Initial data and documents: - Injection mold parameters - Studies on mold heating using hot gas - Experimental design and optimization methods - Experimental equipment for measuring tem
INTRODUCTION
Urgency of the topic
- The research to optimize the shape of the air cover for the heating of the mold cavities with hot air is a potential direction in the field of molds in particular and the precision industry in general
- Optimized air cover reduces heating and cooling times, thereby shortening production cycles
- The optimization of the heating process helps to ensure that the temperature in the mold is evenly distributed, minimizing defects and improving the surface quality of plastic products
- Optimize production processes and use energy efficiently to help minimize environmental impacts, reduce greenhouse gas emissions and industrial wastes.
Scientific and practical significance of the topic
This research improves knowledge about the heating process in plastic injection molds, especially the optimization of temperature when heated by gas scanning Expand your understanding of simulation methods, optimization algorithms, and artificial neural networks, and provide new and important data to scientific databases in this field
Improving the quality of plastic products, as well as finding new methods to improve the quality and output of plastic products is one of the urgent requirements for the plastic industry in Vietnam Therefore, the topic "Research on optimizing air cover to serve the heating process for rectangular injection mold cavities" is proposed to help plastic products become better in quality and uniformity; minimizing breakdowns, saving energy and production costs The research also helps to increase production efficiency by reducing the heating and cooling time of the mold, and the technology can be applied in many industries.
Research objectives of the project
Through the heating method for injection molds with hot air from the outside, the project will focus on the following objectives:
- Air cover design to serve the heating process with hot air for rectangular mold cavities
- Find out and evaluate the method that simulates the mold heating process
- Apply the optimal algorithm to find the optimal shape for the air cover.
Object and scope of study
- Mold Heating Process: Learn about current heating methods, issues related to controlling the temperature in the mold, and the effect of temperature on product quality
- Analysis of air cover technique, how it works and its application in the process of heating the plastic injection mold cavities
- Use optimization tools and methods (such as genetic algorithms, swarm optimization, artificial neural networks, etc.) to improve the efficiency of air cover in mold core heating
- Conduct soft field tests to verify and compare research results
- The topic only focuses on optimizing gas imaging for the heating process for rectangular molds
- The heating process and temperature are surveyed through experiments and simulations using ANSYS software
- Build and train an artificial neural network model and use the GA method to optimize the shape of the airscan
- Perform experiments to verify simulation and optimization results
- Compare the experimental results with the simulation results and the optimal/suboptimal results
The laboratory equipment is provided by the mold laboratory of the University of Technology and Education, Ho Chi Minh City.
Research methods
- Build a air cover shape with a polyline surface
- Simulate the heating process and get data from the simulation
- Analyze the simulation data, use that data to train the ANN and use it for GA
- Re-simulate the heating process with optimized air cover
- Perform the heating process of the mold with hot air with suboptimal air cover and optimized air cover to compare the results.
Structure of the thesis
Chapter 1: Introduction - This chapter presents the urgency of the research topic on optimizing the air cover for the heating process of injection mold cavities It addresses the scientific and practical significance of the topic, research objectives, subjects and scope of the research, research methodology, and the structure of the graduation project
Chapter 2: Overview of research - This chapter provides an overview of the research situation in the field of mold temperature optimization, including domestic and international technologies such as Big Data, IoT, and artificial intelligence in optimizing production processes
Chapter 3: Theoretical basis - This chapter provides the theoretical foundation on the injection molding process, heating methods for injection molds, and the materials used It also introduces the geometric model of the mold, heat transfer simulation using Ansys software, the structure and operation of artificial neural networks (ANN), and the genetic algorithm (GA)
Chapter 4: Simulation process – ANN Temperature prediction – GA Optimization -
This chapter describes the process of simulating mold heating using hot gas, training the artificial neural network to predict temperature, and using the genetic algorithm to find the optimal height The steps include building the simulation model, setting initial conditions, and analyzing the simulation results
Chapter 5: Experiment – Evaluation - This chapter presents practical experiments to verify the simulation and optimization results It compares the performance between the optimized and non-optimized air cover models and evaluates the correlation between experimental and simulation results
Chapter 6: Conclusion - The final chapter summarizes the research results, evaluates the effectiveness of the methods used, and proposes future research directions It emphasizes the importance of optimizing the mold heating process to improve the quality and efficiency of plastic products.
OVERVIEW OF RESEARCH
Overview of research in the field of research
Research on mold temperature optimization in injection molding technology plays an important role in improving product quality, increasing production efficiency, and reducing costs The solidification process, product accuracy, and production cycle time are all directly affected by the mold temperature of the mold cavities Modern methods of temperature control have been used to solve these problems
Liquid cooling, compressed air cooling, resistive heating, and intelligent temperature control systems are some of the modern methods of temperature control The liquid cooling system improves the accuracy and durability of the product by maintaining a stable and uniform temperature in the mold cavity Compressed air cooling cools fast and controls temperature at small details and corners Resistive heating, which is used for fast and accurate heating, helps to optimize production efficiency The intelligent temperature control system uses intelligent sensors and controllers to monitor and regulate the temperature, minimizing product defects [1]
The application of 4.0 technology, the production of new materials and the improvement of the flexibility of temperature control systems are trends in the industry Using Big Data, the Internet of Things, and artificial intelligence helps automate and optimize the temperature control process Research and development of new mold materials have proven to be highly effective These materials have better thermal conductivity and higher heat resistance Finally, an important step has been made in the development of temperature control systems that can be easily adjusted and adapted to the different requirements of production [2]
Optimizing the mold temperature in injection molding technology is critical to improving production efficiency and product quality Smart and modern temperature control methods are being applied more and more widely, bringing many new prospects to the industry.
