1 Introduction

In the current competitive global marketing, the key to manufacturing business success is to build a leaner, more flexible, and more responsive manufacturing environment. Companies’ ability to quickly change from one product to another is a key step towards this achievement. Changeover is one of the most time-consuming, no-value-added activities in several manufacturing environments [1]. Lean manufacturing aims at maximizing the value of the product through the minimization of no-value-added activities and dedicates particular attention to set-up time reduction to get rapid changeover of dies and equipment [2].

Single-Minute Exchange of Die (SMED) is a fundamental tool for improving flexibility in production processes [3]. SMED is part of the lean tools, which aim to reduce no-value-added activities in the production system. This quick change, under 9 min or less (single-digit minute), can be performed by the machine operator and preparer [4]. Briefly, this tool represents the key to reducing production lot sizes, even individual batch sizes (i.e., one-piece flow), thereby improving material flows. It is noteworthy that SMED and quick changeover programs have improved productivity by up to 70% [5].

SMED projects have to be carried out on several workstations but needed resources are missing. Indeed, these are performed following an “oil-spot approach”, initially implementing on a pilot machine to gradually spread to all other machines. It enables the SMED work team to manage expensive and time-consuming operations. Hence, a prioritization of to-be-done actions is required. Initially, it is simple to identify which set-up of the production unit requires attention and improvement, but later it may become more challenging. As in all lean projects, up to a certain point, it is relatively straightforward to obtain improvements, since they are achieved by performing activities that are easy to observe and measure, even to implement. But at some point, improvements are more complicated because they are no longer evident, and this requires more precise and targeted tools. Therefore, the rank of interventions to be performed and the production units on which to perform them play a critical role in making companies flexible in response to rapid market changes. During a SMED project, it is also useful to understand what economic benefits can be achieved and how these impact on costs. Hence, “as-is” data are compared with forecasted information resulting from SMED implementation.

There are several tools and methods in the literature for improving and supporting the SMED implementation. However, most focus on implementation in specific scenarios and, except for a few studies in which authors have proposed rules and guidelines for adapting/designing new machines, tools, and equipment from a “design for changeover” perspective, little effort has been made to develop operational tools to support the analysis of the production system from a changeover perspective.

A structured approach is thus needed that can identify and classify the no-value-added activities, i.e., losses, that occur during set-up operations, and that can provide the right intervention strategy for each of these.

The present paper introduces Set-up Saving Deployment (SSD) which is a novel lean tool that aims to (i) improve set-up efficiency by classifying, analyzing, and removing set-up losses within a changeover process; (ii) quantify possible time savings from improvement actions; and (iii) support decision-making for SMED implementation. The SSD adopts a logical and well-structured approach, derived from Manufacturing Cost Deployment (MCD) [6]. The aim is to achieve an optimal solution for a SMED project in accordance with any budget constraints within the factory. In addition, the tool provides a new basket of tailored set-up efficiency indicators that allow the analysis team to correctly assess set-up efficiency and compare the “as is” condition with the subsequent “to be” condition from an operational perspective.

By simultaneously applying this logical approach to several machines, fixtures, or equipment, it is possible to assess which unit requires specific attention. If properly managed, our new tool is ideal for planning interventions and monitoring the results of projects, while also highlighting their benefits in economic terms. In this sense, it can be seen as the compass of set-up improvement, as it highlights the activities to be made more efficient with reference to their impact on the time and thus the cost of a changeover.

The remainder of this paper is organized as follows. Section 2 overviews the literature on SMED and MCD. Section 3 describes the SSD. Section 4 describes the implementation of SSD in a resin-doming machine, within an important firm in Italy. Finally, Sections 5 and 6 are devoted to discussion, conclusions, and possible future developments, respectively.

2 State of the art: SMED and MCD

The aim of SSD is to identify and classify the losses that occur during set-up operations, quantify possible time savings resulting from improvements, and support decision-making for SMED implementation. SSD adopts a logical and well-structured approach, derived from MCD, to achieve an optimal solution for a SMED project in accordance with any budget constraints within the factory.

SMED and MCD are well-established tools in the scientific literature and industrial practice. Although both SMED and MCD have been combined with other tools to improve effectiveness and efficiency of several processes, none of the studies in the literature formalizes and specifies how they can be integrated effectively.

