in
pipes as well. If drinking water is too acidic, it will dissolve copper and
lead. In modern systems in the U.S. there are relatively few lead pipes,
but many old lead soldered connections. In contrast there are lots of
copper pipes. Excess dissolved copper and lead are unhealthy for
humans. So, drinking water regulations contain provisions for pH
control.

• Drinking water disinfection -- In this country we mostly disinfect

drinking water with chlorine which makes hypochlorous acid (HOCl)
when reacted with water. HOCl dissociates into H+ and a hypochlorite
ion (OCL-). As a bactericide, HOCL is much more potent than OCl-. As
we will discover later, the relative proportions of these two forms are
determined by pH. So pH control by water treatment plant operators is
important for effective disinfection.

• Ammonia toxicity – At lower and middle pH values ammonia occurs in

the form of ammonium ion (NH4+), but at higher pH values it takes the
form of dissolved ammonia gas (NH3). From the point of view of a fish,
NH4+ is just another dissolved salt. In contrast, NH3 is quite toxic. For
this reason, the ammonia allowed in the discharge of wastewater
treatment plants will depend on the pH of the receiving water.

• Metals treatment – Hydroxide salts (e.g., Cu(OH)2) are generally

insoluble. So, if you manipulate the pH to encourage the formation of
these substances, they will precipitate (form a solid) and you can settle or
filter the particles out. This is a treatment option for dissolved metals
which have undesirable health effects.

There are many other applications, but at this point you should have the
impression that pH plays a central role in water quality and health, and for this
reason the study of acids and bases is essential to environmental engineering.

DISSOCIATION OF WATER AND pH

Water itself is both a weak acid and a weak base. Water molecules break down
into hydrogen ions and hydroxide ions in a chemical equilibrium process. This
breaking apart is called dissociation, and it is common to all acids and bases. For
water the reaction is this:

72
 2H2O  H3O+ + OH-

H3O+ is called a hydronium ion. You see that a true H+ ion isn’t made. That
would be a naked proton, which doesn’t occur naturally. Nevertheless, the
system acts as though it produces a hydrogen ion. Accordingly, the shorthand
way of expressing this dissociation reaction is:

H2O  H+ + OH-

The equation for the dissociation constant is:

[ H + ][OH − ]
K=
[ H 2O ]

The dissociation reaction does not occur to a great enough extent to change the
molar concentration of water. Therefore, the molar concentration of water is
essentially constant. (Look at the gas solubility section to remind yourself what
this value is.) So, for convenience, this concentration is incorporated into the
equilibrium constant and a special constant is defined.

Kw = [H+][OH-] = 10-14

Again for convenience, we express hydrogen ion concentrations and K values as

negative base 10 logarithms. This function is called p(x).

pH = -log10[H+]
pOH = -log10[OH-]
pK = -log10K

If we take the logarithms of the water equilibrium equation above, we get:

-logK = -log[H+] + -log[OH-]

pKw = pH + pOH = 14 MZ
Sec 3.7*

The last relationship is one that occurs in all aqueous solutions at all
times. It is a convenient equation to keep in your back pocket for solving
problems.

The negative sign in the p(x) function both helps and trips us up. It allows us to
express very small numbers as positive values. For example, the pK of the
equilibrium constant 5.75x10-10 is 9.24. On the other hand, it leads to some
initial confusion in relating pH and [H+]. A natural tendency is to say that high
73
pH means high [H+]. But this isn’t true, because of the negative sign. Always
keep the pH scale in mind. Low pH values are acidic; high pH values are basic.
Low pH values (1-6) mean [H+] values of 10-1 to 10-6. High pH values (8-14)
mean [H+] values of 10-8 to 10-14.

Practice problem

7-1. If the pH is 5.6, what is the [OH-]? [3.98×10-9 mol/L]

NORMALITY

MZ
Molarity is a measure of the number of molecules per liter of Sec 2.5.1
solution (mol/L). The number of molecules may not, however, be
the best indicator of its chemical reaction potential. For example,
when a mole of H2SO4 is added to water, it completely dissociates and two moles
of H+ result. In contrast, when a mole of HCl is added to water, only one mole of
H+ results. This a 1.0 M solution of H2SO4 is twice as acidic as a 1.0 M solution
of HCl. Since we’re often interested in the strength of a chemical solution, we’ve
invented another unit of measure that expresses it directly. Normality is a
measure of the number of reactants or reaction sites per liter (eq/L). For acid/base
chemistry the reactants are H+ ions (for acids) or H+ reaction sites (for bases).
The relationship between molarity and normality is:

Normality = (z) × (Molarity) or N = zM

where z = the number of H+ ions potentially available or potential H+ reaction

sites per molecule. Thus, z will vary from compound to compound and for
acid/base chemistry, can be determined from the molecular formula. The units of
normality are “equivalents per liter” (eq/L) where an “equivalent” is a mole of
reactants or reaction sites. Given this, the units on “z” in the equation above are
“equivalents per mole” (eq/mol). Keep these units in your calculations.

Two solutions with equal normalities have equivalent chemical strengths, even
though they do not have the same number of molecules (molarities).

Normality is most commonly used in expressing concentrations of strong acids,

that is, acids that we know completely dissociate. The seven strong acids are:

74
 • HCl: Hydrochloric acid • HI: Hydroiodic acid (also
• HNO3: Nitric acid known as hydriodic acid)
• H2SO4: Sulfuric acid • HClO4: Perchloric acid
• HBr: Hydrobromic acid • HClO3: Chloric acid

One other concept associated with normality is the equivalent weight. The
equivalent weight is the mass of acid needed to make a 1.0 N solution (that is, 1.0
equivalent per L).

𝑒𝑒𝑒𝑒 𝑤𝑤𝑤𝑤
𝑁𝑁 = 𝑧𝑧𝑧𝑧 = 𝑧𝑧 � �=1
𝑀𝑀𝑀𝑀

𝑀𝑀𝑀𝑀
𝑒𝑒𝑒𝑒 𝑤𝑤𝑤𝑤 =
𝑧𝑧

If you follow the units, you see that eq wt is in units of g/eq (that is, g/mol ÷
eq/mol).

Practice problems

7-2. What’s the normality of a 2×10-3 M solution of HCl? Of H2SO4? [2×10-3 N

and 4×10-3 N]

7-3. What’s the equivalent weight of phosphoric acid, H3PO4, given that the
molecular weight of phosphorous is 31? [32.7 g/eq]

7-4. What is the pH of a 5×10-3 N solution of nitric acid, HNO3, a strong acid?
[2.3]

MONOPROTIC ACIDS

A monoprotic (“one proton”) acid donates one hydrogen ion in an equilibrium

dissociation reaction. If we call a generic monoprotic acid HA, where A is the
anion part of the acid, we can write the following:

HA  H+ + A-

[ H + ][ A− ]
Ka =
[ HA]

75
The equilibrium constant, Ka will be unique for every different acid. Equilibrium
constants, including Kw, vary somewhat by temperature. In the normal
temperature range in which most environmental engineering occurs, this
variation is small and ignored.

It would be useful to have an equation that expresses the fraction of one or

another of the chemical species in terms of the total concentration min solution.
For a monoprotic acid, we can define α0 and α1 like so:

[ HA] [ A− ]
α0 = α1 =
CT CT

In these equations CT is the total number of moles of A-containing substances.

CT = [HA] + [A-]

So …
α0 + α1 = 1

Starting with the definition of each f and using the equilibrium constant equation,
we can derive equations for each f in terms of K and [H+].

[ HA] [ HA]
α0 = =
CT [ HA] + [ A− ]

Dividing by [HA] gives:

1
α0 =
[ A− ]
1+
[ HA]

Employing the equilibrium equation gives you this:

[ H + ][ A− ]
K=
[ HA]
[ A− ] K
=
[ HA] [ H + ]
1
α0 =
K
1+ +
[H ]

76
By the same process, you can derive α1 which results in the following equation.
(Do it yourself for practice.)

1
α1 =
[H ]
+
+1
K

Notice that both of these equations depend on [H+] which is measured by pH.
So, if the α values are plotted as a function of pH for a hypothetical weak acid
with pK=8, the following plot is generated.

Figure 7-1. General monoprotic acid alpha diagram (distribution diagram)

Observe that the lines cross where pH = pK. When pH = pK, α = 0.5 for both
chemical species. You should verify that this is the case for yourself using the
equations above.

At low pH values (high concentrations of H+), most of CT is in the HA form. At

high pH values (low H+ concentrations) most of CT is in the A- form. You should
check that this is what you would predict from Le Chatelier's Principle.

You can calculate the concentrations of HA and A- if you know CT and the pH.
The pH is used with the equilibrium constant to calculate the α factors.

[HA] = α0CT [A-] = α1CT

77
Example: Chlorination chemistry

When chlorine is added to water, which is commonly done to disinfect it, the
chlorine gas reacts to form hypochlorous acid (HOCl).

Cl2(g) + 2H2O  2HOCl + 2H+

HOCl is a weak acid with a pK of 7.54. The equilibrium reaction is:

HOCl  H+ + OCl-

The distribution diagram (α vs. pH) is:

Figure 7-2. Alpha diagram for hypochlorous acid (HOCl)

As described at the beginning of this chapter, HOCl is a better disinfectant than

OCl-, so water treatment plant operators would like to keep the pH a little on the
acid side.

Practice problem

7-5. At what pH is 85% of the total hypochlorite in the HOCl form? [6.79]

78
Example: Hydrochloric acid (HCl)

Hydrochloric acid (HCl) should not be confused with hypochlorous acid (HOCl).
One big difference is that HCl is a strong acid while HOCl is a weak acid.
Remember, “strong” and “weak” refer to how readily the molecules give up their
H+ ions, not the molarity or normality. The pK of HCl is estimated to be -3.
This means Ka = 103. When the distribution plot (α vs. pH) is created (see
below), it shows that there is no HCl in the pH range of 0 to 14. In other words,
at the pH’s we see in normal life, the HCl is completely dissociated, which is
what we mean by a “strong” acid.

Figure 7-3. Alpha diagram for hydrochloric acid (HCl)

By the way, you are intimately acquainted with HCl, even if you don’t know it.
It is the main component of gastric acid which you secrete in your stomach to
digest your food. It is also called muriatic acid and is sold at swimming pool
supply stores.

Example: Ammonia chemistry

In biological systems, a byproduct of the decay of proteins is ammonium ion.

Ammonium ion is NH4+; ammonia gas is NH3.

Ammonium ion acts like a weak acid (pK = 9.3):

NH4+  H+ + NH3(g)
79
The distribution diagram is shown below.

Figure 7-5. Alpha diagram for ammonium and ammonia (NH4+ and NH3)

At low pH values, the dominant form is NH4+ which is odorless and nontoxic. At
high pH values, NH3 is the dominant form. It can volatilize out of the water to
cause odors. NH3 is also very toxic to fish.

Practice problem

7-6. Knowing that NH3 is the toxic form, at what pH will it be 15% of the total
ammonia (CT)? [8.55]

DIPROTIC ACIDS

A diprotic (“two protons”) acid donates two hydrogen ions in a pair of linked
equilibrium dissociation reactions. If we call a generic monoprotic acid H2A,
where A is the anion part of the acid, we can write the following:

H2A  H+ + HA- HA-  H+ + A2-

[ H + ][ HA − ] [ H + ][ A 2− ]
K1 = K2 =
[ H 2 A] [ HA − ]
80
The equilibrium constants, K1 and K2, will be unique for every different acid.
Similar to monoprotic acids, we can define dissociation fractions (α’s) like so:

[ H A] [ HA− ] [ A2 − ]
α0 = 2 α1 = α2 =
CT CT CT

The subscripts of the α’s correspond to the dissociation (0th, 1st, or 2nd) and the
product of that dissociation is in the numerator. The equations for the α factors
can be derived in a manner similar to what we did above. Here is the derivation
of α0.

