OPM 761 Research Seminar Production Management

Transcription

1 CHAIR OF PRODUCTION MANAGEMENT Prof. Dr. Raik Stolletz OPM 761 Research Seminar Production Management Spring Term 2016 General information The goal of this seminar is to introduce the participants to the conducting of scientific research. It thereby prepares the students for the writing of their Master thesis. The seminar is geared towards students intending to write their thesis at the Chair of Production Management. Participants will explore one of the topics listed below. They will review and critically assess the corresponding scientific literature and present their findings in a written report (15-20 pages) as well as in an in-class presentation (20 min + 20 min discussion). Each participant also acts as a discussant for one of the other presentations. The discussant is responsible for critically assessing the presented work and for opening the ensuing discussion. Applications will be accepted from Monday, November 16th, 2015 until Monday, November 30th, Admission to the seminar will be confirmed by on Friday, December 4th, 2015 and must be reconfirmed by the participant at the kick-off meeting. This meeting will be held on Wednesday, December 10th, 2015 in B4 (17:15-18:45) in room SO 322. During this meeting, general guidelines for conducting a scientific literature review will be discussed. The written reports have to be submitted by Wednesday, April 8th, 2016 as following: Two-fold hard copy version. Electronic version including a copy of the references cited in the report and auxiliary information (tables, data, programming code, etc.). The presentations will be held as a blocked session on Thursday, April 21st, 2015 (exact times and room to be announced). Attendance at the presentations is mandatory. The final grade for the seminar is composed of the following components: Written report (60%), presentation (30%), contribution to discussion (10%). The report and the presentations can be delivered either in English or in German, unless stated otherwise. There is a joint application process for all seminars offered by the chairs of the Area Operations Management. In the spring term 2016, this includes the following seminars: OPM 741: Applied Seminar Supply Chain Management Chair of Logistics and Supply Chain Management, Prof. Dr. Moritz Fleischmann (Topics labeled with L ), OPM 761: Research Seminar Production Management, (Chair of Production Management, Prof. Dr. Raik Stolletz (Topics labeled with P ),

2 OPM 781: Research Seminar Service Operations Management (Chair of Service Operations Management, Prof. Dr. Cornelia Schön (Topics labeled with S ), OPM 792: Applied Seminar Procurement Endowed Chair of Procurement, Prof. Dr. Christoph Bode (Topics labeled with B ). Detailed information on the seminar topics and the link to the online registration tool are available on the home pages of the respective chairs. In their application, students can indicate up to five preferred topics from all seminars. In addition, applicants for OPM 761 must send an with (1) CV, (2) official B.Sc. and M.Sc. grades overviews, and (3) list of courses in the Area Operations that you are currently enrolled in to Jannik Vogel (j.vogel@bwl.uni-mannheim.de), who can also be contacted for general questions concerning the seminar. Topics Catalog P1 Additive manufacturing: Production management challenges Objectives: The rise of additive manufacturing technologies, also known as 3d printing, allows for flexible production of highly customized products. The outcome of this seminar work shall be a structured description of how production management for additive manufacturing differs from the management of established production systems such as flow lines or job shops. The work is expected to include a description of the basic technological variants of additive manufacturing. A comprehensive review of the existing literature on production planning in the context of additive manufacturing should serve as the basis for own discussions on future challenges and arising decision problems in this field. Prerequisites: Basic knowledge of production and operations management (e.g., OPM 561). Basic Paper: Mellor et al. (2014) Abstract: As mass production has migrated to developing countries, European and US companies are forced to rapidly switch towards low volume production of more innovative, customised and sustainable products with high added value. To compete in this turbulent environment, manufacturers have sought new fabrication techniques to provide the necessary tools to support the need for increased flexibility and enable economic low volume production. One such emerging technique is Additive Manufacturing (AM). AM is a method of manufacture which involves the joining of materials, usually layer-upon-layer, to create objects from 3D model data. The benefits of this methodology include new design freedom, removal of tooling requirements, and economic low volumes. AM consists of various technologies to process versatile materials, and for many years its dominant application has been the manufacture of prototypes, or Rapid Prototyping. However, the recent growth in applications for direct part manufacture, or Rapid Manufacturing, has resulted in much research effort focusing on 2

