CSTR Murphy
CSTR Murphy
Basic chemical reactors (plug flow reactor PFR, and continuous stirred tank reactor or CSTR) are studied
considering their behaviour is that of an ideal reactor. Unfortunately, in practice, we often find behaviours
that are far from that considered ideal. These non-ideal behaviours in chemical reactors may be followed by
measuring the output concentration of a non-reactive tracer as a function of time. This was achieved by
studying and thoroughly investigating the behaviour of continuous stirred tank reactors in series, including
their dynamic behaviour and the effect on the progress of chemical reactions carried out in such systems.
This aim was realized by studying the Residence Time Distribution (RTD) concept which can be determined
by the method of injection of an inert chemical called a tracer by either a step input change or pulse input
change on a tank in series model consisting of three tanks in series; this experiment focused on step input
change. The investigation was carried out using the Arm field CEP Mk II Stirred Tank Reactor in series unit,
which allows flexibility in carrying out the required investigations. The deionized water is filled in both tanks
with Potassium chloride diluted in one tank with a concentration of 0.3M—which is allowed to flow through
the three tanks by opening the peristatic pump at a speed of 6.0 to prevent overflow, while the stirrers at 8.0,
as recorded by the speed adjust dial. The initial readings of the conductivity are taken at time t equals zero,
which are 0.11Ms/min; 0.28Ms/min; 0.18Ms/min and the exit stream recorded 0.05Ms/min. Then, the
readings are continuously taken for 5 minutes interval for 45 minutes. The results obtained are presented in
three parts; the analytical method, the graphical method, and the Simpson’s rule. The analytical and graphical
method gave an area under the age distribution curve as 1--from a plot of conductivity versus time of which
E(t) was obtained mathematically and its values plotted against time; while the Simpson’s rule which was
used in a supporting role here, also gave a value of 1 . This strongly agrees with literature—by normalizing
the distribution to obtain unity (Octave Levenspiel). Overall, all three methods for a step input change for
three tanks in series, an exponential curve was obtained from the first tank, followed by s-shaped curves from
the remaining tanks. Lastly the mean value varies inversely as the volumetric flow rate of feed.
The application of the Residence time distribution (RTD) which was also emphasised, characterises the main
flow features in non-ideal flow reactors, which is very useful in determining the mixing behaviour and
conversion rate of a reactor system. A software is recommended to avoid students from making egregious
errors while simulating data.
1
INTRODUCTION
A reactor is a vessel where chemical reaction takes place. To find what a reactor is able to do we need to know
the kinetics, the contacting pattern and the performance equation. The Contacting pattern or how materials
flow through and contact each other in the reactor, how early or late they mix, clumsiness or state of aggregation.
By their very nature some materials are very clumpy-for instance, solids and non-coalescing liquid droplets.
The Kinetics or how fast things happen. If very fast, then equilibrium tells what will leave the reactor. If not so
fast, then the rate of chemical reaction, and maybe heat and mass too, will determine what will happen. The
Performance equation relates input to output.
Basically, there are three types of ideal reactors so-called because their concepts involve assumptions that
simplify their analytical treatment and ensure that for a given reaction, one of these represents the best way of
contacting reactants. These ideal reactors are the batch reactor, the plug flow reactor and the mixed flow reactor.
In the batch reactor, or BR, the reactants are initially charged into
a container, are well mixed, and are left to react for a certain period. The resultant mixture is then discharged.
This is an unsteady-state operation where composition changes with time; however, at any instant the
composition throughout the reactor is uniform.
The first of the two-ideal steady-state flow reactors is variously known as the
plug flow, slug flow, piston flow, ideal tubular, and unmixed flow reactor. We refer to it as the plug flow reactor,
or PFR, and to this pattern of flow as plug flow. It is characterized by the fact that the flow of fluid through the
reactor is orderly with no element of fluid overtaking or mixing with any other element ahead or behind.