[ H 2 A] [ H 2 A]
α0 = =
CT [ H 2 A] + [ HA− ] + [ A2 − ]

Diving through by [H2A] …

1
α0 =
[ HA ] [ A2 − ]
−
1+ +
[ H 2 A] [ H 2 A]

We want to use the equilibrium equations:

[ H + ][ HA − ] [ HA − ] K
K1 = so...... = 1+
[ H 2 A] [ H 2 A] [ H ]
[ H + ][ A 2− ] [ A 2− ] K
K2 = so...... = 2+
[ HA − ] [ HA ] [ H ]
−

If we expand the third term in the denominator of the α0 equation and substitute
in the equilibrium equations, we get …

1
α0 =
[ HA ] [ A2 − ] [ HA− ]
−
1+ +
[ H 2 A] [ HA− ] [ H 2 A]
1 1
α0 = =
 K   K K1  K KK
1 +  1+  +  2+ + 
 1 + 1+ + 1 + 22
 [H ]   [H ] [H ]  [H ] [H ]

By the same process, you can derive α1 and α2. You should do this for yourself.
81
 1
α1 =
[H ]
+
K
+ 1 + 2+
K1 [H ]
1
α2 = + 2
[H ] [H + ]
+ +1
K1K 2 K2

As before,
[H2A] = α0CT [HA-] = α1CT [A2-] = α2CT

The α vs. pH graph will look like two monoprotic acid graphs plotted together.
This isn’t exactly correct, though, because the HA- concentration is influenced by
both reactions. The distribution diagram for a hypothetical diprotic acid is shown
below:

Figure 7-6. Alpha diagram for a general diprotic acid

As with monoprotic acids, the lines cross when pH = pK. Looking at pH 5, it

appears that α0 = α1 and both equal 0.5. This is true for monoprotic acids, but it
isn’t strictly true here because of the existence of α3. As you can see, however,
α3 is so small that it can be neglected. In fact, neglecting one or another of the

82
α’s, depending on the pH, is a great way to simplify a problem. The math gets a
lot easier if you can treat a diprotic acid like a monoprotic acid.

Example: Carbonate System

When CO2 is dissolved into water it forms diprotic carbonic acid, H2CO3.
Carbonic acid dissociates into hydrogen and bicarbonate ions as follows:

H2CO3  H+ + HCO3-

[ H + ][ HCO3− ]
K a1 = = 5.0 × 10− 7 pK1 = 6.3 (T = 25°C )
[ H 2CO3 ]

H2CO3 = carbonic acid which results from dissolving CO2 in water

HCO3- = bicarbonate ion

(Aside: Many authors like to write H2CO3 as H2CO3*. Differentiating between

true H2CO3 molecules and dissolved CO2 gas is analytically almost impossible.
Adding the asterisk is a reminder that the molar concentration represented by the
symbol [H2CO3*] includes both true H2CO3 and dissolved CO2 molecules. This
isn't an important distinction, however, because for all practical purposes,
dissolved CO2 acts as if it has been converted to true H2CO3 molecules.)

A bicarbonate ion can further dissociate into a carbonate ion and H+ as follows:

HCO3-  H+ + CO32-

[ H + ][CO32 − ]
Ka2 = = 5.0 × 10−11 pK 2 = 10.3 (T = 25°C )
[ HCO3 ]
−

The distribution diagram is shown below:

83
 Figure 7-7. Alpha diagram for carbonic acid (H2CO3)

Because CO2 is everywhere, the carbonate chemical system is at work in virtually

all natural waters (and your bloodstream). In the pH range of most natural water
(7-9), the predominant form of carbonate is bicarbonate ion.

Practice problems

7-7. If the pH is 7, what is the dominant carbonate-containing chemical specie,

and what fraction of CT is it? [bicarbonate, HCO3- and α1 = 0.817]

7-8. At what pH will the bicarbonate and carbonate be equal? [10.3]

CARBONATE BUFFERING

"Buffering" describes that phenomenon in which pH changes only a little when a

large quantity of acid or base is added to a water sample. Although H+ (from the
acid) is added, the concentration of acid in the water (measured by pH) doesn't
change proportionately. It stays relatively constant. (This would be like putting
money into your bank account without the balance changing.) In natural waters,
the mechanisms chiefly responsible for this phenomenon are part of the
carbonate chemistry system.

84
In buffering, most of the H+ added with the acid reacts with HCO3- already in the
water to make H2CO3. (The H+ would also react with any CO32- present to make
HCO3- but in the pH range of most natural wasters the amount of CO32- is very
small compared to the concentration of HCO3-.)

CO32- + H+  HCO3-
HCO3- + H+  H2CO3

In this way, the bicarbonate and carbonate ions remove most of the added H+
from solution and the concentration of H+ in the water doesn't change much,
meaning the pH stays relatively constant. This is the buffering effect.

A similar thing happens when a base is added to the water. Without

MZ
buffering, the OH- combines with H+, the H+ concentration drops, and the Sec 3.7.3
pH increases. When bicarbonate is present, the OH- reacts with the
HCO3- rather than the H+ and the pH doesn't change much.

Practice problem

7-9. Write out the carbonate reactions when a base is added instead of an acid.
Does the pH change much?
[HCO3- + OH-  CO32- + H2O and H2CO3 + OH-  HCO3- + H2O
pH does not change much because most of the OH- are consumed by the
carbonate compounds just like the H+ are consumed when an acid is added. The
system is buffered.]

Alkalinity

In the system described above, pH stays relatively constant until all of the
bicarbonate and carbonate ions are used up (i.e., reacted with H+ or OH-). When
that happens, the pH changes fairly quickly with the further addition of acid or
base.

A measure of the amount of carbonate-containing chemical species and the

potential buffering capability of a water sample is the alkalinity. Alkalinity is
usually defined as the ability of the water to neutralize acid. This is because it is
measured in the lab by adding acid to a water sample until the pH makes a
sudden change. The alkalinity test is a titration in which acid is added to a water
sample until the pH reaches about pH 4.5. Look at the carbonate distribution
diagram at this pH (previous page). The predominant chemical specie at pH 4.5
is H2CO3. Virtually all of the HCO3- and CO32- have been converted to H2CO3
85
(α0 = 0.99) by the addition of H+. Further addition of H+ will cause the H+
concentration to increase proportionately. In other words, there is no buffering
below this pH.

Mathematically, alkalinity can be calculated from the molar concentrations of the

species that react with H+. At the end of the titration:

{H+ sources} = {H+ sinks}

The { } nomenclature refers to normality. So, at the end of the titration, the
number of equivalents of H+ from “sources” equals the number from “sinks”.
This can be expanded as follows:

{H+} + {H+ added in acid} = {HCO3-} + {CO32-} + {OH-}

{H+ added} = Alkalinity = {HCO3-} + {CO32-} + {OH-} - {H+}

Often the alkalinity equation is written as:

Alkalinity (eq/L) = [HCO3-] + 2[CO32-] + [OH-] - [H+]

You should be suspicious of this because the units on each side of the equality
don’t match. Remember, the [ ] nomenclature means mol/L.) The technically
correct way to write this formula is:

 eq   1 eq  −  2 eq  2−  1 eq   1 eq  +
 [ HCO3 ] +   [CO3 ] +   [OH ] −   [H ]
−
Alkalinity   = 
 L   mol   mol   mol   mol 

Here it is clear that the 1's and 2 are the z values.

Sometimes the H+ factor is left out of the alkalinity formula altogether. This isn’t
too bad an approximation because in practice [H+] is almost always very small
compared to the other terms in the equation.

Relationship Between CT and Alkalinity

Alkalinity and the total carbonate concentration (CT) look similar, but they aren’t
quite the same.

Alk (eq/L) = [HCO3-] + 2[CO32-] + [OH-] – [H+]

CT = [H2CO3] + [HCO3-] + [CO32-]

86
If you know the pH and the alkalinity, you can solve for CT because …

Alk (eq/L) = (1eq/L)α1CT + (2eq/L)α2CT + (1eq/L)[OH-] – (1eq/L)[H+]

The values for α1, α2, [OH-] and [H+] can be calculated from the pH using
equations discussed earlier in this chapter.

For historical reasons, in practice alkalinity is most often expressed in units of

"mg/L as CaCO3" instead of eq/L. This unit is just another way of expressing
normality. The conversion between eq/L and mg/L as CaCO3 is:

100 g of CaCO3 per mole

1 eq = = 50 g CaCO3 = 50,000 mg CaCO3
2 eq of CaCO3 per mole

The “2” comes from the fact that 2 H+ ions can react with each CO3-. It is easiest
to think of this as a simple conversion factor: 50,000 mg CaCO3 per eq

Practice problems

7-10. If the initial pH is 7.5 and the total carbonate concentration (CT) is 3×10-3
M, what is the alkalinity in meq/L and mg/L as CaCO3? [2.81 meq/L and 140
mg/L as CaCO3]

7-11. If the initial pH is 8.5 and the alkalinity is 120 mg/L as CaCO3, what is the
total carbonate concentration (CT)? [2.38×10-3 M]

Titration curves

Experimentally alkalinity is determined by titrating a water sample

Actually Standard Methods suggests a
to pH 4.5. Thus, you can read alkalinity directly from a titration range of ending values -- pH 4.9 for
curve like the one shown below. For that curve, the alklainuity is low alkalinity samples (< 50 mg/L as
about 2.6 meq/L. CaCO3) and 4.3 for high alkalinity
samples (> 300 mg/L as CaCO3). For
this class, 4.5 is a good average target
Titration curves also can tell you how much acid or base you need value.
to add to move a sample’s pH to some desired value other than 4.5.
For instance, in the curve shown below, it requires about 0.6 meq/L of acid to
move the pH to 7 from a starting value of about 8. You can use this value to
calculate how much acid (or base) you need to add to a given volume of water.
The basic equation is an equality of equivalents.

87
 (# eq needed in sample) = (# eq from the acid)

(NV)sample = (NV)acid

In this equation N is normality and V is volume. If they are flows, you can
substitute Q (volumetric flowrate) for V.

Figure 7-8. Titration curve

Example problems

7-12. Suppose you are on the ASCE water treatment team and you need to move
the pH from the test sample’s starting value down to 7.0. The titration curve you
generated in the lab shows that 0.75 meq/L of acid are needed to do this.
Calculate the volume of 3.0 N acid needed to make this change in the 10-gallon
competition sample. [about 9.5 mL]

7-13. A small manufacturing company wants to locate in Elk Grove and connect
to the city water supply. For one part of the fabrication process the tap water
needs to be reduced to pH 5.5. A titration test with the tap water shows that 4
meq/L is needed to do this. The plant manager plans to feed 2.0 N acid into the
water line. If the desired flow is 20 gpm, what flowrate of acid is needed? [0.04
gpm]

88
ACIDS, BASES AND THEIR SALTS

Adding some salts to water changes the pH while adding others does not. Why
might that be? The answer to this question involves Le Chatelier’s Principle and
knowledge of strong and weak acids. For environmental engineering, there are
applications in disinfection and coagulation.

