Intro

Background:

An unused laboratory-scale, continuous stirred tank reactor (CSTR) was found in the
basement in the Chemical Engineering department. Professor Opitz would like to use this
reactor in her research group but is not sure of how the performance compares to the reactors
she currently uses. She has kindly requested for our team to investigate the performance of
the found reactor in comparison to her existing reactors. This is because she does not want to
spend too much valuable research funding on actual chemical reaction experiments and thus
would prefer ‘preliminary exploratory study’ investigating the theoretical reactor
performance based on reaction data and reactor mixing characteristics, as a function of
agitation speed. The given rate constant for the investigated reaction (A → B) is assumed to
be 0.0036 s-1 at 54.7 °C. (Point at this at the screen)

Aims:

The aims of the practical are to:

• Determine the mixing characteristics of the CSTR reactor using a tracer study.

• Describe the behaviour of the reactor and how it deviates from ideality using a one
parameter model.

• Investigate the performance of the reactor with the given reaction data and discuss the
necessity for agitation within the reactor system.

Lit Review:

So often reactors are idealised for simplicity sake. When a reactor is being designed,
engineers rarely take into consideration issues like by passing and stagnant regions. In reality
these issues occur and they play a vital role in the performance of a reactor. The deviation
from the ideal reactor is explored using the concept of residence distribution time This
concept deals with the amount of time particles spend inside the reactor

Residence time is the period of time species spend inside a reactor or vessel in general
Residence time is typically calculated as the volume of the reactor divided by the volumetric
flow of fluid into the reactor. However, in real life there are several other factors to consider
when finding the effective residence time of a vessel.

V
τ=
v

To find out how much time fluid spends in a reactor one needs to use an easily detectable
substance to feed into the reactor and trace the amount of time it takes until this substance is
detected in the out flow, this substance is call a tracer

One does not use just any tracer as there are certain desirable traits a tracer needs to have and
among those are the following: (POINT AT PRESENTATION)

 Inert, the tracer must not react with other species in the reactor fluid
 Not disturb the inlet flow, it is important that the tracer have minimal impact on the
volumetric flow of the feed to the reactor
 the same fluid properties as pure feed solution
 completely soluble in the fluid.
 Low in concentrations
 Not interactions with the vessel
 Easily detectable

For this EXPERIMENT potassium chloride is used as it meets the above requirements and is
a cheap ionic tracer. It is also easy to detect the change in conduction in the effluent because
fluid conduction is directly proportional to the ionic concentration of a solution.

1.1.1 Pulse tracer experiment

This is one of the two methods of applying a tracer in a residence time distribution
experiment. Here the tracer is added in small amounts such that V pulse<0.01Vreactor
(Levenspiel, 1999). The tracer injection process must take the least amount of time possible
in order to reduce the variance in the outlet concentration. Outlet fluid is analysed for tracer
until the concentration is found to be zero.
 1.1.2 Step tracer experiment
In this experiment tracer is added in the inlet at a constant volumetric flowrate. The
effluent’s tracer concentration is analysed constantly until the outlet concentration is
constant. Tracer injection only stops when once the effluent concentration is constant
(Levenspiel, 1999).

This method is advantageous in that one does not need to know how much tracer was used
in order to find the E(t). Step tracer is relativity more expensive because a lot of tracer is
thus more frequently used compared to the pulse tracer experiment.. For this EXPERIMENT
only the pulse tracer experiment procedure will be considered.

1.2 CSTR residence time distribution

The ideal CSTR has perfect mixing throughout the reactor which means that the
concentration in the reactor is the number of moles over the total volume of the reactor
(Fogler , 1999). In an ideal CSTR vessel the effluent containing tracer is to have the residence
time distribution curve as illustrated by Error: Reference source not found below.

In reality the residence time distribution may differ due to the following reasons:

 Stagnant regions, zones in the reactor that the tracer molecules do not have access to.
This will result in higher E values and low mean residence time than then estimated.
 By passing or short circuiting, tracer molecules do not mix in the reactor instead they
pass to the exit. Results in a sharp early peak in the RTD curve.
 Use of unsuitable tracer. Either the tracer does not have the same fluid properties as
the reactor fluid or it is not inert which results in some of the tracer being lost.

I will now handover to my colleague Precious who will talk you through the experimental
procedure as well as the Healthy safety measures.

Method & HSE:

For this experiment to be performed the apparatus we used is the computer monitor,
conductivity probe, impeller, CSTR, water feed bottle and peristaltic pump, potassium
chloride feed bottle and peristaltic pump. Before starting the experiment, the CSTR vessel
was rinsed with water and it was left empty. The experimental set up was inspected and
ensured that it is connected to the computer and monitor before starting the computer
software .Initially the flowrates of the tracer and water was set at 50ml/min and 100ml/min
respectively and the impeller speed at 50rpm.The conductivity was recorded by the
program software in 30 second intervals and measured by the probe. Press start in order to
pump the tracer into the reactor vessel until the readings of the conductivity reach a
constant value. When a steady value is reached, stop the experiment and rinse the reactor
with water. This experiment was repeated two more times but set at different flowrates.
Protective gear such as googles and lab coats had to be worn for protection against
potassium chloride which can cause irritation if it comes into contact with the skin or the
eye. Avoid inhaling any potassium chloride because it can cause nose and throat irritation
and coughing.It is important to ensure that the pump and the feed bottles are well
connected to avoid any spillage and leakage .Covid 19 regulations had to be practised.

