Manufacturing Line Reconfiguration Guidelines#

1. Introduction#

1.1 Manufacturing Line definition#

A Manufacturing Line is a set of sequential operations established in a factory where parts are processed and components are assembled to make a finished article or where materials are put through a refining process to produce an end-product.

1.2 Reconfigurable Manufacturing Systems (RMS) and Reconfigurable Manufacturing Line (RML) definitions#

Reconfiguration concept

Reconfiguration aims at modifying the manufacturing process to rapidly and cost-efficiently (i) adapt the production to market changes, (ii) increase production flexibility for mass customisation, or (iii) react upon unpredicted events such as machine faults or quality degradation.

Process reconfiguration concept

Process reconfiguration relies on reconfigurable machines and controllers, and methodologies for the systematic design of new process configuration.

Reconfigurable Manufacturing Systems (RMS) concept

Configurability is mainly addressed in the work on reconfigurable manufacturing systems (RMS). Indeed, RMS are a new class of systems characterised by an adjustable structure. They are designed to allow simple changes in their physical structure in order to respond to fluctuation in the latter in terms of volume and product type.

In practice, the question arises as to how to adapt systems that are not designed to be reconfigurable to a new market reality. The latter requires too much investment and often a reconfiguration of the existing system is preferable. The problem of reintegrating the costs of reconfiguring a production system is therefore frequent in the industry. However, this problem is different from that of the initial configuration because of the existing organisation.

The objective of the reconfiguration is to reinstate the cost of the final line, i.e. to reinstate the cost of the new equipment parts to be installed on the line in order to meet the new production process.

Currently, the frequency of new products is increased and the time for designing, building, and ramping-up volumes has been reduced, which means that the manufacturing systems must be built for rapid change in accordance with the market.

For that reason, the reconfigurable manufacturing system (RMS), which can be repeatedly changed in capacity and functionality in a cost-efficient way, has been widely labelled the manufacturing system of the future.

Some insist that an RMS is an intermediate paradigm between Mass Production and Flexible Manufacturing System (FMS), some argue that an RMS is an advanced paradigm whose flexibility must be higher than that of an FMS, and the others think it is not very meaningful to distinguish RMSs from FMSs.

An RMS has an ability to reconfigure hardware and control resources at all of the functional and organizational levels, in order to quickly adjust production capacity and functionality in response to sudden changes in market or in regulatory requirements.

Reconfigurable of transfer or assembly lines

A transfer line must be reconfigured if a new product is to be manufactured by an existing line or if there have been changes in the product characteristics. In such cases, new operations have to be integrated with existing equipment while some previously performed operations are removed.

The problem of reconfiguration arises more often than the problem of designing a new line. As these lines are often manual, the costs of reallocating staff and training them must be carefully considered.

2. Manufacturing Line Reconfiguration Problems and Solutions#

2.1 Assembly Line Balancing Problem (ALBP)#

Problem description

Assembly lines are flow-oriented production systems which are still typical in the industrial production of high quantity standardized commodities and even gain importance in low volume production of customized products.

An assembly line consists of (work) stations \(k\) = 1,…,\(m\) arranged along a conveyor belt or a similar mechanical material handling equipment. The workpieces (jobs) are consecutively launched down the line and are moved from station to station. At each station, certain operations are repeatedly performed regarding the cycle time (maximum or average time available for each work cycle).

The decision problem of optimally partitioning (balancing) the assembly work among the stations with respect to some objective is known as the assembly Line Balancing Problem (ALBP). ALBP consists of distributing the total workload for manufacturing any unit of the product to be assembled among the work stations along the line. Any type of ALBP consists in finding a feasible line balance, i.e., an assignment of each task to a station such that the precedence constraints and further restrictions are fulfilled.

The set \(S_k\) of tasks assigned to a station \(k\) = 1,…,\(m\) constitutes its station load, the cumulated task time is called station time. When a fixed common cycle time \(c\) is given, a line balance is feasible only if the station time of neither station exceeds \(c\). In the case of \(t(S_k)\) < \(c\), the station \(k\) has an idle time of \(c\) - \(t(S_k)\) time units in each cycle. The balancing problem is connected with the decision problem of selecting processing or equipment alternatives.