Some research related to the topic
This is a study of the effect of the temperature of the injection mold on the injection molding product By supplying water with the appropriate temperature in the cooling line of the mold during the cooling process, the proper temperature and deformation of the injection molding item are tested according to the mold temperature In this study, using finite element analysis and experiments, the temperature of the mold was found to have the following effects mm Injection molded item models obtained by supplying 20°C water are found to be a minimum of 0.002 mm, with a deformation of up to 0.761 mm When comparing both conditions, the error rate of the injection molding item obtained by supplying 20°C water in the mold cooling line is about 0.18% lower Then the conclusion is obtained: The lower the temperature of the mold during the cooling process, the less the deformation of the injection molding item
• Li, J., Li, T., Peng, X., Liu, F., Zhou, H., & Jiang, S [4]
In order to improve the heating performance and cavity surface temperature uniformity, an optimal design method has been developed for the heating system in the electric fast heat cycle molding mold First, the electric fast heat cycle molding is simplified into a single heating element for thermal reaction analysis based on the theory of suitable design Second, the response surface uses a backpropagation neural network built on the basis of the original finite element experiments Subsequently, a non-dominant sequencing-II genetic algorithm combined with a polynomial inverse propagation neuron network model was proposed to obtain Pareto optimal solutions Subsequently, the same order prioritization technique with an entropy-based weight-based ideal solution was applied as a multi-attribute decision- making method to select the optimal design point trade-off from the Pareto optimizer In order to achieve the optimal design of the heating system, the heating element optimized for the electric fast heat cycle molding was calculated and finally mapped to the entire mold Cavity surface temperature uniformity increased by 17.1% and heating efficiency increased by 26% The results show that the uniformity of temperature distribution on the mold cavity surface has clearly been improved, and the use of this optimization strategy ensures high heating efficiency
Conventional cooling systems are the most widely used because they are easy to fabricate using conventional machining processes However, uneven cooling is considered one of their weaknesses In addition to conventional systems, there are also suitable cooling systems designed to cool plastic molds faster and more evenly In this study, a suitable cooling system was applied to produce the AZ564 PP plastic bowl product Optimization is carried out to start the machine setup parameters, namely melting temperature, injection pressure, holding pressure and holding time Genetic algorithm and Moldflow methods are used to optimize injection process parameters at minimal cycle times It has been found that, the optimal injection molding process can be achieved by setting the parameters at the following values: TM = 180°C; Injection = 20 MPa; Hold = 16 MPa and hold = 8s, with a cycle time of 14.11s Tests using the right cooling system yield an average cycle time of 14.19 seconds The suitable cooling system studied yields a volumetric shrinkage of 5.61% and a wall shear stress of 0.17 MPa The difference between the cycle time obtained through simulation and testing using a suitable cooling system is negligible (less than 1%) Thus, the combination of optimization of process parameters and simulation using genetic algorithmic methods with
Moldflow can be considered valid
In recent years, plastic injection molds have been widely used to manufacture products in various fields such as aerospace, automotive, medical, electronics, and toys The quality of these products depends on the correctly selected casting parameters In this study, a new package program (NPP)-Injection molding software that calculates various injection molding parameters was developed to mold plastic products obtained by plastic injection molding using an artificial neural network (ANN) model The Delphi programming language is used for development (NPP)-Software The developed software (NPP) was trained and tested using the Levenberg-Marquardt (LM) algorithm, ANN One thousand three hundred data was collected, of which 250 were used to train the network ANN is used to find optimal casting parameters that minimize defects in the injection molding unit, such as volumetric shrinkage, spray time, and cooling time
• Shia-Chung Chen, Pham Son Minh, Jen-An Chang [7]
In this study, a gas-assisted mold surface heating system combined with water-cooling was developed to rapidly control mold temperature during injection molding The effects of GMTC on mold temperature uniformity and fiber exposure were evaluated through experiments and simulations The following conclusions were drawn:
- Increasing gas flow capacity significantly raises the mold surface temperature Starting from an initial mold temperature of 60°C, the surface temperature can reach 147.8°C, 167.2°C, and 229°C with gas flow rates of 100 l/min, 200 l/min, and 300 l/min, respectively, after 6 seconds of heating The heating rate achieves 28°C/s at 300 l/min
- With the mold design illustrated in Fig 5, a higher gas flow rate results in a larger temperature difference between points A and B This difference grows from 9.5°C to 46.6°C as the flow rate increases from 100 l/min to 300 l/min
- Larger gas gaps influence gas speed, decreasing the heating rate near the inlet and increasing it near the outlet Expanding the gas gap from 4 mm to 8 mm reduces the temperature difference between points A and B from 110.5°C to 27°C, enhancing temperature distribution uniformity during heating
- GMTC effectively reduces fiber floating marks and fiber exposure on the product surface due to the elevated mold temperature during the filling and packing stages
The quality of the finished product is affected by a number of factors in the plastic injection molding process Two important process variables in creating a product are melting point and mold The study used the Experimental Design tool in Autodesk Moldflow to examine the impact of melting point and mold temperature on injection molding technology The analytical items are specially made of polypropylene (PP) using a child seat mold According to the simulation analysis results, the melting point has a noticeable impact on both the end time of the packaging process as well as the deviation in the analysis temperature range at t mold of [40, 80]°C and the melting temperature of [180], 220]°C The melting point also shows a noticeable effect not only on sag but also on sink depth and volumetric shrinkage, along with criteria for evaluating the cost of the product
• P S Minh, Tran Minh The Uyen, & T T Do [9]
In this paper, gas-assisted mold temperature control is simulated with various covering properties to improve the cavity temperature during injection molding Separate simulations and experiments have been conducted to evaluate the temperature distribution using an air coating during heating The experimental model is described in figure 2.1 The key findings can be generalized as follows: In sprue heights (h) ranging from 3 mm to 7 mm, the case of 3 mm leads to the highest temperature and 7 mm leads to the lowest temperature In the case of air height (V), the best temperature distribution on the insert surface appears with an air height of 5 mm With a heating time of 20 seconds, GMTC can increase the plate temperature stamp to more than 300°C This result almost meets the requirement for dynamic mold temperature control in the injection molding cycle
• Pham Son Minh, Do Thanh Trung, Nguyen Ho & Phan The Nhan [10]
The heating process for a rectangular mold cavity using external hot gas injection was investigated through both simulation and experiments, yielding several conclusions The experimental results indicated that with external hot gas heating, the mold surface temperature increases rapidly during the first 10 seconds, then rises more slowly over the next 20 seconds before stabilizing after 30 seconds At this stable point, the temperature distribution is relatively uniform, with a variation of approximately 11.5°C This suggests that the external hot gas heating method is fully applicable to the injection molding process Additionally, the comparison between simulation and experimental results revealed minimal discrepancies at all measured points on the mold surface Therefore, ANSYS CFX software can be effectively used to predict the temperature distribution on the mold cavity surface during the hot gas heating process, facilitating the quality assessment of the molded product.
