GRASP for sequencing mixed models in an assembly line with work overload, useless time and production regularity

Abstract

Abstract.A GRASP algorithm is presented for solving a sequencing problem in a mixed-model assembly line. The problem is focused on obtaining a manufacturing sequence that completes the greatest possible amount of required work and fulfils the production regularity property. The implemented GRASP algorithm is compared with other resolution procedures by means of instances from a case study linked to the Nissan's engine plant in Barcelona.Keywords: GRASP; Sequencing; Mixed-model assembly line; Production mix preservation. PreliminaresA mixed-model assembly line is able to manufacture several variants of the same product (e.g. engines for SUVs (Sport Utility Vehicle) and different types of vans) without physical changes at workstations and without significant setup times between consecutive different units. This type of assembly lines presents two categories of problems that are solved traditionally sequentially: (1) balancing problems [1], and (2) product sequencing problems [2]. The first problem type consists of assigning a set of tasks (relating to the product assembly) into a set of workstations arranged in series with the maximum efficiency as possible and fulfilling a set of conditions. Once solved the first problem and given a demand plan and the time to perform the said plan, the second type of problems consists of establishing the manufacturing order of products regarding one or more criteria. The objectives taken into account when the units are sequenced are not necessarily mutually exclusive. Indeed these objectives often respond to several concerns about production [3]. Among them, there are: (o.1) maximise the useless time, completing the maximum number of units and therefore reducing the unnecessary waitings [4]; (o.2) maximise the level of satisfaction of the set of constraints, which are related with spatial components of the products [5]; and (o.3) maintain constant the manufacturing rate of products and the component consumption rate in order to minimise the maximum levels of component stocks [6].

Free and regular mixed-model sequences by a linear program-assisted hybrid algorithm GRASP-LP

“…In view of the most common objectives for sequencing problem and taking as reference the work by Bautista, Alfaro-Pozo and Batalla (2016) [7], this paper lies in a specific sequencing problem, the MMSP-W (Mixed Model Sequencing Problem with Workload Minimization).…”

“…Accordingly, we study a mixed-model sequencing problem with the objective of maximizing productivity, reducing the useless time of operators and regularizing the production by means of preserving the production mix in the manufacturing sequence (pmr). Besides, unlike [7], in this paper the operations can be freely interrupted (i2). Specifically, the objective o.1 is represented by the objective functions (work overload, W, and useless time, U), the objective o.3 is incorporated into the problem by production mix restrictions (pmr) and, finally, the condition i2 is considered by the introduction of some inequalities.…”

“…Similar to [7], the local improvement phase begins with the !(!) sequence, which satisfies the conditions (9).…”

“…The descent algorithms are based on exchange and insertion of products (see [7]) and they are address to exploring sequence cycles in both increasing and decreasing direction. Such procedures are: (i) forward exchange (figure 1), (ii) backward exchange (figure 2), (iii) forward insertion (figure 3), and (iv) backward insertion (figure 4).…”

“…Like [7], this comparison between the performances of all procedures is made through a case study linked with the Nissan's Engine Plant in Barcelona.…”