OEE Evaluation of a Paced Assembly Line through Different Calculation and Simulation Methods: A Case Study in the Pharmaceutical Environment

Modern production systems must guarantee high performance. Increasingly challenging international competition, budget reductions for the health sector and constant technological evolution are just three of the many aspects that drive pharmaceutical companies to continuously improve the productivity of their lines. The scientific literature has for many years been proposing calculation models for estimating the productivity of a machine. One of the most famous, and still used, is overall equipment effectiveness (OEE). This allows the calculation the valuable output considering the six ‘big losses’. The limitations of this approach are noticeable when considering a production line instead of a single machine. Numerous researchers have proposed alternative methods or changes in OEE, to be able to cover the widest spectrum of possible cases. In this study, we wanted to evaluate how such theoretical models related to OEE are actually able to represent the world of tight production flows or whether, in these cases, a more complex type of simulation should be preferred. To do this, we carried out a case study of a production line in the pharmaceutical industry, and the results showed that the simulation approach gives better results because of the peculiarities not considered by the theoretical models.


Introduction
In the industrial sector it is increasingly common to employ methods and tools to measure production performance. There are various reasons why, in recent years, there has been a steady increase in the adoption of these techniques. The main reason is the need to quantify the achievement of the objectives set by the companies and, consequently, to identify areas of improvement [1].
In the literature there are many papers that deal with the measurement of system performance [2] [3] [4] [5]. They emphasize th the ways in features of th the productio performance widely sprea productivity many metho improve the fields [7].
The perfor management particular, it and compare which elemen potential.
As a product machines tha coming ou measuremen machines. Th widely used such indicat approaches a capacity util operation of the causes o resolve them an index th features of a In 1998, Nak equipment e implementat Based on this materials in measure the provides a machine with this differenc losses, whic parameters: quality (Q).
The availabil of producti performance non-measura scraps and re the OEE para will never be The use of O fairly widesp heir historical which they he target mark on field in par from the f ad until at leas is so crucia ds have been performance rmance mea t, focus on qui is essential t e it to the theo nts make it les tion process is at contribute t t of the nt is primaril hroughput ra for this purpo tors are less are preferred lization provi a machine, wi of inefficiencie m. For this rea hat would en production sy kajima [10] int effectiveness ion of total p s concept, the nternational ( efficiency of measure of h respect to th ce are associa ch are usuall availability lity parameter ivity due t efficiency tak able stops, wh eworks. These ameter is less e equal to theo OEE to quantif pread in the l evolution ov had to adapt ket; first of all rticular this sh financial to st the end of t al in product developed re of a line in asures used ick inefficiency to measure ac oretical one, in ss effective tha s generally com to the product same proce ly concerned ate and capac ose [8]. For low effective and [9]. However ide a genera ithout giving es or any gui son the need nclose the m ystem. troduced the (OEE), a key productive ma e semiconduct (SEMI), defin equipment. T the actual p he theoretical o ated with the ly defined in (A), perform r (see Figure 1 to breakdow kes into accou hile the qualit e productivity than one, as oretical produc fy the efficienc production ver the years t to the chan l, globalizatio hifted the focu different asp he 1980s [6]. S tion managem ecently in ord various indus in produc y identificatio ctual perform n order to ide an its hypothe mposed of sev tion of the pro ess, perform with indivi city utilization w production d more struc r, throughput l measure of information a idance on how was felt to de main character concept of ov y element for aintenance (TP tor equipment ned an index The OEE indic productivity one. The caus e so-called six n terms of t mance (Ep) 1) refers to the wns and set unt the minor ty index consi y losses mean actual produc ction. terature of OEE is fou nsidered are u g losses [8]. he production ases, in fact, e machine w in its classical t contexts in everal authors EE, proposing Nakjima. g to Nakajima. n time, the time e production of (OT), the net time (VT) classical OEE roduction line ceutical sector of OEE, for uction models d approaches [13], and to ss [14].
whether these escribing and on process or, imulation, are Furthermore, Jeong and Phillips [15] have shown that OEE is not very suitable in capital-intensive sectors. In these areas, in fact, in order to have an immediate return on investment, the machines should be utilized to their maximum potential. It is therefore appropriate to consider each type of loss, even those related, for example, to the scheduled preventive maintenance (PM) or to the closures of plants during holidays. These elements are not considered in the classical version of OEE, which was conceived in a manufacturing environment. According to Jeong and Phillips, therefore, the right time at which the calculation of OEE should be made is not the loading time but, rather, the total calendar time.
In addition, de Ron and Rooda [16] have introduced a new version of the OEE parameter, in order to measure the actual performance of a production system, without considering all the inefficiencies that come from outside and that do not depend directly on them (operator skills, availability of materials, etc.). They define the E parameter that distinguishes cases in which a machine is inserted and integrated in a production process from those in which it is considered an element in itself. According to the authors, the conditions of starving and blocking, causing slowdowns in an independent machine, should not be considered in the calculation of inefficiencies. The OEE, therefore, measures the performance of a specific machine inserted into a wider production environment. Material handling, the presence of buffers, and production queues, however, significantly impact on its performance; for this reason, as well as having an index that measures the efficiency of each machine, you must also have an index representative of the entire line.
In fact, the main limitation of OEE is that it generally cannot be used for the calculation of the efficiency of an unbalanced production line. For this reason, several authors have proposed modifications to the classical formulation of OEE.
Brandt and Taninecz [17], for example, have introduced a parameter called the overall plant efficiency, which takes into account the efficiency of three elements: the workspace, the people and, of course, the machines.
Braglia et al. [8] have studied how to calculate the efficiency of a production line, introducing a new metric called OEEML (overall equipment effectiveness of a manufacturing line). The main advantage of this method is its possibility of evaluating a global parameter of an entire production line.
Caridi et al. [18] used the OEE parameters to calculate the rate of a balanced paced line without decoupling points, taking into account how the quality parameter impacts negatively on the pace of the line. The main limitation of this approach is that it needs a balanced line.
The analytical approach of OEE, despite the proposed changes, has several limitations, and so several researchers directed their interest towards a different approach [1], namely, the simulation. In literature, in fact, there are many works that demonstrate such interest [19], in particular in the production field [20] [21], for improving line effectiveness [22], for a more efficient plant layout [23], or for management of the entire supply chain [24].
Simulation is defined as the process that allows experiments to be performed on a specifically developed model, rather than on the real system. A simulation model, therefore, is a descriptive model of a process or of a system, built thanks to some of its typical parameters (production speed of a station, production or waiting times, etc.).
As a descriptive model of a real system, it can be used to perform experiments, to evaluate hypothetical changes to the real system, to compare different alternatives, and to urge the system 'in vitro'. This experiment has the advantage of not having any real impact on the system, although many of these simulations are time consuming and require information that is not always readily available [25].
Despite these disadvantages, the simulation is considered to be an indispensable method of problem solving [26] in different application contexts, outstandingly necessary in the following cases [25]: • testing of a complex system; • definition and design of a new system; • heavy investments required for the implementation of a proposed change to a new or existing system; • the need to have a tool that can show the various stakeholders involved the effects of specific solutions for a system.
The use of performance indicators is widespread in all industries, especially in those with a high level of difficulty in achieving high profits [27]. One of these is the pharmaceutical industry, which has high profitability and, at the same time, a remarkable need for high investments. These are linked both to the development phase of a new drug (the time and cost required for the introduction of a new drug into the market are, in fact, extremely long) and to the production phase.