THEORETICAL BASIS
Plastic injection molding process
The basic steps of the plastic injection molding process are described in detail as shown in figure 3.1:
- Initially, raw plastic material, typically in the form of granules or pellets, is fed into the hopper of the injection molding machine This hopper serves as a reservoir from which the plastic material gradually descends into the machine's cylinder The controlled descent ensures a steady supply of material for the molding process
- Inside the cylinder, the raw plastic undergoes a transformation from a solid to a liquid state This process is driven by the rotational and translational motion of a screw within the cylinder The screw's movement not only advances the plastic material but also generates frictional heat Additionally, external heating elements, often in the form of heating resistors wrapped around the cylinder, provide supplemental heat This combined heating action raises the temperature of the plastic particles to a range between 150°C and 320°C, effectively melting them into a viscous liquid state This melted plastic is now ready for injection
- After the initial filling of the mold, the packing stage begins During this phase, additional molten plastic is injected into the mold to compensate for any material shrinkage that occurs as the plastic begins to cool and solidify This packing pressure ensures that the molded part retains its intended shape and size, preventing defects such as sink marks The packing process continues until the plastic at the injection port solidifies, effectively sealing the mold
- Following the packing phase, the cooling stage ensues The cooling process is critical as it allows the molten plastic within the mold to solidify into the final product The mold itself is equipped with a cooling system, often consisting of channels through which a coolant (such as water) circulates This system helps regulate the temperature of the mold, ensuring uniform cooling of the plastic product The cooling time varies depending on the material and the thickness of the part, but it is essential for achieving dimensional stability and ensuring the product reaches the ejection temperature
- Once the plastic product has sufficiently cooled and solidified, the mold opens The molded part is then ejected from the mold cavity by an ejection system, typically consisting of ejector pins or a push latch mechanism This system ensures that the finished product is removed from the mold without damage
All these steps, from feeding, melting, plastic spraying, shaping to cooling and unloading the product, form a complete cycle of the plastic injection molding process
Melting Filling Packing Cooling Mold opening
Figure 3.1 Plastic injection molding process
Heating plastic injection molds
Mold heating is a crucial component in the injection molding process, directly influencing the quality and performance of the final product Various heating methods are employed to achieve the desired mold temperature These methods include using electrical resistors to generate and transfer heat to the mold, circulating hot oil through channels within the mold, blowing hot air through the mold cavities, and applying a magnetic field to induce heat within the mold material itself The primary objective of these heating techniques is to maintain the mold temperature within an optimal range This ensures that the plastic material flows evenly and rapidly, resulting in a high-quality product with precise dimensions and shapes
In contemporary manufacturing, mold temperature control is recognized as one of the key solutions to enhancing the efficiency of the injection molding process Achieving a high mold surface temperature generally leads to better part quality, as the molten plastic fills the mold more completely and with fewer defects However, this also typically increases the cooling time and overall cycle time, which can be a drawback in terms of production speed Conversely, lowering the mold surface temperature can shorten the cooling and cycle times but often at the expense of surface quality Therefore, current research is focused on methods to elevate mold surface temperatures without significantly prolonging the cycle time, balancing quality and efficiency
The gas heating method involves spraying hot air directly onto the heating position of the mold The thermal convection between the hot air and the mold surface facilitates the transfer of heat energy from the air to the mold This results in an increase in surface temperature Although the heating rate with gas heating is generally lower than that achieved with induction heating, this method is versatile and can be applied to a variety of mold materials, including steel, aluminum, and copper Moreover, the characteristic thermal convection of gas heating helps to prevent the mold from overheating, thus protecting it from thermal damage By employing the gas heating method, the mold temperature can be raised sufficiently to improve the flow characteristics of the plastic within the mold, enhancing the overall quality of the molded parts
In summary, while gas heating may not provide the rapid heating rates of some other methods, its ability to be used with different mold materials and its effectiveness in preventing overheating make it a valuable technique in the injection molding process The goal is to increase mold temperature efficiently, ensuring optimal plastic flow and high-quality production without excessively extending the cycle time.
The gas-assisted mold temperature control (GMTC)
In the traditional injection molding process, the sequence of operations follows a consistent pattern After the completion of an injection cycle, where molten plastic is injected into the mold cavity and solidifies into the desired shape, the two halves of the mold open up This action allows the newly formed product to be ejected from the mold cavity using an ejection system Once the product is removed, the mold halves close again, and a new injection cycle commences, repeating the process
However, when the injection molding process is enhanced with hot gas heating for the mold cavity, the procedure involves additional steps aimed at improving the mold's thermal management After the product is ejected from the mold cavity, a hot gas system is immediately activated This system is designed to ensure that the heating between the two mold halves is uniform, which is crucial for maintaining consistent mold temperatures and improving the quality of the next product
The hot gas from the heating system is precisely directed and injected into specific heating positions within the mold cavity These positions are critical as they determine the efficiency and uniformity of the heating process Figure 3.2 illustrates the arrangement of the mold cavity and the placement of the heating source This setup ensures that the heat is distributed evenly across the mold surface, preventing hotspots and ensuring that the entire cavity reaches the optimal temperature required for the next injection cycle
Once the mold cavity has attained the desired temperature, the heating source is retracted from the mold area This retraction is essential to prepare the mold for the next cycle without interference from the heating apparatus Following this, the mold halves close, and the injection machine begins a new cycle This integrated heating process not only helps in maintaining a consistent mold temperature but also enhances the overall efficiency and quality of the injection molding process by reducing thermal gradients and potential defects in the molded parts
Figure 3.2 Mold cavity position during the heating process
Materials
In the process of experimenting and testing the topic of optimizing the temperature distribution of the mold cavities, the test sample material was selected as steel Steel is a popular material used in many industries, including the foundry industry
The use of steel as an experimental sample material has several benefits First, the steel has good thermal conductivity, which allows the temperature to be transmitted efficiently during the casting process Secondly, the steel has high stability and good heat resistance, which helps to ensure that the test specimen is not deformed or affected by high temperatures When using steel as an experimental sample material, it is necessary to pay attention to factors such as steel type, chemical composition and crystal structure These factors can influence the material's temperature properties and response to the molding process
During the experiment, it is necessary to prepare and process steel materials into experimental samples of the necessary size and shape These test samples are then used to perform casting experiments and measure the temperature inside the mold
By using steel material as a test sample, the results and data obtained can help better understand the optimization of temperature distribution in the mold cavity and apply these results to actual casting processes use steel.