3 development of new processes and materials. This study focuses on the implementation process of AM and is motivated by the lack of socio-technical studies in this area. It addresses the need for existing and potential future AM project managers to have an implementation framework to guide their efforts in adopting this new and potentially disruptive technology class to produce high value products and generate new business opportunities. Based on a review of prior works and through qualitative case study analysis, we construct and test a normative structural model of implementation factors related to AM technology, supply chain, organisation, operations and strategy. P2 Shared buffers in manufacturing systems Objectives: Flow lines typically consist of machines in series with buffers located in between. Introducing a buffer that is shared by multiple machines can lead to throughput gains or a reduction of the required total buffer capacity. The outcome of this seminar work shall be a comprehensive literature overview on performance evaluation approaches for manufacturing systems with shared buffers. Moreover, commonalities and differences compared to the use of shared buffers in IT systems have do be discussed. Based on the review a description of related decision problems and a derivation of future challenges are expected. Prerequisites: Knowledge of flow lines (e.g., OPM 561) and queueing models (e.g., OPM 661). Basic Papers: Zhou and Lian (2011) Abstract: In this paper, we consider a two-stage tandem network. The customers waiting in these two stages share one finite buffer. By constructing a Markov process, we derive the stationary probability distribution of the system and the sojourn time distribution. Given some constraints on the minimum loss probability and the maximum waiting time, we also derive the optimal buffer size and the shared-buffer size by minimizing the total buffer costs. Numerical results show that, by adopting the buffer-sharing policy, the customer acceptance fraction and the delivery reliability are more sensitive to buffer size comparing with the buffer-allocation policy. P3 Analysis of the modified offered load approach Objectives: For the analysis of time-dependent queueing systems appropriate performance evaluation methods are needed. One of these methods is the modified offered load (MOL) approach that uses an infinite server queueing model to approximate the performance of a time-dependent queueing system. In the infinite server model the number of busy servers is used to approximate the load of the queueing system. The performance measures are then calculated using stationary models. The student is expected to explain the basic concepts of the MOL-approach and to present existing literature on this method. The approach needs to be implemented in a programming language of his or her own choice. The student should then perform a numerical study that compares this method to a simulation. A simulation tool will be provided by the supervisor. 3

4 Prerequisites: Knowledge of a programming language (C++, Java,...) and basic knowledge of queueing theory (e.g., OPM 661). Basic Paper: Jennings et al. (1996) Abstract: We consider a multiserver service system with general nonstationary arrival and service-time processes in which s(t), the number of servers as a function of time, needs to be selected to meet projected loads. We try to choose s(t) so that the probability of a delay (before beginning service) hits or falls just below a target probability at all times. We develop an approximate procedure based on a time-dependent normal distribution, where the mean and variance are determined by infinite-server approximations. We demonstrate that this approximation is effective by making comparisons with the exact numerical solution of the Markovian M t /M/s t model. P4 A comparison of different overwork measures in queueing systems Objectives: Servers in real queueing systems slow down when they face overwork, a situation in which the system is highly loaded for an extended time period. To measure overwork in queueing systems, various approaches are described in the literature. The load that exceeds the average load over the last periods or the time since the last idle time of a minimum length can be taken as examples. The goal of this topic is to compare different overwork measures using a simulation study. The student should describe and explain the chosen measures from literature thoroughly. A simulation model for a queueing system with the different performance measures should be developed in a simulation software of his or her own choice. A comparison should show the differences of the chosen measures and answer the question, which measures are suited to operationalize overwork in queueing systems. Prerequisites: Basic knowledge in queueing theory and simulation (e.g., OPM 661, OPM 660). Basic Paper: Delasay et al. (2013) Abstract: Servers in many real queueing systems do not work at a constant speed. They adapt to the system state by speeding up when the system is highly loaded or slowing down when load has been high for an extended time period. Their speed can also be constrained by other factors, such as geography or a downstream blockage. We develop a state-dependent queueing model in which the service rate depends on the system "load" and "overwork". Overwork refers to a situation where the system has been under a heavy load for an extended time period. We quantify load as the number of users in the system and we operationalize overwork with a state variable that is incremented with each service completion in a high-load period and decremented with each service completion in a low-load period. Our model is a quasi-birth-and-death process with a special structure that we exploit to develop efficient and easy-to-implement algorithms to compute system performance measures. We use the analytical model and simulation to demonstrate how using models that ignore adaptive server behavior can result in inconsistencies between planned and realized performance and 4

