This document discusses using genetic algorithms to optimize parameters in a pneumatic conveying system. A pneumatic conveying system uses compressed air to transfer materials through pipes. The researcher developed an experimental setup and used genetic algorithms to optimize four parameters: blower speed, venturi feeder, rotary valve, and bend angles. Regression analysis was performed to develop an equation to predict material discharge rates based on these parameters. The genetic algorithm and regression equation can identify optimal parameter conditions for maximum material transfer efficiency and cost effectiveness in pneumatic conveying systems.

1. Abstract--- Pneumatic conveying system is one of the integral part of today’s industrial scenario as it has wide
application from pharmaceuticals to power sectors. Application of various optimization techniques is playing vital
role which seeks identification of the best process parametric condition for optimum material transfer and also cost
effectivity. In this research paper, Genetic Algorithm (GA) has been applied for optimizing of Four process
parameters namely, blower speed, venture feeder, rotary valve and different angle bends, are optimized with
considerations of parameter The aim of this research paper is to find the optimum input parameters values for
maximizing the material discharge rates. The Matlab toolbox is used to develop the GA simulation. GA can be used in
optimization problems such as scheduling, materials engineering, optimal control, and so forth.
Keywords---- Pneumatic Conveying System, Genetic Algorithm, Optimization, Regression Analysis Blower Speed,
Venture Feeder, Rotary Valve, Different Angle Bends
I. INTRODUCTION
Pneumatic Conveying System is one of the important and commonly used process for material transfer in
various industries. As the name suggest pneumatic use of compressed air at respective pressure is carried out in
it for the material transfer like sand, cement, fly ash etc. In this process materials can be transported
conveniently to various destinations by means of a stream of high velocity air through pipe lines. Products are
moved through various tubes via air pressure, allowing for extra vertical versatility. The pneumatic conveying
system is divided into two types based upon the quantity of air used and pressure of the system, viz. dense phase
pneumatic conveying system and dilute pneumatic conveying system. In dense phase pneumatic conveying
system two modes of flow are recognized. One is moving bed flow, in which the material is conveyed in the dunes
on the bottom of the pipeline, or as a pulsatile moving bed, when viewed through a sight glass in a horizontal
pipeline. The other mode is slug or plug type flow, in which the material is conveyed as the full bore plugs
separated by air gaps. Dense phase system is often referred to as non-suspension flow. Dilute-phase systems use
push-pull pressure to guide materials through various entry and/or exit points. It is important to note that either
of the air compressors, vacuums, or blowers can be used to generate the air. This will all depend on what the
engineers think will be the most efficient and economical way of developing the system.
A standard optimization technique using genetic algorithm was developed by [01] to solve different
machining optimization problems such as turning, face milling and grinding [02].
P. M. Pradhan, M. Tech. Department of Mechanical Engineering, National Institute of Technology, Rourkela, Orissa, India. E-mail:
premendramani@gmail.com
Yassin Alkassar, M. Tech. Department of Mechanical Engineering, National Institute of Technology, Rourkela, Orissa, India. E-mail:
eng_yassenkassar@hotmail.com
D. K. Behera, M. Tech. Department of Mechanical Engineering, National Institute of Technology, Rourkela, Orissa, India. E-mail:
deepak227cvrce@gmail.com
P.K Mahto, Assistant Professor, Department of Mechanical Engineering, Sikkim Manipal Institute ofTechnology, Majitar, East Sikkim, India.
E-mail: prem4es@gmail.com
Analysis of Material Discharge Rate of
Pneumatic Conveying System using Genetic
Algorithm
P. M. Pradhan, Yassin Alkassar, D. K. Behera and P.K Mahto
National Seminar on Prospects and Challenges of Electrical Power Industry in India - NSPCEPII 339
ISBN 978-93-84743-43-7 © 2015 Bonfring