2.1 Single-Minute Exchange of Die

Many studies have addressed how to speed up changeover and lead to an overall increase in productivity [1]. There are several benefits or reducing set-up time, such as greater productivity [7]; less waste and rework, reduced inventory and lead time, greater system flexibility [8]; greater machine efficiency, in addition to the decrease in batch sizes [9]; and enhanced expertise of personnel involved in production (operators, mechanics, etc.) and maintenance optimization [10].

Implementation of SMED projects depends on the type of production that is to be analyzed. Various types of case studies involving the conventional SMED application have been presented. For instance, Deros et al. [11] exploited SMED to reduce the set-up time in an automotive battery assembly line. Sahin and Kologlu [12] reduced the machine set-up time on the turning line using the SMED in a bearing manufacturing company. Sousa et al. [13] reduced the changeover time by applying SMED in cork stopper production. Monteiro et al. [14] used SMED to improve the machining process in the metalworking industry. Vieira et al. [15] exploited SMED in the cold profiling process, using a population of five separate pieces of profiling equipment.

There is also an improved version of the SMED, i.e., when the conventional SMED is integrated with other tools, approaches, concepts, or when the steps of the conventional SMED are modified. For example, Patel et al. [16] applied mistake proofing (also known as poka-yoke) to the concept as a strategy to reduce the set-up time. Kumar and Bajaj [17] highlighted the 5S concept for integration into the conventional SMED steps to reduce the set-up time for the mechanical press machine. Ibrahim et al. [18] also investigated applying the 5S concept in the conventional SMED. Stadnicka [19] combined the conventional SMED with other tools including FMEA, survey method, five-whys analysis, Pareto analysis, and statistical analysis. Karasu et al. [20] integrated the conventional SMED with Taguchi methods in an injection molding production. Almonani et al. [21] proposed a novel approach, based again on the traditional SMED, incorporating multiple criteria decision-making techniques (MCDM) within the implementation phase devoted to transform operations from internal to external. Again, Braglia et al. [22] proposed an integrated approach to enhance the conventional SMED. Recently, Pattaro Junior et al. [2] developed a framework based on a practical application of strategies such as improvements by ECRS (eliminate, combine, reduce, and simplify), Standardized Work (SW), and Overall Equipment Effectiveness (OEE).

The impact of design on set-up operations represents another important area of discussion in the SMED literature. The availability of design principles or guidelines reduces the necessity of a posteriori set-up reduction projects. Several authors have already recognized the need for this a priori approach, by proposing and integrating an existing set of design rules [23]. Given that during the design phase the set-up times can be influenced considerably, Reik et al. [24] identified this approach as “Design For Changeover.” Singh and Khanduja [25] completed the set of design rules specifically for foundry dies and tooling. Cakmakci [26] investigated the relationship between the SMED and design “quality” of the equipment. Braglia et al. [27] introduced the concept of “duplication strategy,” which helps practitioners to identify all the items that impact on the changeover process.

The literature outlines several tools and methods for supporting and improving the SMED implementation. However, little effort has been made to develop operational tools to support the analysis of the production system from a changeover perspective. This justifies the development of novel operational tools that help practitioners to highlight not-optimized set-up conditions and to provide the right intervention strategy to remove inefficiencies.

2.2 Manufacturing Cost Deployment

The main purpose of MCD is to establish a systematic cost reduction program [6]. Cost deployment is the fundamental pillar of the World Class Manufacturing (WCM) model, referred to as the “Cost Deployment” pillar [28]. In fact, it cuts across all the other pillars of the WCM and acts as the necessary causal link between the identification of improvement actions in the targeted areas, and the evaluation of the results achieved through the implementation of specific pillars.

MCD uses an accurate and framed procedure, rigorously supported by five matrices:

  • A-Matrix classifies losses within a production system.

  • B-Matrix highlights cause-and-effect relationships among losses.

  • C-Matrix converts losses into manufacturing costs.

  • D-Matrix identifies potential improvements for losses.

  • E-Matrix carries out a cost-benefits analysis to prioritize improvements.