Objective function

The line installation and operating costs as well as the profits mainly depend on the cycle time and the number of stations.

Cost oriented

The objective is to minimize the aggregate wage rate over all stations, while the number of stations is a variable. Production costs per product unit are obtained by multiplying that rate with the given cycle time. The considered objective is equivalent to minimizing the number of stations, if all tasks have the same wage rate.
Station related costs of capital, i.e., each station is assumed to require a constant pre-specified investment.

Profit oriented - Maximizing the total contribution margin per shift. It considers operating expenses, idle time costs, material costs, wages, equipment costs, as well as constant selling prices, buffers where inventory-related cost components are relevant.

Data input

Costs, available machines.

Data output

Work stations and workers assigned to each work station.

Algorithm

Station-oriented construction scheme Task-oriented construction scheme

2.2 Transfer Line Reconfiguration Problem (TLRP)#

Problem description

Tailoring the manufacturing line configuration to meet the production demands. Assign a new set of operations to workstations while reusing as much as possible the existing equipment. The objective of the reconfiguration is to minimize the investment cost necessary to upgrade an existing transfer line. It can be reached by reusing as much as possible the existing equipment.

Despite the fact that reconfigurable manufacturing systems are widely discussed in the academic literature, in industrial practice, users often deal with traditional lines. In such a case, when the product or production volume is modified, the aim of the decision-makers is to readjust the corresponding transfer line that has not been designed for reconfigurability, to the new product/production requirements.

The problem deals with the partial redesign of a transfer line: some machines can be upgraded with new machining modules (multi-spindle heads) which are designed for the modified product; some other modules can be removed. Such decisions have to be made by taking into account the technological and compatibility constraints between operations and equipment.

Objective function

Reduce the investment cost for new equipment by reusing optimally the existing facilities.

Data input

Costs, existing facilities, operations.

Data output

The objective is to assign a new set of operations to workstations and spindle heads while reusing as much as possible the existing equipment.

Algorithm

Mixed-Integer Linear Programming Model (MILP).

2.3 Balancing with Equipment Selection Problem (BESP)#

Problem description

Each device is defined by the set of operations it can perform and its cost. In the general case, several devices are capable of performing the same operation, but only one will be selected at the configuration stage.

Objective function

The optimization problem consists of determining the number of stations and selecting a subset of boxes from a given set to assign to the stations. The criterion to optimise is the line cost. Optimal line design seeks to identify the best configuration of resources and allocation of tasks to satisfy criteria such as maximum throughput or minimum cost. The mathematical model seeks to minimize the ratio of the total investment in machines and buffers to the total throughput of all parts.

Data input

The costs of the stations and of each box are known, the constraints of compatibility between the devices, the constraints of precedence between the operations and the constraint of the cycle time are considered, available machine types, task-precedence-graphs, machine cost, mean-time-between failure (MTBF) and mean-time-to-repair (MTTR) of each machine, throughput demands, maximal allowable investment on the multiple product line (MPL), machine quantity limit, buffer cost per unit size. Also stage key characteristic, in which each manufacturing stage usually has limited capabilities, which are reflected by a group of key characteristics of the stage; when a set of tasks are assigned to a stage, the necessary capabilities must fall in the key characteristics of the stage.

Data output

Tasks assigned to each stage without violating precedence constraints. Machine type used. Machines are required in each stage. Buffer size between each stage. Estimated throughput for each part produced.

Algorithm

MILP model. Genetic Algorithm formulation to capture in string form the configuration and task allocation for a multiple parts line (MPL). Minimal ratio of cost to throughput is used as the criterion for the fitness function. An analytical throughput analysis engine is called during the evaluation of each solution to size and locate buffers, and to consider the effects of machine breakdown.