Geometric models
The choice of the rectangular model as the basis for the die casting has proven to be suitable in many industrial applications This geometric model shown in figure 3.3 allows for the construction of accurate simulations of heat transfer and temperature distribution in the mold In addition, the rectangular model has the advantages of simple calculations, accurate results, and scalability for more complex systems To simplify the simulation process, we first need the technical parameters of the model shown in table 3.1
Thanks to these benefits, the rectangular model is a suitable choice for researching the topic of mold temperature control in many practical applications, helping to simplify the process of designing, simulating and optimizing the system, contributing to improving the quality of molded products and production efficiency
Name Dimession(mm) Height(mm)
3.5.1 Simulating heat transfer using ANSYS software
Accurate control of the mold temperature during the casting process is extremely important to ensure the quality of the final product One of the powerful tools used to simulate and control the heat transfer process in a mold is ANSYS software ANSYS is a multiphysics simulation software and is widely used in many industrial sectors, including the metal foundry industry
CFX uses finite volume (CFD) to solve Navier-Stokes equations and heat transfer equations This method breaks down the simulation domain into small volumes (computational grids) and solves equations for each volume
The finite volume method (CFD) uses a variety of mathematical formulas to describe heat transfer in complex systems Formulas such as the Fourier equation, and the radiation formula help to accurately describe the mechanisms of heat transfer, convection, and radiation
As a result, it is possible to predict the temperature distribution, heat transfer rate and other important parameters in the simulation system
∂T/∂t = α∇²T This formula describes the process of conductive heat transfer in a solid material, with α being the thermal conductivity and T being the temperature
Q = σeT⁴ This formula describes the process of radiant heat transfer, where Q is the heat flux, σ is the Stefan-Boltzmann constant, e is the emissivity, and T is the absolute temperature
∂ρe/∂t + ∇ã( ρve) = ∇ã( λ∇T) + ρQ This formula conserves energy in the system, with ρ being the density, e being the internal energy, v being the flow velocity, λ being the thermal conductivity, T being the temperature, and Q being the energy source
∂ρv/∂t + ∇ã( ρvv) = ∇p + ∇ã( μ∇v) + ρf This formula describes the movement of fluids, including stratigraphic flows and turbulent flows
First, it is necessary to build an accurate geometric model, including details such as the shape, size, and location of parts, inlets, and outputs of the heating process This can be done using various 2D and 3D tools or can be done directly on Ansys simulation software
Accurate simulation results: The more accurate the model, the closer the simulation results to the temperature distribution in the mold cavity are to reality
Efficient optimization process: The GA algorithm will be based on simulation data to find the optimal solution for heating parameters Therefore, the accuracy of the model directly affects the efficiency of the optimization process
Save time and costs: Accurate models help minimize the risk of errors during simulation and optimization, thereby saving time and costs for performing real experiments
The step of building a geometric model is the first and important step in the process of simulating mold cavity temperature using Ansys, ANN and GA Using the right tools and carrying out the modeling process carefully will help ensure the accuracy of simulation and optimization results
After having a geometric model, it is necessary to divide the computational mesh for this model The more detailed the calculation grid, the more accurate the simulation will be, but it will also increase the calculation time and effort Therefore, it is necessary to strike a balance between accuracy and calculation time using optimal meshing techniques
Computational meshing (mesh) is an important step in the process of simulating mold cavity temperature using Ansys The computational mesh consists of small elements (such as triangles, tetrahedra, hexahedrons) used to represent the simulation domain Meshing directly affects the accuracy and efficiency of the simulation:
- Accuracy: The more detailed the mesh, the more accurate the simulation because it can describe in more detail the temperature distribution within the mold cavity
- Efficiency: The more detailed the mesh, the more computational time and effort it takes Therefore, it is necessary to choose the appropriate size and type of mesh
- Model geometry: Models with complex shapes often need more detailed meshing
- Desired Accuracy: The desired level of accuracy of the simulation results will determine the level of detail of the mesh
- Types of grid elements: There are many different types of grid elements, each with its own advantages and disadvantages
- Calculating ability: Computer computing ability is also a factor to consider when choosing the size and type of mesh element
There are many different meshing techniques that can be used, including:
- Manual meshing: Users manually create meshes by identifying nodes and connecting them This technique is time consuming and laborious, but allows for precise pitch control of the mesh
- Automatic meshing: Simulation software automatically creates meshes based on parameters provided by the user This technique is quick and easy, but the accuracy may not be as high as manual meshing
- Adaptive meshing: The mesh is automatically adjusted during the simulation to focus on areas with high temperature gradients This technique allows achieving high accuracy with reasonable computational time
➢ Step 3: Define the initial and marginal conditions
In ANSYS, users need to define initial and boundary conditions for the simulation Accurate definition of initial and boundary conditions in ANSYS is important for the accuracy of cavity temperature simulation These conditions describe the initial state and factors that affect the heat transfer process in the mold cavity, including:
Initial temperature of mold parts: Determine the initial temperature of mold parts before starting the heating process
Temperature and heat transfer coefficient at mold surfaces: Determine the temperature and heat transfer coefficient at surfaces in contact with the surrounding environment, such as air, cooling water
Heat sources: Identify the heat sources that power the heat transfer process, including molten metal temperature and mold heating
Heating time and heating temperature: Determine the heating time and temperature for the mold cavity, affecting the speed and extent of heating
Influence of initial and boundary conditions:
Deviations in the definition of initial and boundary conditions can lead to significant errors in simulation results, including:
Inaccurate simulation results: The temperature distribution in the simulated mold cavity does not match reality
Ineffective optimization process: The GA algorithm may search for an optimal solution that is not consistent with reality because it is based on erroneous simulation data
Wasted time and costs: Wrong simulations can lead to wasted time and costs in performing real experiments
Procedure for defining initial and boundary conditions:
The process for defining initial and boundary conditions in ANSYS typically includes the following steps:
Collect data: Collect information about initial temperature, ambient temperature, heat transfer coefficient, heating capacity, heating time
Determine conditions: Based on collected data, determine specific values for initial and boundary conditions
Import data into ANSYS: Import defined values into ANSYS software
Check and validate: Check the reasonableness and accuracy of the entered conditions
➢ Step 4: Calculate the heat transfer process
The heat transfer calculation step in ANSYS plays a key role in simulating mold cavity temperature After building the geometric model, meshing it and determining the initial and boundary conditions, ANSYS will use differential equations of heat transfer to simulate the temperature distribution in space and time in casting mold This calculation process includes:
- Solving differential equations: ANSYS uses numerical methods such as the finite element method to solve the differential equations that govern heat transfer in the mold cavity
- Calculate temperature at grid points: After solving the differential equations, ANSYS will calculate the temperature at grid points in the model
- Storing results: Calculation results, including temperatures at grid points and other parameters related to the heat transfer process, are stored for use in the next steps ANSYS offers a variety of computational methods to simulate heat transfer, including:
- Steady heat transfer method: This method is used to simulate the heat transfer process when the temperature does not change over time
- Unsteady heat transfer method: This method is used to simulate the heat transfer process when the temperature changes over time
- Combined heat transfer method: This method combines both steady and unsteady heat transfer methods to simulate complex heat transfer processes
The accuracy of heat transfer calculations depends on many factors, including: models, the higher the accuracy of the calculation
- Calculation method: Choosing the appropriate calculation method will affect the accuracy of the results
