Production planning for a ramp-up process with learning in production and growth in demand

This paper presents a production-planning model for a manufacturing process that undergoes a ramp-up period with learning in production and growth in demand. The labour production and demand functions assumed in this paper are validated using available empirical data. A mathematical programming model is developed with numerical examples presented. The results of the paper indicate that the total costs of production can be minimised if the facility produces without interruption during the ramp-up phase and if the production and demand rates are synchronised as much as possible. The latter can be achieved by producing with the lowest possible production rate and by frequently re-structuring the workforce assigned to the production line.     

Strategic ramp-up planning in automotive production networks

In many industries, the frequency of production ramp-ups has increased due to shorter product life cycles and increased product variety. Production often takes place at globally dispersed facilities for customers in multiple markets, for example, in the automotive industry. Therefore, ramp-up planning must be carried out for a network. The planning problem at hand is to simultaneously optimise the allocation of products of a common product platform to plants, the timing of ramp-ups and ramp-downs, ramp-up duration and shape of capacity and production and transportation volumes with respect to the net present value of the profit. We develop a hierarchical planning framework and place strategic ramp-up planning therein. We provide a mixed integer programming model for this planning problem. Following the example of a German car manufacturer in the premium segment, we conduct a numerical analysis to obtain insights into optimal ramp-up and ramp-down decisions. The results show that it could be beneficial to use both steep and flat ramp-up curves. The impact of flexibility in the choice of ramp-up slots and the number of ramp-up curves provided on the objective value is substantial. The strategic ramp-up planning model outperforms sequential planning approaches.     

Optimising lot sizing and order scheduling with non-linear production rates

This research focuses on developing an optimum production schedule in a process with a non-linear production rate. Non-linear production processes may exhibit an increasing production rate as the lot size increases, which results in increasing efficiency in per-unit production. The degree to which this learning is carried forward into the next lot varies by process. Sometimes the learning effect experiences a 100% carryover into the next lot, but other times some learning is forgotten and there is less than a 100% carryover. We consider processes in which the learning effect is completely forgotten from lot to lot. In practice non-linear processes are often treated as linear. That is, the production data are collected and aggregated over time and an average production rate is calculated which leads to inaccuracies in the production schedule. Here we use a discretised linear model to approximate the non-linear process. Production occurs in discrete time periods within which the amount produced is known. This enables a production schedule to be determined that minimises production and holding costs. A dynamic programming model that starts with the latest demand and progresses towards the earliest demand is used to solve the single-product single-machine problem. The model is tested using the production function from the PR#2 grinding process at CTS Reeves, a manufacturing firm in Carlisle, Pa. Solution times are determined for 50, 100, 200, 500, 1500, and 3000 periods.     

Throughput analysis of production systems: recent advances and future topics

Throughput analysis is important for the design, operation and management of production systems. A substantial amount of research has been devoted to developing analytical methods to estimate the throughput of production systems with unreliable machines and finite buffers. In this paper we summarise the recent studies in this area. In addition to the performance evaluation of serial lines, approximation methods for more complex systems, such as assembly/disassembly systems, parallel lines, split and merge, closed-loop systems, etc., are discussed. Moreover, we propose future research topics from the automotive manufacturing systems perspective.     

Governing the dynamics of multi-stage production systems subject to learning and forgetting effects: A simulation study

Managing production systems where production rates change over time due to learning and forgetting effects poses a major challenge to researchers and practitioners alike. This task becomes especially difficult if learning and forgetting effects interact across different stages in multi-stage production systems as rigid production management rules are unable to capture the dynamic character of constantly changing production rates. In a comprehensive simulation study, this paper first investigates to which extent typical key performance indicators (KPIs), such as the number of setups, in-process inventory, or cycle time, are affected by learning and forgetting effects in serial multi-stage production systems. The paper then analyses which parameters of such production systems are the main drivers of these KPIs when learning and forgetting occur. Lastly, it evaluates how flexible production control based on Goldratt’s Optimised Production Technology can maximise the benefits learning offers in such systems. The results of the paper indicate that learning and forgetting only have a minor influence on the number of setups in serial multi-stage production systems. The influence of learning and forgetting on in-process inventory and cycle time, in contrast, is significant, but ambiguous in case of in-process inventory. The proposed buffer management rules are shown to effectively counteract this ambiguity.     

