You are on page 1of 10

ECE-132; No.

of Pages 10
ARTICLE IN PRESS
education for chemical engineers x x x ( 2 0 1 6 ) xxx–xxx

Contents lists available at ScienceDirect

Education for Chemical Engineers

journal homepage: www.elsevier.com/locate/ece

Design of a laboratory experiment for the


performance analysis of a retrofitted tray dryer unit

S.M. Zakir Hossain a,∗ , Noureddine Mansour a , Nahid Sultana b


a Department of Chemical Engineering, University of Bahrain, P.O. Box 32038, Bahrain
b Department of Computer Science, University of Dammam, Building 600, Main Campus, Saudi Arabia

a r t i c l e i n f o a b s t r a c t

Article history: Drying of products and raw materials is a widely used procedure in many manufacturing
Received 10 March 2016 processes and chemical plants. The work described aims to improve the learning experi-
Received in revised form 28 ence of students in a senior level undergraduate chemical engineering laboratory. The tray
September 2016 dryer unit used in the experiment was first retrofitted with a PC, a data acquisition card and
Accepted 12 October 2016 LabVIEW software. Then a 2k full factorial design and analysis of variance (ANOVA) were
Available online xxx used to analyze the operation of the retrofitted dryer by the variation of two parameters:
air temperature and velocity. The interactive effects of these parameters on the drying rate
Keywords: were also evaluated. Air temperature has the most significant effect on the drying rate, while
Tray dryer the air velocity does not have a significant effect. Furthermore, interactive effects were not
Analog system observed for both parameters. The results indicate that the performance of the retrofitted
LabVIEW unit has improved by the incorporation of the intuitive and easy to use interface, which
Computerization allows the students to monitor dynamic data and control the unit in real-time. Results of
Design of experiment the student survey indicate that the level of understanding of process design and optimiza-
tion has increased upon course completion. Most of the students believe that the lab is
challenging and interesting, and would be worthwhile for lifelong learning. Overall, this
approach not only exposes students to a hands-on industrially relevant unit operation, but
it also complement classroom teaching on LabVIEW and statistical design of experimental
concepts.
© 2016 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1. Introduction ties of the solid product may occur. Almost all manufacturing
processes require the drying of a product or raw material and
Drying, which is the final removal of water or another volatile the equipment used in these industrial operations depend
liquid from a wet solid, is performed before packaging some on the particular process and type of material being dried.
products to improve their storage life and reduce the trans- For example, the equipment used for food processing must
portation costs by lowering the weight of a product. Drying is meet tougher requirements than those used for producing fer-
a complex unit operation involving transient transfer of both tilizer. Thus, the drying process is governed by the relative
heat and mass along with several processes, such as physi- importance of several factors including heat sensitivity, poros-
cal or chemical transformations that can change the product ity, bulk density, particle size etc. (Margaris and Ghiaus, 2006;
quality. Physical changes that may occur include: shrinkage, Kotwaliwale et al., 2007).
puffing, crystallization, glass transitions etc. In some cases, A number of chemical engineering laboratory drying exper-
desirable or undesirable chemical or biochemical reactions iments, including microwave drying of sand (Steidle and
that can cause changes in color, texture, odor, or other proper- Myers, 1999), and convection drying of a towel, have been


Corresponding author. Fax: +973 1768 0935.
E-mail address: zhossain@uob.edu.bh (S.M. Zakir Hossain).
http://dx.doi.org/10.1016/j.ece.2016.10.001
1749-7728/© 2016 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Please cite this article in press as: Zakir Hossain, S.M., et al., Design of a laboratory experiment for the performance analysis of a retrofitted
tray dryer unit. Education for Chemical Engineers (2016), http://dx.doi.org/10.1016/j.ece.2016.10.001
ECE-132; No. of Pages 10
ARTICLE IN PRESS
2 education for chemical engineers x x x ( 2 0 1 6 ) xxx–xxx