- Calculation parameters: Calculation parameters such as thermal conductivity, heat capacity, heat transfer coefficient need to be determined accurately
- Calculation ability: The computer's computing ability is also a factor that affects the accuracy of calculation
SIMULATION - PREDICTION OF ANN AND
Simulation model
The selection of an appropriate ANSYS simulation model is crucial for obtaining accurate and reliable results that accurately represent the physical phenomena being analyzed This process involves careful consideration of several factors, including:
Defining the problem and objectives: Clearly define the problem you are trying to solve and the specific objectives you aim to achieve through the simulation This will guide the choice of model type and complexity
Identifying the physics: Determine the governing physical principles involved in the problem, such as heat transfer, fluid flow, structural mechanics, or electromagnetics This will help narrow down the selection of relevant ANSYS modules and solvers
Understanding the geometry and materials: Accurately represent the geometry of the object or system being simulated, including its dimensions, shapes, and material properties This will influence the model's meshing requirements and boundary conditions
Considering boundary conditions: Define the appropriate boundary conditions that represent the physical constraints and external influences acting on the system This may include temperature distributions, heat fluxes, fluid flow conditions, or structural loads Selecting the model type: Choose the appropriate ANSYS model type based on the physics involved and the desired level of detail This could involve selecting a steady-state or transient model, a 1D, 2D, or 3D model, and a specific solver type
Meshing the model: Divide the geometry into small elements or cells to enable numerical approximation of the governing equations The mesh size and quality significantly impact the accuracy and efficiency of the simulation
Running the simulation: Execute the simulation using the chosen model and solver settings Monitor the convergence of the solution and ensure that the simulation runs data, extracting relevant parameters, and comparing them with experimental data or theoretical predictions Interpret the results to gain insights into the physical behavior of the system
Validating the model: Validate the simulation model by comparing its results with analytical solutions, experimental data, or benchmark problems This ensures the model's reliability and credibility
Refining the model: Based on the validation results, refine the model if necessary by adjusting mesh parameters, boundary conditions, or model assumptions This iterative process improves the model's accuracy and applicability
Figure 4.1 Shape and size of the simulation model
In this topic, the mold heating area is designed with an insert plate detail measuring
150 mm x 70 mm shown in figure 4.1 and 4.2 In previous studies in the field of mold surface heating, mold design with insert plate parts is often used to increase the efficiency of the heating process These studies show that the thickness of the insert plate is one of the important parameters, which has a great influence on the heating result for the mold
Figure 4.2 The insert with thickness 1mm
Using insert plates in the design of the plastic injection mold cavity heating zone brings many benefits, including:
-Increased heating efficiency: Inserts help transfer heat faster and more evenly than conventional mold materials, resulting in reduced heating time and improved overall efficiency
-Improved temperature uniformity: The insert helps distribute heat more evenly across the mold surface, reducing the risk of hot or cold spots, contributing to improved product quality
-Increase mold life: Insert plates are often made from materials that are more durable than conventional mold materials, helping to extend mold life and reduce maintenance costs
In previous studies in the field of mold surface heating, there have been many results proving that the thickness of the insert plate has a significant influence on the heating efficiency Accordingly, the optimal thickness of the insert depends on many factors, including:
-Material of mold and insert plate: Each material has different thermal conductivity, affecting the rate of heat transfer through the insert plate
-Size and shape of the mold cavity: The size and shape of the mold cavity affects how heat is distributed in the mold
-Heating method: Different heating methods (resistance, hot oil, hot water) have different heat transfer efficiency
Normally, the insert plate thickness is selected in the range from 5 mm to 20 mm However, it is necessary to conduct simulation or testing to determine the optimal thickness
Simulation process
The use of fine meshes is especially important for simulations that require high accuracy, such as simulating turbulent flows, heat transfer in heterogeneous materials, or simulating stresses in complex structures
However, the finer the mesh, the greater the number of mesh elements, leading to the need for more calculations to solve the problem This can significantly increase calculation time, especially for complex 3D models and unstable simulation problems Using a mesh that is too fine may exceed the computational capabilities of the computer hardware, leading to memory shortage or memory overflow errors
Therefore, it is necessary to carefully consider the appropriate grid size based on the complexity of the problem, the desired accuracy, and the computational capabilities of the system
The choice of mesh fineness depends on many factors, including:
- Geometry complexity: The more complex the geometry, the finer the mesh needs to be used to accurately simulate the details
- Phenomenon to be simulated: Local and nonlinear phenomena require the use of a finer mesh
- Desired precision: The desired level of precision directly affects the mesh size
- Computing capacity: It is necessary to consider the computing capacity of the hardware system to choose a grid of appropriate size
Below are some mesh parameters of the rectangular model and the results are shown in Figure 4.3
The inflation options you provided are suitable for defining an inflation layer in ANSYS for meshing purposes
This option defines the maximum thickness of the entire inflation layer The inflation layer is a refined region around the boundaries of your geometry where the mesh elements are smaller and more dense This is important for capturing accurately the phenomena occurring near the boundaries, such as heat transfer or fluid flow
The growth rate controls the rate of element size increase as you move away from the boundary towards the interior of the domain A growth rate of 1.2 means that the element size will increase by 20% with each layer of inflation This gradual increase helps ensure a smooth transition between the fine mesh near the boundary and the coarser mesh in the interior
This sets the maximum thickness of the inflation layer In your case, the maximum thickness is 0.5 millimeters This value should be chosen based on the characteristic lengths of the features you want to capture accurately
The element size option is not specified in your input This option allows you to define a target element size for the first layer of inflation elements adjacent to the boundary ANSYS
Air temperature Relative presser 0 MPA
Perpendicular to the air heating surface
Figure 4.4 Initial conditions of the simulation process
The simulation model is meshed as shown in Figure 4.2 and the initial conditions of the input and output of the insert plate are shown in Figure 4.3 in an ANSYS set-up environment
To be able to control the insert plate temperature as desired and know the exact temperature at the points to complete the output data set, in the ANSYS result environment, we proceed to create 22 points on the stamp plate with the equation
“maxVal(Temperature)@Point a”, where “a” is the point number from 1 to 22 After heating with a gun at a temperature of 400 C for 20 seconds with an insert whose polyline pitch has not been optimized one after another are 4, 3.5, 3, 2.5, 2 The simulation results are shown in figures 4.4, 4.5 and 4.6 and the temperatures are shown in figure 4.8
Figure 4.5 Results of the simulation process
Figure 4.6 Temperature results of the simulation process at point 11
Figure 4.7 Temperature results of the simulation process at point 19
Figure 4.8 Temperature at 22 Output Points
The temperature table of the simulation process is shown as figure 4.8 which is the temperature result of the original model at 22 points marked on the insert plate
Figure 4.9 Temperature chart of the simulation process when it is not optimal
After finishing the simulation process and determining the temperature of the 22 output points at the 5 initial polyline positions To be able to find the most optimal coordinates, we proceed to create a polyline elevation univariate grid covering table including all possible cases created with those 5 elevations as shown in table 4.1 below Then go to the parameters section of the ANSYS simulation and enter all polyline cases and update to check the temperature at all cases
After completing the simulation process on ANSYS and finding the temperature value of 22 points on the stamp plate in all possible cases of 5 polyline elevation points and also the output temperature data is shown in the table 4.2 below We proceed to use this data as input data of the ANN process.