Actually, there may be lateral mixing of fluid in a plug flow reactor; however, there must be no mixing or
diffusion along the flow paths. The necessary and sufficient condition for plug flow is for the residence time in
the reactor to be the same for all elements of fluid. The other ideal steady-state flow reactor is called the mixed
reactor, the back mix reactor, the ideal stirred tank reactor, the C* (meaning C-star), CSTR, or the CFSTR
(constant flow stirred tank reactor), and, as its names suggest, it is a reactor in which the contents are well stirred
and uniform throughout. Thus, the exit stream from this reactor has the same composition as the fluid within
the reactor. We refer to this type of flow as mixed pow, and the corresponding reactor the mixed pow reactor,
or MFR. These three ideals are relatively easy to treat. In addition, one or other usually represents the best way
of contacting the reactants-no matter what the operation. For these reasons, we often try to design real reactors
so that their flows approach these ideals.
As always, the case in science and engineering, real life situations; an actual reactor equipment shows deviation
from ideal reactors earlier mentioned. Hence real reactors are non-ideal. While it is impossible to completely
map out the velocity distribution within a given reactor vessel, and hence totally account for observed non-
ideality, it is possible to obtain data for feed and effluent streams. Such data are known as end-effect or response
data. The measurable effect observed in the effluent stream due to a planned disturbance in the feed is used to
account for non-ideality in reactors. For this reason, this experiment was performed using the Arm field CEP
Mk 11 stirred tank reactor in series unit, to demonstrate and verify experimentally the responses expected from
stirred tank reactors that demonstrate their non-ideality, as predicted from theory. This objective incorporates
the following:
To study and demonstrate the effect of a step input change caused by a switch from ordinary fluid to fluid with
a definite tracer concentration; to study the effect of this on continuous stirred tank reactors in series.
To study and demonstrate the response of tank concentration to an impulse change carried by instantaneously
introducing a definite amount of tracer into reactor input fluid.
2
THEORY
Basic chemical reactors (plug flow reactor or PFR, and continuously stirred tank
reactor or CSTR) are studied considering their behavior is that of an ideal reactor.
Unfortunately, in practice, we often find behaviors that are far from that considered ideal.
Consequently, working with them, the chemical engineer must be able to handle and
diagnose the behavior of these reactors. At the time of describing the nonideal behavior of a
reactor, three concepts are introduced which are the factors which make up the contacting or
flow pattern, they are:
1. the RTD or residence time distribution of material which is flowing through the
vessel
2. The state of aggregation of the flowing material, its tendency to clump and for a
group of molecules to move about together
3. The earliness and lateness of mixing of material in the vessel.
These three concepts are used to describe the deviations of the mixing assumed in the ideal
models and are considered as attributes of the mixture in non-ideal reactors. One way of
approaching the study of non-ideal reactors is to consider them, in a first approximation, as if
the flow model were the one corresponding to a CSTR or a PFR. However, in real reactors,
the non-ideal flow model implies a minor conversion, so a method that allows for this
conversion loss to be considered must be available. Therefore, a higher level of
approximation implies the use of information about the RTD.
For the sake of this experiment the RTD or residence time distribution of material which is
flowing through the vessel will be of our utmost concern:
3
1. THE PULSE INPUT
This involves injection of an amount of tracer, say M in one shot lasting for the shortest
possible time into the reactor feed stream. The outlet concentration is measured as a function
of time. The effluent concentration time curve is known as the 𝐶𝑝𝑢𝑙𝑠𝑒 curve. If the reactor
vessel has a volume 𝑉𝑚3 and the volumetric flow rate v.
𝑀
Hence, dividing the 𝐶𝑝𝑢𝑙𝑠𝑒 readings by yields E readings.
𝑣
𝑡
If time is measured in terms of mean residence time 𝜗 = 𝑡́
The dimensionless form of the 𝐶𝑠𝑡𝑒𝑝 curve is called F curve. The relationship between the F
and E is:
𝑑𝐹
=𝐸
𝑑𝑡
In summary This kind of experiment consists of introducing into the current entering the
reactor, quickly and at once, an amount N0 of tracer. The output concentration is
subsequently measured as a function of time. The concentration-time curve corresponding to
the effluent is called curve C in RTD analysis.
4
Co Co
Pulse stimulus t pulse response t
Fig 1. The useful information obtainable from the pulse trace experiment.
Pulse response
For the pulse response experiment; if the first tank has a steady flow of v𝑚3 /𝑠 with a volume
of 𝑉1 and time t=0, a pulse of tracer is introduced into the vessel given a concentration of 𝐶𝑜
after even distribution, a material balance can be made at any time t after the introduction of
the tracer.