Salts of Acids -- A salt is anionic compound comprised of a positively charged

cation and a negatively-charged anion. Adding a soluble salt to water can be
represented by:

𝐶𝐶𝐶𝐶 → 𝐶𝐶 + + 𝐴𝐴−

where C+ is the cation and A- is the anion. The dissolved cations and anions are
free to participate in other reactions. One that we’ve discussed in this class is the
acid dissociation reaction:

𝐻𝐻𝐻𝐻  𝐻𝐻 + + 𝐴𝐴−

Applying Le Chatelier’s Principle, we would say that increasing the

concentration of A- constitutes a disturbance that the system “opposes” by
reacting to reduce the A- concentration. From a kinetics point of view, adding A-
increases the rate of the reverse (to the left) reaction so that there is a net
production of HA and a corresponding net reduction of A- and H+. Of course, if
the H+ concentration declines, the pH goes up.

A concrete example of this is sodium hypochlorite (NaOCl). Sodium

hypochlorite is the chemical that makes household bleach (look at the bottle).
Hypochlorite is also a disinfectant. Home swimming pool owners often use
sodium hypochlorite in this way. The two relevant reactions are:

𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 → 𝑁𝑁𝑁𝑁+ + 𝑂𝑂𝑂𝑂𝑂𝑂 − (dissolution)

𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻  𝐻𝐻 + + 𝑂𝑂𝑂𝑂𝑂𝑂 − (dissociation)

According to the upper reaction, adding NaOCl causes the OCl- concentration to
increase, which then causes the dissociation reaction (the lower one) to shift to
the left. This produces more HOCl and reduces the OCl- and H+ concentrations.
Consequently, the pH goes up (because the H+ has decreased). At higher pH
values, most of the disinfectant is in the OCl- form, which isn’t desirable because
HOCl is a better disinfectant than OCl-. To lower the pH, a pool owner will add
acid.

Why doesn’t this happen when we add sodium chloride (NaCl) to water? The

89
reactions are similar.

𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 → 𝑁𝑁𝑁𝑁+ + 𝐶𝐶𝐶𝐶 − (dissolution)

𝐻𝐻𝐻𝐻𝐻𝐻  𝐻𝐻 + + 𝐶𝐶𝐶𝐶 − (dissociation)

Look at the acid. It is hydrochloric acid, which we know is a strong acid and
virtually no undissociated HCl is formed. Therefore, the Na+ and Cl- stay in
solution and the pH doesn’t change.

So, this leads to a general rule: Solutions of the salts of weak acids cause pH to
increase (basic solutions) while the salts of strong acids do not change the pH.

Salts of Bases -- You can apply the same kind of analysis to salts of weak bases.
The dissociation reaction in this case is:

𝐶𝐶𝐶𝐶𝐶𝐶  𝐶𝐶 + + 𝑂𝑂𝑂𝑂 −

The cation (C+) is often a metal such as aluminum or iron. Applying Le

Chatelier’s Principle, we would say that increasing the concentration of C+
constitutes a disturbance that the system “opposes” by reacting to reduce the C+
concentration. From a kinetics point of view, adding C+ increases the rate of the
reverse (to the left) reaction so that there is a net production of COH and a
corresponding net reduction of C+ and OH-. Of course, if the OH- concentration
declines, the H+ concentration increases ([H+][OH-]=Kw), and so the pH goes
down.

The civil engineering application of interest is aluminum sulfate, alum

(Al2(SO4)3). This is a commonly used coagulant in drinking water treatment.
The reactions are:

𝐴𝐴𝐴𝐴2 (𝑆𝑆𝑆𝑆4 )3 → 2𝐴𝐴𝑙𝑙 3+ + 3𝑆𝑆02−

4 (𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑)
3+ − (𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑)
𝐴𝐴𝑙𝑙𝑛𝑛 (𝑂𝑂𝑂𝑂)3𝑛𝑛  𝑛𝑛𝑛𝑛𝑙𝑙 + 3𝑛𝑛(𝑂𝑂𝑂𝑂)

During coagulation, large polymeric aluminum molecules, represented by

Aln(OH)3n, are formed. Increasing the concentration of Al3+ (upper reaction)
increases the rate of the reverse (to the left) process in the bottom reaction so that
there is a net production of Aln(OH)3n and a net reduction of Al3+ and OH-. As
the OH- concentration declines, the H+ concentration increases, and the pH goes
down. The polymeric aluminum hydroxide precipitates to form the floc. It’s
interesting to note that if the pH gets too low, the precipitates don’t form well,
and coagulation/flocculation is not effective. To keep the pH from falling too
low, the water must be well-buffered (remember alkalinity?). Water treatment
90
plant operators monitor alkalinity to track the buffering capacity of water being
treated with alum. If it is too low, they can add alkalinity in the form of lime,
Ca(OH)2.

In contrast, adding sodium sulfate (Na2SO4) does not affect pH.

𝑁𝑁𝑎𝑎2 𝑆𝑆𝑂𝑂4 → 2𝑁𝑁𝑁𝑁 + + 𝑆𝑆02−

4
𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁  𝑁𝑁𝑁𝑁+ + 𝑂𝑂𝑂𝑂 −

Here the sodium hydroxide (NaOH) is a strong base and virtually no

undissociated NaOH is formed, so there is no change in Na+ and OH-
concentrations and no change in pH.

So this leads to another general rule: Solutions of the salts of weak bases cause
the pH to decrease (acidic solutions) while the salts of strong bases do not
change the pH.

Practice Problem

7-14. For each of the following, decide whether the pH will go up, down, or stay
constant when the salt is added to water. Cover the right-hand column until
you’ve decided, then check your answer.
Salt Acid or base, strong or weak pH change
Sodium nitrate, NaNO3 Nitric acid, HNO3, strong None
Sodium acetate, NaCH3COO Acetic acid (vinegar), Up
CH3COOH, weak
Calcium sulfate (gypsum), CaSO4 Lime, Ca(OH)2, strong None
Ferrous sulfate (coagulant), FeSO4 Fe(OH)2, weak Down(1)
Calcium carbonate (limestone), Bicarbonate ion, HCO3-, weak Up
CaCO3
Aluminum sulfate (alum, Al(OH)3, weak Down(1)
coagulant), Al2(SO4)3
Sodium cyanide (used in gold Hydrogen cyanide, HCN, Up
mining), NaCN weak
Ammonium nitrate (fertilizer), Ammonium ion, NH4+, weak Down(2)
NH4NO3
(1) Iron and aluminum combine with hydroxides which lowers the pH.
(2) This one’s tricky. Salts of nitrate don’t affect pH, but here you are also adding NH4+ which is a
weak acid that will lower the pH.

Reference:
https://chem.libretexts.org/Textbook_Maps/Physical_and_Theoretical_Chemistry_Textbook_Maps/
Supplemental_Modules_(Physical_and_Theoretical_Chemistry)/Acids_and_Bases/Acids_and_Bas
es_in_Aqueous_Solutions/Aqueous_Solutions_Of_Salt

91
 Practice Problem Hints

Chap 7 Acid-Base Chemistry

1 Hint: pH+pOH=14 and pOH= -log[OH-].
2 --
3 --
4 Hint: A strong acid gives up all its H+. At this concentration
virtually all the H+ are from the HNO3 rather than from water with
its wimpy Kw of 10-14 (water is a weak acid).
5 Hint: Rearrange the α0 equation to get [H+] = K/(α0-1 -1) =
1.15×10-7
6 Hint: Rearrange the α1 equation to get: [H+] = Ka (α1-1 -1) =
2.84×10-9
7 Hint: The graph tells you the regions in which different forms
dominate. The equation tells you α1.
8 Hint: α’s are 0.5 when pH = pK.
9 --
10 Hint: α1 = 0.933, α2 = 0.001, [OH-] = 3.16×10-7 M
11 Hint: alk = 2.4×10-3 eq/L, α1 = 0.979, α2 = 0.014, [OH-] = 3.16×10-
6
M
12 Hint: 0.75 meq of acid is needed in a 10-gal sample. #eq =
2.84×10-2
13 Hint: use the same NV=NV logic you used before except now it is
NQ=NQ because the problem is based on flows rather than
volumes. #eq/min = 0.303
14 --

92
8. Mineral Solubility
This topic is covered reasonably well in your textbook and few
MZ
additional notes are needed. A couple of comments may be helpful, Sec 3.7.3
though.

Nomenclature

Consider the dissolution of a solid salt.

AaBb(s)  aAb+ + bBa-

For this “reaction”, the equilibrium equation would be:

[𝐴𝐴𝑏𝑏+ ]𝑎𝑎 [𝐵𝐵𝑎𝑎− ]𝑏𝑏

𝐾𝐾 =
[𝐴𝐴𝑎𝑎 𝐵𝐵𝑏𝑏 (𝑠𝑠)]

The chemical activity of a solid phase material in contact with a solution is 1.0,
so the denominator disappears, and the equilibrium constant is called the
solubility product.

[𝐴𝐴𝑏𝑏+ ]𝑎𝑎 [𝐵𝐵𝑎𝑎− ]𝑏𝑏 = 𝐾𝐾𝑠𝑠𝑠𝑠 when the system is at equilibrium

We know that not all solutions are in equilibrium, which means that the equality
in the equation above does not always hold true. For convenience, chemists have
a couple of names for the left side of the equation above.

[𝐴𝐴𝑏𝑏+ ]𝑎𝑎 [𝐵𝐵𝑎𝑎− ]𝑏𝑏 𝑖𝑖𝑖𝑖 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑡𝑡ℎ𝑒𝑒 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 (𝑄𝑄)

Example:
Al2(SO4)3  2Al3+ + 3(SO42-) Q = [Al3+]2 [SO42-]3

Note that Q is not a constant like Ksp is. When calculating Q use the actual
concentrations of the ions in solution. Then, compare the value of Q with the Ksp
value to decide whether a solution is in equilibrium.

• For saturated solutions (equilibrium), Q = Ksp.

• For unsaturated solutions, Q < Ksp.
• For supersaturated solutions, Q > Ksp.

Another term you might encounter is the saturation index (Ω).

93
 𝑄𝑄
Ω=
𝐾𝐾𝑠𝑠𝑠𝑠

• For saturated solutions (equilibrium), Ω = 1.

• For unsaturated solutions, Ω < 1.
• For supersaturated solutions, Ω > 1.

Temperature -- Ksp can be obtained from a table (or online) and is constant for a
given compound and temperature. Generally speaking (there are exceptions), the
solubility of salts increases as temperature increases (i.e., the opposite of gases).
To get more salt to dissolve, you heat the solution. If you cool down a warm
saturated solution, it will become supersaturated and could produce a precipitate.

Applications -- There are several reasons environmental (and civil) engineers are
interested in this topic. Unsaturated solutions (e.g., rain or rivers) can dissolve
engineering materials over time. In drinking water treatment processes,
operators deliberately manipulate Q to cause precipitation of an undesirable
solute through the common ion effect. Here are some typical applications.

• Removal of metals from treated water – Metal hydroxides and

carbonates (e.g., Cu(OH)2 or CuCO3) tend to be relatively insoluble,
so the metals (which are toxic to fish) can be removed from solution
by precipitation.

