asiye aydilek | Gulf University for Science and Technology (original) (raw)

Papers by asiye aydilek

World Academy of Science, Engineering and Technology, International Journal of Industrial and Manufacturing Engineering, Jun 6, 2016

The two stage assembly flowshop scheduling problem has a lot of applications, and hence, it has r... more The two stage assembly flowshop scheduling problem has a lot of applications, and hence, it has recently received more attention in the scheduling literature. The performance measure of total tardiness is important as the fulfillment of due dates of customers has to be taken into account while making scheduling decisions. To the best of our knowledge, the problem with this performance measure has not been addressed so far, and hence, it is addressed in this paper. Different algorithms are proposed for the problem. The proposed algorithms are; an insertion algorithm, a genetic algorithm, two versions of simulated annealing algorithm (SA), and two versions of cloud theory-based SA. Moreover, the proposed insertion algorithm (PIA) is combined with the rest of the algorithms resulting in a total of eleven algorithms. Computational analysis, by using a non-parametric statistical test, indicates that one of the versions of the SA combined with the PIA performs better than the rest of the algorithms. The PIA helps in reducing the error of the SA by about seventy percent. It is worth to state that the performance of the combined algorithm is neither possible to achieve by the insertion algorithm alone nor by the simulated annealing alone no matter how much more computational time is given to the each.

Proceedings of the International Conference on Industrial Engineering and Operations Management

We address a no-wait flow shop scheduling problem where the setup times are treated as separate a... more We address a no-wait flow shop scheduling problem where the setup times are treated as separate and uncertain. The objective is to minimize maximum lateness which is one of the most essential performance measures for customer satisfaction and loss of goodwill. The considered problem is common in manufacturing environments like metal, food and pharmaceutical industries where some of the operations must not be interrupted. In many cases, setup time, the time required to set up a resource, is either ignored or assumed to be constant. However, these assumptions might lead to inefficient planning. Therefore, we relax these assumptions and consider uncertain but bounded setup times. The problem has been addressed in the scheduling literature where dominance relations were presented. In the current paper, we propose new dominance relations and show that the newly proposed dominance relations are around 100% more efficient than the existing. One of the main strengths of the newly proposed dominance relations is that their performances get better as the uncertainty in setup time increases. These dominance relations can also be used for initial sequence generation or improving the final solution as well as finding the optimal solutions for small size problems. Moreover, we also propose new and efficient heuristics as potential solutions for the problem and also for similar problems. Based on extensive computational experiments, we show that our new heuristics are around 90% more efficient than the existing ones in the literature under the same computational time.

International Journal of Industrial Engineering Computations

We address a manufacturing environment with the no-wait constraint which is common in industries ... more We address a manufacturing environment with the no-wait constraint which is common in industries such as metal, plastic, and semiconductor. Setup times are modelled as uncertain with the objective of minimizing maximum lateness which is an important performance measure for customer satisfaction. This problem has been addressed in scheduling literature for the two-machine no-wait flowshop where dominance relations were presented. Recently, another dominance relation was presented and shown to be about 90% more efficient than the earlier ones. In the current paper, we propose two new dominance relations, which are less restrictive than the earlier ones in the literature. The new dominance relations are shown to be 140% more efficient than the most recent one in the literature. As the level of uncertainty increases, the newly proposed dominance relation performs better, which is another strength of the newly proposed dominance relation. Moreover, we also propose constructive heuristics...

World Academy of Science, Engineering and Technology, International Journal of Industrial and Manufacturing Engineering, 2015

World Academy of Science, Engineering and Technology, International Journal of Mathematical and Computational Sciences, 2016

Engineering Optimization, 2020

The two-machine flowshop scheduling problem to minimize total completion time with separate setup... more The two-machine flowshop scheduling problem to minimize total completion time with separate setup times is addressed. Setup times are modelled as uncertain within an interval where only the lower and upper bounds are known. Eighty-one different versions of a newly developed constructive algorithm are proposed. Computational experiments to evaluate the performance of the proposed algorithm are conducted in two stages. In the first stage, 81 versions of the algorithm are compared with each other and the top seven versions are selected. In the second stage, the performances of the top seven versions are compared with the performance of the best existing known algorithm for the deterministic setup times solution in the literature. The computational results reveal that errors of the top seven (out of 81) versions of the algorithm are less than 0.005. All computational results are statistically verified. Therefore, the proposed algorithm has excellent performance.