2. II. EXPERIMENTAL METHODOLOGY AND CONDITIONS
The experimental setup designed consists of a blower, Venturi meter, inclined manometer, rotary valve,
hopper, and a container. The pipes are kept horizontal for some distance and then the bends of different radius
are connected by means of socket. The two extreme ends, i.e. the inlet and the outlet of the pipe are connected to
an inclined manometer by means of flexible pipes having equal diameters to determine the pressure at the inlet
and outlet. The working fluid used in inclined manometer is kerosene.
Fig. 1 Pneumatic Conveying System
 Blower
 Hooper
 Rotary Valve
 Venturimeter
 Inclined manometer
 Bend pipe
 Container
Selection of Process Parameters and their Levels
The design of experiment was carried on for various input parameter i.e. Venturimeter angle, Rotary Valve,
Blower Speed and Bend angle. Various process parameter and there levels were prepared on based upon the
availability of resources and the experimentation was carried out.
Table 1: Shows the Input Parameter Along with their Levels
III. RESULT AND DISCUSSION
After undergoing selection of various process parameter and their level, experiments were performed for the
transfer of materials from one point to other at varying levels of input parameter to see the experimental
discharge from the conveyor system.
Table 2: Experimental Results for Discharge with Varying Process Parameters
Expt.
No
Process parameter level Measured discharge
X(1) X(2) X(3) X(4)
1 5 70 12000 30 1.480
2 5 110 14000 45 1.475
3 5 150 16000 60 1.470
4 10 70 14000 60 1.475
5 10 110 16000 30 1.500
National Seminar on Prospects and Challenges of Electrical Power Industry in India - NSPCEPII 340
ISBN 978-93-84743-43-7 © 2015 Bonfring

3. 6 10 150 12000 45 1.475
7 15 70 16000 45 1.500
8 15 110 12000 60 1.470
9 15 150 14000 30 1.490
IV. DEVELOPMENT OF REGRESSION EQUATION FOR DISCHARGE
Regression equation was developed for the system based upon the input parameters using Minitab 16 and
the predicted discharge was obtained based upon the regression equation.
Regression Analysis
Regression analysis is a statistical technique which is used for modeling and analyzing trends between
dependent and independent variable. Regression analysis helps us to predict data within and outside the range.
If the prediction data is within the range it is known as interpolation and if the prediction data is outside the
range than it is known as extrapolation.
Here, statistical regression technique has been used to model the equation using Analysis of Variance
(ANOVA).
The objective consists of adjusting the parameters of a model function to best fit a data set. A simple data set
consists of n points (data pairs) (xi, yi) i = 1, ..., n, where xi is an independent variable and yi is a dependent
variable whose value is found by observation. The model function has the form f(x, β), where the m adjustable
parameters are held in the vector β. The goal is to find the parameter values for the model which "best" fits the
data.The least squares method finds its optimum when the sum, S, of squared residuals
S= 

n
i
ir
1
2
(1)
Is a minimum. A residual is defined as the difference between the actual value ofthe dependent variable and
the value predicted by the model.
ri = yi - f (xi, β) (2)
An example of a model is that ofthe straight line. Denoting the intercept as β0 and the slope as β1, the model
function isgiven by
f(x, β) = β0 + β1 x (3)
A data point may consist of more than one independent variable. For an example, when fitting a plane to a
set of height measurements, the plane is a function of two independent variables, x and z, say. In the most
general case there may be one or more independent variables and one or more dependent variables at each
data point.
The minimum of the sum of squares is found by setting the gradient to zero. Since the model contains m
parameters there are m gradient equations.
mj
r
r
S
j
i
i
i
j
,....,1,02 





 
(4)
From equations (1) and (4), the gradient equation can be written as:
mj
xf
r
j
i
i
i
,....,1,0
),(
2 


  
 (5)
The gradient equations apply to all least squares problems.
Each particular problem requires particular expressions for
The model and its partial derivatives. A regression model is a linear one when the model comprises a linear
combination of the parameters, i.e.



1
)(),(
j
ijji
xxf  (6)
National Seminar on Prospects and Challenges of Electrical Power Industry in India - NSPCEPII 341
ISBN 978-93-84743-43-7 © 2015 Bonfring

4. Here the coefficients, j
 , is function of i
x .
)(
),(
ij
j
i
Ji
x
xf
x 





 (7)
In case the Least Square Estimate (or Estimator, in the Context of a Random Sample),β is given,
yXXX
TT 1
)(


 (8)
The following regression equation has been developed based on the input parameters. This regression
equation is achieved by using statistical Minitab software. Fig 2, the snapshot view of the results of regression
analysis from Minitab software is shown.
Fig. 2. Snapshot of regression analysis done on Minitab
The equation developed for material discharge is as follows:
Discharge = 1.45 + 0.00117 Venturimeter angle - 0.000083Rotary valve + 0.000004 blower speed–
0.000611 differentbend angle. (9)
Comparison of Experimental and Regression Predicted Value:
A Plot was plotted between the experimental and regression predicted value for discharge. As per Fig3. The
graph of Experimental and regression predicted value were overlapping with each other, therefore regression
predicted value can be used for future prospect.
National Seminar on Prospects and Challenges of Electrical Power Industry in India - NSPCEPII 342
ISBN 978-93-84743-43-7 © 2015 Bonfring