Many intangible benefits can be achieved by applying MCD. First of all, a critical feature of MCD is that it can be used to display all inefficiencies affecting a production process in a structured and straightforward manner. Furthermore, MCD concentrates on areas where the major inefficiencies are located, thus offering opportunities for greater efficiency and effectiveness in reducing and eliminating them. Finally, MCD simplifies the selection of improvements to be implemented to reduce/remove the root causes of such losses.

Due to its several benefits and thanks to its structured step-by-step procedure, MCD has been recently integrated with other tools and adapted to production contexts different from the automotive industry. Carmignani [29] developed a structured approach to assess the Supply Scrap Management Process (SSMP), and Abisourour et al. [30] proposed an integration of Value Stream Mapping (VSM) with cost deployment. Bertolini et al. [31] developed Project Time Deployment to identify the critical losses affecting an Engineer-To-Order (ETO) project, focusing on the business processes where causal losses occur, and providing opportunities for greater efficiency and effectiveness by reducing or even eliminating them. On the other hand, the MCD assessment model was extended, with appropriate modifications, to different industrial realities. Braglia et al. [32] studied a modified MCD tool designed to cope with the inefficiencies of ETO manual assembly tasks. Furthermore, MCD-based efficiency tools have been proposed from an energy perspective in the process industry [33] and material consumption for manufacturing processes [34].

3 The proposed tool: Set-up Saving Deployment

Set-up Saving Deployment (SSD) is a new lean tool whose purpose is to classify and analyze losses that occur during set-up operations and quantify possible time savings from improvement actions and support decision-making for SMED implementation. The implementation of this tool has significant advantages when applied to a production system consisting of several machines or production lines subject to frequent or long set-up processes. In fact, it compares the different production changeovers and identifies those that can lead to greater time savings. For example, the pressing department in a job or flow shop production system may be a suitable candidate for SSD implementation. The SSD adopts a logical and well-structured approach to achieve an optimal solution for a SMED project in accordance with any budget constraints within the factory. The flowchart of the proposed tool is reported in Fig. 1.

Fig. 1
figure 1

The flowchart of the proposed SSD tool

Full size image

3.1 Set-up Saving Deployment losses classification structure

In order to correctly identify the possible time savings that may result from the implementation of SMED improvements, we present a possible classification of set-up losses. This classification detects the losses that arise during a set-up and enables the analysis team to assess the temporal impact of each loss. Figure 2 shows the structure of the losses, which highlights where a loss occurs, and the portion of time lost. The actual set-up time is defined as the time recorded before the implementation of improvements and provides the current condition. The ideal set-up time is the condition that the analysis team expects to be achieved by implementing a SMED project on the selected machine tool/production line. The gap between the actual set-up time and the ideal set-up time is the result of multiple losses. Figure 2 shows the three categories of losses that can be adopted:

  • Process Losses (PLs). This category includes deviations from set-up standard procedures or the absence of the standard itself. This results in performing activities while the machines are stopped (internal operations) but should actually be performed while they are running (external operations). PLs involve, for example, a lack of personnel for set-up operations, missing equipment and parts, and their non-optimized transport to production units (e.g., machine tools or production lines).

  • Design Losses (DLs). This category includes losses due to the current non-optimal design of machines and set-up processes. These can be solved by a partial redesign to convert activities from internal to external. Examples of DLs are the lack of advanced preparation of operating conditions, the lack of reference systems for part positioning, and the non-standardization of key functions.

  • Internal activity Losses (ILs). This category involves losses due to the non-optimal design of internal activities. These losses are partially fixable by techniques that speed up and simplify internal set-up operations. For instance, this category includes a disorganized working area, the non-use of non-standardized parts, and functional fast clamping system.

Fig. 2
figure 2

The SSD losses classification structure

Full size image

To better understand this classification, some examples of losses are reported in Table 1. To reinforce standardization, each loss is coded through a loss ID. The first two letters refer to the loss category to which the loss belongs, while the next three letters identify the specific loss. Furthermore, this coding simplifies integration with management software to provide real-time visibility of all assets to determine which machines in the plant are not operating at peak efficiency. In the following paragraphs, this coding is omitted to increase the readability during the description of the various steps of the tool.