2.4 Balancing with Choice of Machine Problem (BCMP)#

Problem description

This problem has been formulated for the parallel machining lines. The workstations are arranged in sequence, but each of them can be composed of several identical machines. The same operations are duplicated on all the machines of the same station. This means that several parts are processed at the same time on the same workstation. The parallel machines have a local cycle time equal to the number of machines multiplied by the target cycle time of the line. Thus at each cycle time, only one piece leaves the station and only one piece returns to the station. This configuration allows operations with an operating time that exceeds the line cycle time to be carried out. The machines that can be installed in parallel do not use regular spindle housings and their design is not customised. Instead, each machine is characterised by its type which determines its technical capabilities (e.g. the number of machining axes). A cost is associated with each machine type according to the functionalities provided. To install the tool required for each operation, a tool change time is necessary. Thus the time taken to perform two operations is not equal to the time taken to perform these operations but depends on the sequence in which they are performed. This is because the tool movement, tool change and workpiece rotation times are not insignificant. The position of the workpiece on the worktable at each station is another decision variable. It defines the accessible surfaces for machining.

Objective function

The optimization problem consists in dividing the operations on the working stations and determining the type and the number of parallel machines installed on each station as well as the positioning of the piece on each station while respecting the given constraints and reducing the total cost of the line.

Data input

Machine lines, workstations, machines in workstations, machine operations, product time processing, machine local cycle time, machine technical capabilities, machine cost, tool change time.

Data output

Operations sequence. The position of the workpiece on the worktable at each station is another decision variable. It defines the accessible surfaces for machining.

Algorithm

Mixed-Integer Linear Programming Model (MILP). Branch and Bound. Dynamic programming. Heuristic: ranked positional weight, computer method for sequencing operations for assembly lines. Meta-heuristic: Evolutionary algorithms, ant colony, synergetic annealing and taboo method. Multi-objective optimisation: a priori (weighted sum method, goal programming), a posteriori (non-dominated sorting genetic algorithm, Total order preference by similarity to the ideal solution TOPSIS) and progressive method (stem, MIN-MAX, STEUER’S METHOD).

2.5 Balancing with special machine design: Moving Table Machine Problem (BMTMP)#

Problem description

Special machines such as mobile table machines and rotary table machines. These machines can be used separately or grouped together in a clocked machining line where the movement of the workpiece is organised from one machine to another. When designing this type of line, the balancing problem concerns not only the number of machines and the type of each special machine used, but also its detailed configuration in terms of working position and multi-spindle housings.

On a moving table machine, each workpiece is machined by passing through m successive working positions. The advancement of the workpiece from one position to the next is achieved by moving the moving table on which the workpiece is fixed. Thus, only one workpiece is present on the machine throughout the cycle time. Therefore, only the position where the workpiece is located can be active at any one time. The multi-spindle units of the active position are executed simultaneously.

At the end of this execution, the workpiece is advanced either to the next position or if finished to the first position for unloading. A new cycle starts for a new piece and so on. In order not to disturb the movement of the moving table, only two machining units can be installed at each working position: one on the left of the table and one on the right. A third machining unit can be added in front of the table at the last working position only.

In this way, in each working position (except the last one) it is possible to install either two horizontal boxes or one horizontal and one vertical box, and in the last position it is possible to have at most two horizontal and one vertical box or three horizontal boxes. As for the number of working positions, this type of machine rarely has more than 3, so that the space occupied by the machine and the working time are acceptable.

Objective function

When designing a moving table machine, the optimisation objective may be to reduce either the cost of the machine or the cycle time, as the configuration with only one active workstation makes machining the workpiece slower.

Data input

Moving table machines, cost of the machine, number of working positions.

Data output

Number of working positions and multi-spindle units to be installed on each working position to carry out all the machining operations

Algorithm

Reduction to a constrained shortest path problem.

Digraph generation and searching for a shortest path.

2.6 Balancing with special machine design: Rotary Table Machine Problem (BRTMP)#

Problem description

Special machines such as mobile table machines and rotary table machines. These machines can be used separately or grouped together in a clocked machining line where the movement of the workpiece is organised from one machine to another. When designing this type of line, the balancing problem concerns not only the number of machines and the type of each special machine used but also its detailed configuration in terms of working position and multi-spindle housings.