• Removal of hardness – Water “hardness” is the sum of the Ca2+ and

Mg2+ ions. These naturally-occurring minerals can precipitate out as
CaCO3 and Mg(OH)2 in drinking water systems and boilers. The
precipitate is called “scale”. To keep this from happening at the
customer’s end, engineers can cause it to occur in treatment plants
instead.

• Seashells and coral reefs – Sea creatures need excess carbonate to

make their CaCO3 shells. As the ocean’s pH drops due to more CO2
dissolving into the water from the atmosphere, there is concern that
CO32- is being converted to HCO3- which will make it more difficult
for shellfish and coral to make their shells.

• Service life of concrete structures − Water that contains few

dissolved minerals will dissolve those minerals out of solids that it
contacts. In other words, if the water is unsaturated, concrete pipes,
tanks (see photo), piers, canals, and dams will dissolve.

94
Figure 8-1. Settling tank at the
Fairbairn Water Treatment Plant. The
aggregate below the water line is
exposed because the cement in the
outer layer has dissolved over the 40+
years that the plant has been in
operation. (Johnston)

A final word – All the discussion about hydroxides and carbonates should alert
you to the fact that there is significant interaction between the acid-base-pH
materials in the previous section and mineral solubility.

Practice problems

8-1. Imagine you have a 1-L solution with 2×10-5 moles of Ca2+ and 3×10-5 mol
CO32-. To this, you add 1×10-3 mol of sodium carbonate, Na2CO3. Assuming the
Na2CO3 dissolves completely, would you expect to see any limestone formation
(i.e.,precipitation of CaCO3)? Ksp for CaCO3 is 4.96×10-9 at 25°C. [Yes, the Ksp
value is exceeded]

8-2. Suppose the Cu2+ concentration in a solution is 1×10-7 M. How high would
you have to raise the pH to get Cu(OH)2 to precipitate? Ksp for Cu(OH)2 is
2.2×10-20. [7.7]

“Solubility Product” is not the same as “Solubility”

The solubility of a material is the maximum mass of solid (solute) that will
dissolve in a given volume (or mass) of liquid (solvent). It is expressed as g/100
mL or g/L or g/100 g of solvent. When the maximum amount of solute has been
dissolved at a particular temperature, the solution is saturated. In contrast, the
solubility product is a number equal to the ion product (Q) when the system is at
equilibrium, i.e., the equilibrium constant.

95
Calculating solubility uses the same math as the other solubility problems but
with one twist. In a solubility test, you typically test only the compound of
interest. Therefore, you know the stoichiometric relationship between the ions
and can use that in the Ksp equation.

For example, suppose you want to calculate the solubility (in mg/L) of calcium
fluoride, CaF2. You run a lab test in which you measure how much CaF2 can be
dissolved in a liter of water. Here’s the reaction and the equilibrium equation:

CaF2 = Ca2+ + 2F- Q = [Ca2+] [F-]2 = Ksp

Because the CaF2 is the only source of Ca2+ and F-, for every mole of CaF2 that
dissolves, you get one mole of Ca2+ and 2 moles of F-. You can plug these
expressions into the Ksp equation and solve.

Assume T=25° C and the corresponding Ksp is 1.46×10-10. The unknown number
of moles of CaF2 is designated as “x”.

𝐾𝐾𝑠𝑠𝑠𝑠 = [𝐶𝐶𝐶𝐶2+ ][𝐹𝐹 − ]2 = (𝑥𝑥 )(2𝑥𝑥 )2 = 4𝑥𝑥 3

1
𝐾𝐾𝑠𝑠𝑠𝑠 3 10−4 𝑚𝑚𝑚𝑚𝑚𝑚
𝑥𝑥 = � � = 3.32 ×
4 𝐿𝐿
𝑚𝑚𝑚𝑚𝑚𝑚 𝑔𝑔 𝑔𝑔 𝑚𝑚𝑚𝑚
�3.32 × 10−4 � �78 � = 0.026 = 26
𝐿𝐿 𝑚𝑚𝑚𝑚𝑚𝑚 𝐿𝐿 𝐿𝐿

Be careful. That there are 2 F’s for every Ca is true only in the special case
where the ions come exclusively from the salt added to the water, as happens in a
laboratory solubility test. This is not the general case. In normal water samples
there are multiple sources of Ca2+ and F-, so their relationship is cannot be
assumed to be 2:1.

In problems where you are not doing a solubility test, you do not know the
stoichiometric ratio between ions. So, write your ionic product equation in
general terms with no assumed relationships between A and B, like so:

Q = [Ab+]a [Ba-]b which could be >, =, or < Ksp

96
 Practice Problem Hints

Chap 8 Mineral Solubility

1 Hint: After the addition the total [CO32-] is 1.03×10-3 M. The
product of the concentrations is 2.06×10-8 > 4.96×10-9
2 Hint: Based on the Ksp equation for equilibrium [OH-] = 4.7×10-7 M
(don’t forget the square).

97
9. Coagulation/Flocculation

As you recall from Stokes Law, the settling velocity of a particle depends heavily
on its size.

𝑔𝑔�𝜌𝜌𝑝𝑝 − 𝜌𝜌𝑓𝑓 �𝑑𝑑𝑝𝑝2

𝑣𝑣 =
18𝜇𝜇

v = settling velocity
George Gabriel Stokes
g = gravitational constant
(1819-1903)
ρp, ρf = density of the particle and the fluid
dp = diameter of the particle
µ = viscosity of the fluid

Very small particles (colloids) don’t settle fast enough to make removal by
sedimentation practical. Building settling tanks with detention times of days or
weeks, however, would be very expensive. One strategy to increase settling rates
is to cause small particles to clump together in larger structures called “flocs”.
Because the flocs are relatively large, they can be settled in shorter times,
meaning smaller tanks.

In most natural waters, colloidal particles won’t clump together, even though
there is an attractive force known as van der Waals force. This is a molecular
cohesion force that acts over a short distance. If two particles
MZ
collide, these forces will cause them to adhere to each other. This material is covered
Unfortunately, particles are prevented from approaching each in Section 8.5*.
other by electrical repulsion forces arising from like charges
(usually negative) on the surfaces of the particles (see figure below). These
charges can arise from several mechanisms such as lattice imperfections in clay
minerals, adsorption of ions, and particle dissolution of surface layer minerals.
In any case, the surface charges attract oppositely charged ions in a loose cloud
that surround the particles and keep them from approaching each other close
enough for the van der Waals forces to work. A suspension of colloids that don’t
aggregate naturally is called a stable suspension. We want to create a
destabilized suspension so that the particles will aggregate and settle.

98
 Figure 9-1. Particles repel each other because of their surface charges (source:
Viessman and Hammer)

In coagulation, we use chemicals to either: (1) reduce the particles’ surface

charges, (2) physically bind the particles to each other using large molecules
called polymers, or (3) form precipitates (solids) which sweep the small particles
out of the water as the precipitates settle. Mixing is an important aspect of this
process. A rapid mix is needed initially to bring the chemical into contact with
the all particles. Afterwards, a slow mix is needed to cause the particles to
contact each other and form flocs. The slow mixing process is usually called
flocculation.

Design and operation of coagulation/flocculation processes (the words are joined

this way because in practice we rarely use one without the other) depend on jar
testing. The success of chemical processes depends on many factors such as the
nature of the particles (e.g., organic or mineral), the number and size distribution
of particles, temperature, pH, and the presence of dissolved substances that
interfere with precipitation. Rather than measure all these parameters
continuously, some of which can change relatively quickly with time, it is more
practical to perform a miniature process in a jar and scale up the results.
Similarly, providing the proper amount of mixing is not easy to predict. If too
little mixing is provided, there will be insufficient contact between particles to
build large flocs. If too much mixing is provided, the hydrodynamic forces in the
water will break the flocs up. How much is “too much” depends on the strength
of the flocs, which in turn depends on the chemistry and interactions between the
particles and the coagulants (i.e., the chemistry). Again, doing the mixing in a jar
test is more practical than trying to predict the best mixing from measurements
and models in real time.

In this lab, we will perform a typical jar test for choosing a coagulant dose and a
mixing intensity, measured by a parameter called the “velocity gradient”.

99
COAGULATION

There are many coagulants on the market, most with proprietary formulations.
Three broad categories include aluminum-based chemicals, iron-based
chemicals, and organic polymers. The aluminum and iron chemicals form
polymer-like precipitates. In this lab we will use nonproprietary aluminum
sulfate, also known as alum. Our alum crystals include 14 waters of hydration,
so the chemical formula is:

Al2(SO4)3⋅14H2O (MW = 594)

When alum is added to water we get many different compounds, depending on

the pH and other chemical parameters, including alkalinity as shown in the
reaction below. If enough alum is added, you get an Al(OH)3 precipitate.

Al2(SO4)3⋅14H2O + 6HCO3- → 2Al(OH)3↓ + 14H2O + 3SO42- + 6CO2

The waters of hydration don’t really participate in the reaction. The downward
pointing arrow means the aluminum hydroxide is a solid precipitate that settles.
As this precipitate settles, it collides with other particles which stick to the
precipitate and settle also. This phenomenon is called sweep floc because the
precipitate “sweeps” the colloidal particles out of the water column as they sink.
Notice also that alkalinity (shown as HCO3-) is consumed in this reaction. In
fact, the alkalinity drops by about 0.5 mg as CaCO3 for every mg of alum added.
If the water’s alkalinity is too low, the precipitate will not form.

The reaction shown above is a simplified representation of what really happens.

Crystalline Al(OH)3 does not form. Instead, a variety of hydrated aluminum
compounds result of the general form: (Alm(OH)n)+/-. The subscripts m and n can
vary and the resulting charge can be positive or negative, depending on the
relative values of m and n. Some of these molecules can adsorb to the surfaces
of some particles, neutralizing their charges. Other hydrated compounds form
charged polymer-like molecules that interact with the particles. These polymers
are long, string-like molecules that can neutralize the surface charges of the
particles and/or physically tie them together to make flocs large enough to settle
(see following figure).

100
Figure 9-2. Particles bound together by polymers (either Al-based, Fe-based, or
organic) can grow large enough to settle (source: Viessman and Hammer)

FLOCCULATION AND MIXING

Coagulation is the means of destabilizing the colloids (i.e., making them

chemically “sticky”). Flocculation is the process of causing the particles to bump
into each other and thereby growing flocs.

Velocity Gradient and Power Equations

Velocity gradient is a measure of the intensity of mixing in a system. The name

derives from the idea that a well-mixed system (one with a lot of turbulence) has
large changes of fluid velocity with respect to distance (the derivative of the
velocity with respect to distance or gradient). In contrast, in ideal laminar flow,
where there is no mixing, the adjacent streamlines are moving at identical speeds
and directions, so the velocity gradient is zero.

Practically, the velocity gradient is measured by the power input per unit volume
(P/V) divided by the viscosity, raised to the 0.5 power.

1/ 2
 P 
G =  
 µV 

where:
G = velocity gradient (s-1)
P = power dissipated in the water (W or ft-lb/s)
µ = viscosity (N-s/m2 or lb-s/ft2)
V = volume (m3 or ft3)

101
Think about this equation for a moment. Maybe picture a blender in your mind.
As you increase the power (P↑), the mixing intensity increases. If you put the
same power into a smaller volume (V↓), the mixing intensity also goes up. If
you put the same power into a volume of something thick and gooey like yogurt
or syrup (µ↑), you get less intense mixing than you do with water.

You can use several methods to put mixing power into water. You can bubble
gas through it. You can run the water through a hydraulic jump or down a rough
surface. You can also use propellers and paddles.