5. Fig. 3. Discharge Vs Experiment No. for Experimental and Regression Predicted Value
Table 3 Comparative Result of Discharge Based on Experimental and Regression Equation
Observation ExperimentalResults Predicted ResultsfromRegressionEquation
1 1.480 1.47971
2 1.475 1.475225
3 1.470 1.47074
4 1.475 1.47523
5 1.500 1.49824
6 1.475 1.469755
7 1.500 1.498245
8 1.470 1.46976
9 1.490 1.49277
V. OPTIMISATION BASED ON GENETIC ALGORITHM
Genetic algorithm was developed and invented by John Holland in 1975, in his well-known book- Adaptation
in natural and artificial systems from Darwin’s theory of survival of the fittest where he describehow to use the
principle of natural evolution to optimization problems and built the first genetic algorithms. The main principle
of genetic algorithms are genetics and evolution [03]. GA comprises of a set of individual elements (i.e.
population) and a set of individual of biologically inspired operators defined over the population itself. As, per
the theory of evolution only the most fittest element can only survive and generate offspring, with their
biological heredity . In computing term genetic algorithm maps a problem on to a set of strings, every individual
string represents a potential solution. The GA manipulates the most promising strings searching for improved
solutions. A GA operates typically through a simple cycle of four stages:
 Creation of a “population” of strings,
 Evaluation of each string,
 Selection of “best” strings, and
 Genetic manipulation, to create the new population of strings.
Evaluation
It measures the fitness of each individual solution in the population and assigns it a relative
Value based on the defining optimization (or search) criteria. Typically in a non-linear programming scenario,
this measure will reflect the objective value of the given model. The selection procedure randomly selects
individuals of the current population for development of the next generation. Various alternative methods have
been proposed but all follow the idea that the fittest have a greater chance of survival [04].
1.45
1.46
1.47
1.48
1.49
1.5
1.51
1 2 3 4 5 6 7 8 9
Discharge
Experiment
Experimental
Value
Regression
Predicted
Value
National Seminar on Prospects and Challenges of Electrical Power Industry in India - NSPCEPII 343
ISBN 978-93-84743-43-7 © 2015 Bonfring

6. 5.2 Crossover
The crossover procedure takes two selected individuals and combines them about a crossover point thereby
creating two new individuals. Simple (asexual) reproduction can also occur which replicates a single individual
into the new population [04].
5.3 Mutation
The mutation procedure randomly modifies the genes of an individual subject to a small mutation factor,
introducing further randomness into the population. This iterative process continues until one of the possible
termination criteria is met: if a known optimal or acceptable solution level is attained; or if a maximum number
of generations have been performed; or if a given number of generations without fitness improvement occur.
Generally, the last of these criteria applies as convergences lows to the optimal solution.
Population size selection is probably the most important parameter, reflecting the size and complexity of the
problem. However, the trade-off between extra computational `efforts with respect to increased population size
is a problem specific decision to be ascertained by the modeler, as doubling the population size will
approximately double the solution time for the same number of generations. Other parameters include the
maximum number of generations to be performed, a crossover probability, a mutation probability, a selection
method and possibly an elitist strategy, where the best is retained in the next generation’s population.
Unlike traditional optimization methods, GA is better at handling integer variables than continuous variables.
This is due to the inherent granularity of variable gene strings within the GA model structure. Typically, a
variable is implemented with a range of possible values with a binary string indicating the number of such
values; i.e. if x1 [0, 15] and thegene string is 4
Characters (e.g. ―1010‖) then there are 16 possibilities for the search to consider. To model this as a
continuous variable in- creases the number of possible values significantly. Similarly, other variable information
which aids the search considerably are upper and lower bound values. These factors can affect convergence of
the model solutions greatly.
Fig. 4: Flowchart of Genetic Algorithm [04]
VI. RESULT OF OPTIMIZATION
The optimal parametric setting of Venturimeter, rotary valve, bower speed and bend angle for maximum
discharge for is shown at table 4.
Table 4: Optimization result of discharge achieved in GA
Venturimeter
angle,
Degree
Rotary Valve
rpm
Blower Speed
rpm
Bend angle
degree
Best Fitness of
discharge
5.092 129.2 12900.238 59.951 1.460
National Seminar on Prospects and Challenges of Electrical Power Industry in India - NSPCEPII 344
ISBN 978-93-84743-43-7 © 2015 Bonfring

