13 research outputs found
In-Process Chatter Detection in Surface Grinding
No full textThe chatter causes the poor surface finish during the surface grinding. It is therefore necessary to monitor the chatter during the process. Hence, this research has proposed the in-process chatter detection in the surface grinding process by utilizing the dynamic cutting forces. The ratios of the average variances of three dynamic cutting forces have been adopted and applied to identify the chatter during the surface grinding process to eliminate the effects of the cutting conditions. The effects of the cutting conditions on the chatter are also studied and analyzed. The algorithm has been proposed to detect the chatter regardless of the cutting conditions. The verification of the proposed system has been proved through another experiment by using the new cutting conditions. The experimental results have run satisfaction. It is understood that the chatter can be avoided during the in-process surface grinding even though the cutting conditions are changed
Comparison of in-Process Cutting State Detection in CNC Turning Using Different Neural Network Systems
No full textThe aim of this research is to propose and compare the in-process detection systems of the cutting states of the continuous chip, the broken chip and the chatter for the carbon steel in CNC turning process by utilizing the sensor fusion, which are the force sensor, the sound sensor, the accelerometer sensor and the acoustic emission sensor. The new six parameters proposed for the inputs of the neural network systems, which are the enegy spectral densities of three dynamic cutting forces, sound signal, accelation signal, and the standard deviation of acoustic emission signal. All signals of parameters have been integrated via the different neural network systems by using the pattern recognition and the percertron technique to detect the cutting states, which are. Among the cutting states of chip formation and chatter, the broken chip is required for the reliable and stable cutting system. The experimentally obtained results showed that the in-process detection system using the neural network with the pattern recognition technique can be effectively used to detect the cutting states with the higher accuracy and reliability more than the one with the perceptron technique.</jats:p
Grouping eucalyptus species in kraft pulp process for cost reduction
No full textThe objective of this research is to study the level of the important factors that can decrease total cost of pulp production. First of all, experts and experienced users identify the factors that affect the total production cost by applying the principle of 4M 1E cause and effect diagram. Then the primary factors were chosen based on 80% of their significance and tested by hypothesis for two population means. It was found that at the 95% confidence level the significant factors that have effects on the total production cost are amount of Effective alkali in white liquor and Kappa number. However, the proportion of easy delignification according to Eucalyptus species is considered as a significant factor based on various studies. Box-Behnken experiment is designed with respect to 3 mentioned factors and 3 levels of each factor. The response surface method (RSM) is employed to determine the non-linear relation between the total cost as the response and the proportion of easy delignification, amount of Effective alkali in white liquor and Kappa number. To minimize the total cost, the optimal values of each factor are 75% of easy delignification, 112 grams per liter of Effective alkali in white liquor and 13.5 of kappa number. Under this optimal condition, the average total cost per ton of Eucalyptus is 13,393.91 Baht which is significantly less than the total cost of 15,517.06 Baht per ton before improvement
Development of Efficient Washing System for Reduction of Oil Contamination on Machining Parts
No full textThe objective of this study is to develop an efficient washing system to remove cutting oil from machining part surface. The proposed washing system consists of two processes: the dipping process and the modified automatic ultrasonic washing process. The automatic ultrasonic washing process is redesigned and developed to reduce operating cost and increase productivity from the previously developed machine. For this proposed system, experiments have been performed to determine the washing conditions that yield satisfactory proportion of defectives due to oil contamination. Under the suggested operating conditions, the proportion of defectives due to oil contamination is reduced from 12.8% to 1.78%, which leads to $16,800 defective cost reduction. The proposed washing system yields 42.9% increase in washing productivity. Furthermore, it as has more standard procedure than the current washing process.</jats:p
Intelligent Monitoring and Prediction of Surface Roughness in Ball-End Milling Process
No full textIn order to realize the intelligent machines, the practical model is proposed to predict the in-process surface roughness during the ball-end milling process by utilizing the cutting force ratio. The ratio of cutting force is proposed to be generalized and non-scaled to estimate the surface roughness regardless of the cutting conditions. The proposed in-process surface roughness model is developed based on the experimentally obtained data by employing the exponential function with five factors of the spindle speed, the feed rate, the tool diameter, the depth of cut, and the cutting force ratio. The prediction accuracy and the prediction interval of the in-process surface roughness model at 95% confident level are calculated and proposed to predict the distribution of individually predicted points in which the in-process predicted surface roughness will fall. All those parameters have their own characteristics to the arithmetic surface roughness and the surface roughness. It is proved by the cutting tests that the proposed and developed in-process surface roughness model can be used to predict the in-process surface roughness by utilizing the cutting force ratio with the highly acceptable prediction accuracy.</jats:p
