YANIS, MUHAMMAD and Mohruni, Amrifan Saladin and Sharif, Safian and Yani, Irsyadi (2019) PERFORMANCE EVALUATION OF GREEN MACHINING ON THIN-WALLED Ti6Al4V USING RESPONSE SURFACE METHODOLOGY AND ARTIFICIAL NEURAL NETWORKS. Doctoral thesis, Sriwijaya University.
![[thumbnail of RAMA_KODE_03043621520005.pdf]](http://repository.unsri.ac.id/style/images/fileicons/text.png) |
Text
RAMA_KODE_03043621520005.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (7MB) | Request a copy |
|
Text
RAMA_KODE_03043621520005_TURNITIN.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (7MB) | Request a copy |
Preview |
Text
RAMA_KODE_03043621520005_0011096407_ 0025127104_01_front_ref.pdf - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (1MB) | Preview |
|
Text
RAMA_KODE_03043621520005_0011096407_ 0025127104_02.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (1MB) | Request a copy |
|
Text
RAMA_KODE_03043621520005_0011096407_ 0025127104_03.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (649kB) | Request a copy |
|
Text
RAMA_KODE_03043621520005_0011096407_ 0025127104_04.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (3MB) | Request a copy |
|
Text
RAMA_KODE_03043621520005_0011096407_ 0025127104_05.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (186kB) | Request a copy |
|
Text
RAMA_KODE_03043621520005_0011096407_ 0025127104_06_ref.pdf - Bibliography Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (185kB) | Request a copy |
|
Text
RAMA_KODE_03043621520005_0011096407_ 0025127104_07_lamp.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (1MB) | Request a copy |
Abstract
Milling of thin-walled Ti6Al4V components is very substantial in the modern aerospace industry. The machining operation achieved by the utilization of minimum quantity lubrication (MQL) technique. Neat coconut oil was chosen as a good local source cutting fluid. The aim of this study was to evaluate the machining performance of Ti6Al4V on dependent variables (surface roughness Ra and cutting force Fc). RSM and ANN were developed to determine modeling predictions and optimization. Variations in the value of independent variables (cutting speed Vc, feed rate fz, radial ar and axial ax depth of cut) based on the CCD method (Central Composite Design) consist of 30 test data. Machining uses coated and uncoated carbide tools. The best mathematical equation results based on RSM for surface roughness prediction using coated tools was quadratic model, and using uncoated tools was linear model. The best mathematical equation results for cutting force prediction using coated tools was quartic model, and using uncoated tools was quadratic model. Optimal conditions for the minimum dependent variable according to RSM were for coated tool Vc = 113.9 m/min, fz = 0.04 mm/tooth, ar = 0.27 mm, ax = 5 mm that obtained Ra = 0137 µm and Fc = 25.29 N. Optimal conditions for uncoated tool Vc = 125 m/min, fz = 0.04 mm/tooth, ar = 0.25 mm, ax = 5 mm that obtained Ra = 0.161 µm and Fc = 14.89 N. The best accuracy prediction on ANN with back propagation obtained was Levenberg-Marquardt (LM) algorithm. Network structure to achieve the lowest MSE value for surface roughness with coated and uncoated tools were 4-10-1 and 4-13-1, respectively. Network structure to achieve the lowest MSE value for cutting force with coated and uncoated tools were 4-8-1 and 4-10-1, respectively. Based on the MSE value, the accuracy prediction of surface roughness using ANN was better than RSM with coated and uncoated at 62.27% and 93.05%, respectively. The accuracy prediction of cutting force using ANN was better than RSM with coated and uncoated at 99.17% and 96.61%, respectively. The MSE of RSM and ANN both surface roughness and cutting force shows that the prediction were close to the results of the experiment. Low feed rates and depth of cut resulted in low surface roughness, but high cutting speeds reduced surface roughness. Reduction in depth of cut and feed rate resulted in low cutting forces, but the effect of cutting speed is very small. All dependent variables were lower on non-thin walled machines compared to thin-walled machines. The values of cutting force on coated tool are higher than uncoated tool, whereas the surface roughness value of coated tool was lower than uncoated tool and this tendency occurs both in thin-walled and non-thin-walled. All machining conditions used in this study did not cause chatter.
Actions (login required)
 |
View Item |