𝑐1
𝑑𝐶1 1 𝑜
∫ = ∫ 𝑑𝑡
𝑐0 𝐶1 𝑡́1 0
Which integrates to
1
𝐶1 ́1
𝑡
= 𝑒
𝐶0
𝐶
The area under the 𝐶1 vs time curve is 𝑡́1, so normalizing the curve yield the curve:
0
1
𝑡́1 𝐸́1 = 𝑒 𝑡́1
For the second tank, where 𝐶1 𝑒𝑛𝑡𝑒𝑟𝑠 𝑎𝑛𝑑 𝐶2 leaves a material balance yields.
𝑡 𝑡́1
𝑡́1 𝐸́1 = 𝑒1
𝑡2
Hence for N tanks:
𝑁−1 1
𝑡 1
𝑡́1 𝐸́1 = [𝑡́ ] (𝑁−1)!
𝑒 𝑡́1
2
5
The response of a first order system consisting of three tanks in series to an impulse change
in input is an exponential curve from the first tank for the duration of the impulse followed by
an exponential decay back to the restore input level after the impulse is over. The other tanks
show a sluggishly moderated response delayed by their respective transfer lags but ultimately
return to the restored input level.
2. STEP RESPONSE
The output F curve from a series of N ideal stirred tanks is, in its various forms given by the
following equation.
From the equations, the response of a first order system consisting of three tank in series is an
exponential curve from the first tank followed by S- shaped curves from the remaining tanks.
In simpler notation; if
A- Concentration in the tank at time t after time input step change
B- Concentration of input, then
1
𝐴 = 𝐸(1 − 𝑒 𝑡́1 )
6
Co
t t
In comparison with a pulse input, the step input has the following advantages:
1. A step change is usually easier to achieve.
2. A material balance is usually easier to achieve.
Selection of a Tracer
Accurate determination of RTD in a vessel requires proper selection and introduction of a
tracer. Here, we consider characteristics and examples of tracers. Ideally, a tracer should have
the following characteristics:
7
1. The tracer should be stable and conserved, so that it can be accounted for by a
material balance relating the response to the input; if the tracer decays (e.g., a radio-
labeled tracer), its half-life should be such that t1,2 (tracer) > 25t (fluid).
2. The analysis for tracer should be convenient, sensitive, and reproducible.
3. The tracer should be inexpensive and easy to handle; this is particularly important for
a step input, in which relatively large quantities of tracer may be required.
4. The tracer should not be adsorbed on or react with the surface of the vessel.
Alternatively, the tracer should be chemically and physically similar to the fluid
flowing, so that any adsorption (or diffusion) behavior may be replicated.
Examples of tracers
1. gas-phase tracers such as He, Ne, and Ar used with thermal conductivity detectors;
2. pH indicators such as phenol red and methylene blue;
3. electrolytes such as K+ and Na+ used with electrical conductivity detectors or
specific-ion electrodes;
4. dyes (e.g., India ink) used with color intensity;
5. radioactive isotopes such as 3H, 14C, l8O; 51Cr-labeled red-blood cells used to
investigate hepatic blood flow; isotopes of iodine, thallium, and technetium used to
investigate cardiac blood flow; stereoisomers and structural analogs used for
diffusion-limited processes (e.g., f-versus d-glucose in biological systems).
8
DESCRIPTION OF APPARATUS
The Armfield CEP-MKII Stirred Tank Reactors in Series Unit is designed to follow the
dynamics of the perfectly mixed multistage process. Dynamic behaviour can be studied as
can multistage chemical reaction. Bench mounted and self-contained, the unit requires only to
be connected to a single-phase electrical supply for operation. There are three reactor vessels
connected in series, each containing a propellor agitator driven by a variable speed electric
motor. Two reagent vessels and two variable speed feed pumps feed reagents into the first
reactor in line. For certain experiments the feed can be connected to the third reactor and a
dead time coil, also positioned on the vacuum formed plinth. Each reactor and the exit port of
the dead time coil are fitted with conductivity probes for monitoring the process.
Conductivity is displayed on a digital meter on the console through a selector switch and all
four probes can be connected to the optional Armfield data logging accessory.
FIG 4: STOPWATCH
A stopwatch is a handheld timepiece designed to measure the amount of time that elapses
between its activation and deactivation. The timing functions are traditionally controlled by
two buttons on the case. Pressing the top button starts the timer running, and pressing the
button a second time stops it, leaving the elapsed time displayed.
