maedeh mosayeb motlagh; Parham Azimi; maghsoud Amiri
Abstract
This paper investigates unreliable multi-product assembly lines with mixed (serial-parallel) layout model in which machines failures and repairing probabilities are considered. The aim of this study is to develop a multi-objective mathematical model consisting the maximization of the throughput rate ...
Read More
This paper investigates unreliable multi-product assembly lines with mixed (serial-parallel) layout model in which machines failures and repairing probabilities are considered. The aim of this study is to develop a multi-objective mathematical model consisting the maximization of the throughput rate of the system and the minimization of the total cost of reducing mean processing times and the total buffer capacities with respect to the optimal values of the mean processing time of each product in each workstation and the buffer capacity between workstations. For this purpose, in order to configure the structure of the mathematical model, Simulation, Design of Experiments and Response Surface Methodology are used and to solve it, the meta-heuristic algorithms including Non-Dominated Sorting Genetic Algorithm II (NSGA-II) and Non-Dominated Ranked Genetic Algorithm (NRGA) are implemented. The validity of the multi-objective mathematical model and the application of the proposed methodology for solving the model is examined on a case study. Finally, the performance of the algorithms used in this study is evaluated. The results show that the proposed multi-objective mathematical model is valid for optimizing unreliable production lines and has the ability to achieve optimal (near optimal) solutions in other similar problems with larger scale and more complexity.IntroductionA production line consists of a sequence of workstations, in each of which parts are processed by machines. In this setup, each workstation includes a number of similar or dissimilar parallel machines, and a buffer is placed between any two consecutive workstations. In production lines, the buffer capacity and processing time of machinery have a significant impact on the system's performance. The presence of buffers helps the system to maintain production despite possible conditions or accidents, such as machinery failure or changes in processing time. Previous research has investigated production lines without any possibility of machinery failure, referred to as "safe production lines." However, in real production lines, machinery failure is inevitable. Therefore, several studies have focused on "uncertain production lines,"assuming the existence of a probability of failure in a deterministic or exponential distribution. This research examines uncertain production lines with a combined layout, resulting from the combination of parallel deployment of machines within each workstation, if necessary, and serial deployment of workstations. The objective of this research is to determine the optimal values (or values close to optimal) of the average processing time of each product in each workstation, as well as the volume of buffers, as decision variables. The approach aims to maximize the system's output while minimizing the costs associated with reducing the processing time of workstations and minimizing the total volume of buffers between stations. Moreover, simulation can be applied without interrupting the production line or consuming significant resources. In this research, due to the high cost and time involved, implementing the proposed changes on the system is not cost-effective for investigating the changes in the production system's output rate. Therefore, the simulation technique has been utilized to optimize the production line.Research methodThe present study aims to develop a multi-objective mathematical model, based on simulation, to optimize multi-product production lines. In the first step, the structure of the multi-objective mathematical model is defined, along with the basic assumptions. To adopt a realistic approach in the model structure, the simulation technique has been employed to address the first objective function, which is maximizing the output rate of the production line. To achieve this, the desired production system is simulated. The design of experiments is used to generate scenarios for implementation in the simulated model, and the response surface methodology is utilized to analyze the relationship between the input variables (such as the average processing time of each product type in each workstation and the buffer volume between stations) and the response variable (production rate).ResultsTo implement the proposed methodology based on the designed multi-objective programming model, a case study of a three-product production line with 9 workstations and 8 buffers was conducted. Subsequently, to compare the performance of the optimization algorithms, five indicators were used: distance from the ideal solution, maximum dispersion, access rate, spacing, and time. For this purpose, 30 random problems, similar to the mathematical model of the case study, were generated and solved. Based on the results obtained, both algorithms exhibited similar performance in all indices, except for the maximum dispersion index.ConclusionsIn this article, the structure of a multi-objective mathematical model was sought in uncertain multi-product production lines with the combined arrangement of machines in series-parallel (parallel installation of machines in workstations if needed and installation of workstations in series). The objective was to determine the optimal values of the average processing time of each type of product in each workstation and the buffer volume of each station, with the goals of maximizing the production rate, minimizing the costs resulting from reducing the processing time, and the total volume of inter-station buffers simultaneously. To investigate the changes in the output rate of the production system, due to the high cost and time, it was deemed not cost-effective to implement the proposed changes on the system. Therefore, the combination of simulation techniques, design of experiments, and response surface methodology was used to fit the relevant metamodel. In the proposed approach of this research, taking a realistic view of production line modeling, the probability of machinery failure, as well as the possibility of repairability and return to the system, were considered in the form of statistical distribution functions. Additionally, all time parameters, including the arrival time between the parts, the start-up time of all the machines, the processing time, the time between two failures, and the repair time of the machines, were non-deterministic and subject to statistical distributions. Finally, to solve the structured mathematical model, two meta-heuristic algorithms (NSGA-II) and (NRGA) were considered.