The Quarterly Review of Economics and Finance, 2020

We investigate the existence of representative agent under various heterogeneities in a recursive... more We investigate the existence of representative agent under various heterogeneities in a recursive utility framework. We provide the analytical solution of household allocations. We numerically explore whether we can find a representative agent whose income is the aggregate income of the society and whose allocations are the aggregate allocations of the society under heterogeneity in the parameter of risk aversion and/or parameter of intertemporal substitution, or discount rate or survival probability. We find that there is no representative agent when the heterogeneity is in survival probability. From the data, we know that there is heterogeneity among people in terms of survival probability. For instance, women tend to live longer than men, rich people tend to live longer than poor people. We conclude that it may be better to use heterogeneous agents models than the representative agent models under the recursive utility framework. Our results support the idea that we do need the heterogeneous agent models indeed.

Journal of Industrial & Management Optimization, 2017

In this paper, we consider a manufacturing system with twomachine no-wait flowshop scheduling pro... more In this paper, we consider a manufacturing system with twomachine no-wait flowshop scheduling problem where setup times are uncertain. The problem with the performance measure of maximum lateness was addressed in the literature (Computational and Applied Mathematics 37, 6774-6794) where dominance relations were proposed. We establish a new dominance relation and show that the new dominance relation is, on average, about 90% more efficient than the existing ones. Moreover, since it is highly unlikely to find optimal solutions for problems of reasonable size by utilizing dominance relations and since there exist no heuristic in the literature for the problem, we propose constructive heuristics to solve real life problems. We conduct extensive computational experiments to evaluate the proposed heuristics. Computational experiments indicate that the performance of the worst proposed heuristic is at least 20% better than a benchmark solution. Furthermore, they also indicate that the best proposed heuristic is about 130% better than the worst one. The average CPU time of the heuristics is significantly less than a second. 2010 Mathematics Subject Classification. 90B36.

Applied Mathematics and Computation, 2020

The no-wait flowshop scheduling problem on m machines with separate setup times is addressed to m... more The no-wait flowshop scheduling problem on m machines with separate setup times is addressed to minimize total tardiness with an upper bound on makespan. Conditions for a dominance rule are established. Then, a new simulated annealing algorithm utilizing block insertion and block exchange operators is proposed, which we call a block simulated annealing algorithm. The proposed block simulated annealing algorithm is combined with an iterated search algorithm where the block simulated annealing algorithm explores the search space for a smaller total tardiness while the iterated search algorithm satisfies the constraint on the makespan. The proposed combined algorithm is called PA. Moreover, six closely related and well performing algorithms in the literature are modified to the problem, and PA is compared with these six algorithms. Extensive computational experiments reveal that PA reduces the error of the best modified algorithm by more than 50% for the same CPU times. Furthermore, the results are statistically tested, and thus, PA is recommended.

Journal of Economics and Finance, 2019

We investigate both of analytical and numerical solutions of retirees' spending and investment de... more We investigate both of analytical and numerical solutions of retirees' spending and investment decisions. We use a dynamic and realistic recursive utility setting which includes the standard expected utility setting as a special case. We find that recursive utility is superior to expected utility in terms of predicting retirees' consumption data. In addition to stock and bond investment decisions, we explicitly include housing decision. Our setting includes the setting without housing as a special case. We estimate retiree decisions numerically through simulations. We provide both of analytical and numerical comparative analysis which shows that some of the analytical dependencies are found to be weak numerically. For example, although marginal propensity to consume depends on the parameter of intertemporal substitution analytically, this dependence is found to be weak numerically. These differences show the importance of providing both of analytical and the numerical solutions. Our analytical solution could be useful for future studies to estimate some model parameters, to evaluate different elderly related policies, to quantify the welfare effects of different decisions and to analyze the parameter related issues such as the interchangeability of some parameters.

International Journal of Production Research, 2017

This paper addresses a two-stage assembly flowshop scheduling problem with the objective of minim... more This paper addresses a two-stage assembly flowshop scheduling problem with the objective of minimising maximum tardiness where set-up times are considered as separate from processing times. The performance measure of maximum tardiness is important for some scheduling environments, and hence, it should be taken into account while making scheduling decisions for such environments. Given that the problem is strongly NP-hard, different algorithms have been proposed in the literature. The algorithm of Self-Adaptive Differential Evolution (SDE) performs as the best for the problem in the literature. We propose a new hybrid simulated annealing and insertion algorithm (SMI). The insertion step, in the SMI algorithm, strengthens the exploration step of the simulated annealing algorithm at the beginning and reinforces the exploitation step of the simulated annealing algorithm towards the end. Furthermore, we develop several dominance relations for the problem which are incorporated in the proposed SMI algorithm. We compare the performance of the proposed SMI algorithm with that of the best existing algorithm, SDE. The computational experiments indicate that the proposed SMI algorithm performs significantly better than the existing SDE algorithm. More specifically, under the same CPU time, the proposed SMI algorithm, on average, reduces the error of the best existing SDE algorithm over 90%, which indicates the superiority of the proposed SMI algorithm.