reported in the literature (Nollert, 2002). A bench-scale exper-


imental drying apparatus (Moor, 2001) and the statistical
treatment of drying data (Prudich et al., 2003) have also been
reported. However, to the best of our knowledge, studies on
simple upgrading of the drying unit and the incorporation of
a statistical experimental design element in laboratory drying
experiments have not been reported. A tray dryer is considered
in this study as it is one of the most common unit opera-
tions found in the process industry (Colak and Hepbasli, 2007;
Corzo et al., 2008; Ghasemkhani et al., 2016; Aviara et al., 2014).
Besides, students can study aspects of fluid mechanics, sur-
face chemistry, solid structure, and mass and heat transfer
associated with the general drying process using a tray dryer.
Fine particles of the solid substance that needs drying is evenly Fig. 1 – Armfield UOP 8 tray dryer unit where, temperature
distributed on the tray, which is heated through conduction and relative humidity sensors: before (1a) and after (1b) the
or radiation using hot air swept across the tray. However, for trays, electronic balance (2), air-fan with speed controller
many years, students have conducted experiments on the (3), air outlet duct section (4), PC running the LabVIEW
unit using instruments which are virtually obsolete in mod- software (5).
ern industry, making it very tedious and time consuming
to perform any experiment. For example, wet and dry bulb on a response variable (drying rate). The most significant
temperatures of air are measured using an aspirated psy- parameter and the interactive effects of the drying process
chrometer and data are manually collected using a stop watch. are also reported. Overall, hands-on experimental learning
An advanced laboratory setup using new equipment is a costly through the application of LabVIEW and statistical experi-
option for academic institutions. An alternative low-cost solu- mental design can enhance the understanding of the course
tion is to retrofit the existing laboratory containing proven subject by the students. The retrofitted unit can be a potential
systems with computer-controlled data acquisition systems industrially relevant pivotal tool for a wide range of experi-
and to develop custom-written software to suit the existing ments, including the performance analysis of drying of food
experimental modules. and biological materials in real-time.
Therefore, one of the objectives of this study was to over-
come the constraints of the existing system and to make 2. Experimental setup and procedure
use of modern teaching and learning tools by incorporating
computer-based educational tools for student use, either in 2.1. Description of the apparatus
the lectures or in the laboratory or both (Ertuggrul, 2000).
Accordingly, the tray dryer unit was retrofitted locally with An Armfield UOP 8 type tray dryer unit was used in this study
a PC, USB-120LS data acquisition card (Measurement Com- and the retrofitted version of the unit is shown in Fig. 1. The
puting), and custom-written LabVIEW (Laboratory Virtual tray dryer consists of an air duct mounted on a frame standing
Instrument Engineering Workbench) software to suit the exist- on the floor to provide a comfortable working height for the
ing laboratory experiments. LabVIEW was used because it is operator. Air is drawn into the duct through a mesh guard by a
easy to learn, adapt, and apply. Some of its features include motor-driven axial-flow fan impeller with a controllable speed
a graphical user interface (GUI), built-in engineering-specific (3) to produce a range of air velocities. The air passes over a
libraries of software functions, hardware interfaces, data anal- computer-controlled electrically heated element to provide a
ysis, and visualization. Software allows the users to select variation in air temperature up to a maximum of 80 ◦ C at low
their own input values and manipulate them in a manner air velocities. The air passes into the central section of the duct
that resembles a real laboratory (Essik, 2009; Xu, 2015; Agarwal where four trays containing the material to be dried (e.g., wet
et al., 2013; Popović et al., 2011; Wang et al., 2007). sand) are suspended in the air stream. The trays are mounted
Another objective of this study was to develop a labora- on a support frame, which is attached to an electronic balance
tory where students are exposed to statistical experimental placed above the duct on which the total weight is continu-
design (DOE) concept in a quasi-industrial setting. The DOE ously indicated (2). The trays are inserted or removed from
concept is an important statistical technique for developing the duct through a latched side door with a glass panel for
a black box model correlating various process variables or viewing purposes. After passing over the drying trays, the air
factors. The technique also uses statistics to obtain meaning- is discharged to the atmosphere through an outlet duct section
ful relationships through a minimum number of experiments where a digital anemometer is used to measure the air velocity
(Montgomery, 2001; Montgomery and Runger, 2006; Aftanasar (4). Temperature and relative humidity of the heated air before
et al., 2015; Box et al., 1978). Students can easily evaluate and after the trays are monitored continuously using appropri-
which subset of the process variables has the most signifi- ate sensors (1a and 1b). All dynamic data are collected through
cant effect on process performance or response variable by a PC running the LabVIEW software (5), which is programmed
performing controlled tests and analyzing the results. They specifically for this experiment.
can also investigate the interactive effects among the process
variables. Accordingly, a retrofitted tray dryer unit was used 2.2. Instrumentation and data acquisition
to perform a real drying process and analyze the results. A
2k full factorial design and analysis of variance (ANOVA) were The air temperature and the relative humidity upstream and
used to evaluate the operation of the dryer unit by manipulat- downstream of the trays are measured (Fig. 1a and b) using
ing two main variables: i.e., air temperature and air velocity duct type temperature and humidity sensors (SE-MFD3110)
with a range of 0–100% for relative humidity (±2% RH) and

Please cite this article in press as: Zakir Hossain, S.M., et al., Design of a laboratory experiment for the performance analysis of a retrofitted
tray dryer unit. Education for Chemical Engineers (2016), http://dx.doi.org/10.1016/j.ece.2016.10.001
ECE-132; No. of Pages 10
ARTICLE IN PRESS
education for chemical engineers x x x ( 2 0 1 6 ) xxx–xxx 3

to communicate with the data acquisition card. Mimic dia-


grams were developed using LabVIEW to acquire data, monitor
the unit, and change the set point of the computerized tem-
perature controller. The data acquired by the computer are
temperature and humidity before and after drying the mate-
rials in the trays and the weight of the material. All data logged
are exportable to general purpose spreadsheets (i.e., MS Excel).
Fig. 3 shows a snapshot of a mimic diagram of the unit. The
temperature set point is accessed by pressing on the “Heater”
tab. The sampling time to control the temperature is entered
via the home page. The “Datalog” button must be pressed in
order to initiate data logging using software by clicking “Start
Logging”, which samples data every minute. The data gathered
Fig. 2 – I/O signals are connected to computer via DAQ card. are exported automatically to an MS Excel file. The graphs of
temperature/humidity before and after drying the material in
−40 ◦ C to 80 ◦ C (±0.3 ◦ C) for temperature, respectively. The ana- the trays and their weight can be viewed by clicking “Position
log voltage output ranges are 0–10 V for both temperature and 1”, “Position 2”, and “Load Cell” buttons, respectively.
humidity sensors. The weight is determined and monitored
using an electronic load cell (2) MODUS-MODEL323 with a
range of 0–5 kg, and the analog voltage output representing 2.3. Determination of the rate of drying
the weight on the load cell is sensed for continuous moni-
toring of the weight. The fan and its control system was not The upgraded tray dryer unit is used for the ‘hot air drying’
changed. The fan is operated using the on/off switch and the lab experiment. The rate of drying is an important parameter
fan speed is controlled using the knob (3) as shown in Fig. 1. that must be determined in this experiment and the deriva-
However, the malfunctioning obsolete control system of the tive of water content with respect to time is required for its
heating unit was upgraded and replaced with a computerized calculation. The moisture content in the solid (in kg water/kg
on/off control system. The temperature of the heating ele- dry solid) is determined through:
ment can be varied up to 80 ◦ C and is monitored continuously
using LM35 series sensors, which are accurate temperature W − WS
Xt = (1)
sensors based on integrated circuits. The outlet air velocity WS
(4) is monitored using a digital thermo-anemometer (EXTECA
Instruments) with a range of 0.4–30 m/s and a resolution of where W is the total weight of solid at different times in the
0.01 m/s. The PC is interfaced to the tray dryer using a Mea- drying period (kg of wet solid) and Ws is the weight of the dry
surement Computing USB-120LS DAQ card. The card provides solid in (kg).
8 analog input (AIs) and 2 analog outputs (AOs). A signal con- Then, the free moisture content in unit kg dry solid is cal-
ditioning circuit was designed and installed at the interface culated by subtracting the equilibrium water content from the
between the card and the tray dryer to match the signals moisture content of water at any time as follows:
to/from the tray dryer with the ranges of the DAQ card.
As shown in Fig. 2, all I/O signals are connected to the
computer via the DAQ card and LabVIEW software is used X = Xt − X∗ (2)