ANN training process
- A neural network is a sophisticated computational model that mimics the structure and functionality of the human brain Renowned for its ability to tackle intricate problems and manage vast quantities of data, neural networks have emerged as vital tools across numerous domains They play critical roles in fields such as image and speech recognition, financial forecasting, and medicine This report aims to provide a comprehensive overview of the structure, training process, and common applications of neural networks
- The Neural Network Fitting application in MATLAB serves as a highly effective tool for addressing input-output fitting challenges using two-layer neural networks This application facilitates the setup and training of neural networks, enabling users to easily assess the performance of their networks and optimize the data fitting process The user-friendly interface and robust capabilities of this tool make it an invaluable resource for those seeking to implement neural networks in their projects
- In today's technological landscape, the use of artificial neural networks (ANNs) has become increasingly prevalent for solving input-output data matching problems MATLAB offers the powerful Neural Network Fitting Tool (nftool), which simplifies the process of setting up, training, and evaluating neural networks This paper will provide a detailed account of the steps involved in initializing a neural network training program using MATLAB It will cover everything from the initial setup to the fine-tuning of the network, ensuring optimal performance and accurate results
- Neural networks are a subset of artificial intelligence (AI) and machine learning, designed to emulate the neural activity of the human brain These networks possess the capability to learn from data and make informed predictions or decisions based on their learning process By simulating brain activity, neural networks can identify patterns, recognize trends, and provide insights that are critical for decision-making in various applications Their ability to learn and adapt makes them indispensable in the realm of AI and machine learning, continually advancing the frontiers of technology and innovation
Neural networks, inspired by the human brain's structure and function, are powerful computational models designed to solve complex problems and process large amounts of data They have become essential tools in various domains such as image and speech recognition,
(nftool) provides a user-friendly interface for setting up, training, and evaluating neural networks for input-output fitting problems By entering 'nftool' in the command window, users can launch this tool, which guides them through loading and preprocessing data, defining the network structure, training the network, and assessing its performance Typically, the tool uses a two-layer neural network, involving an input layer, a hidden layer, and an output layer, and employs training algorithms like Levenberg-Marquardt backpropagation Throughout the training process, users can monitor metrics such as Mean Squared Error (MSE) to evaluate how well the network is learning and adjusting its weights The Neural Network Fitting Tool allows for the real-time visualization of these metrics, enabling users to make necessary adjustments and fine-tune the network for optimal performance Once training is complete, the tool provides detailed plots and metrics to help users assess the network's efficacy and ensure that it generalizes well to new data If the initial performance is unsatisfactory, users can iteratively adjust parameters and retrain the network until the desired performance is achieved The ability to export the trained network to the MATLAB workspace further enhances its utility, allowing for the prediction of outputs for new, unseen data
Following the architech ANN of Figure 4.10 Neural networks find applications in various fields due to their ability to learn from data and make predictions In image and speech recognition, Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) are used to identify patterns and convert spoken language into text In financial forecasting, neural networks model complex relationships to predict stock prices and market trends Medical diagnosis leverages neural networks to analyze patient data and medical images, aiding in disease detection and prognosis In natural language processing, neural networks perform tasks such as language translation and sentiment analysis Additionally, in robotics and automation, neural networks enable intelligent decision-making and autonomous operations MATLAB's nftool, with its robust capabilities and ease of use, makes neural networks an indispensable asset in these advanced computational tasks, driving innovation and technological advancement
Figure 4.10 Neural network model structure
A neuron is the basic unit of calculation of a neural network Each neuron receives an can be a sigmoid, fish, ReLU, or other functions depending on the requirements of the model
Neural networks consist of many classes of neurons organized at different levels:
- Input Layer: Receive input signals from data
- Hidden Layer: Processing signals and extracting features Neural networks can have many hidden layers, creating a deep neural network
- Output Layer: Provides the final result of the neural network
Each connection between neurons has a weight, which determines how much the signal is affected Weights are adjusted during network training to optimize network efficiency
• Application of matlab in the use of ANN artificial intelligence optimization
- MATLAB is a powerful tool for the development and optimization of artificial intelligence models, particularly artificial neural networks (ANNs) It offers a comprehensive suite of libraries and tools that provide a flexible and efficient environment for building, training, testing, and deploying ANN models MATLAB's user-friendly interface and robust computational capabilities make it an ideal choice for researchers and engineers working in the field of AI The integration of various toolboxes, such as the Deep Learning Toolbox, enhances MATLAB's capability to handle complex ANN optimization tasks, enabling users to leverage advanced algorithms and techniques to improve model performance
- One of the primary applications of MATLAB in ANN optimization is in the initial development and training of neural network models MATLAB provides a variety of functions and tools that simplify the process of designing neural network architectures, selecting appropriate algorithms, and configuring training parameters The platform supports different types of neural networks, including feedforward networks, convolutional neural networks (CNNs), and recurrent neural networks (RNNs) With MATLAB, users can easily experiment with different network configurations and training methods to identify the most effective approach for their specific application
- MATLAB also excels in the testing and validation phase of ANN development The environment allows users to rigorously evaluate their models using various metrics and visualization tools This capability is crucial for assessing the performance of neural networks and ensuring they generalize well to new, unseen data MATLAB's visualization tools, such as confusion matrices and performance plots, help users gain insights into model behavior and identify areas for improvement Additionally, MATLAB supports cross-validation techniques that are essential for robust model evaluation and preventing overfitting
- Another significant advantage of using MATLAB for ANN optimization is its ability to deploy trained models efficiently MATLAB offers tools for converting models into deployable code, which can be integrated into various applications, including embedded systems and web services This feature is particularly valuable for real-time applications where computational efficiency and reliability are paramount The seamless integration with other MATLAB toolboxes and external libraries further enhances the versatility of MATLAB in deploying ANN models across different platforms and environments Overall, MATLAB's extensive capabilities in ANN development, optimization, testing, and deployment make it a comprehensive and powerful tool for advancing artificial intelligence and machine learning applications
STT Pitch of the point
With large data of the table 4.2 volumes and weighted weights on Input and Output data values being the same, saving time and using high-tech algorithms is extremely important in generating predicted results that are consistent with calculations and experiments Therefore, the use of the ANN algorithm is extremely reasonable Choosing to use the ANN – GA optimization tool is the most optimal tool in MATLAB – a neural network optimization tool that focuses on optimizing data and predicting results
Typing the nftool starting the ANN network programing
Figure 4.11 Neural network optimization tool interface