A Study on the Single Caprylic Acid Fractionation and Centrifugal Separation of Equine Rabies Immunoglobulin
Get PDFThis study proposes alternative caprylic acid precipitation and centrifugal separation for the equine rabies Immunoglobulin manufacturing process. The objective is to determine the optimal setting associated with the centrifugal machine and the optimal amount of caprylic acid for the maximum process yield (%). The experiments were designed based on the central composite design and performed to analyze the relationship of three factors which are the caprylic acid (1%-5%V/V), the rotation speed (7,500-12,500 rpm), and centrifugal time (20-40 min) on the yield of the process. For the first time, the prediction model as a second-degree polynomial regression is presented and developed by a response surface method (RSM) with R2 approximately 51%. RSM model also reveals that the process yield is affected by the concentration of caprylic acid and the amount of time to centrifuge the precipitated plasma but not by the rotation speed of the centrifugal machine. With the predicted process yield of about 12.97%, the optimal setting by RSM suggests the concentration of caprylic acid at 2.82% and the centrifugal time at 28 minutes
Process Yields Improvement of Filter Presses in Rabies Immunoglobulin Production
Get PDFThe objective of this study is to improve the performance of the Equine Rabies Immunoglobulin (ERIG) process by using the caprylic acid as a single precipitated agent and the replacement plan for the filter press machine to support higher demand in ERIG. The experiments based on the face-center central composite design are performed to investigate the relationship of yield recovery with 1. the concentration of caprylic acid used in the purification process, 2. the flow rate, and 3. the pore size of the filter press machine used in the filtration process. The regression analysis shows no relationship on linear, quadratic, or interactions between yield recovery and the three factors. Fortunately, according to the Analysis of Variance (for comparing means), there is a significant effect of interaction between the flow rate and pore size and the main effect of concentration of caprylic acid on the yield recovery at a 90% confidence level. From the interaction plot, at a flow rate of 10.5 ml/s with a filter media pore size of 6-15 micron, the process has the maximum yield recovery of 15.5% that is significantly superior to those of the other two sizes of filter media at the same flow rate. With a flow rate of 16.8 ml/s, the yield recovery is not significantly different for any pore size. At a flow rate of 4.2 ml/s, the yield recovery is higher when using the pore size of 4-9 and 6-15 micron than those from 5-12 micron. Nevertheless, with a 90% confidence interval for average yield recovery by Tukey comparison, the average yield recovery received from 16.8 ml/s at every pore size and 4.2 ml/s at 4-9 micron are not statistically different. Therefore, with the current size, 5-12 micron, of filter media used in the current process, the flow rate is suggested to be 16.8 ml/s whereas, the caprylic acid concentration of 1% is preferred due to the highest yield recovery compared to other concentration levels and the lowest production cost
Data-Driven Solutions for Backcalculating Elastic Moduli of Flexible Pavements from FWD Test
Get PDFTraditional methods for calculating pavement layers elastic moduli from falling weight deflectometer (FWD) tests often rely on computationally intensive iterative processes and lack struggle to capture complex variable relationships. This article highlights the utilization of machine learning (ML) algorithms, which include artificial neural networks (ANN), long-short-term memory (LSTM), and random forests (RF), to predict the elastic moduli of multi-layered flexible pavement based on FWD test. All ML algorithms were developed using synthetic databases derived from the exact stiffness matrix scheme, which was employed for the analysis of multi-layered pavements under axisymmetric surface loading. The development of ML models involves preprocessing of data, hyperparameter optimization, and performance evaluation. The input variables consist of the FWD surface deflections, the magnitude of applied loading, and the layer thicknesses, while the output variables represent the predicted layered elastic moduli of the pavement structure. The ANN and LSTM models capture complicated relations more effectively than the RF model in the backcalculation of the layered elastic modulus based on the FWD test. Among the two, LSTM achieves higher accuracy, with the average values across all layer moduli of R2 and MAPE being 99.04% and 2.41%, respectively, in the test set. The applicability of LSTM model is further demonstrated by comparing with the backcalculated elastic modulus based on the FWD field experiments performed on the infrastructure of roads in Thailand. Furthermore, a sensitivity analysis reveals that deflections near the center of loading predominantly impact the predictions of upper layer moduli, while the moduli of lower layers are influenced by deflections across all geophones