9
Fig 5: round bottom flask
10
PROCEDURE
All connections to the CSTR were done properly and the CSTR connected to Mains.
3litres of 0.1M Potassium Chloride solution was prepared by dissolving 22.365g of
KCL in 3litres of water and used to fill one of the reagent feed vessels.
The oiler feed vessel was half filled with demineralized water.
The water was poured into each of the reactors until they were full to the standpipe
overflow levels.
The salt feed pump and the water feed pump were set to 6.0 and 8.0 respectively as seen
from the speed dial.
The water feed pump was switched on and the data logger started collecting the
conductivity readings for each of the reactors.
The time at which the conductivity readings changed for each reactor was taken note of.
This indicated that the tracer element had entered the reactors.
The reaction was left to go on and the conductivity readings of the three reactors was
collected over a 45 minutes period at an interval of 5 minutes.
The logged data was used to plot a graph of conductivity against time for each of the
three reactors.
11
RESULTS AND CALCULATIONS
VOLUME OF REACTORS
Calculations
FOR REACTOR 1
𝑡𝑖𝐶𝑖∆𝑡𝑖:
At t = 0 mins; 0 × 0.04 × 0 = 0
At t = 5 mins; 5 × 3.08 × 5 = 77
12
At t = 10mins; 10 × 4.32 × 5 = 216
𝑪𝒊∆𝒕𝒊:
At t = 0 mins; 0.04 × 0 = 0
Ti 𝒕𝒊𝑪𝒊∆𝒕𝒊 𝑪𝒊∆𝒕𝒊
0 00.00 0.00
5 77 15.4
10 216 21.6
15 435 29.00
20 627 31.35
25 875.75 35.15
30 1150.5 38.35
35 1442 41.20
13
40 1750 43.75
45 2075.25 46.05
0.00106029
v= = 0.00003692m3/min.
28.72
time
(mins) R2
0 0.08 0 0
5 0.55 13.75 2.75
10 1.08 54 5.4
15 1.73 129.75 8.65
20 2.43 243 12.15
25 3.16 395 15.8
30 3.88 582 19.4
35 4.58 801.5 22.9
40 5.26 1052 26.3
45 5.91 1329.75 29.55
∑𝑖 𝑡𝑖𝐶𝑖∆𝑡𝑖 =4600.75 ∑𝑖 𝐶𝑖∆𝑡𝑖 =142.9
0.00106029
v= =0.000032193m3/min
32.196
14
TABLE 4: RESIDENCE TIME TABLE FOR REACTOR 3
time
(mins) R3
0 0.09 0 0
5 0.1 2.5 0.5
10 0.16 8 0.8
15 0.27 20.25 1.35
20 0.4 40 2
25 0.53 66.25 2.65
30 0.65 97.5 3.25
35 0.7 122.5 3.5
40 0.85 170 4.25
45 0.92 207 4.6
734 22.9
0.00106029
v= =0.00003307m3/min
32.0524
time
R1 R2
(mins) R3
0 0.04 0.08 0.09
5 3.08 0.55 0.1
10 4.32 1.08 0.16
15 5.8 1.73 0.27
20 6.27 2.43 0.4
25 7.03 3.16 0.53
30 7.67 3.88 0.65
35 8.24 4.58 0.7
40 8.75 5.26 0.85
45 9.21 5.91 0.92
15
TABLE 5: TABLE OF AREA UNDER THE CURVE—from the plot of conductivity
versus time
10
9
8
7
concentration
6
5 R1
4 R2
3
R3
2
1
0
0 5 10 15 20 25 30 35 40 45
time (mins)
CALCULATION OF RTD
E(t) =
The area under the curves was calculated using Simpson method.
=
{yo + 4y1 + 2y2 + 4y3 + 2y4 +…….yn}
From the graph of conductivity R(1) against Time (t) Between the range of 30 to 0, n = 6, b
=30, a=0;
= = =5
= {0.04 + 4{3.08+5.38+7.03 } + 2{4.32 +6.27 } + 7.67 = 132.18
By interpolating between the ranges of values (45 to 30), the average height is gotten to be;
40 8.75
37.5 X
35 8.24
=
X = 9.20
From range of 45 to 30 = 9.20 –7.67 = 1.53
= 1.53 × (45 - 30) = 22.95
Area =
= 126.33 + 22.95 = 149.28 gmin/m3
16
At t = 0min
E(t) = = 2.6792×10-4 m3/gmin
At t = 5min
E(t) = =0.020632 m3/gmin
At t= 10min
E(t) = 0.0289m3/gmin
TABLE 6
REACTORS CONDUCTIVITY.
