Modeling of Aluminium – Flyash Particulate Metal Matrix Composites using Fuzzy Logic

Document Sample
Modeling of Aluminium – Flyash Particulate Metal Matrix Composites using Fuzzy Logic Powered By Docstoc
					                                                     (IJCSIS) International Journal of Computer Science and Information Security,
                                                     Vol. 9, No. 3, March 2011




    Modeling of Aluminium – Flyash Particulate
    Metal Matrix Composites using Fuzzy Logic
                                 1
                                  R.Elangovan and 2Dr.S.Purushothaman


          1                                                                   2
              R.Elangovan, Research Scholar,                                      Dr.S.Purushothaman, Principal,
         Department of Production Engineering,                           Sun College of Engineering and Technology,
Vinayaka Missions University, Salem, India-636 308                     Sun Nagar, Erachakulum, Kanyakumari District –
                                                                                           629902, India
                                                                            Email: dr.s.purushothaman@gmail.com



Abstract--This paper models the tension and bend test               furniture, and engine blocks in the automotive, small
data using fuzzy logic and radial basis function (RBF)
                                                                    engine and electro mechanical industry sectors.
artificial neural network. The data have been collected
experimentally working on Aluminium–flyash metal                    Flyash can be classified into two categories,
matrix composite. The fuzzy logic estimates change in
                                                                    precipitator and cenosphere. Precipitator flyash is a
height in tension test and change in buldge diameter in
bend test better when compared to RBF.                              solid and has a density of about 2- 2.5 gm /cm3.
                                                                    Cenosphere flyash is hollow and has a lower density
Keywords: Radial basis function, fuzzy logic, tension test,
bend test, scanning electron microscopy                             of about 0.6 gm/cm3. Flyash have been used as the
                                                                    reinforcing particulates in aluminium matrix [4-6]. Flyash
                   I.    INTRODUCTION
                                                                    has received attention as reinforcing phase as it is found to
      Composite materials are engineered materials                  increase the hardness, tensile strength and wear resistance of
made from two or more constituent materials with                    aluminium metal matrix composites.
significantly different physical or chemical properties
and which remain separate and distinct on a                                Particulates are the most common and cheapest
macroscopic level within the finished structure. Metal              reinforcement materials. These produces the isotropic
Matrix Composite (MMC) consists of a metallic                       property of MMC's which shows promising application in
matrix combined with a reinforcing material. The                    structural fields The Al - flyash - silicon carbide hybrid
matrix    materials     are   Aluminium,     Magnesium,             matrix composites have a good potential for use as wear
Titanium[1-3]. The reinforcing materials can be                     resistant materials. Flyash particulate improves properties
flyash, Silicon Carbide, Graphite, and Alumina                      such as hardness, wear resistance and compressive strength.
                                                                    If the composite is to be used in a structural application, the
      Al-flyash composite has low density. This                     modulus, strength and density of the composite will be
composite has potential applications in covers,                     important which requires high modulus, low density
shrouds, casings, manifolds, valve covers, garden




                                                              220                               http://sites.google.com/site/ijcsis/
                                                                                                ISSN 1947-5500
                                                      (IJCSIS) International Journal of Computer Science and Information Security,
                                                      Vol. 9, No. 3, March 2011




reinforcement like flyash. If the composite is to be used in         material is stirred again and again and poured into the
wear resistant applications, hardness is important.                  moulds. It solidifies. The solid is cut into shapes, and
      II         MATERIALS AND METHODS                               the surface is cleaned.