Fig. 3 – Snapshot of the mimic diagram of the unit.

Please cite this article in press as: Zakir Hossain, S.M., et al., Design of a laboratory experiment for the performance analysis of a retrofitted
tray dryer unit. Education for Chemical Engineers (2016), http://dx.doi.org/10.1016/j.ece.2016.10.001
ECE-132; No. of Pages 10
ARTICLE IN PRESS
4 education for chemical engineers x x x ( 2 0 1 6 ) xxx–xxx

where X is the free moisture in (kg water/kg dry solid), Xt is the


Table 1 – Treatments for 2k full factorial design.
moisture content at time t (kg water/kg dry solid) and X∗ is the
equilibrium water content in (kg moisture/kg dry solid). Treatment Air temperature Air velocity

The experimental data are fit to the relationship between 1 −1 −1


the water content in unit kg dry solid versus time (dx/dt). Then, 2 0 0
it can be used to determine the rate of drying as follows: 3 −1 1
4 1 1
−Ls
 dX  5 1 −1
R= × (3) 6 1 −1
A dt
7 −1 1
where R is the drying rate in (kg of water per h m2 ), Ls is the 8 1 1
9 0 0
weight of dry solid in (kg), A is the exposed surface area in
10 −1 −1
(m2 ), dX is the change in water content in kg H2 O per kg dry 11 −1 −1
solid and dt is the change in time in hours. 12 1 −1
13 −1 1
2.4. Design of experiments 14 0 0
15 1 1
Design of experiments (DOE) is a systematic method to
determine the relationship between the factors affecting a Table 2 – Levels and ranges of each parameters.
process and the output of that process (Montgomery, 2001;
Factors Levels
Montgomery and Runger, 2006; Aftanasar et al., 2015; Box et al.,
1978). In other words, DOE is used to find cause-and-effect + − 0
relationships. This information is needed to manage process ◦
Air temperature ( C) 45 35 40
inputs in order to analyze or optimize the output. In this study, Air flow rate (m/s) 3.33 2.5 3
a 2k full factorial design with two levels (high and low) and
one center point was used to analyze the effect of two input
the end of each run, and then for the experiment was repeated
parameters, air temperature and air velocity on the drying
with a different combination of air speed and temperature.
rate (response variable). A total of 15 experimental runs or
treatments (in triplicate) were conducted. The behavior of the
2.6. Approach for data collection
process is described by an equation (Eq. (4)), that accounts for
the independent factors and all interaction terms regardless
From the design of the experiments, it is obvious that a total
of their significance.
of 15 experiments are required and it is impossible to com-
plete all these experiments in one laboratory session due to

N

N

N
y = ˇ0 + ˇi xi + ˇij xi xj (4) time constraints. Each laboratory session held once a week,
comprising of a lecture and experimentation, is three and
i=1 i=1 J=1
half hours long for each student group. The instructor pro-
where y = predicted response, N = number of factors, vides instructions about the instruments and experimental
xi = coded variables, ␤o = intercept term, ␤i = linear effect, procedure for the first thirty min and then students start the
and ␤ij = interacting effect. experiment. Thus the entire experiment was completed over
For statistical calculations, the true variables/factors Xi the course of several laboratory sessions and the data gathered
were coded as xi according to the following equation: were shared among all groups of students. In the most recently
completed term there were a total of 26 students working in
xi = (Xi − Xo )/X, i = 1, 2, 3...N (5) groups of five. Each group was assigned a set of three exper-
iments at different operating conditions to generate a total
where xi = coded value of the true variable Xi ; Xo = the value of of 15 data points. By assigning several runs to each group,
Xi at the center point and X = the step change. the instructors were able to ensure that the experiments were
Table 1 lists the treatments employed, while Table 2 lists randomized.
the range and level of the two independent variables studied.
Treatments were performed in a fully random order in order 2.7. Survey of the students and statistical analysis
to avoid bias. MINITAB (V-14) was used for DOE analysis.
A survey was conducted to evaluate the response of the stu-
2.5. Experimental procedure dents. The survey form consists of twenty questions which
is shown in Appendix A. Out of these, thirteen questions are
Sand was filtered first with a mesh, thus removing larger peb- rated on a point scale of 1 = poor, 2 = fair, 3 = good, 4 = very
bles, to ensure the uniformity of the solid to be dried. The four good, and 5 = excellent. Three questions are rated as ‘yes
trays must be rinsed with water and dried well before con- and no’ while the remaining questions are to obtain gen-
ducting experiments. Then the weight of the empty trays was eral comments on this specific lab experiment. Data (student
determined and recorded. Air speed and heater temperature responses) were analyzed using the statistical package SPSS,
are set according to the design of experiment and treatments version 17. Descriptive statistics (i.e., frequencies, percent-
(Table 2). A 590 g sample of sand was weighed for each treat- ages, means, and standard deviations) were reported for all
ment, and 100 mL of water was poured over the sand. The tray Likert-type statements. Chi-square (X2 ) test was used to deter-
was placed in the equipment, and dynamic data (tempera- mine the relationship between the responses to the question
tures, relative humidity, weight, air speed etc.) were recorded ‘students level of understanding at the end of the course’ (Q11)
using the data logger for 30 min. The equipment was reset at with the responses to the question ‘lab intellectually challeng-