5 can lead to suboptimal, unstable, or oscillatory staffing decisions. P5 Time-dependent performance approximation: Comparing the stationary backlog-carryover approach with the pointwise stationary fluid flow approximation Objectives: The reliable evaluation of the time-dependent performance of queueing systems is a crucial requirement in multiple fields of applications. This seminar focuses on two approximation approaches - (1) the stationary backlog-carryover approach and (2) the pointwise stationary fluid flow approximation - which both are based on steady-state queueing models. The student is expected to describe and compare the ideas of the two approaches and to analyze their approximation qualities in a numerical study for different parameter settings. A Java implementation of the two approaches as well as of a simulation model, serving as a benchmark, can be provided if required. Prerequisites: Basic knowledge in queueing theory (e.g., OPM 661); basic programming knowledge (e.g., Java) is an advantage. Basic Papers: (1) Stolletz (2008) and (2) Chen et al. (2013) Abstract: (1) This paper proposes a new approach for the time-dependent analysis of stochastic and non-stationary queueing systems. The analysis of a series of stationary queueing models leads to a new approximation of timedependent performance measures. Based on a stationary backlog-carryover (SBC) approximation of the time-dependent expected utilization, different approximations of the time-dependent expected queue length and the number of customers in the system are discussed. Limiting results are given for the case of constant rates. The accuracy of the SBC approach is shown for non-stationary M(t)/M(t)/c(t) queueing systems with timedependent and piecewise constant arrival rates. In numerical experiments we demonstrate the reliability of this approach and compare it with the (lagged) stationary independent period by period (SIPP) approach. In addition, the approximation is applied to temporarily overloaded systems that cannot be analyzed by the variants of the SIPP approach. (2) Long truck queue is a common problem at big marine container terminals, where the resources and equipment are usually scheduled to serve ships prior to trucks. To reduce truck queues, some container terminals adopt terminal appointment system (TAS) to manage truck arrivals. This paper addresses two implementation scenarios of TAS: static TAS (STAS) and dynamic TAS (DTAS). First, a non-stationary M(t)/E k /c(t) queueing model is used to analyse a terminal gate system, and solved with a new approximation approach. Then, genetic algorithm is applied to optimise the hourly quota of entry appointments in STAS for the derived queueing model. Lastly to relax the assumption of knowing the truckers preferred arrival pattern in STAS, we propose the concept of DTAS, which is much easier to apply and can assist individual trucker in making appointment by 5