Table 1 Set-up loss types
Full size table

3.2 The decomposition of a set-up process

Once the set-up losses have been classified, it is fundamental to investigate at which phase the losses occur within a set-up process. Set-up processes appear to vary depending on the type of operation and the type of equipment being used. Yet when these processes are analyzed from a different viewpoint, it can be seen that set-ups comprise a sequence of phases. In traditional set-up changes, there are five fundamental phases: preparation, extraction, mounting, establishing control settings, and first-run capability [35]:

  • Preparation (P) — This phase ensures that all parts and tools are where they should be and that they are functioning properly. Also included in this step is the period after processing when these items are removed and returned to storage, machinery is cleaned, etc.

  • Extraction (E) — This involves the removal of parts and tools after the completion of processing.

  • Mounting (M) — This involves the attachment of the parts and tools for the next lot.

  • Establishing control settings (C) — This step refers to all the measurements and calibrations that must be made to perform a production operation, such as centering, dimensioning, and measuring temperature or pressure.

  • First run capability (F) — This includes any adjustments (re-calibrations, additional measurements) required after the first trial pieces have been produced.

The above five phases are a first-level decomposition. We do not divide the set-up phases up into elementary operations, because such a second-level decomposition structure depends on product complexity and consequently changes considerably from case to case.

3.3 The Set-up Saving Deployment steps

SSD uses a precise and well-structured procedure that is rigorously developed in three sequential steps, each of which is supported by building a specific matrix:

  • Step 1. Identification and evaluation of time savings through the L-Matrix.

  • Step 2. Correlation of time savings through TC-Matrix and calculation of “as is” set-up efficiency.

  • Step 3. Evaluation and prioritization of improvement actions through ECE-Matrix and calculation of “to be” set-up efficiency.

The following section describes in detail each step of the SSD.

3.3.1 Step 1

The first step of the SSD implies building the loss matrix (L-Matrix) but before doing so, a pre-selection of the machines that can be applied to the tool is required. As already mentioned, the implementation of SMED follows an oil-spot approach, since time and economic resources are limited, a pre-selection of candidate production units (i.e., machine tools or production lines) is necessary. The analyst team has to consider all the aspects that affect the identification of which production units are suitable candidates for implementing a SMED project, such as the number of set-ups performed on the single production units and the criticality in terms of relevance within the production system of the units themselves (bottlenecks, special order realization, brand production, etc.). Once the set of production units has been identified, an L-Matrix can be constructed for each production unit (Fig. 3). This matrix classifies losses highlighting where the loss occurs during the set-up process. Each single loss type is reported in the rows of the matrix, while the set-up phases where losses occur are shown in the columns. In order to better understand, losses are clustered in accordance with our three categories (i.e., PL, DL, and IL).

Fig. 3
figure 3

The L-Matrix

Full size image

Note that each element of the L-Matrix reports the expected time-saving which is defined as “a prediction of the time that can be gained by implementing improvement action on the specific loss”. Losses are ranked by means of colors (red, yellow, or green) based on their magnitude in terms of time-saving value. Each class is bounded by time-saving threshold values decided by the analysis team in order to present the priorities of intervention. Specifically, red refers to very big losses, yellow to big losses, and green to small losses. Analysts can adopt any criteria to assign colors, for instance by calculating the quartiles of the data distribution of time-savings, and can also adopt a different number of colored classes.

3.3.2 Step 2

The Time Correlation Matrix (TC-Matrix), shown in Fig. 4, clarifies the correlation between time-savings identified in the L-Matrix. Ignoring this correlation may prevent the analysis team from properly assessing the set-up efficiency and thus, selecting the right SMED project to implement first. In this respect, a loss, at a specific set-up phase, is said to be direct if its correction by a specific improvement action can save time by eliminating/reducing both the loss itself and other losses, thus called indirect losses, that occur at a specific phase of the set-up process. For instance, the standardization of a die decreases the mounting time (direct loss) and enables the implementation of fast clamping systems that significantly reduce the time required for positioning and locking (indirect loss).