Please cite this article in press as: Zakir Hossain, S.M., et al., Design of a laboratory experiment for the performance analysis of a retrofitted
tray dryer unit. Education for Chemical Engineers (2016), http://dx.doi.org/10.1016/j.ece.2016.10.001
ECE-132; No. of Pages 10
ARTICLE IN PRESS
education for chemical engineers x x x ( 2 0 1 6 ) xxx–xxx 5

Fig. 4 – Drying curves at constant drying conditions where, (a) the water content (kg H2 O/kg dry solid) vs. time (h), and (b)
the drying rate (kg H2 O/h m2 ) vs. water content (kg H2 O/kg dry solid).

ing and stimulating’ (Q12), and the responses to the question


Table 3 – Effect estimate summary and analysis of
‘useful in the future and lifelong learning’ (Q20). A p-value variance (ANOVA) for the drying rate (coded unit).
<0.05 was considered statistically significant at the 5% level,
Term Effect Coef SE Coef T P
while 0.05 < p-value < 0.1 was considered significant at the 10%
level. Constant 0.9300 0.1336 6.96 0.000
Block 1 0.1360 0.1690 0.80 0.444
Block 2 −0.1480 0.1690 −0.88 0.407
3. Observations and discussion T 0.6867 0.3433 0.1336 2.57 0.033
V −0.5767 −0.2883 0.1336 −2.16 0.063
T×V −0.6233 −0.3117 0.1336 −2.33 0.058
3.1. Application of LabVIEW and the design of
experiment (DOE) to the tray dryer unit S = 0.462855, R-Sq = 89.77%, R-Sq(adj) = 87.10%.

The tray dryer unit described in this article was developed


and implemented over the past two years in a chemical engi-
the results of previous research conducted on drying fruits
neering senior laboratory course at the University of Bahrain.
and vegetables (Colak and Hepbasli, 2007; Shalini et al., 2008;
An existing analog tray dryer unit was retrofitted successfully
Sacilik and Elicin, 2006). Overall, drying curves follow the gen-
using a PC, USB-120LS data acquisition card, and LabVIEW
eral trends of drying curves, in agreement with the theory of
software by a group of students. The upgraded tray dryer unit
the drying process (Seader and Henley, 2006).
is now very simple, intuitive, and user-friendly for students
In the statistical experimental design, two-level full facto-
to monitor dynamic data, which also allows them to conduct
rial design and the analysis of variance (ANOVA) techniques
several lab experiments in real-time. Some of these experi-
were implemented using MINITAB (V.14) statistical software.
ments are: (i) demonstration of drying rate regimens, (ii) heat
The levels and treatment were mentioned in detail in the
and mass transfer analogies, (iii) plotting drying curves, (iv)
experimental section. The experiments were conducted in
influence of drying time on the equilibrium moisture content,
triplicate to account for variability. The ANOVA was performed
(v) influence of the particle size, (vi) influence of the air speed,
for 95% confidence interval which is shown in Table 3. The
(vii) influence of the air temperature, (viii) application of psy-
coefficients of the equation and their values are also listed in
chrometry in the drying, etc. For demonstration purposes, the
Table 3. Based on the DOE and the results of the experiments,
effect of the two drying parameters, air temperature and air
the linear regression model for predicting the drying rate can
velocity on the drying rate was investigated and the results
be written as follows:
analyzed based on the statistical experimental design.
Fig. 4 depicts the drying curves of wet sand at constant
drying conditions for one of the experimental data sets. Dry- y = 0.93 + 0.34T − 0.29V − 0.31TV (6)
ing curves describe the drying characteristics of a particular
product under specific conditions. Fig. 4a shows that water or The determination coefficient (R2 ) was evaluated to test
moisture content of sand decreases with time, while Fig. 4b the fit of the model. The adjusted determination coefficient
depicts the drying rate in (kg H2 O/h m2 ) versus the water con- (adj. R2 ) of 87.10%, indicates that the model can explain 87.1%
tent in (kg H2 O/kg dry solid). The water content of the wet of the variability and only 12.9% of total variations were not
sand decreases with drying time for all conditions of temper- explained by the model, which demonstrates the goodness of
ature and air velocity. However, the drying time varies with the the fit and confirms the adequacy of the regression model.
drying conditions, the drying time being lower at higher dry- The p value is less than 0.05 for the parameter air tempera-
ing temperature. The relationship between the moisture loss ture, indicating that this factor has a strong significant effect
and the drying time is apparently non-linear, with the initial on the drying rate and thus the null hypothesis is true for
moisture loss higher due to the release of free moisture, as this case. While the p value is greater than 0.05 for param-
compared to the latter part of drying. Drying rate as a function eter, air velocity, rejecting the null hypothesis, indicating that
of the water content also exhibits non-linear behavior. In this this parameter has an insignificant or low effect on the dry-
study, the drying rate was observed to decrease with decreas- ing rate. The data also indicate that there is little or no (since
ing moisture content (Fig. 4b), which is in agreement with p > 0.05) interactive effect between these two parameters.