A machine with a rotating rotary table also has several working positions, but it can process several workpieces simultaneously: one per position. A working position corresponds to a position of the rotary table. Position 0 is always used for loading/unloading operations, and no machining operations are assigned to it. On the other hand, at any different work position, we can assign a maximum of two machining units, one horizontal and one vertical, which are active simultaneously, to process two different sides of the workpiece.

All the units in all the working positions are switched on in ternary mode: they carry out the machining operations on the workpiece in the working position where they are installed. Even though the machining times may vary from one work position to another, the movement of all the pieces on the machine is carried out in ternary regeneration by rotating the table in one step. Thus, each workpiece moves to the next working position.

Objective function

The problem of optimising the configuration of rotary table machines consists in reducing the total cost of the machine by determining the number of working positions and multi-spindle units to be installed on each working position to carry out all the machining operations and respecting the technological and technical constraints given.

Data input

Rotary table machines, cost of the machine, number of working positions and multi-spindle units.

Data output

Number of working positions and multi-spindle units to be installed on each working position to carry out all the machining operations

Algorithm

Reduction to a constrained shortest path problem.

Digraph generation and searching for a shortest path.

2.7 Inbound Reconfigurable Transportation Systems (IRTS)#

Problem description

Inbound transportation systems are one of the major application areas
of the reconfigurability concepts, designed as multiple independent
modules for the implementation of alternative inbound logistic system
configurations. The problem consists of synchronising the use of the
RTS’s set of mechatronic modules in order to transport all the parts
to be processed towards their respective destinations in the
production plant, that is the online part-routing problem (OPRP).
According to the OPRP’s definition, parts are transported by a set of
available pallets and are supposed to enter the system at a given
rate. Each part is characterised by an entry point, a destination and
a work plan, which is assigned depending on the part’s type. Workplans
define the number, type and ordering of the working stations each part
requires to be processed on, ultimately defining the part’s high-level
routing within the RTS. A solution for the OPRP must ensure that all
the parts are properly worked (according to their workplan) as well as
guarantee the routing consistency so that all paths satisfy the
temporal constraints related to the mechanical properties of the
transportation modules as well as those of their subcomponents. Also,
the routings must be collision-free so as to guarantee the correct
execution of all pallets’ motions, avoiding all physical clashes.
Finally, routings must be efficient thus being as short (hence as
fast) as possible.
This problem can adopt the classical approach or present certain
adaptations such as: The total number of tasks that have to be
assigned to the agents for each incoming pallet is not known in
advance, or not all the agents participate to a given auction, but
only those that may potentially acquire the task at that particular
auction round. It remains however true that all the agents may
potentially participate in the auctions while the pallet path is being
constructed.

Objective function

To minimize task allocation costs for a transportation module.

Data input

Working stations, transportation modules, parts to be processed, pallets, costs, connectivity function for the transportation modules, set-up times, distances that each pallet must travel to reach its destination

Data output

Routing decisions: the best route for each transportation module.

Algorithm

Agent-based algorithm that combines global and local optimisation criteria.

2.8 Static Facility Layout Problem (SFLP)#

Problem description

Objective function

To minimize the amount/cost of material flow from the drop-off point of the facility to the pick-up point of the facility (being the pick-up point the location from which parts enter facilities and the drop-off point the location to which parts leave facilities).

Data input

Plant dimensions, parts routes and quantities (flows), distance traveled by each part from the drop-off of facility to the pick-up of facility, area constraints on the plant, area allocated to each machine on the floor, possible clearance between facilities, (facilities must not overlap).

Data output

The layout of each facility.

Algorithm

Mixed Integer Programming Problems

2.9 Dynamic Facility layout problem (DFLP)#

Problem description

The placement of the facilities in the plant area, often referred to as ‘‘facility layout problem’’, is known to have a significant impact on manufacturing costs, work in process, lead times and productivity. Although most layout problems are implicitly considered static, there is a growing stream of dynamic layout problems. Dynamic layout problems take into account possible changes in the material handling flow over multiple periods. In this respect, the planning horizon is generally divided into periods that may be defined in weeks, months, or years. For each period, the estimated flow data remains constant. A layout plan for the dynamic layout problem consists of a series of layouts, each layout is associated with a period. This type of problem is frequently addressed from a discrete formulation, such is, the plant site is divided into rectangular blocks with the same area and shape and each block is assigned to a facility (if facilities have inequal areas, they can occupy different blocks).