     A. Materials
                                                                         B      Methods
      A hollow pipe is taken first with a dimension of
                                                                         B.1 Fuzzy Logic
500 mm long 70mm radius and 8 mm thickness.
                                                                                Fuzzy logic has rapidly become one of the
Some quantity of water is added to the foundry sand
                                                                     most      successful    of    today's     technologies       for
to increase the adhesives and collapsibility of the
                                                                     developing sophistication in technologies. Fuzzy
sand. After that, a hollow pipe of 40 mm diameter
                                                                     logic addresses applications perfectly as it resembles
and 500mm long is inserted vertically into the
                                                                     human decision making with an ability to generate
standing bigger hollow pipe. The gap between the
                                                                     precise solutions from certain or approximate
pipes is filled with the prepared foundry sand. It is
                                                                     information. It fills an important gap in engineering
rammed well so as to form the shape of the pattern.
                                                                     design methods left vacant by purely mathematical
Then it is kept under the sunlight so that it will set
                                                                     approaches, and purely logic-based approaches in
well. After that, the pattern is removed and the cavity
                                                                     system design. While other approaches require
is formed on the mould.
                                                                     accurate equations to model real-world behaviors,
      The pure aluminium ingots arc cut into small
                                                                     fuzzy design can accommodate the ambiguities of
pieces. The coal is heated in the furnace. After the
                                                                     real-world human language and logic. It provides
sufficient heat is applied blower is switched on.
                                                                     both an intuitive method for describing systems in
Crucible is placed in the furnace. When sufficient
                                                                     human terms and automates the conversion of those
amount of heat is obtained in the crucible, aluminium
                                                                     system specifications into effective models.
pieces are put in. When the crucible reaches about
                                                                         B.2 Radial basis function (RBF)
600oC, the aluminium pieces melts into liquid. Slag is
                                                                                Radial basis function is a supervised neural
removed. When the molten metal gets into a semi
                                                                     network. The network has an input layer, hidden
solid condition, flyash is added to it and is mixed
                                                                     layer (RBF layer) and output layer. The features
with stirrer. Then it is kept cooled so as to become a
                                                                     obtained are used as inputs for the network and the
solid composite.
                                                                     target values for training is based on the values
      In stir casting process, the aluminium is melted
                                                                     change in height or change in buldge diameter.
at a controlled temperature and the desired quantity
                                                                         Training RBF is done as follows,
of flyash is added to the molten metal. The molten
                                                                          1.    Distance between pattern and centers are
metal is stirred continuously to create a vortex to
                                                                                found.
force the slightly lighter particle into the melt.
                                                                          2.    An RBF matrix whose size will be (np X
Continuous stirring is done to disperse the flyash
                                                                                cp). where ‘np’ is the number of patterns
particulate as uniformly as possible in a short time.
                                                                                used for training and ‘cp’ is the number of
For stirring the flyash and molten metal, three blade
                                                                                centers which is equal to ‘np’.
propeller type is used. The mixed matrix is then
                                                                          3.    Final weights are calculated.
transformed into a preheated transfer ladle. The




                                                               221                                http://sites.google.com/site/ijcsis/
                                                                                                  ISSN 1947-5500
                                                            (IJCSIS) International Journal of Computer Science and Information Security,
                                                            Vol. 9, No. 3, March 2011