When the 'nftool' command is initiated, the neural network training tasks will appear in the MATLAB interface The project team configured the properties of the neural network, including the data partition, the training method, and the performance function The first step involves random data splitting, which randomly divides the data into training, validation, and testing sets to ensure the model's generalizability The figure 4.11 shows The training method employed is the Levenberg-Marquardt algorithm, known for its powerful optimization capabilities This algorithm efficiently minimizes the error by adjusting the network's weights The performance function used is the mean squared error (MSE), which quantifies the difference between the predicted and actual values, serving as a critical measure of the model's accuracy
To train the network, the "train" function is utilized, which requires input and output data This training process enables the neural network to learn how to predict output values from the given input data by iteratively adjusting the weights to minimize the error The Levenberg-Marquardt algorithm ensures that the training is both efficient and effective, helping the network converge to an optimal solution Throughout this process, the network continually refines its predictions, improving its performance as measured by the reduction in MSE
After the training is complete, the project team evaluates the network's performance by comparing the predicted output with the actual output This evaluation is crucial as it provides insight into how well the network has learned from the data and how accurately it can make predictions The primary metric used for this evaluation is the MSE index, which reflects the average squared difference between the predicted and actual values A lower MSE indicates a better-performing model By examining the MSE and other performance metrics, the team can determine the effectiveness of the neural network and make any necessary adjustments to improve its accuracy and reliability in predicting outcomes based on new input data
Table 4.4 Table of algorithms for performing data prediction
Performance Mean Squared Error (MSE)
Optimization process by ANN – GA genetic algorithm
Genetic algorithms (GA) are a powerful and flexible tool for solving complex optimization problems Based on the principles of natural selection and genetics, GA has the ability to find solutions that are globally optimal, maintain diversity in populations, and can be applied to many different types of problems in many fields
CONSTRUCT OBJECTIVE FUNCTION FOR GA
Figure 4.27 Block diagram tranning ANN -GA
After performing network training and training with ANN, we will perform ANN-GA optimization Combining artificial neural networks (ANNs) with genetic algorithms (GA) brings many significant benefits in solving complex problems in machine learning and optimization Combining an artificial neural network (ANN) with a genetic algorithm (GA) offers many outstanding benefits, including the ability to solve complex problems, global search, increased optimization efficiency, reduced computational complexity, and wide applicability As seen in figure 4.27, there is a significant trend in this combination provides a robust and flexible approach to solving optimization problems in a variety of areas, enhancing the performance and accuracy of prediction and optimization systems
The "network" data that is tranned from the ANN network will be carried out and output the transmitted data in combination with the GA genetic optimization algorithm
❖ GA Multiobjective Optimization Using Genetic Algorithm is a multi-objective optimization method using a genetic algorithm (GA) This method aims to find optimal solutions to problems with multiple goals that need to be optimized simultaneously The genetic algorithm simulates natural evolution to find the best solutions based on the set criteria
4.4.2 Setting up the GA genetic algorithm
• Before implementing the GA Multiobjective Optimization Using Genetic Algorithm optimizations, we need to make some settings:
➢ Setting up the goal function:
A target function is one or more functions that we need to optimize In the case of multi-objective optimization, we have many different objective functions that need to be optimized simultaneously
Defining variables determines variables that are factors or parameters that we can adjust to optimize the target function
Constraints are the conditions or limits that variables must follow during the optimization process
➢ Set the input variable number:
Determine the number of decision variables involved in the optimization process
• First, to run the GA optimization program, we do:
- Create a target function: objfucntion = @(x) sim(net, x')
Table 4.10 Creating the Lower Margin for the Input Variable
Table 4.11 Creating the Lower Margin for the Output Variable
- Target functions or constraints related to decision variables are not linear functions This means that the relationships between variables and target functions can be complex and may include exponential functions, logarithms, or other nonlinear functions Nonlinear optimization problems are often more difficult to solve than linear ones due to their complex nature They can have many local extremes, and there is no specific solution method that can guarantee finding the global extreme
- "objfunction" is a target function that will be used during optimization This function takes the x input variables and uses the net neural network to compute the output This output will be used to evaluate the quality of the x input variables during optimization
- sim: is a MATLAB function used to simulate a neural network This function calculates the output of the net neural network for the input values x
- NET: A neural network that has been trained before This network is used to predict the output value based on the input value x
- x': The ' (single comma) is a transpose in MATLAB It converts the x-row matrix into a column matrix to match the input format of the sim function
In the optimization process using Genetic Algorithms (GA), population size is a crucial parameter A large population size helps maintain diversity within the population, preventing premature homogenization This diversity allows the GA to explore more potential solutions within the search space By maintaining high genetic diversity, the GA is better equipped to conduct a global search, thereby reducing the risk of becoming trapped in local optima
A larger population increases the likelihood of discovering better potential solutions range of genetic variations, enhancing the algorithm's ability to identify and exploit beneficial traits Consequently, a larger population contributes to the robustness of the search process, facilitating the identification of optimal solutions
Establishing the correct population size is integral to the GA optimization process It ensures that the algorithm maintains sufficient diversity and achieves effective search performance while converging on the best possible solutions After optimizing the data with an appropriately sized population, we obtain a set of solution numbers and their corresponding objective values, reflecting the success of the GA in finding optimal solutions
Figure 4.32 Values ANN-GA Optimization
- In the initial generations, genetic diversity was notably high as the populations
During this phase, the average distance between individuals ranged from 1.2 to 2.4 As the generations progressed, the average distance gradually decreased, yet it still remained within the range of 1.2 to 2.0 This indicates that while the population began converging towards better solution regions, it retained sufficient diversity to continue exploring new solutions
- High genetic diversity in the early generations plays a critical role in exploring a wide solution space, thereby increasing the likelihood of discovering optimal solutions The graph depicting this process demonstrates an efficient search, as evidenced by the decreasing average distance, which still maintains the necessary level of diversity As depicted in figure 4.33 this balance ensures that the population stabilizes around optimal solutions over time The graph also serves as a tool for assessing and tracking changes in genetic diversity within the population across generations Maintaining high genetic diversity is crucial as it enhances the ability to explore the search space thoroughly and prevents premature convergence By monitoring genetic diversity, one can ensure the optimization process remains efficient and effective
- The graph provides valuable insights into the efficiency of the optimization process by showing how the population converges towards optimal solutions while maintaining necessary diversity This monitoring aids in adjusting optimization parameters to achieve the best performance, ensuring a robust and reliable search process
Figure 4.34 Fitness Of each Imdividual
The fitness levels of individuals within the population at a specific stage in the Genetic Algorithm (GA) optimization process reveal significant diversity in their adaptability This variation in fitness levels is a positive indicator of genetic diversity within the population, which is essential for the efficient functioning of the GA Such diversity allows for a more effective natural selection process, where different traits and solutions are tested and the most adaptable ones are retained Individuals with lower fitness values, approximately around 200, are the least adaptable and thus less likely to be selected for future generations On the other hand, individuals with higher fitness values, nearing 500, represent the most adaptable and are considered the best performers within the population