Please cite this article in press as: Zakir Hossain, S.M., et al., Design of a laboratory experiment for the performance analysis of a retrofitted
tray dryer unit. Education for Chemical Engineers (2016), http://dx.doi.org/10.1016/j.ece.2016.10.001
ECE-132; No. of Pages 10
ARTICLE IN PRESS
6 education for chemical engineers x x x ( 2 0 1 6 ) xxx–xxx

Main Effects Plot (data means) for Drying rate


Air temperature Air flow rate Point Ty pe
1.50 Corner
Center

1.25
Mean of Drying rate

1.00

0.75

0.50
-1 0 1 -1 0 1

Fig. 5 – The main effects plot of drying rate (R).

Interaction Plot (data means) for Drying rate


2.25 Air temperature Point Type
-1 Corner
0 Center
2.00
1 Corner

1.75

1.50
Mean

1.25

1.00

0.75

0.50

-1 0 1
Air flow rate

Fig. 6 – Interaction plot for drying rate.

Normal Probability Plot of the Standardized Effects


(response is Drying rate, Alpha = .05)
99
Effect Ty pe
Not Significant
95 Significant

90 F actor N ame
A A ir temperature
80 A B A ir flow rate

70
Percent

60
50
40
30
20

10

1
-3 -2 -1 0 1 2 3 4
Standardized Effect

Fig. 7 – Normal probability plot of the effects.

Fig. 5 shows the effect of each of the parameters (e.g., tem- of each factor affect the response differently. The steeper the
perature and air velocity) on the drying rate. Different levels slope of the line, the greater the magnitude of the main effect.

Please cite this article in press as: Zakir Hossain, S.M., et al., Design of a laboratory experiment for the performance analysis of a retrofitted
tray dryer unit. Education for Chemical Engineers (2016), http://dx.doi.org/10.1016/j.ece.2016.10.001
ECE-132; No. of Pages 10
ARTICLE IN PRESS
education for chemical engineers x x x ( 2 0 1 6 ) xxx–xxx 7

Table 4 – Means and std. deviations of responses for each of the thirteen statements in the student survey.
Questions Mean Std. deviation

Q1. The course description and the objectives of the lab are clear. 4.42 0.703
Q2. The lab makes you aware of safety consideration, instrumentation and process optimization. 4.58 0.504
Q7. Level of understanding about DOE, factor screening and optimization prior to take this course. 1.15 0.368
Q8. Instructor is enthusiastic about teaching the course. 4.27 0.962
Q9. Several lectures given by the Instructor were helpful for understanding the basics of DOE and its application. 4.19 0.801
Q10. Have you learned the practical usage of DOE, factor screening, and process optimization? 4.42 0.643
Q11. Overall, level of understanding about DOE, factor screening and optimization at the end of the course. 4.19 0.849
Q12. The DOE lab is intellectually challenging and stimulating. 4.31 0.788
Q14. Is upgraded tray dryer unit suitable for DOE study? 4.62 0.496
Q15. Is the user interface (developed by using Labview) intuitive and easy to use? 4.23 0.652
Q17. How do you rate the quality of the DOE lab? 4.58 0.504
Q18. The course is interesting. 4.35 0.689
Q20. This type of lab experience would be useful in the future and lifelong learning 4.12 0.711