Fig. 4
figure 4

The TC-Matrix

Full size image

The TC-matrix places the direct losses and their set-up steps in the rows, while the columns report indirect losses and the step where they occur. The non-empty cells provide the temporal evaluation of the correlation. In order to estimate the set-up efficiency, two different contributions are assessed. The direct saving time (\({DST}_{{d}_{h},z}\)) is the time saved by tackling the direct loss \({d}_{h}\) in the set-up phase z. The indirect saving time (\({IST}_{{d}_{h},z}\)) is the sum of all the indirect savings (\({T}_{{d}_{h},z}({i}_{j},k)\)) due to the implementation of the action that eliminates/reduces the indirect losses \({i}_{j}\), in the set-up phase k, related to the direct loss \({d}_{h}\) occurring in the set-up step z. The total saving time (\({TST}_{{d}_{h},z}\)) is thus evaluated by adding the aforementioned contributions. Analytically, it is expressed as follows:

$${TST}_{{d}_{h},z}={DST}_{{d}_{h},z}+{IST}_{{d}_{h},z}=T({d}_{h},z)+\sum_{j,k=1}^{J,K}{T}_{{d}_{h},z}({i}_{j},k)$$
(1)

Using the L and TC matrices, it is possible to highlight how and where losses occur and the correlation in terms of time savings between direct and indirect losses.

At this point, we propose a new lean indicator, called Set-Up Efficiency (SUE), which summarizes the information reported in the TC-Matrix and provides the analysis team with a means to quickly assess set-up efficiency and compare different machines:

$$\mathrm{Set}-\mathrm{Up}\;{\mathrm{Efficiency}}_{as-is}={\mathrm{SUE}}_{as-is}=\frac{\mathrm{Ideal}\;\mathrm{set}-\mathrm{up}\;\mathrm{time}}{\mathrm{Actual}\;\mathrm{set}-\mathrm{up}\;\mathrm{time}}$$
(2)

As stated above, the gap between Ideal and Actual set-up times is the result of many losses which progressively increase the time associated with carrying out activities. The subscript “as-is” underlies that the indicator represents the current process efficiency. By removing all the losses reported in L and TC matrices, the efficiency of the changeover can be maximized. However, in an industrial environment, there may be constraints that restrict the number of improvement actions that can be implemented, such as budget. SUE must thus be re-evaluated after selecting a subset of improvement actions that can be executed through the following ECE-matrix. In order not to burden the notation, the subscript as-is will hereafter be omitted.

The lower the SUE value is from the ideal value of 1, the greater the need for attention for the specific set-up process. If the team deals with equivalent machine tools or production lines, SUE becomes a critical discriminant factor.

By evaluating the SUE of several set-up processes, the analysis team can select which SMED project should start first. This is because Actual set-up time does not provide any information on how the process is performed, and neither does the OEE [36]. In fact, OEE recognizes set-up processes as losses that decrease the availability of the production unit [37], but it does not provide any further information about set-up performance. An Actual set-up time of 10 min could be worse than an Actual set-up time of 1-h set-up if the former takes ideally only 1 min to execute and the latter needs an hour.

The SUE can also be obtained as the product of three separate indicators, namely, Standard Set-up Efficiency, Design Set-up Efficiency, and Internal Set-up Efficiency. This is shown in the following Formula (3):

$$\mathrm{SUE}=\mathrm{Standard}\;\mathrm{Set}-\mathrm{Up}\;\mathrm{Efficiency}\times\mathrm{Design}\;\mathrm{Set}-\mathrm{Up}\;\mathrm{Efficiency}\times\mathrm{Internal}\;\mathrm{Set}-\mathrm{Up}\;\mathrm{Efficiency}$$
(3)

where:

$$\mathrm{Standard}\;\mathrm{Set}-\mathrm{Up}\;\mathrm{Efficiency}=\frac{\mathrm{Standard}\;\mathrm{set}-\mathrm{up}\;\mathrm{time}}{\mathrm{Actual}\;\mathrm{set}-\mathrm{up}\;\mathrm{time}}$$
(4)
$$\mathrm{Design}\;\mathrm{Set}-\mathrm{Up}\;\mathrm{Efficiency}=\frac{\mathrm{Target}\;\mathrm{set}-\mathrm{up}\;\mathrm{time}}{\mathrm{Standard}\;\mathrm{set}-\mathrm{up}\;\mathrm{time}}$$
(5)
$$\mathrm{Internal}\;\mathrm{Set}-\mathrm{Up}\;\mathrm{Efficiency}=\frac{\mathrm{Ideal}\;\mathrm{set}-\mathrm{up}\;\mathrm{time}}{\mathrm{Target}\;\mathrm{set}-\mathrm{up}\;\mathrm{time}}$$
(6)