Objective function

To determine a layout for each period in the planning horizon, while minimizing the sum of the material handling costs, for all periods, and the sum of the rearrangement costs between time periods. Rearrangement costs have to be considered when facilities need to be moved from one location to another. These models can also look minimize traffic congestion.

Data input

Plant dimensions, number of facilities, parts routes and quantities (flows), distance travelled by each part from the drop-off of the facility to the pick-up of the facility, area constraints on the plant, and distance from one location to another.

Data output

The layout of each facility.

Algorithm

Mixed Integer Programming Problems.

2.10 Multi-Facility Layout Problem (MFLP)#

Problem description

This problem involves the physical organization of departments inside several facilities, considering two different concerns: the location of departments within a group of facilities, and the location of departments inside each facility itself.

Objective function

For this problem, it is used a dynamic multi-objective approach. The objectives of the model are: the minimization of costs (material handling inside facilities and between facilities, and re-layout); the maximization of adjacency between departments; and the minimization of the “unsuitability” of department positions and locations, where the unsuitability measure is an objective proposed to measure the fitness between the characteristics of the existing locations and the requirements of the departments.

Data input

Total material handling costs (internal and external), the reconfiguration costs, the adjacency between departments (due to the importance of locating high-affinity departments at the same facility), the unsuitability between departments and locations, flow (product quantity) to move between departments, transport cost of a unit of material per unit of distance (inside facilities and between facilities), material handling costs, transportation capacity between facilities, and weights for the multi-objective function. On the other hand, the constraints in the model are those usually considered for the facility layout problem: no department overlapping; each department is assigned only to one position; and the department size fits into the location area. In this case, a new constraint has been considered to bound the transportation capacity between facilities in each planning period.

Data output

Location areas for the different facilities in each planning period.

Algorithm

Quadratic Programming Problem with multiple objectives and unequal areas, allowing layout reconfigurations in each planning period.

2.11 Reconfigurable Manufacturing Systems for Scalability Planning (RMS-SP)#

Problem description

Scalability is a key characteristic of reconfigurable manufacturing systems, which allows system throughput capacity to be rapidly and cost-effectively adjusted to abrupt changes in market demand. Scalability planning methodology for reconfigurable manufacturing systems that can incrementally scale the system capacity by reconfiguring an existing system. Adding or removing machines to match the new throughput requirements and concurrently rebalancing the system for each configuration, accomplishes the system reconfiguration.

Objective function

The objective of scalability planning is to minimize the number of machines needed to meet a new market demand.

Data input

Configuration information: Number of stages = L; number of machines in each stage = Ni, where i = 1,2, …, L; maximum number of machines allowed in each stage Mi, i = 1,2, …, L.
Stage characteristics: Each manufacturing stage has capabilities that are defined by a set of characteristics of the stage
Manufacturing tasks: Task precedence and characteristics
Machine reliability information: Machine reliability can be expressed by two parameters: MTBF (Mean Time Between Failure), and MTTR (Mean Time to Repair).
Demand: The system must be reconfigured so its new capacity will fulfill the new demand Dnew.

Data output

Two decision variables need to be determined:(1) machine allocation array, M[i], which determines how many machines are to be added to the system and where to add them, and (2) task allocation array, T[i], which describes how the tasks should be reallocated when the new machines are added to or removed from the system.

Algorithm

Genetic algorithm.

2.12 Reconfigurable Manufacturing Systems for Machine Selection (RMS-MS)#

Problem description

Reconfiguring and designing a manufacturing system consists of two different tasks: the first task consists to determine the set of machines to be involved in the production process, while the second one concerns the definition of the layout of the selected machines. This problem focuses on machines selection in RMS design, the layout design goes beyond this problem. Machines selection problem for RMS could be described as follow: a new production system is being built, and the designer of this new system has already established a list of candidate machines. Each candidate machine possesses a set of required functionalities and is able to perform at least one of the needed machining operations for the future product/family of products. The issue is to decide, from the set of m known candidate reconfigurable machines, n machines that will be selected to form the desired RMS, with n <= *m. We consider that a machine can perform an operation only if it possesses the required functionalities in the form of tools and allowed motions. We assume also that every machine incurs some functional costs such as configuration change cost, tool usage cost and machine usage cost.