The data indicate that drying rate increases with increasing 100%
temperature but decreases with increasing air velocity at all 90%
temperatures used. Fig. 6 shows the interactive effect of the excellent
80%
two input parameters, temperature and air velocity, on the 70% very good
drying rate. The data show the absence (or low) of interac- 60% good
tive effects on the drying rate, which is in agreement with the 50% fair
ANOVA results. Fig. 7 depicts the normal probability plot of 40% poor
the effects and all the effects that lie along the line are neg- 30%
ligible and the important effects are far from the line. The 20%
results indicate that the most important factor in the drying
10%
process is temperature (A), which is consistent with the data
0%
reported previously (Colak and Hepbasli, 2007; Shalini et al., 1 2 7 8 9 10 11 12 14 15 17 18 20
2008). This is probably due to the air temperature having a Question number
strong and directly proportional relationship with the drying
Fig. 8 – Percentage of responses for each of the thirteen
rate as the temperature gradient works as the driving force for
questions in the student survey. Most of the students
the heat transfer which enhances the rate of moisture evapo-
agreed with a rating of very good or excellent for all the
ration. However, a higher air velocity decreases the residence
questions expect question 7 (Table 4 shows all the
time of hot air, lowering the amount of heat transferred from
questions).
air to the wet material. Therefore, the rate of drying decreases
with increasing air velocity (Fig. 6). These observations support
the theory of drying phenomena (Seader and Henley, 2006). of the thirteen statements are plotted in Fig. 8. Most of the stu-
Overall, the upgraded unit is equipped with a computer- dents agreed (with a rating of very good or excellent) that they
controlled front panel that allows students to monitor learned about design of experiments, regression, and opti-
dynamic data and control the unit in real-time. The measured mization of an industrially relevant unit operation from this
values of temperature, humidity, air velocity, drying time, and course. The responses to questions 4, 5, and 6 indicate that,
weight are displayed in real-time and are exported automati- although most of the students have completed an undergrad-
cally to an MS Excel file. This lab experiment has also provided uate statistic course, they have not been exposed to DOE. Only
opportunities for students to achieve ABET (ABET, 2004) Cri- a few students have completed an advanced statistics course
terion of student outcome, SO—b (ability to design, conduct, (as a special topic). Even though their theoretical knowledge
analyze, and interpret experiments) and ABET (ABET, 2004) about DOE is high, they lack practical experience in using DOE.
Criterion of SO—d (ability to perform as part of a team). The overall level of understanding about DOE, factor screen-
ing, and optimization prior to taking this course (Q7) is low.
3.2. Student evaluation and learning The students found that both MATLAB and MINITAB are use-
ful tools for this exercise (Q13). They also found that LabVIEW
Each semester, student evaluations are normally completed is a useful tool for acquiring and monitoring data for this lab-
at the end of a course, conforming to the mandatory policy oratory experiment and they do not need any prior knowledge
at the University of Bahrain. These evaluations are in general about the software (Q16).
related to the overall quality of a course. In addition to the eval- Fig. 9 shows how students’ overall, level of understanding
uation, a survey was also conducted with specific questions about DOE, factor screening and optimization at the end of
related to the targeted learning aspects for this laboratory. the course (Q11) affects on the responses of Q12 (the DOE lab
The results of the survey are preliminary, considering that the is intellectually challenging and stimulating) and Q20 (this lab
population surveyed is small. Table 4 shows the values of the experience would be useful in the future and lifelong learn-
mean and standard deviation of responses for each of the thir- ing). The results (in both Fig. 9a and b) indicate that the
teen Likert-type statements in the student survey. The values percentage responses to the Q11 with rating of excellent, very
of the mean ranged from 4.12 to 4.62 except for statement good and good are about 46%, 27% and 27%, respectively. Out
7, indicating that the other twelve statements are negatively of the 46% (with a rating of excellent), as shown in Fig. 9a,
skewed, with over 50% of the respondents giving a rating of majority (∼42%) strongly believe (very good 11.54% + excellent
very good or excellent. The percentages of responses for each 30.77%) that the DOE lab is intellectually challenging and

Please cite this article in press as: Zakir Hossain, S.M., et al., Design of a laboratory experiment for the performance analysis of a retrofitted
tray dryer unit. Education for Chemical Engineers (2016), http://dx.doi.org/10.1016/j.ece.2016.10.001
ECE-132; No. of Pages 10
ARTICLE IN PRESS
8 education for chemical engineers x x x ( 2 0 1 6 ) xxx–xxx

~46% ~46%
35 a 25 b
Q20: This type of
30.77 Q12: The DOE lab ~27%
lab experience
30 is intellectually ~27%
20 19.23 19.23 would be useful
~27% challenging and in the future and
25 stimulating 15.38 15.38 lifelong learning
~27% 19.23
Percent

15

Percent
20 Good
15.38 Very Good
11.54 11.54 Good
15
11.54 11.54 Excellent
10
7.69 Very Good
10 7.69
5 Excellent
5 3.85
0 0 0 0
0 0
Good Very Good Excellent Good Very Good Excellent
Q11: Overall, students' level of understanding about DOE, factor Q11: Overall, students' level of understanding about DOE, factor
screening and optimization at the end of the course screening and optimization at the end of the course

Fig. 9 – Chart of the relationship between the responses to the questions where, (a) the responses to the Q11 vs. the
responses to the Q12, and (b) the responses to the Q11 vs. the responses to the Q20.

stimulating. While, about 35% strongly believe (very good formance analysis of drying of food and biological materials in
15.36% + excellent 19.23%) that this lab experience would be a dynamic environment. Furthermore, a hands-on tutorial in
useful in the future and lifelong learning, as shown in Fig. 9b. experimental design, regression, and optimization of a pro-
These survey data indicate that the students who have high cess system is a great tool to capture the imagination and
level of understanding for Q11, they also have the high level of enthusiasm of students. The experiment also helps students
understanding as well as deep thinking for questions 12 and to design and optimize any complex industrial process where
20, as expected. However, the p-values of the chi-square (␹2 ) a number of parameters with different levels have to be opti-
test indicate that the responses to the Q12 is significantly asso- mized. The survey results indicate that the students found
ciated (p-value = 0.042) with the responses to the Q11, while the experience both challenging and worthwhile. Thus, this
the responses to the Q20 is not significantly associated (p- experience may lead them toward life-long learning of pro-
value = 0.24) with the responses to the Q11. cess design and optimization. Overall, this novel learning and
In order to evaluate the understanding and the problem teaching approach has a great potential for the improvement
solving skills of the students, an assignment related to this of other chemical engineering lab experiments. Also, there is
particular experiment was given, which is shown in Appendix an option to integrate LabVIEW software with any statistical
B. In addition, each group of students are evaluated and graded package, and thereby LabVIEW alone can be a very useful soft-
based on a written report and oral presentation. Overall, stu- ware package for both retrofitting and design of experiments.
dents provided positive feedback to the experimental learning Work in this direction is under way.
approach as they enjoyed working with this on-line tray dryer
experiment and they would like to recommend this course to Appendix A.
their peers (Q19). The students found the laboratory experi-
ment to be challenging, and most of them believe that this CHENG 425: Chemical Engineering Lab II.
course would be useful for lifelong learning of process design
and optimization.
Questionnaire on design of a laboratory experiment for the
performance analysis of a retrofitted tray dryer unit.
4. Conclusions
Survey objectives:
In this study, a tray dryer unit was retrofitted with a PC, The aim of this survey is to assess students’ opinion about DOE
USB-120LS data acquisition card, and LabVIEW software, and lab at the end of the course. There are no correct or incorrect
the statistical experimental design (DOE) concept in a quasi- answers and no foreseeable risks associated with this survey.
industrial setting was introduced. To the best of our knowledge
this is the first report that describes both upgrading a tray dryer Instructions:
unit and conducting experiments using the same unit. The
effectiveness of the retrofitted unit has been augmented by its • Your participation in this study is completely voluntary. It
simplicity and the user friendly interface that allows students will take approximately 4–5 min to complete.
to monitor dynamic data and control the unit in real-time. • Your survey responses will be strictly confidential and data
In addition, a 2k full factorial design was used to investigate from this will be reported only in the aggregate. Your infor-
the effects of air temperature and air velocity on the drying mation will be coded and will remain confidential. If you
rate. The ANOVA data and the other results indicate that the have questions at any time about the survey or the proce-
effect of temperature on the drying rate is significant, while dures, you may contact me.
the effect of air velocity is insignificant. Also there are no inter- • Please do not discuss or help the student(s) filling the
active effects between the two factors. The upgraded unit may questionnaire, and delay the discussions until after the
be useful for a wide range of experiments, including the per- questionnaire is submitted.