Time (min) R1 (ms) E(t) m3/gmin
0 0.04 2.6792×10-4
5 3.08 0.020632
10 4.32 0.02890
15 5.38 0.03603
20 6.27 0.04200
25 7.03 0.04709
30 7.67 0.05138
35 8.24 0.05519
40 8.75 0.05861
45 9.21 0.06170
17
= × (0.045)2 × 0.2295 = m3
t=
= 28.72min
= = /29.82
= 5.084×10-5 m3/min
TABLE 7.
REACTORS CONDUCTIVTY 3.
Time (min) R2 (ms) E(t) m3/gmin
0 0.08 1.66×10-3
5 0.55 0.011378
10 1.08 0.022341
15 1.73 0.035788
20 2.43 0.050269
25 3.16 0.065370
30 3.88 0.080265
35 4.58 0.094746
40 5.26 0.108812
45 5.91 0.122259
CALCULATION OF RTD
18
E(t) =
The area under the curves was calculated using Simpson method.
=
{yo + 4y1 + 2y2 + 4y3 + 2y4 +…….yn}
From the graph of conductivity R(2) against Time (t) Between the range of 30 to 0, n = 6, b
=30, a=0;
= = =5
= {0.08 + 4{0.55+ 1.73+3.16} + 2{1.08 + 2.43} + 3.88}
= 32.74
By interpolating between the ranges of values (45 to 30), the average height is gotten to be;
40 5.26
37.5 X
35 4.58
=
X = 4.92
From range of 45 to 30 = 4.92 – 3.88 = 1.04
= 1.04× (45 - 30) = 15.6
Area =
= 32.74 + 15.6 = 48.34 gmin/m3
At t = 0min
E(t) = = 1.66×10-3 m3/gmin
At t = 5min
E(t) = = 0.011377741 m3/gmin
At t= 10min
E(t) = = 0.0223417 m3/gmin
The rest values of R2 was calculated and tabulated.
THE MEAN RESIDENCE TIME (t)
t= =
V = r2h
D = 9 cm = 0.09 m, r = (0.09/2) = 0.045 m, h 22.95cm = 0.2295 m.
= × (0.045)2 × 0.2295 = m3
Using the equation above, the mean residence time was calculated to be = 32.12 min.
= = m ÷ 32.12
= 4.55×10-5 m3/min
19
RTD for Reactor 2.
From the graph 2. E(t) = 0.0028x + 0.0028
RTD=Area under the curve= =[0.0028(x/2) + 0.0028x + C]450
RTD=0.0028(45)2/2+0.002845=2.705 sq. unit.
0 0.09 7.063×10-3
5 0.10 7.85×10-3
10 0.16 0.01257
15 0.27 0.02119
20 0.40 0.03139
25 0.53 0.04159
30 0.63 0.04944
35 0.76 0.05964
40 0.85 0.06671
45 0.92 0.07222
CALCULATION OF RTD
E(t) =
The area under the curves was calculated using Simpson method.
=
37.5 X
35 0.76
=
X = 0.875
From range of 45 to 30 = 0.875 – 0.63 = 0.245
= 0.245 × (45 - 30) = 3.675
Area =
= 9.067 + 3.675 = 12.742 gmin/m3
At t = 0min
E(t) = = 7.063×10-3 m3/gmin
At t = 5min
E(t) = = 7.85×10-3 m3/gmin
At t= 10min
E(t) = = 0.01257 m3/gmin
The rest values of R3 was calculated and tabulated.
THE MEAN RESIDENCE TIME (t)
t= =
V = r2h, m3
Using the equation above, the mean residence time was calculated to be =34.15 min.