Formula 3 helps to interpret t the causes behind set-up inefficiency. While SUE is a global assessment of current performance concerning global efficiency, each of the three components of SUE pinpoints specific aspects of the process that can be targeted for improvement:

  • Standard Set-Up Efficiency (Formula 4) considers all losses related to deviations from standard procedures, or the absence of the standard itself. A low value of this indicator stresses that most of the elementary operations are performed with the machine stopped, although they can be operated with the unit running.

  • Design Set-Up Efficiency (Formula 5) evaluates the time lost due to elementary operations that are performed externally but could be performed internally by redesigning the process or machine. A poor value of this indicator pinpoints that equipment, such as the height of a die, and the functional conditions necessary for operating the equipment itself, such as the temperature of a die, are not properly standardized.

  • Internal Set-Up Efficiency (Formula 6) considers time losses caused by internal operations that could be improved by better management and machine redesign. A low value of this indicator highlights the need to reorganize the internal operations and alter the positioning and locking systems.

The reading of these indicators is simple: the farther the KPI value deviates from the ideal value of 1, the greater the need for attention within the cluster defined by the indicator. For instance, a low Standard Set-up Efficiency value, compared with other indicators, suggests a more careful allocation of internal and external operations.

With this perspective, the SSD provides an evaluation of the current set-up efficiency that enables the analysis team to correctly select which SMED projects to start first, and furthermore, it pinpoints which aspects need improving.

3.3.3 Step 3

By using L-Matrix and TC-Matrix and the set of lean set-up indicators, the analysts can identify which SMED project may generate more time savings than those currently being investigated. SSD is able to support the decision-making process for the SMED implementation by selecting a series of appropriate corrective actions to cope with each loss previously detected. It can eliminate losses through the most effective measures, reducing changeover times and simplifying the tasks to be carried out.

We propose two kinds of improvements: organizational-procedural and engineering. Organizational-procedural improvements entail management interventions such as the implementation of lean management tools, while engineering entails machine/production line redesign to convert many external operations into internal and thus improve the internal operations themselves. The set of techniques presented here can be modified depending on the case being analyzed.

The ECE-Matrix (Fig. 5) places losses and the set-up phases where they occur in the rows, while the improvements are reported in the columns. Furthermore, the ECE-Matrix provides an index for prioritizing the implementation of corrective actions. This index, named ECE (Efficacy, Cost, and Ease), addresses the time impact saving capability (i.e., Efficacy), the cost to sustain to roll out the intervention (i.e., Cost), and the ease of implementation in terms of time and needed resources (i.e., Ease). Given a direct loss, each possible technique is evaluated according to these three criteria by assigning a score from 1 to 5. The higher the score given with respect to a factor, the greater the capacity to satisfy the property associated with that factor. The product of the three scores, i.e., the ECE, represents a qualitative estimate, ranging from 1 to 125, of the value of the technique to tackle that direct loss:

$${ECE}_{{d}_{h},z,{t}_{j}}={Ef}_{{d}_{h},z,{t}_{j}}\times {C}_{{d}_{h},z,{t}_{j}}\times {Ea}_{{d}_{h},z,{t}_{j}}$$
(7)

where:

Fig. 5
figure 5

The ECE-Matrix

Full size image

  • Efficacy factor (\({Ef}_{{d}_{h},z,{t}_{j}}\)) expresses the mitigation power of improvement technique \({t}_{j}\) on the direct loss \({d}_{h}\) in the set-up step z. Ef = 1 means low impact, while Ef = 5 means complete time saving.

  • Cost factor (\({C}_{{d}_{h},z,{t}_{j}}\)) expresses the economic weight of cost that should be sustained to adopt the improvement technique \({t}_{j}\) on the direct loss \({d}_{h}\) in the set-up step z. C = 1 means high cost, while C = 5 means low cost.