Different power equations are needed for different types of mixers. They are all
based on multiplying the drag force on the paddle by the angular velocity. For
horizontal shaft mixers like that shown in the photo to the right, the equations on
page 213 can be used.

Figure 9-3. Horizontal mixers at the

Fairbairn WTP (Johnston)

In a typical jar test apparatus, the paddles are vertical-shaft, solid blades (see
photo below). For this device, the power equation is:
K T n 3 D 5γ
P=
g
where:
P = power (ft-lb/s)
n = rotational speed (rev/s, not rpm)
D = diameter of paddle (ft)
γ = specific weight of water (lb/ft3)
g = acceleration due to gravity (32.17 ft/s2)
KT = empirical constant = 2.4 for Phipps and Bird jar test mixer used in this lab
(KT was calculated from data provided by Phipps and Bird.)

102
Because of the empirical constant KT, the power calculations must be done in
U.S. units.
To convert the power to SI units, use: 1.3558 W per ft-lb/s.

Figure 9-4. Jar test apparatus (Johnston)

What’s the right G value? If you don’t provide enough mixing (a low G value),
the particles don’t contact each other frequently enough, which limits floc
growth. If you provide mixing that is too intense (a high G value), the
hydrodynamic shear forces tear the floc up. One beneficial side effect of high G
value is that the floc that survive, although small, are dense and therefore settle
quickly for their size. To take advantage of this, some water plants use tapered
flocculation in which the G value is tapered from a relatively high to a relatively
low value. In that configuration, dense floc created in the upstream mixing
stages become nuclei for growing large floc in downstream stages.

One thing to keep in mind is that the power calculated from these equations is the
power delivered to the water by the flocculation device. It is not the power that
must be purchased from the local utility. To determine the power requirement
for the process, the mechanical efficiency must be considered.

Power to water = (mechanical efficiency, η) × (Power from the utility)

In large treatment plants, jar tests are performed regularly so that operators can
adjust the coagulant dose and flocculator speed. If the source water is a river
whose characteristics are changing rapidly (such as in a rainstorm), jar testing
might be done as often as hourly.

103
Practice Problems

9-1. For a 2-L jar test apparatus with a 2-inch impeller spinning at 20 rpm and a
20°C water temperature ….
a. Calculate the power. [2.23×10-5 ft·lb/s (fps)]
b. Calculate the velocity gradient. [3.9 s-1]

9-2. For a 25’x75’x10’ deep flocculation tank, calculate the hydraulic and
electrical power (in kW) required to provide a velocity gradient of 30 s-1. The
temperature is 50°F and the mechanical efficiency of the mixing equipment is
70%. [mixing power = 461 ft·lb/s; electrical power = 0.892 kW]

References:

McCabe, W. L., J. C. Smith, and P. Harriott, Unit Operations of Chemical Engineering,

5th ed., McGraw-Hill, 1993.
Rushton, J. H., "Mixing of Liquids in Chemical Processing", Industrial and Engineering
Chemistry, American Chemical Society, v 44, n 2, p 2931, Dec 1952.
Rushton, J. H. and J.Y. Oldshue, "Mixing -- Present Theory and Practice", Chemical
Engineering Progress, v 46, n 4, p 161, April 1953.
Reynolds, T. D. and P.A. Richards, Unit Operations and Processes in Environmental
Engineering, PWS Publishing Company, 1996.
Wagner, E.G., "Jar Test Instructions, Conduct of Jar Tests and the Important Information
Obtained", booklet from Phipps and Bird, 8741 Landmark Rd., Richmond, VA 23228,
July 1993.
Viessman, W., Jr. and M.J. Hammer, Water Supply and Pollution Control, 6th ed.,
Addison-Wesley, 1998.
http://www.marcsteinmetz.com/pages/wasser/ewasser_minis.html
http://site.iugaza.edu.ps/frabah/files/2010/02/Water_Treatment_Lecture_2.pdf
http://iwawaterwiki.org/xwiki/bin/view/Articles/CoagulationandFlocculationinWaterand
WastewaterTreatment#HFlocculation

104
 Practice Problem Hints

Chap 9 Coagulation
1 Hint (a): n = 0.333 rps; D = 0.167 ft; γ = 62.31 lb/ft3 (interpolate
from the table in chapter 2 or use the water properties spreadsheet
provided online).
Hint (b): 1.3558 W per ft·lb/s. Watch your units.
2 Hint: µ = 2.73×10-5 lb·s/ft2 (interpolate from the table in chapter 2
or use the water properties spreadsheet provided online); 1.3558 W
per ft·lb/s; 1.3558 W per ft·lb/s; efficiency is the fraction of
electrical power delivered as mechanical mixing power.

105
10. Biochemical Oxygen Demand

All natural environments teem with microbial life. A typical agricultural soil can
contain as many as a million bacteria per gram! For these microorganisms,
organic material is food. Like virtually all higher forms of life (i.e., you and me),
many of these microbes use oxygen to metabolize their food. We call this
aerobic metabolism. As microbes degrade organic substances, they pull oxygen
from the local environment in direct proportion to the amount of organic
material metabolized. For example, glucose can be converted to CO2 and H2O
according to the following reaction:

C6H12O6 + 6O2  6CO2 + 6H20

This reaction could represent combustion, or it could represent microbes

consuming the glucose. In either case, the important thing to note is that 6 moles
of oxygen are required for every mole of glucose consumed. A stoichiometric
relationship like this is called the theoretical oxygen demand
(ThOD). MZ
This material is covered
in Section 5.4*.
There's lots of oxygen in the air, but the amount dissolved in water
is very limited (recall the saturation values you measured in the gas
transfer lab). If enough organic materials are present, microbes can consume all
of the available oxygen and the fish will die. This is why engineers are interested
in measuring organic materials. Biodegradable organic pollution leads to oxygen
depletion in rivers, streams, and lakes.

Let's think about how we might measure organic material. If the material was a
single substance, such as glucose, we could chemically isolate and measure it in
mol/L or mg/L. But suppose we wanted to measure a mixture of organics. It
might be possible to isolate and measure each compound through chemical
means, but this wouldn't be practical or even particularly useful. Wastewater and
natural waters like rivers and lakes may contain hundreds of organic substances.
Analyzing for all of these compounds would be a huge task. One way around
this is to use a "lumped" parameter like Total Organic Carbon (TOC). In a
TOC machine, all of the organic substances are oxidized to CO2 which is then
measured by a CO2 sensor. This tells us the total amount of organic material, but
not the amounts of individual substances. Neither does it tell us anything about
the chemical characteristics of these substances, which can be very important to
environmental processes. For instance, both plastic and sugar are organic and
would show up in a TOC test, but the environmental effects of dumping sugar
and plastic into a river are very different. This is because sugar is biodegradable,
106
and plastic is not (at least not in a reasonable time period). Microbes consuming
sugar would pull dissolved oxygen (DO) out of the river water. Plastics, being
relatively inert, would not be consumed by the microbes and would therefore not
affect the DO.

An alternative to directly measuring biodegradable organics is to measure the

effect they cause, the depletion of oxygen. This is the strategy used in the
Biochemical Oxygen Demand (BOD) test. This test was developed in Britain
in the early part of the 20th century. It is a simplified physical model of what
happens when organic waste is added to a stream. In the test, a liquid containing
organic wastes (either full strength or diluted) is placed in
a bottle (i.e., a batch reactor). The dissolved oxygen (DO)
is measured, and the bottle is sealed. After some
incubation time (usually 5 days), the DO of the sample is
measured again. Because the bottle is sealed, the
difference in the DO values represents the oxygen used by
microbes in degrading the waste. The change in DO is
proportional to the mass of organic substances in the
bottle that were aerobically consumed during the
incubation period. Biodegradable organic materials can
be said to represent a potential demand for oxygen that
occurs when they are degraded by microbes. This is the
origin of the term “biochemical oxygen demand".

Note the importance of microbes. Organic material that is not biodegradable

does not have an oxygen demand. It is the microbial population that actually
consumes the DO. If the microbes aren’t there because the water sample is
sterilized, no DO would be consumed even if biodegradable organic material is
present.

The calculation of the BOD is based on a mass balance for a batch reactor:

(Mass of BOD) = (Mass of O2 consumed) = (Mass of O2 in bottle at t = 0) –

(Mass of O2 at t = t)

Vsample(BOD) = Vbottle(DO0) – Vbottle(DOt)

( DO0 − DOt )Vbottle ( DO0 − DOt ) DO0 − DOt

BOD = = =
Vsample (Vsample / Vbottle ) P

where DO0 and DOt = dissolved oxygen concentrations at t = 0 and t = t

P = dilution factor = (Vsample)/(Vbottle, 300 mL)

107
If the waste is strong, all the oxygen will be consumed. In this case, the DO in
the bottle will go to zero before the end of the incubation period and we won’t be
able to read the test results. To avoid this, we dilute the waste. If we over-dilute
it, the change of DO will be too small to be reliably measured. Details about
choosing the right dilution can be found in the lab procedures.

BOD calculated from the equation above is based on microbial activity taking
oxygen from the water. When this happens, we say that the biochemical oxygen
demand is being “exerted” (i.e., the potential demand is being converted into an
actual consumption of oxygen). A typical graph of DO in the BOD bottle as a
function of time might look like Figure 10-1 below. As shown, after a long time,
the bacteria consume all of the organic material, and the DO stops dropping. The
total DO drop at this point represents the "ultimate BOD" (also called BODU or
sometimes BODL for "limiting BOD"). Twenty to thirty days is a long time to
wait, so usually the bottles are opened and measured after 5 days, and the test
results are reported as the "5-day BOD" or BOD5. The story goes that 5 days was
chosen originally as the incubation time because that is the longest travel time of
any river in Britain from its headwaters to the sea. Although the choice of
incubation time is somewhat arbitrary, 5 days is still the standard incubation
time. One reason that we don’t use longer time is that after 5-10 days, ammonia
in the water sample is converted to nitrate (nitrification), which also removes
oxygen from the water. Keeping the incubation time a little short assures that the
oxygen loss we measure is due to the oxidation of organic compounds rather than
ammonia.

To convert between BOD5 and BODU, we need to know something about the
kinetics of the BOD test. Suppose we set up a number of BOD bottles on day 0,
and then periodically over the next month, we open the bottles one-by-one and
measure the DO in each. If we use these DO values in the BOD equation shown
above, we would see an increasing BOD result as shown in Figure 10-2. In other
words, the measured BOD1 < BOD2 < BOD3 < BOD10 < BOD 20 because ∆DO1 <
∆DO2 < ∆DO3 < ∆DO10 < ∆DO20.

108
 Figure 10-1. Change of oxygen as a function of time

Figure 10-2. Results of BOD tests done at different times (“BOD exerted”)

Remember, the measured BOD represents that mass of organic material

consumed by bacteria in that time period. Therefore, the measured values of
BOD start at 0 and end up at the ultimate value. When you compare Figure 10-2
with Figure 10-1, you see that Figure 10-2 is actually Figure 10-1 flipped over.