       4.   During testing the performance of the RBF                      test provides information on proof stress, yield point,
            network, RBF values are formed from the                        tensile strength, elongation and reduction in area
            features obtained from inputs and processed                    (Table 1).
            with the final weights obtained during
            training. Based on the result obtained, the                         B.     Bend test
            test data is classified.                                                   Testing will be done using a UTM.
                                                                           Capacity of UTM is 40 tonnes.
III         EXPERIMENTAL WORK AND RESULTS                                  Diameter of rod                          =24mm
Chemical composition of flvash                                             Bending pan radius                       =16mm
(Data collected from Tuticorin thermal power plant                         Length of the rod                        =265mm
Tamil Nadu, India)
Sio2                             -     60.62 %,                                         IV RESULTS AND DISCUSSIONS
Al2O3                            -       21.93%
                                                                                A. Estimating change in height during tension
Fe2O3 + Fe304                                          -
                                                                                       test
7.12%
                                                                           Figure 1 shows estimation of change in height by
CaO                                         -     2.28%
                                                                           Fuzzy logic for the tension test data. The estimation
MgO                                         -     0.85 %
                                                                           is accurate and concides with the original data
So 4                                        -     Traces
                                                                           collected. Figure 2 shows estimation of change in
Loss on ignition                                       -
                                                                           height by RBF for the tension test data. The
0.72%
                                                                           estimation slightly deviates and not close with the
Bulk Density                                -     0.86gm/cc
                                                                           actual data.
Fineness                                    -     0.075mm in
m2 / kg                                                                    Figure 3 presents comparisons of the performance of
Melting point of aluminium is               - 660° C                       RBF and fuzzy logic in estimating the change in
Casting period                              -2 ½ hrs                       height. The performance of Fuzzy logic is superior to
Stirring period                             20 minutes                     performance of RBF. However, depending upon the
                                                                           type of data used, the performance of RBF will
      A. Tension test                                                      improve further.
            A standard test piece is taken.. The gauge
                                                                                B      Estimating buldge diameter during bend
length is maintained by gripping at either end by
                                                                                test
suitable      apparatus     in       a     universal       testing
machine(UTM). The UTM slowly exerts an axial pull                          Figure 4 shows estimation of change in buldge
so that the specimen is stretched until it breaks. The                     diameter during bend test by Fuzzy logic for the
bend test data. The estimation is almost accurate and                      for the bend test data. The estimation slightly
coincides with the original data collected. Figure 5                       deviates and not close with the actual data
shows estimation of change buldge diamater by RBF                          Figure 6 presents comparisons of the performance of
                                                                           RBF and fuzzy logic in estimating the buldge




                                                                     222                               http://sites.google.com/site/ijcsis/
                                                                                                       ISSN 1947-5500
                                                                       (IJCSIS) International Journal of Computer Science and Information Security,
                                                                       Vol. 9, No. 3, March 2011




                                                                 Table 1 Tension test performance
                                               Aluminium +                                       Aluminium +                                       Aluminium +
                                                 5% flyash                                            10% flyash                                    15% flyash
Maximum ultimate                  15.94 KN                                   16.14KN                                               17.85KN
load
Deflection                  at    9.9mm                                      11.7mm                                                13.3mm
maximum load
Maximum                           7.5mm                                      9.2mm                                                 10.3mm
displacement                at
fracture
Percentage                  of    8.3%                                       10.2%                                                 12.4%
elongation
Scanning Electron
Microscope
photograph




 

                                             Table 2 Scanning electron microscope photograph Tension test performance
                          Aluminium +                                          Aluminium +                                                       Aluminium +
                           5% flyash                                            10% flyash                                                       15% flyash




                 13                                                                                           13
                                                                           Target                                                                                      Target
                 12                                                        Estimated                          12                                                       Estimated


                 11                                                                                           11

                                                                                                              10
                 10
    Height, mm




                                                                                                 Height, mm




                                                                                                              9
                 9
                                                                                                              8
                 8
                                                                                                              7
                 7
                                                                                                              6

                 6                                                                                            5

                 5                                                                                            4
                  0         0.5          1        1.5        2       2.5                3                      0     0.5       1        1.5           2          2.5                3
                                                Load, kg                         x 10
                                                                                       4                                              Load, kg                                     4
                                                                                                                                                                             x 10



                      Figure.1 Change in height estimated by Fuzzy logic                                           Figure.2 Change in height estimated by RBF




                                                                                           223                                         http://sites.google.com/site/ijcsis/
                                                                                                                                       ISSN 1947-5500
                                                                                        (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                        Vol. 9, No. 3, March 2011




                          13                                                                                                          38
                                                                                       Rbf
                          12                                                           Fuzzy
                                                                                                                                      36
                                                                                       Actual
                          11                                                                                                                                                                       Target
                                                                                                                                      34
                                                                                                                                                                                                   Estimated
  Change in Height,mm




                                                                                                            Buldge diameter, mm
                          10
                                                                                                                                      32
                          9
                                                                                                                                      30
                          8
                                                                                                                                      28
                          7

                          6                                                                                                           26


                          5                                                                                                           24