European Journal of Operational Research, 2018

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service... more This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Highlights  No-wait flowshop scheduling problem is addressed with respect to both makespan and total tardiness  An algorithm, combination of simulated annealing and insertion algorithm, is proposed  Five existing algorithms are adapted to the considered problem  The proposed algorithm reduces the error of the best existing algorithm (among the five algorithms adapted) about 60 %

Applied Mathematical Modelling, 2017

This paper addresses a manufacturing system consisting of a single machine. The problem is to min... more This paper addresses a manufacturing system consisting of a single machine. The problem is to minimize the number of tardy jobs where processing times are uncertain, which are within some intervals. Minimizing the number of tardy jobs is important as on-time shipments are vital for lowering cost and increasing customers' satisfaction for almost all manufacturing systems. The problem is addressed for such environments where the only known information is the lower and upper bounds for processing times of each job since the exact processing times may not be known until all jobs are processed. Therefore, the objective is to provide a solution that will perform well for any combination of feasible realizations of processing times. First, a dominance relation is established. Next, several versions of an algorithm, incorporating the dominance relation, are proposed. The computational analyses reveal that the error of one of the versions of the algorithm is at least 60% smaller than the errors of the other versions of the algorithm. Besides, the performance of this version is very close to the optimal solution, i.e., on average, 1.34% of the optimal solution.

Journal of Industrial and Production Engineering, 2016

The objective of minimizing the number of tardy jobs is important as it is directly related to th... more The objective of minimizing the number of tardy jobs is important as it is directly related to the percentage of on-time shipments, which is often used to rate managers’ performance in many manufacturing environments. To the best of our knowledge, the assembly flowshop scheduling problem with this objective has not been addressed so far, and thus is addressed in this paper. Given that the problem is NP-hard, different heuristics are proposed for the problem in this paper. The proposed heuristics are genetic algorithm (GA), improved genetic algorithm (IGA), simulated annealing algorithm with three different neighborhood structures (SA-1, SA-2, SA-3), Dhouib et al.’s simulated annealing algorithm (DSA), and an improved cloud theory-based simulated annealing algorithm (CSA). The heuristics are evaluated based on extensive computational experiments and all the heuristics were run for the same computational time for a fair comparison. The experiments reveal that the overall average errors of DSA, GA, IGA, CSA, SA-1, SA-2, SA-3 were 20.53, 13.49, 11.64, 3.27, 2.81, 1.92, and 0.56, respectively. Therefore, the proposed heuristic of SA-3 reduces the error of DSA, GA, IGA, CSA, SA-1, SA-2 by about 97, 96, 95, 83, 80, and 71%, respectively. All the results are statistically confirmed.

We first investigate a household's asset allocation, housing and consumption decisions with a... more We first investigate a household's asset allocation, housing and consumption decisions with a multi-period model under recursive utility function. We obtain both the analytical and numerical solutions for the optimal policies. We focus just on the period after retirement. In light of the empirical evidence that the elderly people do not down-size or liquidate their housing, we assume that the household buys a house initially and then keeps it until he dies. Our analysis indicates that the composition of the liquid portfolio between stocks and bonds do not depend on the elasticity of intertemporal substitution or the weight of housing in utility parameters. We find that optimal decisions of the household depend on intertemporal elasticity of substitution, risk aversion, weight of housing in utility and characteristics of the financial assets. Our calibrated model can generate empirically documented consumption patterns of the elderly homeowners. Second, we analyze and quantify th...