  • Ease factor (\({Ea}_{{d}_{h},z,{t}_{j}}\)) expresses the simplicity, in terms of resources and time, of the improvement technique \({t}_{j}\) that are needed to reduce/eliminate the direct loss \({d}_{h}\) in the set-up step z. Ea = 1 means low-level ease, while Ea = 5 means high-level ease.

After determining the highest ECE index, different implementation strategies can be identified. For example, selected techniques can be implemented strictly following the order of priority obtained with the ECE. A company could decide to implement, whenever possible, only one improvement in different phases, without rigorously following the priority order. Alternatively, the analysis team could adopt different criteria, such as a cost–benefit analysis on the subset of improvements selected according to the previous ECE ranking. The policy that a company decides to adopt will promote the implementation process of one strategy over another to obtain the most benefits.

We believe that the simplest and most operational way to use SSD is to rank the improvement techniques according to their index value, also considering the chain of direct–indirect losses previously detected in the TC-matrix, and then implement the first n improvement actions (where n is a number set a priori by the analysis team considering the budget). Once the n improvement actions to be implemented have been selected, it is possible to assess the “to be” set-up efficiency (\({\mathrm{SUE}}_{to-be}\)), which represents the expected efficiency achievable with the existing constraints of the SMED project. Using the same criterion, the other intermediate indicators can be recalculated.

4 Case study

We now report how our SSD was applied to an Italian company that produces self-adhesive resin labels. The company was chosen for two main reasons (Fig. 6). Firstly, we had already collaborated with the managers, which guaranteed a direct face-to-face relationship with them throughout the study and gave us full access to data on the set-up operations. Secondly, given that the company operates in a very competitive scenario, and considering that changeover is one of the most time-consuming activities, it needs to dedicate particular attention to set-up time reduction in order to reduce the lead time and to increase the overall quality of its products.

Fig. 6
figure 6

The resin-doming machine

Full size image

The company was thus a perfect environment for testing SSD. The company felt that SSD would be a useful tool to classify and analyze losses that occur during set-up operations, quantify possible time savings from improvement actions, and support decision-making for SMED implementation.

As this was the first implementation of the tool, it was decided to focus on a consolidated machine. A resin-doming machine designed for the production of self-adhesive resin labels for the fashion and automotive sectors was selected. The SSD was carried out by a team composed of engineering, maintenance, production, and academic personnel. As part of this task, the company collected and classified the information required for the implementation of the SSD following the steps presented in this paper.

4.1 Description of the resin-doming machine

In order to better describe the operational aspects of the SSD, the resin-doming machine and the corresponding changeover process were thoroughly analyzed. Doming consists in applying a layer of transparent two-component polyurethane resin to a two-dimensional label to create a three-dimensional effect that makes the product much more attractive, while at the same time, the UV-resistant film provides additional protection. This process is performed by an automated line made of several components including:

  • Manual station for loading and unloading the sheets

  • Metering pump

  • Mixer

  • Ovens

  • Control system

  • Linear robot

  • Vacuum pump

  • Motion system

  • Frame with needles to pour the resin onto the labels

As the line is highly automated, the operator has to load the sheets into the manual station and remove them once the process is complete. The resin is poured, row by row, onto the labels through several needles mounted on a metal bar positioned on a rectangular frame. The frame is moved by a linear robot by means of linear guides.

4.2 Description of the set-up process

The machine undergoes two set-ups. We filmed all the details of the changeover and noted that it took the operator an average of 17 min to perform the set-up process. The set-up process is as follows:

  • Washing. Washing is an automated process involving washing mixers, tubes, and needles using a methylene chloride solution. During this nothing can be done on the machine. It must be performed within 5 min of stopping the machine, as the resin immediately begins to cure inside the mixer, tubes, and needles. This is also a problem during breakdowns and micro stops.

  • Tool removal. This involves disconnecting the tubes from the mixer head by unscrewing them with pliers. If necessary, the operator must replace the needles by removing them from the metal bar.