= = ÷ 31.82
= 4.59×10-5 m3/min
Hence the Percentage of element that have spent 0 to 45min in Reactor 1 (R1), Reactor 2 (R2)
and Reactor 3 (R3) are 1.795, 2.705 and 1.6515 sq unit
21
DISCUSSION
We used this experiment to analyze and extensively investigate the behavior of continuous stirred tank reactors
in series, including their dynamic behavior and the impact on the development of chemical reactions carried out
in such systems. The following goals were pursued: to investigate and demonstrate the effect of a step input
change caused by a switch from ordinary fluid to fluid with a definite tracer concentration (potassium chloride
KCL); and to investigate and demonstrate the effect of this on continuous stirred tank reactors in series. Second,
to investigate and illustrate how tank concentration responds to an impulse change caused by injecting a certain
quantity of tracer (KCL) into the reactor input fluid. The Arm field CEP Mk 11 stirred tank reactor in series unit
was used in this experiment to illustrate and validate experimentally the reactions expected from stirred tank
reactors that reveal its non-ideality, as predicted from theory. Filling the reagent feed containers with 3 liters of
0.1M KCL solution was the initial stage in the experiment. There are two methodologies to observe in the CSTR
in series experiment: the influence of step change input and the effect of pulse input.
However, we will just address the influence of the step input technique in this discussion. The distinction
between these two approaches is that step change input entails continually feeding the salt solution KCL into
the reactors until the conductivity of the first and third reactors is equal. To test the effect of step up input, we
supplied 3 liters of 0.1M KCL solution into the reagent feed vessels, then continued the experiment by supplying
the reactor with deionized water through a peristatic pump, distributing the salt solution evenly across all three
reactors.
The feed is supplied at a speed of 6.0 on the speed adjust dial via the reactors and 8.0 on the speed adjust dial
through the stirrers. The system is operating isothermally, with the temperature of each reactor set to room
temperature (29.75°C). We measured the conductivity of each reactor every 5 minutes for 45 minutes in this
experiment, yielding 10 distinct values starting at time zero. At t= 0, the corresponding conductivity values as
measured by the detector were 0.11Ms/min, 0.28Ms/min, 0.18Ms/min, and 0.05Ms/min, with the exit stream
recording 0.05Ms/min.
Part A contains tabular measurements as well as calculations for calculating the residence duration using the
analytical technique. The first, second, and third reactors had residence times of 28.72, 32.196, and 32.0524
minutes, respectively. The volumetric flow rates for the first, second, and third reactors were
0.00003692m3/min, 0.000032193m3/min, and 0.00003307m3/min, respectively, with a reactor volume of
0.00008528m3.
As more salt solution is introduced into the reactors, the conductivity of the mixture rises with time, as shown
in Part-A. Hence the Percentage of element that have spent 0 to 45min in Reactor 1 (R1), Reactor 2 (R2) and
Reactor 3 (R3) are 1.795, 2.705 and 1.6515 sq unit
CONCLUSION
Overall, the experiment yields findings that correspond with theory with a minimum of divergence, according
to considerable debate. While attempts are made in real reactors to approximate ideality, the residence time
distribution accounts for non-ideality and aids in exit conversion calculations. Experiment results using the tank
22
in series model corroborate theoretically expected responses for the step-input approach. This is extremely
helpful in forecasting a reactor system's mixing behaviour and overall conversion rate.
RECOMMENDATION
Fully computer controlled and supplied with educational software specific to each reactor type with
simple interfacing to the (user-supplied) computer by a USB interface. This can be used for an
accurate result and also for exposure of students to the software since the world is going
digital.
23
REFERENCES
1. Levenspiel, O. (1999); “Chemical Reaction Engineering”, 3rd Edition John Wiley & Sons, Inc.
Singapore. Pp 257-275, 321-334.
2. Smith J.M. (1981); “Chemical Engineering Kinetics” 3rd Edition McGraw Hill Inc., USA. (p
268-275, 283-286).
3. Sprivav R.P.S. (2000); “Elements of Reaction Engineering”, Khanna Publishers, Delhi.P (p
215-223, 238-240.
4. Schmidt, Lanny D. (1998). “The Engineering of Chemical Reactions. New York; Oxford
University Press. ISBN 0-19-510588-5
5. Robert H. Perry, Don W. Green, (1998) “Perry’s Chemical Engineers’ Handbook”, 2nd edition
McGraw Hills.
6. INTRODUCTION TO CHEMICAL REACTION ENGINEERING AND KINETICS Ronald
W. Missen Charles A. Mims Bradley A. Saville chapter 19
24