Please cite this article in press as: Zakir Hossain, S.M., et al., Design of a laboratory experiment for the performance analysis of a retrofitted
tray dryer unit. Education for Chemical Engineers (2016), http://dx.doi.org/10.1016/j.ece.2016.10.001
ECE-132; No. of Pages 10
ARTICLE IN PRESS
education for chemical engineers x x x ( 2 0 1 6 ) xxx–xxx 9

1. The course description and the objectives of the lab are a. Definitely not recommend.
clear. b. Unlikely to recommend.
• poor • fair • good • very good • excellent c. Recommend with reservations.
2. The lab makes you aware of safety consideration, instru- d. Likely to recommend.
mentation and process optimization. e. Recommend with enthusiasm.
• poor • fair • good • very good • excellent
3. Please indicate the primary reason you took this course 20. This type of lab experience would be useful in the future
(you may have a number of reasons but please select only the and lifelong learning.
most important one). • poor • fair • good • very good • excellent

a. I took this class only because it was required. Appendix B.


b. I was interested in the subject matter.
c. I wanted to take a course from this professor. CHENG 425: Chemical Engineering Lab II, Assignment #1, Due
d. I took this class because my first choice was full or didn’t date: December 21, 2015.
fit my timetable. An article describes agricultural experiments for finding
treatment or factor combinations that lead to the best crop
4. Have you completed any undergraduate statistic course growth rate. The design factors are the three types of fertiliz-
before? ers: A, B and C. Each factor is run at two levels and the design
• yes • no is replicated twice. The response variable is the crop growth
5. If your answer is yes in question 3, what is the title of the rate. The growth rate data are shown in Table B1.
course? Calculate the followings:

6. If your answer is yes in question 3, did the course cover i) Magnitude of main effects of A, B, and C.
DOE, factor screening and process optimization? ii) Magnitude of interactions of AB, AC, BC, and ABC.
• yes • no iii) Find the most significant variables and interactions.
7. Level of understanding about DOE, factor screening and iv) Write the appropriate regression model.
optimization prior to take this course. v) show main effects and interactions plots.
• poor • fair • good • very good • excellent
8. Instructor is enthusiastic about teaching the course. Solution:
• poor • fair • good • very good • excellent
9. Several lectures given by the Instructor were helpful for i) From Table B2, the magnitude of A, B and C are −101.62,
understanding the basics of DOE and its application. 7.37, and 306.13 respectively.
• poor • fair • good • very good • excellent ii) From Table B2, the magnitude of AB, AC, BC and ABC are
10. Have you learned the practical usage of DOE, factor −24.88, −153.63,and −2.12, 5.62 respectively.
screening, and process optimization? iii) From Table B2, it is clear that the largest effects are for A, C
• poor • fair • good • very good • excellent and AC interaction. Both A and C are highly significant (p < 0.05).
11. Overall, level of understanding about DOE, factor The AC interaction is also highly significant (p < 0.05).
screening and optimization at the end of the course.
• poor • fair • good • very good • excellent
Table B1 – Crop growth rate data.
12. The DOE lab is intellectually challenging and stimulat-
Run Coded factors Growth rate
ing.
poor • fair • good • very good • excellent A B C Replicate 1 Replicate 2
13. Which statistical software you use for data analysis in
1 −1 −1 −1 550 604
this lab?
2 1 −1 −1 669 650
3 −1 1 −1 633 601
14. Is upgraded tray dryer unit suitable for DOE study? 4 1 1 −1 642 635
• poor • fair • good • very good • excellent 5 −1 −1 1 1037 1052
15. Is the user interface (developed by using Labview) intu- 6 1 −1 1 749 868
itive and easy to use? 7 −1 1 1 1075 1063
8 1 1 1 729 860

• poor • fair • good • very good • excellent


Table B2 – Effect estimate summary and analysis of
variance (ANOVA) for the agricultural experiments.
16. Do you need any previous experience on LabVIEW to
operate the unit? Term Effect Coef SE Coef T P

Constant 776.06 11.87 65.41 0.000


• yes • no A −101.62 −50.81 11.87 −4.28 0.003
B 7.37 3.69 11.87 0.31 0.764
C 306.13 153.06 11.87 12.9 0.000
17. How do you rate the quality of the DOE lab?
A*B −24.88 −12.44 11.87 −1.05 0.325
• poor • fair • good • very good • excellent A*C −153.63 −76.81 11.87 −6.47 0.000
18. The course is interesting. B*C −2.12 −1.06 11.87 −0.09 0.931
• poor • fair • good • very good • excellent A*B*C 5.62 2.81 11.87 0.24 0.819
19. How strongly would you recommend this course to your
S = 47.4612, R-Sq = 96.61%, R-Sq(adj) = 93.64%.
peers?