On the stability and bullwhip effect of a production and inventory control system

This paper focuses on the discrete-time automatic pipeline, inventory and order-based production control system (APIOBPCS), a well-established production and inventory control model. The feedback mechanism within the replenishment rule enables the model to mitigate the bullwhip effect, but introduces a stability problem. In this research, a comprehensive stability analysis is conducted for arbitrary lead times using difference equation theory. On the basis of stability, a state space approach is advocated to analyse the impact of replenishment parameters, demand processes, and lead times on the robustness of the bullwhip effect. The stability results demonstrate that the production control system can easily be destabilised without incorporating the work-in-progress (WIP) feedback loop. Furthermore, it reveals that the stability problem for long lead times can be simplified with the stability condition independent of the lead time. The results obtained in this study provide useful guidelines for the selection of replenishment parameters to guarantee stability and mitigate the bullwhip effect.     

Selection of a pull production control system in multi-stage production processes

In this paper, we conduct an analytical comparison of three pull production control systems: Kanban, CONWIP and Base-stock in multi-stage production processes. First, we compare the three control systems in a multi-stage serial production process. Then, we compare them in multi-stage assembly production processes, and present guidelines that allow us to select the best system. As a result, we show which structural parameters decide the superiority of one control scheme to the others, and how they are related. A key for superiority is a configuration of parameters, such as processing times and number of cards employed in the system. We show that there is no general superiority amongst the analysed concepts. Finally, we verify the effect of variability on the system performance, and generalise the analytical results of deterministic cases by conducting numerical experiments.     

Approximate analysis of serial flow lines with multiple parallel-machine stations

This paper studies serial flow lines, in which each station consists of multiple identical reliable parallel machines. The parallel machines of different work stations are not necessarily identical, viz., station processing times are assumed to be exponentially distributed with non-identical mean service rates. Initially, a model consisting of two stations with multiple parallel machines and an intermediate buffer is solved analytically, by developing a recursive algorithm that generates the transition matrix for any value of the intermediate buffer capacity. This model is used as a decomposition block for solving larger lines. More specifically, the decomposition block is solved via exact Markovian analysis and then the decomposition equations and an algorithm that simultaneously solves them are derived in order to evaluate the performance measures of large production systems with multiple parallel-machine stations. Numerical results are provided for large production lines with up to 1000 workstations. These results are compared against simulation and the average percentage error is found to be very small.     

Optimising lot sizing with nonlinear production rates in a multi-product multi-machine environment

In a variety of discrete manufacturing environments, it is common to experience a nonlinear production rate. In particular, our interest is in the case of an increasing production rate, where learning creates efficiencies. This leads to greater output per unit time as the process continues. However, the advantages of an increasing production rate may be offset by other factors. For examples, JIT policies typically lead to smaller lot sizes, where the value of an increasing production rate is largely lost. We develop a general model that balances the impact of various competing effects. Our research focuses on determining lot sizes that satisfy demand requirements while minimising production and holding costs. We extend our prior work by developing a multi-product, multi-machine method for modelling and solving this class of production problems. The solution method is demonstrated using the production function from the PR#2 grinding process for a production plant in Carlisle, PA. The solution heuristic provides solution times that are on average only 0.22 to 0.55% above optimum as the solution parameters are varied and the ratio of heuristic solution times to optimal solution times varies from 18.16 to 14.15%.     

Modelling ramp-up curves to reflect learning: improving capacity planning in secondary pharmaceutical production

The experience gained during production ramp-up leads to an increase of the effective production capacity over time. However, full utilisation of production capacity is not always possible during ramp-up. In such cases, the experience gained and hence the available effective capacity are overestimated. We develop a new method, which captures ramp-up as a function of the cumulative production volume to better reflect the experience gained while producing the new product. The use of the more accurate and computationally effective approach is demonstrated for the case of secondary pharmaceutical production. Due to its regulatory framework, this industry cannot fully exploit available capacities during ramp-up. We develop a capacity planning model for a new pharmaceutical drug, which determines the number and location of new production lines and the build-up of inventory such that product availability at market launch is ensured. Our MILP model is applied to a real industry case study using three empirically observed ramp-up curves to demonstrate its value as decision support tool. We demonstrate the superiority of our volume-dependent method over the traditional time-dependent ramp-up functions and derive managerial insights into the selection of ramp-up function and the value of shortening ramp-ups.     