6 providing real-time estimation of waiting time based on existing appointments. Our analysis reveals DTAS can significantly increase the system flexibility. P6 Optimizing truck arrival patterns for container terminals Objectives: Innovative demand management mechanisms have emerged in the past decade to improve the performance of truck handling operations at seaport container terminals. The underlying objective of these mechanisms is to shift truck arrivals from peak to off-peak periods. The student is expected to discuss the corresponding decision problem and to implement its optimization model in an optimization environment such as GAMS. Furthermore, the student should conduct a numerical study and particularly analyze the impact of the variability of the service process on the optimal solution. Prerequisites: Basic knowledge in queueing theory (e.g., OPM 661), knowledge of a modeling language (e.g., OPM 662). Basic Papers: Chen et al. (2011) Abstract: An analytical point-wise stationary approximation model is proposed to analyze time-dependent truck queuing processes with stochastic service time distributions at gates and yards of a port terminal. A convex nonlinear programming model is developed which minimizes the total truck turn time and discomfort due to shifted arrival times. A two-phase optimization approach is used to first compute a system-optimal truck arrival pattern, and then find a desirable pattern of time-varying tolls that leads to the optimal arrival pattern. Numerical experiments are conducted to test the computational efficiency and accuracy of the proposed optimization models. P7 Application of Clearing Functions in Production Planning Objectives: The relationship between the number of orders produced in a period and the load of the machine is non-linear. This is ignored in most of the planning models as they use linear programming. Clearing functions are a way to overcome this problem. They model the relationship between load and output by a piecewise-linear function. In the seminar, the concept of clearing functions should be explained. Applications in practice as well as advantages and disadvantages of this approach are to be discussed. Prerequisites: Basic knowledge in production management (e.g., OPM 561), basic knowledge of planning in stochastic manufacturing systems (e.g., OPM 661), and a general mathematical understanding. Basic Paper: Asmundsson et al. (2009). Abstract: A fundamental difficulty in developing effective production planning models has been accurately reflecting the non-linear dependency between workload and lead times. We develop a mathematical programming model for production planning in multiproduct, single stage systems that captures the non- 6

7 linear dependency between workload and lead times. We then use outer linearization of this nonlinear model to obtain a linear programming formulation and extend it to multistage systems. Extensive computational experiments validate the approach and compare its results to conventional models that assume workload-independent planning lead times. P8 Order release planning using iterative linear programming and simulation approaches Objectives: In most of the literature about scheduling, production capacity does not depend on the work in process. However, in practice it can be observed that workload and throughput are related. One way to model this relationship is to apply traditional scheduling approaches iteratively. The aim of this seminar work is to give an overview over the current state of the art in this topic. Here, different solution approaches, the underlying assumptions, and their advantages and disadvantages shall be discussed. Prerequisites: Basic knowledge in production management (e.g., OPM 561), basic knowledge of planning in stochastic manufacturing systems (e.g., OPM 661), and a general mathematical understanding. Basic Paper: Irdem et al. (2010). Abstract: Production planning models that aim at determining optimal release schedules for production facilities face a fundamental circularity. In order to match supply to demand in an optimal manner, they must recognize the cycle time that elapses between material being released into the plant and its emergence as finished product. However, it is well known from queuing models that the mean cycle time increases nonlinearly with resource utilization, which is determined by the release schedule. To address this circularity, a number of authors have suggested algorithms which iterate between a linear programming model that determines releases for a set of flow time estimates, and a simulation model that evaluates the production realized from that release schedule. We present computational experiments examining the behavior of two such algorithms. We find that the convergence behavior of one is significantly more consistent than that of the other, and explore insights that may lead to improved algorithms. P9 Centralized workforce planning in a health care environment Objectives: Hospital managers are faced with rising turnover and shortages of qualified nursing staff. At the same time they are under increased pressure to simultaneously increase patient care and satisfaction while reducing costs. One method to improve the service quality or reduce cost is to pool nurses from different departments and optimize their shift plans simultaneously. The goal of this seminar thesis is to implement an already existing nurse scheduling approach in GAMS (or any equivalent optimization software) and provide an alternative measure of service quality. A brief numerical analysis should highlight the main benefits of workforce 7