Please cite this article in press as: Zakir Hossain, S.M., et al., Design of a laboratory experiment for the performance analysis of a retrofitted
tray dryer unit. Education for Chemical Engineers (2016), http://dx.doi.org/10.1016/j.ece.2016.10.001
ECE-132; No. of Pages 10
ARTICLE IN PRESS
10 education for chemical engineers x x x ( 2 0 1 6 ) xxx–xxx

iv) The regression model with significant effects and interac-


tion: y = 776.06 − 50.81A + 153.06C − 76.81AC.

v)

Montgomery, D.C., Runger, G.C., 2006. Applied Statistics and


Main effects: A and C; interaction: only AC.
Probability for Engineers, 3rd ed. John Wiley and Sons.
Montgomery, D.C., 2001. Design and Analysis of Experiments, 5th
References ed. John Wiley and Sons, New York.
Moor, A., 2001. A benchscale drying laboratory illustrating
ABET, 2004. “Criteria for Accrediting Engineering Programs,” combined heat and mass transfer. In: ASEE Mid-Atlantic
Engineering Accreditation Commission. www.abet.org. Section Conference, Rowan University, April.
Aftanasar, M.S., Zalina, A., Ngah, U.K., 2015. Design of experiment Nollert, M., 2002. An easy heat and mass transfer experiment for
(DOE) for microstrip transmission line in RF and microwave transport phenomena. Chem. Eng. Educ. 36 (1), 56–59.
undergraduate course. Comput. Appl. Eng. Educ. 23 (April (5)), Popović, B., Popović, N., Mijić, D., Stankovski, S., Ostojić, G., 2011.
658–666. Remote control of laboratory equipment for basic electronics
Agarwal, A., Uppaluri, R., Verma, A., 2013. LabVIEW based courses: a labVIEW-based implementation. Comput. Appl.
e-learning portal for virtual mass transfer operations Eng. Educ. 21, 110–120.
laboratory. CSI Trans. ICT 1, 75–90. Prudich, M., Ridgway, D., Young, V., 2003. Integration of statistics
Aviara, N.A., Onuoha, L.N., Falola, O.E., Igbeka, J.C., 2014. Energy throughout the undergraduate curriculum: use of the senior
and exergy analyses of native cassava starch drying in a tray chemical engineering unit operations laboratory as an
dryer. Energy 73, 809–817. end-of-program statistics assessment course. In: 2003 ASEE
Box, G.E.P., Hunter, W.G., Hunter, J.S., 1978. Statistics for Annual Conference, Nashville, June.
Experimenters: An Introduction to Design and Model Sacilik, K., Elicin, A.K., 2006. The thin layer drying characteristics
Building. John Wiley and Sons, New York. of organic apple slices. J. Food Eng. 73, 281–289.
Colak, N., Hepbasli, A., 2007. Performance analysis of drying of Seader, J.D., Henley, E.J., 2006. Separation Process Principles, 2nd
green olive in a tray dryer. J. Food Eng. 80, 1188–1193. ed. John Wiley and Sons, US.
Corzo, O., Bracho, N., Vaı́squez, A., Pereira, A., 2008. Optimization Shalini, R., Ranjan, A., Kumar, N., 2008. Studies on the drying
of a thin layer drying process for coroba slices. J. Food Eng. 85, characteristics of apple pomace on tray dryer. In: 16th Int.
372–380. Drying Symposium (IDS 2008), Hyderabad, India, 9–12,
Ertuggrul, N., 2000. Towards virtual laboratories: a survey of November, pp. 1636–1640.
labVIEW-based teaching/learning tools and future trends. Int. Steidle, C., Myers, K., 1999. Demonstrating simultaneous heat
J. Eng. Educ. 16 (2), 1–10. and mass transfer with microwave drying. Chem. Eng. Educ.
Essik, J., 2009. Hands-on Introduction to LabVIEW for Scientists 33 (1), 46–49.
and Engineers. Oxford University Press. Wang, W.B., Li, J.Y., Wu, Q.J., 2007. The design of a chemical virtual
Ghasemkhani, H., Keyhani, A., Aghbashlo, M., Rafiee, S., instrument based on labVIEW for determining temperatures
Mujumdar, A.S., 2016. Improving exergetic performance and pressures. J. Autom. Methods Manag. Chem. 2007, 1–7.
parameters of a rotating-tray air dryer via a simple heat Xu, K., 2015. The design concept of a virtual experiment teaching
exchanger. Appl. Therm. Eng. 94, 13–23. platform for digital logic based on labVIEW. Int. J. Hybrid Inf.
Kotwaliwale, N., Bakane, P., Verma, A., 2007. Changes in textural Technol. 8 (2), 229–236.
and optical properties of oyster mushroom during hot air
drying. J. Food Eng. 78, 1207–1211.
Margaris, D.P., Ghiaus, A.G., 2006. Dried product quality
improvement by air flow manipulation in tray dryers. J. Food
Eng. 75, 542–550.

Please cite this article in press as: Zakir Hossain, S.M., et al., Design of a laboratory experiment for the performance analysis of a retrofitted
tray dryer unit. Education for Chemical Engineers (2016), http://dx.doi.org/10.1016/j.ece.2016.10.001

You might also like