Perhaps this is a useful analogy. If 12 pizzas were ordered for an ASCE student
chapter meeting. After 10 minutes, 7 of the pizzas might be consumed and 7
empty boxes produced (see Figure 10-3). If we call the pizza demand “PD”, then
PD10 = 7 pizzas, but the total number consumed at the end of the meeting, what
we call the ultimate pizza demand (PDU), is 12 pizzas. In a BOD test, we don’t
109
know the concentration of biodegradable organics in the water, but we can
measure the amount consumed (as mg/L of oxygen) over time. The BOD test is
analogous, therefore, to counting the empty pizza boxes. PD10 = 7 means we saw
7 pizzas consumed in 10 minutes. PD20 = 10 means we saw 10 pizzas consumed
in 20 minutes. From data such as these we can use the equations discussed below
to extrapolate to the ultimate number of pizzas delivered (PDU).

Figure 10-3. Pizzas at a ASCE student chapter meeting.

While it is not always true, most often we can fit a first-order decay model to the
BOD concentration in the bottle. The equation of the line in Figure 10-2 can be
derived like so: The BOD consumed in time, t is BODt. This is the value
measured in the test. The BOD in the bottle at t=0 is the total BOD that can be
consumed. This is the ultimate BOD (BODU). The BOD remaining in the bottle
is modeled with an exponential decay equation. Therefore …

𝐵𝐵𝐵𝐵𝐵𝐵𝑡𝑡 = 𝐵𝐵𝐵𝐵𝐵𝐵𝑠𝑠𝑡𝑡𝑎𝑎𝑎𝑎𝑎𝑎 − 𝐵𝐵𝐵𝐵𝐵𝐵𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟

𝐵𝐵𝐵𝐵𝐵𝐵𝑡𝑡 = 𝐵𝐵𝐵𝐵𝐵𝐵𝑈𝑈 − 𝐵𝐵𝐵𝐵𝐵𝐵𝑢𝑢 𝑒𝑒 −𝑘𝑘𝑘𝑘
𝐵𝐵𝐵𝐵𝐵𝐵𝑡𝑡 = 𝐵𝐵𝐵𝐵𝐵𝐵𝑈𝑈 �1 − 𝑒𝑒 −𝑘𝑘𝑘𝑘 �

If you plug "5" in for "t", you can use this equation to convert between BOD5 and
BODU.

It is important that you understand the difference between BODt and BODU.
BODU is a measure of the total biodegradable organic content of the water. If
you add more biodegradable organic material to water, BODU will increase.
BODt is the amount of BOD exerted in a t-day test and is proportional to the
amount of organic material actually degraded by microbes in that time. After 20-
30 days, all the BOD in the water is exerted and the measured BODt equals
BODU. In shorter times, only some of the BOD is exerted, and some remains in
the water.

Figure 10-4 shows three water samples with three different BODU values. Notice
that the curves all “flatten out” at different levels. Notice also that they seem to
approach these levels at different rates (red fast and green slow). In the equation

110
the rate is represented by “k”. So, each BOD measurement has associated with it
a total quantity (BODU) and a rate (k). In the case illustrated in Figure 10-4, the
rates were chosen so that the three samples showed the same BOD5, even though
the BODU’s were different. Knowing the BOD5 is not enough; to estimate BODU
you need to know the degradation rate also.

Figure 10-4. Different wastes with different BODU values can show the same
BOD5 because of differing reaction rates

Because different organic substances degrade at different rates, one has to be

careful when comparing samples based on BOD5. Look at the following data.

Sample A: Urban sewage, BOD5 = 100 mg/L

Sample B: Urban sewage, BOD5 = 200 mg/L
Sample C: Industrial wastewater, BOD5 = 100 mg/L

It is reasonable to say that Sample B contains about twice as much organic

material as Sample A because typical urban sewages have similar rate constants.
On the other hand, it might be a mistake to say that Sample A and Sample C have
the same BODU (even with the same BOD5) because these organic materials are
different, and they probably degrade at different rates. What you can say,
though, is that the impact on the DO in a stream receiving Samples A and C will
be the same after 5 days, but maybe not long term.

Note that the first order model does not always fit BOD data. Don't be surprised
if in real life, you get data that doesn't plot out in such neat curves. Like any
mathematical model, use the first order equation with care.

111
Practice problems

10-1. In a 5-day BOD test, the initial DO concentration was 7.8 mg/L and the
final concentration was 2.5. If 100 mL of sample was added to the bottle, what is
the BOD5? [about 16 mg/L]

10-2. In a BOD test you don’t want to run out of DO in the bottle, so usually the
test is run with a diluted sample. Suppose past data show that the BOD5 of your
sample is in the vicinity of 185 mg/L. If the starting DO is the bottle is 8.0 mg/L
and the lowest DO allowed in the bottle is 1.0 mg/L, what volume of sample
should be used? [11.4 mL]

10-3. Calculate the lower and upper limits of BOD that can be measured validly
using 200 mL of sample in a 300-mL bottle. Remember that the minimum ∆DO
allowed is 2.0 mg/L and that the lowest allowable final DO is 1.0 mg/L. Assume
the DO of the diluted sample at time zero is 8.00 mg/L. [3.0 to 10.5 mg/L]

10-4. If the measured BOD5 is 30 mg/L and you know that the rate constant is
0.3/d, calculate the BODu. [38.6 mg/L]

10-5. The standard incubation tome is 5 days, but it could be different. If the
measured BOD3 is 30 mg/L and you know that the rate constant is 0.3/d,
calculate the BODu. [50.6 mg/L]

There is one last kinetic issue to consider. Suppose we put a water sample with
biodegradable organics and microbes into a batch reactor and supply it with
oxygen in the form of air bubbles. Now if we take a sample out each day and test
the sample in a BOD5 test, we would see that the BOD5 of the original water
declines over time as shown below:

Day 1 BOD test bottle

Day 2
BOD test bottle

BOD test bottle

Day t

112
 Figure 10-5. Organic matter (expressed as BOD) remaining in the water as a
function of time

What’s happening? We said earlier that BOD is a measure of the organic

material in the water. So, if we see declining BOD test results, we are observing
declining concentrations of organic materials due to their consumption by the
microbes. In other words, the BOD content of the water declines over time.
Organic biodegradation is often successfully modeled as a first-order reaction. If
we call the organic content of the water L, then:

dL
= − kL
dt

If we perform a mass balance on our hypothetical batch reactor and integrate the
differential (see class notes from mass balance), we get:

L = L0 e-kt

This equation is illustrated in Figure 10-5. Note that L = BOD (organic)

remaining in the water at any time. The first order model can be used to predict
organic content in complete mix and plug flow reactors just as we showed in
class. Be warned, however, that not all organics decay according to a first order
model. Don't be surprised by real-life data following a different pattern.

113
Nomenclature

The mathematics and concepts associated with BOD are not difficult. What is
confusing sometimes is the terminology. When engineers talk about the “BOD”
of a water sample, they are using that term to represent organic content. In their
minds they are thinking about the organic material (“L”) and how it is consumed
by microbes in a first-order decay process. They do this because organic material
is what causes undesirable effects in rivers and lakes, and municipal wastewater
treatment is largely focused on removing BOD. The only time you use the
equation beneath Figure 10-2 is when measuring BOD in a lab.

Is “L” BOD5 or BODU? Usually, it is
BODU. However, if the k value is
constant, then the relationship
between BOD5 and BODU is constant
and both BOD5 and BODU decay in a
similar manner.
The BOD described in this chapter is oxygen demand caused by
organic substances. Often this is called “CBOD” (carbonaceous
BOD). As you will learn in class, oxygen can also be removed from
water by bacteria converting ammonia into nitrate. This called
NBOD (nitrogenous BOD). NBOD reactions will happen in a BOD
bottle, but usually not until a number of days have passed because
the bacteria grow slowly. By stopping the BOD test at 5 days, we
avoid NBOD reactions interfering with the CBOD data.

Practice problems

10-6. If the BODu is 120 mg/L and the decay coefficient is 0.27/d, how much
BOD is remaining in the water after 10 days? [8.1 mg/L]

10-7. In a lab test, 30 mL of a waste was used in a standard 300-mL BOD bottle.
The initial DO of the mixed sample was 8.6 mg/L. After 5 days, the DO was 4.7
mg/L. After 30 days the DO was 2.4 mg/L, holding steady with time. Calculate
the BOD5, the BODu, and the BOD remaining in the water after 5 days. [39, 62,
and 23 mg/L]

10-8. If the measured BOD5 of a waste is 250 m/L and the ultimate BOD is 350
mg/L, what is the rate constant? [0.25 d-1]

10-9. A factory with a wastewater flow of 15 ft3/s (cfs) and a BOD5 of 550 mg/L
discharges into the local creek. Upstream of the discharge point, the BODu of
the creek is 2.4 mg/L. The creek’s flow is 172 cfs. The BOD rate constants (k)
are 0.30/d for the factory waste and 0.17/d in the creek. Calculate the BODU in
the water after 11 days travel downstream. [9.1 mg/L]

10-10. The wastewater treatment plant at Johnstown will utilize an activated

sludge process to remove BOD. As part of the design team your task is to
calculate the capacity of the aeration machinery that must be provided. Aeration
114
capacity is expressed in lb of oxygen delivered per day. Records at similar plants
using the bubble technique you envision show that the mechanical aeration
system must provide 8 pounds of oxygen for each pound of BOD5 removed from
the primary effluent (see figure below). Calculate the aeration system capacity
needed. [about 36,000 lb/d, though you’d want to add a safety factor in real life]

Most of the oxygen input escapes with the bubbles

Treated
Primary water to
effluent secondary
O2 transfer
from bubble
settling

Aeration tank in which organic

Air waste (BOD) is converted to CO2
and more bacteria which are then
removed in secondary settling.

Data:
 Flow = 4.0 MGD.
 Primary effluent BOD5 data: 130, 120, 140, 93, 160, 130, 150, 170, 110, 97
 To be conservative, your boss recommends that you base your calculation on
the upper limit of the 90th percentile confidence interval (C.I.) for the mean
of the primary effluent BOD.
 For design purposes, a BOD of 10 mg/L is the target value for the treated
water.
 Helpful conversion factor: 8.34 (lb/d) per (MGDmg/L)
Mass (lb/d) = Q(MGD) × C(mg/L) × 8.34 (applies also to ∆Q or ∆C)

115
 Practice Problem Hints

Chap 10BOD
1--
2Hint: Solve for V in the BOD test equation.
3Hint: the maximum ∆DO is 8 – 1= 7 mg/L
4--
5Hint: Exchange 3 for 5 in the equations used in the last problem.
6Hint: straight first order decay
7--
8Hint: Convert between BOD5 and BODu
9Hint: BODu of the factory waste = 708 mg/L (using the
BOD5/BODu conversion), BODu of mix = 59 mg/L (this is the
value at t=0 in the creek), BODu downstream comes from first
order decay
10 Hint: For the primary effluent, the mean BOD is 130 and the upper
end of the 90th percentile confidence limit is 145 mg/L. The mass
of BOD removed is a bit over 4500 lb/d.

116
11. Reactor Analysis

In the lecture part of the class you learned about ideal batch, complete mix, and
plug flow reactors, and how to predict their performances under various
conditions. Of course, real world reactors are rarely ideal. A blender might be a
good approximation of an ideal batch reactor. But a natural pond is not an ideal
complete mix reactor, and a river is not an ideal plug flow reactor. Even
engineered systems like pipes or mixed tanks are rarely ideal.