OEE and O
OEE is a ver become a r production s according to time the unp availability, e The corporat reason for th obtain a synt a machine.
One of the m the unavaila determinatio information involves a continuous essential to h hardware sys [28].
As is known reducing the of scraps tha particular, is does not t considered p performance The line includes the presence of some buffers, which are able to decouple the different stations. Important in particular are the buffers adjacent to the bottleneck. The buffer upstream allows exploitation of the greater productivity of the machines preceding the bottleneck (A and B) and ensures that the bottleneck is never starving (problems such as failures to upstream machines set ups, etc., generating a lack of material for the bottleneck, are avoided). The buffer downstream, instead, avoids the bottleneck never being in the condition of blocking. In fact the critical station can work even when the downstream machine is no longer able to work (for a failure or any setup), continuing to produce and to store the pieces worked right in the downstream buffer. The goal of the buffers adjacent to the bottleneck, therefore, is to avoid the blocking of production for reasons dependent from other working stations.
The line has the main decoupling buffer just upstream of the machine C. It is supervised by the control logic that constantly checks his level of filling and stops, if necessary, either the upstream or downstream machine. The machines at the bottom of the line have high productivity that is obviously not fully exploited because the bottleneck sets its lower pace to the entire line.

Results
The methods presented in the third section were applied to the packaging line of a pharmaceutical company in order to assess its performance.
The application of each method was preceded by a phase of production log analysis, necessary to obtain useful data for the calculation of productive performance. This phase required considerable effort, mainly related to the need to determine the times of production and those of machine downtime. These values were derived by analysing the codes associated with the states of the machines, although this relation is not always trivial and immediate because the difficulty of the interpretation of some codes is somewhat cryptic.
The classic OEE was the first model applied. Starting from the log data of the production line, we prepared the necessary information for the calculation of �, � � and � and, in particular, the load times, failures, setup times, and the time lost due to non-measurable stops and to loss of quality.
The database used included an opening time of the factory of approximately three months (89 days) from which, given non-working days (holidays, Sundays) and work shifts, we had a loading time of 48 days.
We then performed the availability parameter calculation, considering only the efficiency losses related to faults and setups. Thus, for each machine of the line, we could evaluate the operating time. Table 1 shows the values gained. To evaluate the OEEML [8] it was necessary to separate inefficiencies due to individual machines of the line from the external ones (blocking, starving and preventive maintenance). This difference has an impact mainly on the values of Ep, which were obtained by eliminating the efficiency losses related to blocking and starving. To take into account preventive maintenance, however, we analysed the maintenance plan of the line, from which we could derive the values of � �� . Table 3 shows the values of OEEM for each machine of the line.  Table 3. Availability, performance efficiency, quality, PM availability and OEEM for each machine in the line, obtained by applying the method proposed by Braglia et al. [8] To compare packaging l creation of a required a ca and through the informati level of detai We used sto was built in its specific p considers the which are t materials (la location, and     Table   4 shows th both for the re may not notice ortness of the observe a non the different m m the analysi ds are compar in the phases in the simu