Extending assembly line balancing problem by incorporating learning effects

In the ramp-up phase, or time to volume of new products, pronounced learning effects are observed. They are present especially on assembly lines producing mass-goods because of a high number of repetitions of the tasks. Shortening the ramp-up phase and reaching the steady-state production as soon as possible generates main advantages for firms that introduce new products. Moreover, a careful planning of the ramp-up stage is getting even more important in view of shorter product life cycles and a growing importance of the ‘time to payback’ financial indicators. Former studies on incorporation of learning effects into assembly line balancing have limited applicability, because they rely on unrealistic assumptions. We model learning effects, based on general and realistic assumptions, as an extension of the Simple Assembly Line Balancing Problem. We propose exact and heuristic solution procedures and perform extensive computational tests. We found that for instances similar to the problems, which arise in firms, the duration of the learning stage can be reduced by up to 10% if our specialised methods are applied.     

Delayed reconfigurable manufacturing system

Reconfigurable manufacturing systems (RMS) is a new manufacturing paradigm aiming at providing exactly functionality and capacity needed and exactly when needed. Reconfiguration is the main method to achieve this goal. But, the reconfiguration is an interruption to production activities causing production loss and system ramp-up problem and the ‘exact functionality’ may increase the reconfiguration efforts and aggravate the production loss and the ramp-up time. Therefore, a special RMS – delayed reconfigurable manufacturing system (D-RMS) is proposed to promote the practicality of RMS. Starting from the RMS built around part family with the characteristic of delayed differentiation, whose reconfiguration activities mainly occur in the latter stages of manufacturing system and the former stages have the potential to maintain partial production activities to reduce production loss during reconfiguration. Inspired from this, the basic structure of RMS is divided into two subsystems, subsystem 1 is capable of maintain partial production with a certain more functionality than needed, subsystem 2 reconfigure to provide exactly functionality and capacity of a specific part exactly when needed. And then, the benefits of D-RMS are analysed from inventory and ramp-up time aspects. Finally, a case study is presented to show the implementation process of D-RMS and validates the practicability of D-RMS.     

Lead time prediction in unbalanced production systems

Whether job due dates are set internally or externally, it is critical for the shop floor controller to be able to accurately predict job lead times. Previous research has shown that utilizing information on the congestion levels along a job's route is more valuable than overall shop congestion levels when predicting job lead times. While this information is easily attained in a simulation model, in industrial applications the task may be considerably more difficult, especially when lot splitting is used to accelerate material flow. We examine the effectiveness of three lead time estimation procedures which utilize different shop information in bottleneck-constrained production systems where lot splitting is practiced under a variety of experimental conditions. The results indicate that accurate lead time estimates can be obtained using information pertaining solely to the bottleneck work centre when the bottleneck is at an entry work centre. This offers operations managers a substantial ease in implementation over previously reported methods. The results also show that the operations managers interested in accurately estimating job lead times are well advised to take advantage of the excess capacity at non-bottleneck work centres by performing additional setups, and take measures that reduce bottleneck shiftiness.     

Setup tasks scheduling during production resettings

Production waste reduction during production resettings is critical in the industry as it is the key point of flexibility increase. One way to improve production waste is to assign setup tasks efficiently to operators. This article describes a possible model of this problem adapted to a real-life application. This model is based on an unrelated parallel machines problem that takes into consideration the skills of the operators and the production line structure. The nature of the data and some industrial constraints help us simplifying this problem into an assignment problem. As often in the industry, the objective is to maximise production. Because most production lines are complex series parallel lines, it is often possible to maintain production even if not all machines are running. This particularity makes the criterion hard to express as it depends on the line structure. In this article we describe three heuristics to solve this problem: a hill climbing algorithm, a genetic algorithm and a memetic algorithm that combines the advantages of the two previous algorithms. The neighbourhood used for these algorithms is based on multiple exchanges of tasks between operators.     

The equivalence of duplicate automated serial workstations and two-workstation tandem systems and its use in serial production line analysis

In automated serial production systems it is sometimes necessary to maintain two identical workstations in series, with jobs receiving processing on only one of these two workstations. This is required because a single workstation may not have sufficient capacity to support the target throughput rate and due to cost, space, and sequencing reasons, the workstations are not arranged in parallel. We show how these duplicate automated serial workstations can be approximated by a two-workstation tandem system with zero buffer. The parameters of the tandem system are simple functions of the duplicate workstation parameters. In isolation we show that the tandem system is equivalent to the duplicate workstation system. When part of a larger serial production system, the equivalence no longer holds however, simulation testing shows that a two-workstation tandem system approximation to the duplicate workstation system is very accurate. This extends the analysis capabilities of analytical tools and simulations that are tailored for serial production systems, but do not explicitly model duplicate workstations.     