The adaptability and diversity of the population at any given time are crucial metrics for assessing the evolutionary progress and the potential of the population to discover optimal solutions High levels of both adaptability and diversity suggest that the population is evolving well, continuously exploring the solution space and avoiding premature convergence The graph depicting fitness levels clearly demonstrates this diversity, showing a wide range of fitness values This diversity is beneficial for the evolutionary process as it indicates that the GA is effectively maintaining a variety of solutions, which is critical for finding the best possible outcomes
The presence of individuals with a wide range of fitness values ensures that the natural selection process within the GA operates efficiently This variation allows for the selection of the most adaptable individuals, helping the population to evolve and improve over successive generations The fitness distribution graph provides important insights into the current state of the population, highlighting areas where adjustments to GA parameters may be necessary
By monitoring this distribution, it is possible to tweak the parameters to maintain or enhance the optimization efficiency, ensuring that the GA continues to perform effectively
EXPERIMENT – EVALUATION
Temperature Measurement Experiment
The heating process will use the following tools:
Figure 5.1 Makita HG6530 Heat Blower Gun
Table 5.1 Parameters of Makita HG6530 hot air blower
Figure 5.1 and table 5.1 describe the heating gun and its parameter table
Figure 5.2 is a Fluke thermal camera enhancing user productivity with 320 x 240 resolution Infrared images are taken by capturing smaller temperature differences from a distance
Figure 5.3 The optimized air cover used for experiments with the optimal parameters obtained after running GA
Figure 5.4 Optimized air cover model
Figures 5.3 and 5.4 depict the optimized air cover used for the experiments with the optimal parameters obtained after running the GA
Figure 5.5 The non-optimized air cover used for experiments
Figure 5.6 Non-optimized air cover model
Figures 5.5 and 5.6 depict the non-optimized air cover used for the experiments with the optimal parameters obtained after running the GA
Table 5.2 is the experimental parameters including input temperature and heating time
Fix the insert plate to the mold cavity as described in the figure 5.7
Figure 5.7 Inserting the insert into the mold
Turn on the gun and adjust the temperature to 400°C (figure 5.8)
Figure 5.8 Adjusting the gun temperature to 400°C
After the gun has reached 400°C, proceed to place the air scan on the stamp fixed on the mold Then use a gun to blow air into the inlet hole to carry out heating This process is illustrated in the figure 5.9
After about 20 seconds, pull out the gun and shoot when it comes out of the stamp and use the infrared camera to capture the temperature distribution on the insert
❖ This process applies to both non-optimized and optimized air cover
Figure 5.11 Results measured after a 20-second heating period when using suboptimal gas imaging
In the figure 5.11, after a period of 20 seconds, the maximum measured temperature was 130.4°C When using a non-optimized air cover (a flat profile), the temperature is mainly concentrated in the central area and is not widely distributed on the stamp
Figure 5.12 Results measured after a 20-second heating period when using an optimized gas scan
In the figure 5.12, the results measured after 20 seconds when using the optimized air cover (polyline profile) showed a higher temperature, about 172.6 degrees Celsius This temperature is 40 degrees celsius higher than the non-optimal air cover The temperature zone is more widely distributed and appears to be more even.
Evaluation of experimental results
Figure 5.13 Graph comparing the temperature of the mold cavity when using a polyline- shaped air cover and a flat profile air cover
The graph 5.13 above represents the temperature change by distance (mm) in the mold for two cases: using optimized air cover (polyline profile) and non-optimal (flat profile) The black line, which represents the temperature when using optimised air imaging, shows that the highest temperature reaches around 170°C in the center position, then gradually decreases towards the ends In contrast, the red line, which represents the temperature when using an unoptimized gas scan, shows that the highest temperature reaches about 130°C at the same location and also decreases to both ends
With the same input temperature of 400°C and a heating interval of 20 seconds, the graph clearly shows that the temperature of the insert plate when using the optimized air cover is 170°C, which is about 40°C higher than the non-optimal air cover This makes the heating process faster, shortening the production time when using optimized air cover.
CONCLUSION
• In this project, we used a combination of artificial neural networks (ANNs) and genetic algorithms (GA) to optimize the parameters of the air cover to improve the heating process of the rectangular plastic injection mold The research process is carried out through two main stages: simulation and experimental
- The simulation results show that the use of the ANN-GA model can effectively predict and optimize the heating process, which helps to determine the optimal conditions for achieving high temperatures and more uniform temperature distribution in the mold cavity
- The experimental results show that when using optimal air cover (polyline profile), the measured temperature in the mold reaches 172.6°C, which is about 40°C higher than when using non-optimal air cover (flat profile) This shows that the optimization of air cover not only improves the temperature, but also increases the uniformity of the temperature in the mold cavity, thereby improving product quality and production efficiency
• Research has proven that the use of ANN-GA to optimize air cover for plastic injection mold cavities heating is effective and practical Experimental results have shown a marked improvement in temperature and temperature uniformity in the mold when using optimal gas imaging
• This method not only helps to reduce heating time and improve product quality, but also opens up a new direction in applying ANN-GA technology to other industrial processes to optimize production efficiency and quality
- In addition to ANN-GA, other optimization algorithms such as Particle Swarm Optimization (PSO) or Differential Evolution (DE) can be tested and applied to find the most optimal solution for the heating process
- Expand the research to apply optimization methods for different types of molds in plastic production, in order to find the optimal parameters for each specific type of mold
• These development directions will contribute to improving production efficiency and product quality, and open up new opportunities in research and application of optimization technology in the plastic manufacturing industry
[1] Z Wei et al., “Review of conformal cooling system design and additive manufacturing for injection molds,” Math Biosci Eng., vol 17, no 5, pp 5414–5431, 2020
[2] A Torres-Alba, J M Mercado-Colmenero, D Diaz-Perete, and C Martin-Doủate,
“A new conformal cooling design procedure for injection molding based on temperature clusters and multidimensional discrete models,” Polymers (Basel), vol 12, no 1, p 154,
[3] J.-H Han and Y.-C Kim, “Study on effects of mold temperature on the injection molded article,” Arch Metall Mater., vol 62, no 2, pp 1271–1274, 2017
[ ] J Li, T Li, X Peng, F Liu, H Zhou, and S Jiang, “Optimal design of heating system for electrical rapid heat cycle mold based on multi-objective optimization, multiple- attribute decision-making, and conformal design theory,” Adv Mech Eng., vol 10, no 8, p
[ ] S Y Martowibowo and A Kaswadi, “Optimization and simulation of plastic injection process using genetic algorithm and moldflow,” Chin J Mech Eng., vol 30, no 2, pp 398–406, 2017
[6] C ệzek and Y H ầelık, “Calculating molding parameters in plastic injection molds with ANN and developing software,” Mater Manuf Process., vol 27, no 2, pp 160–168,
[7] S.-C Chen, P S Minh, and J.-A Chang, “Gas-assisted mold temperature control for improving the quality of injection molded parts with fiber additives,” Int Commun Heat
Mass Transf., vol 38, no 3, pp 304–312, 2011
[8] T Pham Quang, “A study on the temperature optimization of mold and melt using Design of Experiments for children’s chair,” Key Eng Mater., vol 969, pp 59–64, 2023
[9] P S Minh, Tran Minh The Uyen, and T T Do, “Numerical study on gas-assisted mold temperature control with the application of air cover for improving the cavity temperature in injection molding process,” in th International Conference on Green Technology and Sustainable Development (GTSD), 2020
[10] Pham Son Minh, Do Thanh Trung, Nguyen Ho, and Phan The Nhan, “Evaluation of heating process for the rectangular mold cavity by external gas-assisted heating”, JTE, vol
[11] Tran Minh The Uyen, P S Minh, and T T Do, “Experimental study on external