8 pooling. Prerequisites: Basic knowledge of planning in stochastic service systems (e.g., OPM 661), knowledge of a modeling language (e.g., OPM 662). Basic Paper: Wright and Mahar (2013). Abstract: Nursing managers are faced with rising turnover and shortages of qualified nursing staff. At the same time they are under increased pressure to simultaneously increase patient care and satisfaction while reducing costs. In this study, we examine the impact of centralizing scheduling decisions across departments in a hospital. By pooling nurses from multiple units and scheduling them in one model, improved costs and reduced overtime result. Reduced overtime improves schedules for nurses. Improved satisfaction levels can positively impact turnover rates among nurses. Our results show that by using a centralized model, nursing managers in hospitals can improve the desirability of nurse schedules by approximately 34% and reduce overtime by approximately 80% while simultaneously reducing costs by just under 11%. P10 Hierarchical optimization of raw sugar production Objectives: Producers in the agricultural industry rely on increasing productivity in order to stay competitive on the world market. In this seminar thesis, the production of raw sugar is investigated. In particular, a steady supply of sugar canes must be ensured to smoothen the production rate of the sugar mills and to minimize quality losses of the agricultural product. The goal of this seminar thesis is to implement an already existing hierarchical cultivation and harvesting planning approach in GAMS (or any equivalent optimization software) and provide an extensive numerical analysis on the impact of external factors, such as the sucrose content reduction rate. Prerequisites: Knowledge of a modeling language (e.g., OPM 662). Basic Paper: Grunow et al. (2007). Abstract: Based on a case study from Venezuela, the production of raw sugar is investigated. Ideally, sugar mills operate at a constant production rate. However, safety stocks of the raw material cannot be maintained as sugar cane quality deteriorates very rapidly. Sugar cane is therefore continuously sourced in diverse quantities and qualities from hundreds of geographically dispersed haciendas and supplied to the milling process. Furthermore, due to weather conditions changing throughout the year, tight time windows must be observed for harvesting. The approach presented in this paper aims at preserving a constant supply while minimizing the associated costs. The entire planning problem is structured in a hierarchical fashion: (1) cultivation of the haciendas, (2) harvesting, and (3) dispatching of the harvesting crews and equipment. The corresponding optimization models and solution procedures are introduced and applied to the case study problem. 8

9 References Asmundsson, J., R. L. Rardin, C. H. Turkseven, and R. Uzsoy (2009). Production planning with resources subject to congestion. Naval Research Logistics (56), Chen, G., K. Govindan, Z.-Z. Yang, T.-M. Choi, and L. Jiang (2013). Terminal appointment system design by non-stationary m (t)/e k/c (t) queueing model and genetic algorithm. International Journal of Production Economics 146(2), Chen, X., X. Zhou, and G. F. List (2011). Using time-varying tolls to optimize truck arrivals at ports. Transportation Research Part E: Logistics and Transportation Review 47(6), Delasay, M., A. Ingolfsson, and B. Kolfal (2013). Modeling Load and Overwork Effects in Queueing Systems with Adaptive Service Rates. Grunow, M., H.-O. Günther, and R. Westinner (2007). Supply optimization for the production of raw sugar. International Journal of Production Economics 110(1), Irdem, D., N. Kacar, and R. Uzsoy (2010). An Exploratory Analysis of Two Iterative Linear Programming & Simulation Approaches for Production Planning. IEEE Transactions on Semiconductor Manufacturing 23(3), Jennings, O. B., A. Mandelbaum, W. A. Massey, and W. Whitt (1996). Server Staffing to Meet Time-Varying Demand. Management Science 42(10), Mellor, S., L. Hao, and D. Zhang (2014). Additive manufacturing: A framework for implementation. International Journal of Production Economics 149, Stolletz, R. (2008). Approximation of the non-stationary m (t)/m (t)/c (t)-queue using stationary queueing models: The stationary backlog-carryover approach. European Journal of operational research 190(2), Wright, P. D. and S. Mahar (2013). Centralized nurse scheduling to simultaneously improve schedule cost and nurse satisfaction. Omega 41(6), Zhou, W. and Z. Lian (2011). A tandem network with a sharing buffer. Applied Mathematical Modelling 35(9),