Dynamic adjustment of dispatching rule parameters in flow shops with sequence-dependent set-up times

Decentralised scheduling with dispatching rules is applied in many fields of production and logistics, especially in highly complex manufacturing systems. Since dispatching rules are restricted to their local information horizon, there is no rule that outperforms other rules across various objectives, scenarios and system conditions. In this paper, we present an approach to dynamically adjust the parameters of a dispatching rule depending on the current system conditions. The influence of different parameter settings of the chosen rule on the system performance is estimated by a machine learning method, whose learning data is generated by preliminary simulation runs. Using a dynamic flow shop scenario with sequence-dependent set-up times, we demonstrate that our approach is capable of significantly reducing the mean tardiness of jobs.     

Queueing network modelling and lead time compression of pharmaceutical drug development

Designing an optimized pharmaceutical drug development process is an important problem in itself and is of significant practical and research interest. Drug development lead time is a critical performance metric for a pharmaceutical company. In this paper, we develop a multiclass queueing network model to capture the project dynamics in drug development organizations that involve multiple, concurrent projects with contention for human/technical resources. We explore how drug development lead times can be reduced using efficient scheduling and critical mass-based resource management. The model captures important facets of any typical drug development organization, such as concurrent execution of multiple projects, contention for resources, feedback and reworking of project tasks, variability of new project initiations and task execution times, and certain scheduling issues. First, we show, using a class of fluctuation smoothing scheduling policies, that development lead times can be compressed impressively, without having to commit additional resources. Next, we show that critical mass-based project teams can compress lead times further. The model presented, though stylized, is sufficiently generic and conceptual, and will be of much value in new drug development project planning and management.     

Performance evaluation of single and multi-class production systems using an approximating queuing network

Performance evaluation, and in particular cycle time estimation, is critical to optimise production plans in high-tech manufacturing industries. This paper develops a new aggregation model based on queuing network, so-called queue-based aggregation (QAG) model, to estimate the cycle time in a production system. Multiple workstations in serial and job-shop configurations are aggregated into a single-step workstation. The parameters of the aggregated workstation are approximated based on the parameters of the original workstations. Numerical experiments indicate that the proposed QAG model is computationally efficient and yields fairly accurate results when compared to other aggregation approaches in the literature.     

Lot-sizing for a single-stage single-product production system with rework of perishable production defectives

We consider a single-stage single-product production system. Produced units may be non-defective, reworkable defective, or non-reworkable defective. The system switches between production and rework. After producing a fixed number (N) of units, all reworkable defective units are reworked. Reworkable defectives are perishable or can become technologically obsolete. We assume that the rework time and the rework cost increase linearly with the time that a unit is held in stock. Therefore, N should not be too large. On the other hand, N should not be too small either, since there are set-up times and costs associated with switching between production and rework. For a given N, we derive an explicit expression for the average profit (sales revenue minus costs). Using this expression, the optimal value for N can be determined numerically. Moreover, it is easy to perform a sensitivity analysis, as we illustrate. RID="*" ID="*"The research of Dr. Ruud H. Teunter has been made possible by a fellowship of the Royal Netherlands Academy of Arts and Sciences. The research presented in this paper is part of the research on re-use in the context of the EU sponsored TMR project REVersed LOGistics (ERB 4061 PL 97-5650) in which take part the Otto-von-Guericke Universitaet Magdeburg (D), the Erasmus University Rotterdam (NL), the Eindhoven University of Technology (NL), INSEAD (F), the Aristoteles University of Thessaloniki (GR), and the University of Piraeus (GR). We thank the anonymous referees for their many helpful comments. Correspondence to: R. H. Teunter     

Characteristics affecting management of design information in the production system design process

Although it has been argued that the design of production systems is crucial, there is a general lack of empirical studies analysing and identifying resources and capabilities required for an efficient production system design process. One of these resources is the critical role attributed to design information and one such capability is how the design information is managed. To address this research gap, this paper reports the results from two in-depth case studies in the automotive industry focusing on the management of design information in the production system design process. Our results show that design information management needs to be understood as a multidimensional concept having three dimensions: acquiring, sharing and using design information. By focusing on the three dimensions, six characteristics affecting the management of design information when designing the production system are identified. The characteristics are information type, source of information, communication medium, formalisation, information quality and pragmatic information.