The equations you derived in class usually require knowledge of the volume of
the reactor. For instance, in the mass balance for a steady state complete reactor
with a first order decay reaction, the reactor term is multiplied by the volume as
shown below

𝑑𝑑𝑑𝑑
= 𝑄𝑄𝑖𝑖 𝐶𝐶𝑖𝑖 − 𝑄𝑄𝑒𝑒 𝐶𝐶𝑒𝑒 + 𝑟𝑟𝑟𝑟
𝑑𝑑𝑑𝑑
0 = 𝑄𝑄𝐶𝐶𝑖𝑖 − 𝑄𝑄𝑄𝑄 − 𝑘𝑘𝑘𝑘𝑘𝑘
0 = 𝐶𝐶𝑖𝑖 − 𝐶𝐶 − 𝑘𝑘𝜃𝜃𝐻𝐻

The ratio of the volume and the flow (V/Q) is the

theoretical hydraulic detention time, θH, also known as MZ
See example 4.3 on
the hydraulic residence time, HRT. (Both terms can be page 125.
used interchangeably.) HRT is an essential term because
the extent of reaction that occurs depends on the time the water is in the reactor.

In an ideal complete mix reactor, the whole volume is engaged in the reaction,
but in real reactors there may be “dead” zones or volumes which are
hydraulically distinct from the rest of the reactor. These can cause the effective
volume of the reactor to be smaller than what you might think from the
dimensions.

It’s easy to imagine how these might occur in a natural lake or pond. Geometry
may prevent water in the main part of a pond from interacting with deep holes or
inlets or bays along the shore. When warm water enters at the top of a cold lake,
it will float on top of the colder water and not mix in. In this case, the cold water
is the dead zone. A similar situation occurs when a freshwater flow enters a
saltwater bay or estuary. The denser salty water forms a dead zone below the
fresh water. Dead zones occur in engineered devices as well, and for the same
reasons -- geometry and density differences.

117
Because of dead zones the effective volume (and hence the actual HRT) is not
always what you expect. When a dead zone is present and water passes through
a smaller-than-expected effective volume, we say that a condition of short-
circuiting exists. The water is bypassing the main volume in the same way that a
short-circuiting electrical current bypasses an electrical circuit. You can also
think of it as the water taking a short-cut out of the tank.

Figure 11-1. Non-Ideal reactor: In this case the bottom is the dead volume, and
the flow short-circuits across the top.

Tracer analysis of a complete mix reactor

We can measure the characteristics of real water (and air) systems with tracers.
Tracers are usually conservative (non-reacting), dissolved chemicals whose
concentration can be easily determined. By measuring the movement and
concentration of a tracer, engineers can determine how well different hydraulic
models (batch, complete mix, and plug flow) fit real systems.

One of the characteristics measured is the actual HRT. This is particularly useful
for natural water bodies where the geometry is often uncertain. Knowing the
HRT and the flow, we can calculate the effective volume.

𝑉𝑉𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
𝐻𝐻𝐻𝐻𝐻𝐻𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 = 𝜃𝜃𝐻𝐻 =
𝑄𝑄

Two common ways to test a complete mix reactor is with a constant tracer feed
or a washout analysis. In a constant tracer feed test, a reactor starts full of clean
water and then a constant flow containing a conservative constituent such as salt
or dye (the tracer) is introduced. The constituent is conservative, but the problem
is not steady state. In this case, the concentration in the reactor increases from
zero to a value equal to the influent concentration. You can see this in Figure 11-
2. The equation for the effluent concentration is:

−𝑡𝑡�
𝐶𝐶 = 𝐶𝐶𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 �1 − 𝑒𝑒 𝜃𝜃𝐻𝐻 �

118
 Figure 11-2. Examples of constant tracer feed and washout tracer data. The
effluent graphs show results from two reactors with different HRTs.

The washout tracer test is the opposite. We start with a reactor full of
conservative constitute and introduce a clean flow. The constituent effluent
concentration will drop to zero according to the following exponential function.

−𝑡𝑡�
𝐶𝐶 = 𝐶𝐶𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑒𝑒 𝜃𝜃𝐻𝐻

The equation above (nonsteady-state CMFR for no reaction and Cin = 0) is

relatively easy to derive. You should do this as an exercise. Your derivation
should parallel MZ Example 4.5 (page 128) with and integration limits of 0 to t,
and C0 to C (concentration exiting the reactor at time t). MZ
In this equation, θH = V/Q. As you can see, this can be For the derivation of the
rearranged to look like a first order decay function effluent equation, see
example 4.5 on page 128.
except that where you expect to find, you have 1/θH.
Applying some algebra, you can get …

𝐶𝐶 1
𝑙𝑙𝑛𝑛 � � = − � � 𝑡𝑡
𝐶𝐶0 𝜃𝜃𝐻𝐻

119
Figure 11-3. A laboratory scale washout tracer test. The effluent samples collected over
time are shown in a rack. Starting in the upper left of the rack, notice how the samples
show less and less blue color over the first three rows. The dark vial in the lower left
corner is the original (time = 0).

120
Practice problems

11-1. When drinking water treatment plants chlorinate water, the chlorine must
remain in contact with the water for at least 30 minutes before being delivered to
customers. A water plant operator wishes to know if his existing 10,000-gallon
storage tank can provide this HRT. He does a wash-out tracer test with fluoride
salt (conservative). The initial concentration is 12 mg/L and after an hour of
feeding non-fluoridated water through the tank at 20 gpm, the concentration has
dropped to 2 mg/L. Does the tank have the required HRT? [Yes, the HRT is
33.5 minutes]

11-2. Upstream of a pond (75,000 m3 with a downstream outlet stream) a factory

has discharged XYZ to the creek feeding the pond for many years. A new
wastewater treatment system is installed which removes XYZ totally from the
factory wastewater. If the creek flowrate is 0.5 m3/s and there is no other source
of XYZ in the system, how long will it take for the pond concentration to drop to
10% of its original value? [4 d]

121
 Practice Problem Hints

Chap 11 Reactor Analysis

1 Solve for HRT using C=C0e-t/HRT
2 Assume the pond is completely mixed. C/C0 = 0.10

122
12. Dissolved Oxygen Modeling:
The DO Sag Curve

Environmental models can be either physical or mathematical. An example of a

physical model is the Bay-Delta model maintained by the Corps of Engineers in
Sausalito. Physical models are based on the principles you studied in Engr 132
(Buckingham Pi theorem and others). They are good for modeling transport, but
not so good at modeling chemical processes. They're also not very portable. The
Bay-Delta model, for instance, is over an acre in size. With the increase in the
availability of computers, mathematical models have become much more
common. They are based on mass balance equations and
equations describing the behavior of various physical and MZ
This material is covered
chemical processes. in Section 7.7*.

Figure 12-1. Part of the SF Bay Model (Wikipedia)

Environmental models are very useful for: 1) predicting the effects of

engineering activities on the environment, and 2) evaluating different alternative
actions. In this second use, engineers can play "what if" games with the model --
what if the project was twice as large, or what if we treated the waste twice as
much, or what if we moved the project somewhere else, etc. One of the first and
simplest environmental models was derived by Streeter and Phelps in 1925. It is
a one-dimensional model of oxygen concentrations in a river.

The Streeter-Phelps equation (also known as the "dissolved oxygen sag"

equation) is based on a mass balance which is affected by two processes. One is

123
that oxygen is removed from water by the degradation of organic materials. In
other words, the biochemical oxygen demand of an organic waste is satisfied by
oxygen taken from the water. The second process is "reaeration" by oxygen
transfer into the water from the atmosphere.

DERIVATION OF THE DISSOLVED OXYGEN SAG (STREETER-PHELPS)

EQUATION

Consider an element of water in a stream. We will track the element as it moves

downstream with the general flow of water.

Oxygen in atmosphere

dV
QC Q(C+dC)
BOD reaction

Figure 12-2. Mass balance control volume

Setting up the differential equation

The mass balance equation written out in words is:

Accumulation = In - Out + Reaction

Looking at the element with differential volume dV, the mass flow in is QC and
the mass flow out is Q(C+dC) where dC is the change that occurs in the element.
There is also mass flow into the element from the atmosphere (called reaeration)
and oxygen consumption by the BOD reaction (called degradation). The overall
mass balance is:

𝑑𝑑𝑑𝑑
𝑉𝑉 = (𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑖𝑖𝑖𝑖) − (𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑜𝑜𝑜𝑜𝑜𝑜)
𝑑𝑑𝑑𝑑
− (𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑜𝑜𝑜𝑜 𝑂𝑂2 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑏𝑏𝑏𝑏 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏)𝑉𝑉
+ (𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑜𝑜𝑜𝑜 𝑂𝑂2 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑡𝑡ℎ𝑒𝑒 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎ℎ𝑒𝑒𝑒𝑒𝑒𝑒)𝑉𝑉

For steady state, accumulation is zero.

124
  dC   dC 
0 = QC − Q(C + dC ) +   dV +   dV
 dt d  dt r
 dC   dC  dV
QdC =   dV +   dV but = dt
 dt  d  dt  r Q
 dC   dC 
dC =   dt +   dt
 dt  d  dt  r
dC  dC   dC 
=  + 
dt  dt  d  dt  r

Let's look at the reaeration and biodegradation two rates. The rate of O2 removal
by biodegradation is:

 dC 
  = ( Rate of BOD decay) = − kd L
 dt  d

where
L = ultimate BOD of organics remaining in the water at any time t
kd = first-order degradation rate constant. kd is related to, but is not numerically
the same as, kBOD, the rate constant in the BOD test.

The rate of O2 transfer from the atmosphere (called reaeration) is:

 dC 
  =k r (C s − C )
 dt  r

where
C = dissolved oxygen concentration (DO)
Cs = dissolved oxygen concentration at saturation, predicted from Henry's Law
kr = reaeration rate constant. This value will be related to the characteristics of
the stream.

Note that this is the same form of equation we used in the gas transfer lab. Now,
substitute the rate expressions into the mass balance.

dC  dC   dC 
V =  V+   V
dt  dt  d  dt  r

125
 dC
= − k d L + k r (C s − C )
dt

Mathematically, it is convenient to express everything in terms of the deficit, D,

rather than concentration, C.

D = (Cs - C)

𝑑𝑑𝑑𝑑 𝑑𝑑(𝐶𝐶𝑠𝑠 − 𝐶𝐶) 𝑑𝑑𝑑𝑑

= = �0 − � = 𝑘𝑘𝑟𝑟 𝐿𝐿 − 𝑘𝑘𝑟𝑟 (𝐶𝐶𝑠𝑠 − 𝐶𝐶 )
𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑

This is the basic differential equation that must be solved.