7. VII. CONCLUSION
Genetic Algorithms have advantages for a number of problems that prove intractable on other well-known
algorithms. We have discussed the design of the GA, and shown how it has been incorporated in finding the best
suitable solutions. Several extensions have been identified. First, in order to support a wider range of
experimental models, more flexible implementations of the GA need to be developed. These include different
crossover mechanisms, mutation schemes, selection schemes and replacement schemes. We also need to support
more data types. Second, the current implementation does not handle work on a wider set of the test functions.
Conditions such as fitness function noise, populations with low diversity, and multimodal search spaces need to
be addressed. An adaptive mechanism could also be added to allow the search parameters to be changed during
run time.
1. The application of genetic algorithm to the optimization of looped natural gas pipe networks
proved to be an efficient method for determining the optimum pipe sizing to minimize the
network cost.
2. The nature of the developed algorithm makes it possible, not only to identify an optimum
solution, but also to enumerate the most promising solutions that may exist spread in the
workspace. These promising solutions may have a more expensive cost than the identified
optimum solution but they may have another advantage from the reliability or maintainability
point of view.
3. The comparison of results with the heuristic human based approach gives surprising pipe
combinations that has a reduced network cost price and were far away of designer suggestions.
Application of the algorithm reveals the facts that sometimes using the highest available
diameter at some branches makes it possible to use the minimum available diameters at other
branches and that the optimum solution may exist within this new pipe combinations.
4. The application of the genetic algorithm is not suggested to replace the human judgment
entirely, but its great importance is to replace the time consuming trial and error procedure
previously adopted by designers and engineers. The developed algorithm provides the decision
maker with a set of promising solutions that are to be judged and verified.
5. It is clear that as the complexity of the network increases, the ability of the algorithm to achieve
a more competitive cost reduction compared to human based approach increases.
REFERENCE
[1] R. Saravanan, G. Sekar, M. Sachithanandam, ―Optimization of CNC machining operations subject to constraints Using genetic algorithm
(GA)‖, In: International Conference on Intelligent Flexible Autonomous Manufacturing Systems, Coimbatore, India, 472--479 (2000).
[2] R. Saravanan, M. Sachithanandam, ―Genetic algorithm (GA) for multi variable surface grinding process Optimizationusing a multi
objective function model‖, Int J Adv Manuf Technol. 17, 330—338 (2001).
[3] Thomas Weise, Global Optimization Algorithms Theory and Application, Version: 2009-06-26.
[4] Ishwer Shivakoti, Sunny Diyaley, Golam Kibria, B.B. Pradhan ― Analysis of Material Removal Rate using Genetic Algorithm Approach,
International Journal of Scientific & Engineering Research Volume 3, Issue 5, May-2012, ISSN 2229-5518.
Mr. Premendra Mani Pradhan, He is currently pursuing M. Tech in Mechanical Engineering(Machine Design &
Analysis) from NIT, Rourkela along with this he hold the position as Assistant Professor in Mechanical Engineering
Department in Sikkim Manipal University, India.
Mr. Yassin Alkassar, He is currently pursuing M.Tech in Mechanical Engineering (Machine Design & Analysis) from
NIT, Rourkela. He joined NIT, Rourkela under ICCR Scholarship. He did his Bachelor of Science in Mechanical Design
Engineering from Damascus University, Syria. He worked as a teaching Assistant at Damascus University, Syria.
National Seminar on Prospects and Challenges of Electrical Power Industry in India - NSPCEPII 345
ISBN 978-93-84743-43-7 © 2015 Bonfring

8. Mr. Deepak Kumar Behera, He is currently pursuing M.Tech in Mechanical Engineering (Machine Design & Analysis)
from NIT, Rourkela. He has good knowledge in Machine Design and Thermal Engineering. He is currently researching
in Industrial Tribology. He did his Bachelor Technology in Mechanical Engineering from C.V. Raman College of
Engineering at Bhubaneswar, India.
Mr. Premchand Kumar Mahto, He is Assistant Professor in Mechanical Engineering Department in Sikkim Manipal
University, India.He did his post-graduation in Heat power from BIT, Sindri.
National Seminar on Prospects and Challenges of Electrical Power Industry in India - NSPCEPII 346
ISBN 978-93-84743-43-7 © 2015 Bonfring