Calculations:

 To calculate the required values for the F curve, each measured conductivity value
was divided by the initial concentration, which is assumed to be the maximum
conductivity that was measured. These values were then plotted against time to
produce the F curve. After the F curve reaches 1 for the first time, all data points
were artificially adjusted to 1 as it is assumed that all the tracer has been recovered,
in other words, the f curve plateaus after reaching a value of 1 for the first time.
 The F curve is the cumulative distribution function, and therefore the e curve is the
derivative of the f curve. The data for the E curve was calculated by taking the
 absolute value of the difference between two consecutive points on the F curve and
dividing that value by the interval, which is 0.5.
 The data points of time times E were calculated next, which is simply the time data
point multiplied by the relative E curve data point. These points are used to find t bar
which is the integral from zero to infinity of the tE curve. For this integration, the
trapezoidal method for n equally spaced points was used. The integration and
differentiation formulas are displayed here.
 To calculate tau for an ideal reaction, the initial volumetric flowrate is required. This
is equal to the sum of the initial flows of KCL and water which varies from run to run.
Tau is then calculated by dividing the reactor volume by this volumetric flowrate.
 Lastly, two different single parameter models were used to determine the
conversion in each run. The first model explored was the segregation model. The
conversion formula is displayed on the board, and for experimental conversion t bar
is used in place of tau in the equation. It is assumed that the formation of the RTD is
not affected by temperature, as the graphs are derived from a ratio of
concentrations at the same temperature. A change in temperature will not affect the
ratio, and therefore, an RTD at the temperature of the given rate constant is used,
despite being at a greater temperature than that which was operated at.
 The next model explored is the tanks in series model. Firstly, the variance was
calculated using the variance formula which is displayed on the board. The number
of tanks required is then calculated by dividing the square of t bar by variance, as per
the formula. For an ideal CSTR, it is assumed that 1 tank is necessary. The conversion
is calculated using the formula, where tau n is equal to t bar for the experimental
conversion.

Results and Analysis:

Now we will analyse the results and graphs produced from the data provided and collected.
Looking at run 1 where there was 100mL/min of Potassium Chloride solution and
100ml/min of water at 50 rpm, it is evident that t bar is less than tau. This suggests that
there is a zone of dead volume thus the active volume decreases, decreasing t bar. The dead
volume was found to be 28% of the total volume.
A similar scenario was discovered for case two where the potassium chloride solution was
decreased to 50 mL/min. 41% of the volume was calculated to be dead volume, a greater
percentage than run 1.

A greater dead volume suggests a reactor that is operating further away from idealality.

The tanks in series model calculates a lower number of tanks for run 1 than run 2, which
therefore also suggests that run 1 is more ideal. The difference between run one’s
calculated conversion is smaller than the difference between run 2s calculated conversion
however run twos calculated conversion is higher.

Lastly, the segregation model produces a similar relationship to the tanks in series model as
the ideal and experimental conversions for run 1 are more similar to that of run 2, which
further reinforces that run one is operating more ideally. This is expected as the flow rate in
run 1 is higher than run 2.

As expected the reactor RTD curves differ from the theoretical curves.

The plotted e curves are displayed on the board. The ideal curve is plotted in orange and as
you can see, neither of the runs are operating ideally, however, although it is harder to
graphically analyse the relationships due to dispersed data, one may infer that run 1 is more
like the ideal curve. The multiple decaying peaks in the e curves suggest that there may be
internal recirculation.

NECESSITY FOR STIRRING

Conclusions:

The data collected allows us to analyse optimal mixing configurations and optimal flowrates
being fed into the CSTR. From the runs and the single parameter models it was observed
that the second run obtained the greatest conversion compared to the other two runs
which is surprising seeing as this run had less tracer than the other two runs and operated at
a lower flowrate of 50rpm and has the largest dead volume percentage. However, run 1 at
0.1L/min and 50rpm has the most ideal mixing configuration as its value for t bar is the
closest to tau. Run 3 also had a t bar close to tau but not as close as run 1. It was found that
increasing stirring speed does not decrease the percentage of dead volume. Impeller speed
also does not have a large affect on conversion as conversion is largely based flowrates.

Recommendations:

Recommendations for this experiment include, running the impeller at greater speeds to
further understand the relationship between impeller speed and conversion. Including
baffles in the CSTR to minimize vortex formation and reduce dead volume zones. Lastly
would be to introduce different impeller types to see how impeller blades and shapes affect
mixing.