Solving the differential equation

Solving the differential equation requires the use of integrating factors. First
substitute the BOD decay equation for L.

dD
= kd L0e − k d t − kr D
dt

dD
+ kr D = kd L0e − k d t
dt

This arrangement matches the standard form:

dz
+ Pz = Q
dx
So, use the integrating factor:

e ∫ kr dt = e k r t
dD k r t
dt
( ) ( ) ( )
e + k r D e k r t = k d L0 e − k d t e k r t

d
dt
( )
De k r t = k d L0 e ( k r − kd )t

126
Separate variables and integrate.

� 𝑑𝑑(𝐷𝐷𝑒𝑒 𝑘𝑘𝑟𝑟𝑡𝑡 )   =   ��𝑘𝑘𝑑𝑑 𝐿𝐿0 𝑒𝑒 (𝑘𝑘𝑟𝑟−𝑘𝑘𝑑𝑑)𝑡𝑡 �𝑑𝑑𝑑𝑑

𝑘𝑘𝑑𝑑 𝐿𝐿0 (𝑘𝑘 −𝑘𝑘 )𝑡𝑡
𝐷𝐷𝑒𝑒 𝑘𝑘𝑟𝑟𝑡𝑡   =   𝑒𝑒 𝑟𝑟 𝑑𝑑   + 𝐾𝐾
𝑘𝑘𝑟𝑟 − 𝑘𝑘𝑑𝑑

Apply the boundary condition that at t = 0, D = D0

𝑘𝑘𝑑𝑑 𝐿𝐿0
𝐷𝐷0 𝑒𝑒 0   =   � � 𝑒𝑒 0   +  𝐾𝐾
𝑘𝑘𝑟𝑟 − 𝑘𝑘𝑑𝑑
𝑘𝑘𝑑𝑑 𝐿𝐿0
𝐾𝐾  =   𝐷𝐷0   −   � �
𝑘𝑘𝑟𝑟 − 𝑘𝑘𝑑𝑑
𝑘𝑘𝑑𝑑 𝐿𝐿0 𝑘𝑘𝑑𝑑 𝐿𝐿0
𝐷𝐷𝑒𝑒 𝑘𝑘𝑟𝑟𝑡𝑡   =   � � 𝑒𝑒 (𝑘𝑘𝑟𝑟−𝑘𝑘𝑑𝑑)𝑡𝑡   +   𝐷𝐷0   −   � �
𝑘𝑘𝑟𝑟 − 𝑘𝑘𝑑𝑑 𝑘𝑘𝑟𝑟 − 𝑘𝑘𝑑𝑑
𝑘𝑘𝑑𝑑 𝐿𝐿0 𝑘𝑘𝑑𝑑 𝐿𝐿0
𝐷𝐷  =   � � 𝑒𝑒 −𝑘𝑘𝑑𝑑 𝑡𝑡   +   𝐷𝐷0 𝑒𝑒 −𝑘𝑘𝑟𝑟𝑡𝑡   −   � � 𝑒𝑒 −𝑘𝑘𝑟𝑟𝑡𝑡
𝑘𝑘𝑟𝑟 − 𝑘𝑘𝑑𝑑 𝑘𝑘𝑟𝑟 − 𝑘𝑘𝑑𝑑

Finally, after grouping like terms, the integrated equation is:

𝑘𝑘𝑑𝑑 𝐿𝐿0
𝐷𝐷  = � � (𝑒𝑒 −𝑘𝑘𝑑𝑑 𝑡𝑡 − 𝑒𝑒 −𝑘𝑘𝑟𝑟𝑡𝑡 ) + 𝐷𝐷0 𝑒𝑒 −𝑘𝑘𝑟𝑟𝑡𝑡
(𝑘𝑘𝑟𝑟 − 𝑘𝑘𝑑𝑑

where
L0 = ultimate BOD in the water at t = 0
D0 = deficit at t = 0.

Note that both L0 and D0 are the values in the stream after any external waste
streams have been mixed in.

This equation is known as the DO sag equation (for its distinctive shape) or the
Streeter-Phelps equation, after the gentlemen who first published it in 1925.

APPLICATION OF THE DO SAG EQUATION IN A SIMPLE

ENVIRONMENT

Of course, the fish aren't interested in the deficit. They want to know the
concentration. This can be calculated from:

127
 𝑘𝑘𝑑𝑑 𝐿𝐿0
𝐶𝐶  =   (𝐶𝐶𝑠𝑠 − 𝐷𝐷)   =   𝐶𝐶𝑠𝑠 − �� � (𝑒𝑒 −𝑘𝑘𝑑𝑑 𝑡𝑡 − 𝑒𝑒 −𝑘𝑘𝑟𝑟𝑡𝑡 ) + 𝐷𝐷0 𝑒𝑒 −𝑘𝑘𝑟𝑟𝑡𝑡 �
𝑘𝑘𝑟𝑟 − 𝑘𝑘𝑑𝑑

Be careful of brackets. It’s not …

𝑘𝑘𝑑𝑑 𝐿𝐿0
𝐶𝐶  =   𝐶𝐶𝑠𝑠 − � � �𝑒𝑒 −𝑘𝑘𝑑𝑑 𝑡𝑡 − 𝑒𝑒 −𝑘𝑘𝑟𝑟 𝑡𝑡 � + 𝐷𝐷0 𝑒𝑒 −𝑘𝑘𝑟𝑟 𝑡𝑡
𝑘𝑘𝑟𝑟 − 𝑘𝑘𝑑𝑑

When this equation is plotted with typical values, the curve in Figure 12-3 is
generated. Do you see the sag?

Figure 12-3. DO sag curve

We can use the DO sag equation to find the concentration at any point
downstream of a waste discharge as long as we know the travel time to that point,
which is easy to calculate if the stream velocity is known.

To protect fish, engineers often want to know the minimum oxygen concentration
that can be expected as a result of a waste discharge. This is called the critical
point (critical from the point of view of the fish). The travel time to the lowest
point in the sag is the critical time. Using the critical time (tc), we can easily
calculate the critical deficit (Dc), and the minimum DO concentration.

To find the critical time and deficit, differentiate the Streeter-Phelps equation and
set it equal to zero.

128
 dD d  k L  
dt
( )
= 0 =  d 0  e − k d t − e − kr t + D0 e − k r t 
dt  (k r − k d  
 k L 
[ ]
0 =  d 0  − k d e − kd t + k r e − kr t − D0 k r e − kr t
 kr − kd 

If you divide through by e− k r t , bring the right-hand term to the other side of the
equal sign, take the natural log of both sides, and solve for t = tc, you get:

 1   kr  D (k − k d ) 
t c =   ln  1 − 0 r 
k
 r − k d   kd  k L
d 0 

To solve for the critical deficit, Dc, you could plug this expression for tc into the
Streeter-Phelps equation, but that would give you a massive algebra headache.
It's easier to realize that at the bottom of the sag, dD/dt = 0. This allows us to go
back to the differential equation.

dD
= k d L0e −kd t − k r Dc = 0
dt

k 
Dc =  d  L0 e −kd t
 kr 

Now, the minimum concentration Cmin can be calculated from: Cmin = Cs - Dc

Note that tc and Dc are single values at a specific An extended example will be
location. If we want the DO at any other location, we solved in lab. Check MZ (Sec 7.7.4
would use the Streeter-Phelps equation itself. and 7.7.5) as well.

129
USING THE DO SAG EQUATION IN A MORE COMPLEX ENVIRONMENT

One shortcoming to the Streeter-Phelps

equation is that can predict conditions
downstream only as long as physical
parameters like kd and kr remain constant,
and there are no external flows into the
stream. In real rivers, kd and kr depend on
temperature, depth, and velocity, which
aren’t constant everywhere. Also,
tributaries periodically enter real rivers.

How can we deal with changing conditions? Let’s start by dividing the river into
small pieces, called "reaches", such that conditions inside each reach are
constant. When this is the case, we can use Streeter-Phelps inside each reach.
We do have to make some adjustments going from reach to reach, however.
Consider the following two-reach river.

kd1, kr1 kd2, kr2

L1 C2 from DO L2 = L1*exp(-kd1*t) C3 from DO

Sag eqn Sag eqn

Figure 12-4. Relationships between river reaches – no external input

The water entering Reach 1 contains BOD (L1) and dissolved oxygen (C1) which
allows calculation of the initial deficit (D1). We would use these as inputs to the
DO Sag equation and calculate the dissolved oxygen concentration at the
downstream end of the reach (C2). What comes out of Reach 1 becomes the
input to Reach 2. So, C2 is used to calculate the initial deficit for Reach 2. The
BOD for Reach 2 is the BOD remaining in the water after traveling through
Reach 1. We calculate that from the first order decay equation, L2 = L1*exp(-
kd1*t) where t is the travel time in Reach 1. Now we can use the DO Sag
equation with the values of kd and kr that are appropriate for Reach 2.

We can use the same method to account for new inputs to the river such as
tributary streams. All we have to do is define the reaches so that external flows
occur only on reach boundaries, as shown below.

130
 External input: Q, L, C

kd1, kr1 kd2, kr2

L1 C2 from DO L2 = L1*exp(-kd1*t) mixed with L external

Sag eqn

Figure 12-5. Relationships between river reaches – including external input

The only difference is that now we must mix in the external inputs using the
mixing equation:

(C1Q1 ) + (C2Q2 )
Cmix =
(Q1 + Q2 )

where 1 and 2 refer to any two streams.

A spreadsheet model based on this concept can easily be written. (That’s part of
the lab assignment.)

One thing you should know about computer models is that they are only as good
as the equations and coefficients used. Usually, before a model is used to predict
something that hasn't yet happened, it is tested against past conditions. Then, if
necessary, coefficients or equations are adjusted so that the model reproduces
existing data. This is called "calibrating" the model. In your computer
assignment, you are asked to calibrate your spreadsheet model and then use it to
solve a waste loading problem.

LIMITATIONS OF THE DO SAG EQUATION

As with all mathematical models, certain simplifications have been made in the
derivation of the DO sag equation. Here's a partial list of limitations:

1. Steady state -- Streams aren't steady state. Flows, velocities, geometries, and
temperatures all vary with time. Dividing the stream into smaller reaches
reduces this limitation, but steady state conditions are still assumed inside
each reach. To the extent that the reach is not steady state, inaccuracies will
be introduced.

2. Plug flow -- Streams aren't really plug flow. The geometries of natural
streams are not regular -- there are wide spots, pools, narrow chutes, sand
bars, rocks -- so the flow doesn't move as a plug.
131
3. Algae -- The model doesn't include algae which are a very important source
of oxygen. The effects of algae are very dependent on sunlight, which
changes through the day. Modeling algae accurately would require a
nonsteady-state model.

4. Nitrification – Biological oxidation of ammonium/ammonia is a mechanism

for consuming DO that is not considered in the model. (There are versions of
the Streeter-Phelps equation in the literature that include nitrification.)

5. Constant degradation constant -- The model assumes that all the oxygen
demand is from suspended organisms (i.e., bacteria living in the water
column like they were in the BOD bottle) and that they are uniformly
dispersed in the water column. Consequently, in the model the rate of BOD
removal doesn’t depend on location. BOD removal in real systems is more
complicated and several factors affect BOD removal rates.

First, a significant fraction of BOD may be in particulate form that can settle
out. Therefore, different parts of the stream (slow-moving and fast-moving)
will see different BOD removal rates due to settling only. Beyond this,
where organic particles settle, the growth of benthic (bottom-dwelling)
organisms is stimulated (lots of food  lots of microbes). This increases the
biological BOD removal (and DO consumption) rate at that location because
there are more organisms there.

Second, most natural bacteria live attached to surfaces in "biofilms" -- slimy

coatings on rocks or soil particles. Consequently, BOD removal rates will
vary by the ratio of biofilm surface area to water volume -- the more surface
area per volume, the greater the biofilm effect. If the stream is deep and free
of rocks (section A in the figure), the influence of biofilms will be relatively
small compared to a location where the water is flowing through and around
a lot of rocks (section B) and has little empty volume.

Figure 12-6. Two contrasting bottom configurations

By the way, we’ve seen this surface area-to-volume effect in adsorption

(adsorbent particle size) and gas transfer (gas-water interface area). In some
wastewater treatment plants, wastewater is treated by being poured over beds
132
of rocks with high microbial populations. These devices have a much higher
BOD treatment rates than in open ponds where the treatment depends on a
less dense concentration of microbes suspended in the open water.

133