Objective function

The selection of the machines is based on two main criteria respectively minimum total cost and minimum total time.

Data input

Number of product features, features of the considered product, number of available machines, available machines, number of available configurations for each machine, tools available for machine, configurations of each machine, number of operations of feature, operations of a particular feature, total number of all the operations, matrix of the total approach direction offered by the machine and required for a particular operation, matrix of the tool required by a particular operation, operations precedence matrix for a particular subpart, cost of using each machine, cost of using a tool, configuration change cost for each machine, tool change cost for each machine, configuration change time for each machine, tool change time for each machine, the processing time for each operation.

Data output

The total cost and the total completion time.

Algorithm

Multi-objective optimization and Non-dominated sorting genetic algorithm (NSGA-II) are proposed to solve the problem (NSGA-II is considered one of the well-known multi-objective evolutionary algorithms).

2.13 Celular Reconfigurable Manufacturing Systems for Machine Selection (CRMS-MS)#

Problem description

This problem involves the design and loading of Cellular Reconfigurable Manufacturing Systems in the presence of alternative routing and multiple time periods. These systems consist of multiple reconfigurable machining cells, each of which has Reconfigurable Machine Tools and Computer Numerical Control (CNC) machines. Each reconfigurable machine has a library of feasible auxiliary machine modules for achieving particular operational capabilities, while each CNC machine has an automatic tool changer and a tool magazine of a limited capacity. The approaches to solving the problem consist of two phases: the machine cell design phase which involves the grouping of machines into machine cells, and the cell loading phase which determines the routing mix and the tool and module allocation.

Objective function

The cell design problem can be modelled as an Integer Linear Programming formulation, considering the multiple process plans of each part type as if they were separate part types, and with the objective to find clusters of machines and process plans so that the maximum number of operations of each process plan family can be performed within the cell. Once the manufacturing cells are formed, a Mixed Integer Linear Programming model can be developed for the cell loading problem, considering multi-period demands for the part types, and minimising transportation and holding costs while keeping the machine and cell utilisations in each period, and the system utilisation across periods, approximately balanced.

Data input

Sequence of operations for each process plan, part types, process plans of each part type, CNC machines and RMT, the maximum number of cells to be formed, machine/process plan binary incidence matrix, weighting parameter that measures the importance of voids versus exceptional elements, demand for each part type in each period, tool types and auxiliary module types, number of tool slots required by each tool type in each CNC machine, tool magazine capacity of each CNC machine, maximum number of auxiliary modules for each RMT, number of available units of each tool type, number of available units of each auxiliary module type in each period, tool life of each tool type, average life of each auxiliary module type, processing time of each operation of each process plan of each part type at each machine, maximum workload available for each machine in each period, maximum inter-cell utilisation imbalance, maximum intra-cell utilisation imbalance, maximum inter-period utilisation imbalance, unit intercellular transportation cost, unit holding cost in each period, additional input data from cell design like number of inter-cell movements along each route of each part type, if each machine type belongs to each cell, and number of machines in each cell.

Data output

Number of cells to be formed, machine type to be assigned to each cell, number of process plans assigned to each cell, quantity of each part type to be assigned to each route in each period, inventory of each part type at the end of each period, number of tools of each type assigned to each CNC machine in each period and number of auxiliary modules of each type assigned to each RMT in each period, workload of each machine in each period, and total workload of machines in each cell in each period.

Algorithm

The cell design problem can be modelled as an Integer Linear Programming formulation. Once the manufacturing cells are formed, a Mixed Integer Linear Programming model can be developed for the cell loading problem. Although these problems can be optimally solved using optimisation models and approaches, for the efficient solution of larger problems, heuristics approaches are needed.