Roozbeh . Azizmohammadi; Maghsoud .Amiri; Reza Tavakkoli- Moghadam; Hamid Reza. Mashatzadegan
Volume 14, Issue 42 , October 2016, , Pages 103-121
Abstract
A redundancy allocation problem is a well-known NP-hard problem thatinvolves the selection of elements and redundancy levels to maximize thesystem reliability under various system-level constraints. In many practicaldesign situations, reliability apportionment is complicated because of thepresence of ...
Read More
A redundancy allocation problem is a well-known NP-hard problem thatinvolves the selection of elements and redundancy levels to maximize thesystem reliability under various system-level constraints. In many practicaldesign situations, reliability apportionment is complicated because of thepresence of several conflicting objectives that cannot be combined into asingle-objective function. A stele communications, manufacturing and powersystems are becoming more and more complex, while requiring shortdevelopments schedules and very high reliability, it is becoming increasinglyimportant to develop efficient solutions to the RAP. In this paper, a newhybrid multi-objective competition algorithm (HMOCA)based oncompetitive algorithm (CA) and genetic algorithm (GA) is proposed for thefirst time in multi-objective redundancy allocation problems. In the multiobjectiveformulation, the system reliability is maximized while the cost andvolume of the system are minimized simultaneously. Additionally, ay RSMis employed to tune the CA parameters. The proposed HMOCA is validatedby some examples with analytical solutions. It shows its superiorperformance compared to a NSGA-II and PAES algorithms. Finally, theconclusion is given
Reza Abasi; Jamshid Salehi Sadaghiani; Maghsoud Amiri
Volume 13, Issue 38 , October 2015, , Pages 79-98
Abstract
There are several different factors in industrial processes, which may have an impact on final product specifications. In this paper, the nonlinear multi-objective mathematical modeling for the 5 qualitative characteristics of polyethylene terephthalate (PET) as one of the most widely used products in ...
Read More
There are several different factors in industrial processes, which may have an impact on final product specifications. In this paper, the nonlinear multi-objective mathematical modeling for the 5 qualitative characteristics of polyethylene terephthalate (PET) as one of the most widely used products in petrochemical industry is done by 9 Key parameters affecting the production process. For this purpose, Response Surface Methodology that is the combination of experiments design and statistical tests is applied along with Artificial Neural Network and Metaheuristic algorithms.
Ali Mohtashami*
Volume 12, Issue 33 , July 2015, , Pages 97-124
Abstract
This paper presents a multi-objective mathematical model for redundancy allocation in production systems. In many of the production and assembly lines, process times, time between failures and repaired times are generally distributed. The proposed method of this paper is able to consider time dependent ...