International Journal of Production Research, 2015

ABSTRACT A wide range of uncertainties exists in some real-world production environments which re... more ABSTRACT A wide range of uncertainties exists in some real-world production environments which result in uncertain setup and/or processing times. Factors such as crew skills, shortages in equipment and resource breakdowns can be the sources of these uncertainties. This study considers a two-machine production flowshop scheduling problem where both setup and processing times are treated as uncertain variables. The objective is to minimise makespan which is an effective way of resource utilisation. There exists a dominance relation in the literature for the two-machine flowshop scheduling problem with uncertain setup and processing times. However, the dominance relation has not been evaluated. In this study, we evaluate the existing dominance relation. Moreover, a new dominance relation is established and shown to be more effective than the existing one. Furthermore, twenty-five implementations of a polynomial time algorithm are developed. Extensive computational experiments are conducted to evaluate the performance of the implementations of the algorithm. The computational experiments indicate that the overall gap (error) of the best implementation of the algorithm is less than 0.3% when compared to the optimal solution. Moreover, the performance of this implementation of the algorithm is the best one when compared to the remaining implementations for all the considered experimental environments. Additionally, the performance of this implementation of the algorithm is shown to be insensitive to the uncertainty in setup times.

Computers & Operations Research, 2014

The single resource scheduling problem is commonly applicable in practice not only when there is ... more The single resource scheduling problem is commonly applicable in practice not only when there is a single resource but also in some multiple-resource production systems where only one of the resources is bottle neck. Thus, the single resource (machine) scheduling problem has been widely addressed in the scheduling literature. In this paper, the single machine scheduling problem with uncertain and interval processing times is addressed. The objective is to minimize mean weighted completion time. The problem has been addressed in the literature and efficient heuristics have been presented. In this paper, some new polynomial time heuristics, utilizing the bounds of processing times, are proposed. The proposed and existing heuristics are compared by extensive computational experiments. The conducted experiments include a generalized simulation environment and several additional representative distributions in addition to the restricted experiments used in the literature. The results indicate that the proposed heuristics perform significantly better than the existing heuristics. Specifically, the best performing proposed heuristic reduces the error of the best existing heuristic in the literature by more than 75% while the computational time of the best performing proposed heuristic is less than that of the best existing heuristic. Moreover, the absolute error of the best performing heuristic is only about 1% of the optimal solution. Having a very small absolute error along with a negligible computational time indicates the superiority of the proposed heuristics.

International Journal of Production Research, 2013

The link between makespan and the profitability and competitiveness of a firm is addressed first.... more The link between makespan and the profitability and competitiveness of a firm is addressed first. We then study the problem of minimising makespan in a two-machine flowshop with setup times. Jobs have random setup times that are bounded within certain intervals. The distributions of job setup times are not known. We propose a polynomial time algorithm that generalises Yoshida and Hitomi's algorithm. The algorithm uses a weighted average of lower and upper bounds for setup times. Different combinations of weights result in nine different versions of the algorithm. The computational results indicate that one of the versions, with equal weights given to the lower and upper bounds of setup times, performs much better than the others. Next, the performance of this best version is compared with that of the optimal solution, which is obtained by Yoshida and Hitomi's algorithm applied to the problem after setup times have been realised. Computational analysis shows that the overall average absolute error of the best algorithm is 0.03%, and this decreases in size as the number of jobs increases. The analysis also shows that the proposed best version yields robust results regardless of setup-time distributions and the range of setup times.

The International Journal of Advanced Manufacturing Technology, 2007

In this paper, we address the two-stage assembly flowshop scheduling problem with a weighted sum ... more In this paper, we address the two-stage assembly flowshop scheduling problem with a weighted sum of makespan and mean completion time criteria, known as bicriteria. Since the problem is NP-hard, we propose heuristics to solve the problem. Specifically, we propose three heuristics; simulated annealing (SA), ant colony optimization (ACO), and self-adaptive differential evolution (SDE). We have conducted computational experiments to compare the performance of the proposed heuristics. It is statistically shown that both SA and SDE perform better than ACO. Moreover, the experiments reveal that SA, in general, performs better than SDE, while SA consumes less CPU time than both SDE and ACO. Therefore, SA is shown to be the best heuristic for the problem.

Computers & Operations Research, 2009

We address the two-stage assembly scheduling problem where there are m machines at the first stag... more We address the two-stage assembly scheduling problem where there are m machines at the first stage and an assembly machine at the second stage. The objective is to schedule the available n jobs so that total completion time of all n jobs is minimized. Setup times are treated as separate from processing times. This problem is NP-hard, and therefore we present a dominance relation and propose three heuristics. The heuristics are evaluated based on randomly generated data. One of the proposed heuristics is known to be the best heuristic for the case of zero setup times while another heuristic is known to perform well for such problems. A new version of the latter heuristic, which utilizes the dominance relation, is proposed and shown to perform much better than the other two heuristics.