                          4                                                                                                           22
                           0          0.5         1         1.5          2       2.5             3                                      0             0.5        1         1.5        2         2.5                  3
                                                          Load, kg                        x 10
                                                                                                4
                                                                                                                                                                        Load, Kg                                     4
                                                                                                                                                                                                              x 10



                          Figure.3 Change in height estimated by RBF and Fuzzy logic                                                            Figure.4 Change in buldge diameter estimated by Fuzzy


                          38                                                                                                              38

                          36                                                                                                              36

                          34                                                                                                              34
                                                                                              Target
    Buldge Diameter, mm




                                                                                                                     Buldge Diameter,mm                                                                 Rbf
                          32                                                                  Estimated                                   32                                                            Fuzzy
                                                                                                                                                                                                        Actual
                          30                                                                                                              30

                          28                                                                                                              28

                          26                                                                                                              26

                          24                                                                                                              24

                          22                                                                                                              22
                            0           0.5           1         1.5          2          2.5                                                 0          0.5        1        1.5         2         2.5                 3
                                                              Load, kg                               x 10                                                                Load, kg                                    4
                                                                                                                                                                                                              x 10


                                Figure.5 Change in buldge diameter estimated by RBF
                                                                                                            Figure.6 Change in buldge diameter estimated by RBF and Fuzzy logic




diameter. The performance of Fuzzy logic is superior                                                                                                                  REFERENCES
to performance of RBF. However, depending upon
                                                                                                                                           [1] P. K. Rohatgi, J. K. Kim, R. Q. Guo, D. P. Robertson and M.
the type of data used the performance of RBF will                                                                                          Gajdardziska-Josifovska, “Age Hardening Characteristics of
improve further.                                                                                                                           Aluminium Alloy – Hollow Fly Ash Composites”, Metallurgical
                                                                                                                                           & Materials Transaction A, Vol. 33A, (2002), 1541-1547.
                                              V   .CONCLUSIONS                                                                             [2] T. Matsunaga, J. K. Kim, S. Hardcastle and P. K. Rohatgi,
                                                                                                                                           “Crystallinity and Selected Properties of Fly Ash Particles”,
                          This paper presents the outputs of experimental                                                                  Materials Science & Engineering: A, Vol. A325, (2002), 333-343.
work performed on Al- flyash with 5%, 10% and                                                                                              [3] J. Sobczak, Z. Slawinski, N. Sobczak, P. Darlak, R. Asthana
                                                                                                                                           and P. K. Rohatgi, “Thermal Fatigue Resistance of Discontinuous
15% flyash as volume fractions. The bend test,
                                                                                                                                           Reinforced Cast Aluminium-Matrix Composites”, Journal of
tension test have been performed and data have been
                                                                                                                                           Materials Engineering and Performance, Vol. 11, (2002), 595-
collected to train fuzzy logic and radial basis function                                                                                   602.
neural network. As per this paper, Fuzzy logic                                                                                             [4] J. Bienias, M. Walczak, B. Surowska and J. Sobczak,
                                                                                                                                           “Microstructure and Corrosion Behaviour of Aluminum Fly Ash
performs well when compared to RBF.



                                                                                                       224                                                                http://sites.google.com/site/ijcsis/
                                                                                                                                                                          ISSN 1947-5500
                                                          (IJCSIS) International Journal of Computer Science and Information Security,
                                                          Vol. 9, No. 3, March 2011




Composites”, Journal of Optoelectronics and Advanced Materials,          [6] S. Kolukisa, A. Topuz and A. Sagin, “The Production and
Vol. 5, (2003), 493-502.                                                 Properties of Fly Ash Containing Aluminium Matrix Composite
[5] X. F. Zhang, D. J. Wang and G. Xie, “Manufacturing of                Materials”, Practical Metallography, Vol. 40, (2003), 357-368.
Aluminium/Fly Ash Composites with Liquid Reactive Sintering
Technology”, Acta Metallurgica Sinica, Vol. 15, (2002), 465-470.




                                                                   225                                 http://sites.google.com/site/ijcsis/
                                                                                                       ISSN 1947-5500