Read More
This paper presents a multi-objective mathematical model for redundancy allocation in production systems. In many of the production and assembly lines, process times, time between failures and repaired times are generally distributed. The proposed method of this paper is able to consider time dependent parameters as general distribution functions by using the hybrid approach of simulation and response surface methodology. The objectives of the mathematical model are maximizing production rate, minimizing total cost and maximizing quality. In order to solve the proposed mathematical model, non-dominated sorting genetic algorithm and multiple objective particle swarm optimization are used. Numerical results indicate the effectiveness of both algorithms for generating non-dominated solutions. Moreover, comparative results indicate the superiority of the Non-dominated sorting genetic algorithm.
Hossein Khanaki; Mahdi Azizmohammadi; Masoud Vakili; Saeed Khan Mohammadian
Volume 11, Issue 30 , October 2014, , Pages 153-179
Abstract
AbstractIn this paper, the critical parameters of a method of welding with shieldinggas arc welding (GMAW) are discussed; this method is an important processin creating high quality metal permanent connections in various industries,including the automobile industry to improve the quality of stemdiameter ...
Read More
AbstractIn this paper, the critical parameters of a method of welding with shieldinggas arc welding (GMAW) are discussed; this method is an important processin creating high quality metal permanent connections in various industries,including the automobile industry to improve the quality of stemdiameter welding parameters. One of the most useful techniques for modelingand solving the problems is Response Surface Method. In this paper,considering five most important factors such as speed welder, torch anglewith the work piece, electrode diameter, wire speed, gas consumption ,andCO2 levels as input variables, can be controlled independently from thelevel of response, the relationship between the input variables and the responsevariables were determined using linear regression. Then optimumvalue for each factor was calculated using non-linear programming model to evaluate the results obtained along with the comparison of output of theSimulation Annealing Algorithm.In this study, both qualitative and quantitative variables are considered toevaluate and optimize all response variables regarding that these variablesare not the same, and then fuzzy set theory and LP metric are used to findanswers for multi-objective optimization methods.
maghsoud amiri; laya olfat; amir hasanzadeh
Volume 11, Issue 29 , July 2013, , Pages 89-112
Abstract
A major cause of supply chain deficiencies is the bullwhip effect. This phenomenon refers to demand variability increases as one move up the supply chain. Supply chain managers experience this variance amplification in both inventory levels and orders. Other side, dampening variability in orders may ...
Read More
A major cause of supply chain deficiencies is the bullwhip effect. This phenomenon refers to demand variability increases as one move up the supply chain. Supply chain managers experience this variance amplification in both inventory levels and orders. Other side, dampening variability in orders may have a negative impact on customer service due to increase in inventory variance. The contribution of this paper is the estimation of the bullwhip effect and stock functions with using of response surface method under causes of the bullwhip effect in a three stage supply chains consisting of a single retailer, single wholesaler and single manufacturer with both centralized and un-centralized chains and show that with Considering the importance of the bullwhip effect reasons and interactions between them, the percent of optimal reasons and interactions between them and the level of desirability of bullwhip effect and net Stock will be illustrated by a model
Mehdi Yazdani; Mahshid Aioobi; Amin Ghoroori
Volume 8, Issue 21 , June 2011, , Pages 131-142
Abstract
Some real world problems include determining optimum values of lput variables in order to obtain the desired levels of output variable response surface variable). One of the applicable techniques which are set for modeling and solving such problems is Response Surface methodology (RSM). In this paper, ...
Read More
Some real world problems include determining optimum values of lput variables in order to obtain the desired levels of output variable response surface variable). One of the applicable techniques which are set for modeling and solving such problems is Response Surface methodology (RSM). In this paper, the effect of three controllable lput factors: temperature, the density of sodium cyanide, and ampere n the determined response surface level i.e. the thickness of the electroplating cover of a hook screw is checked by Design Of experiments (DOE). After the execution of experiments and the ^cognition of effective factors, according to the application of response surface methodology, the relationship between the variables f effective input factors and the response surface variable is stermined with the help of nonlinear regression model. Then, the Optimum value of each variable in nonlinear model is obtained by goal pogromming method.