  • Tool substitution. This involves fitting the new mixing head by first screwing it in manually and then using a spanner. Next, the operator has to select the correct number of needles and tubes, check the length of the tubes to ensure an even flow of resin, and insert the needles into the new bar, avoiding tube entanglement.

  • Program loading. If there is one, the work program is loaded from the machine or must be written from scratch. In any case, the operator has to set the “zero point” of the linear robot, as differences of a few millimeters can occur due to centering errors during the previous print. The operator positions the needles on the first dot to be resined according to the last set of coordinates, and then repositions the needles on the label dots where resining is to begin. Hence, a test cycle is required.

  • Test cycle. This involves placing a sheet of silicone paper on the line and starting the machine, which runs a full cycle. The operator checks the position and height of the needles and the correct amount of resin on the label.

  • Pallet preparation. This last task entails preparing the pallets for the forthcoming job.

By monitoring and analyzing OEE, we observed that the greatest cause of inefficiency was due to the set-up process, which accounted for between 30 and 40% of the total losses. Table 2 shows the set-up operation, the phases into which the process was broken down, and the average time required by each activity.

Table 2 Set-up time structure
Full size table

4.3 Application of the SSD tool to the doming machine

The result of the SSD application is presented below, with a detailed description of each matrix. The set-up losses considered in the application of the SSD tool were classified by the analysis team according to three different levels of criticality (Fig. 7). Red cells depict time savings of > 180 s, yellow between 90 and 180 s, and green cells represent minimal time savings of < 90 s. This first screening criterion regards the amount of time saved as qualitatively estimated by the team.

Fig. 7
figure 7

Case study L-Matrix

Full size image

The TC-Matrix (Fig. 8) was constructed highlighting the relationship between losses whose resolution in a specific set-up phase (direct losses) enables the resolution of other losses in the same set-up phase or outside it (indirect losses). By managing direct losses with tailored corrective actions, other improvements can be implemented that eliminate indirect losses and increase time savings. In addition, the TC-Matrix shows the estimated time that the analyst team expected to save by removing losses. In particular, the team identified a direct–indirect loss chain related to the outsourcing of tool removal and replacement that can represent a major time-saving opportunity.

Fig. 8
figure 8

Case study TC-Matrix

Full size image

It is then possible to summarize all the information of the TC-Matrix by evaluating the \({\mathrm{SUE}}_{as-is}\) and other intermediate indicators.

$${\mathrm{SUE}}_{as-is}=\frac{1070-880}{1070}=18\mathrm{\%}$$
(8)
$$\mathrm{Standard}\;\mathrm{Set}-\mathrm{Up}\;{\mathrm{Efficiency}}_{as-is}=\frac{950}{1070}=89\%$$
(9)
$$\mathrm{Design}\;\mathrm{Set}-\mathrm{Up}\;{\mathrm{Efficiency}}_{as-is}=\frac{250}{950}=26\%$$
(10)
$$\mathrm{Internal}\;\mathrm{Set}-\mathrm{Up}\;{\mathrm{Efficiency}}_{as-is}=\frac{190}{250}=76\%$$
(11)

As the indicators show, the set-up process is affected by a non-set-up-oriented design, justified by a low value of Design Set-Up Efficiency which makes it difficult to perform operations while the machine is running and, as a result, significantly increases the time required for the entire changeover process.

At this point, the presence of a non-empty cell in a TC-Matrix element implies the need to implement corrective actions to achieve the estimated time savings reported in the cell. The experts in the analysis team are responsible for identifying and classifying a series of corrective actions to manage specific losses (Table 3). The set-up losses are shown in the rows of the ECE-Matrix (Fig. 9), while the improvement actions with the corresponding ECE value are in the columns. Based on the ECE score, improvement actions can then be selected.

Table 3 List of the improvement techniques
Full size table
Fig. 9
figure 9

Case study ECE-Matrix

Full size image

Tables 4, 5, and 6 summarize the technical data used to assign the three ECE scores and thus assess the ECE value for each element in the ECE-Matrix. Specifically, Efficacy is assessed as a function of the Actual set-up time (Table 4), Cost according to the costs required for implementation (Table 5), and Ease according to the expected time required to implement corrective actions (Table 6).

Table 4 Ratings for the Efficacy value
Full size table
Table 5 Ratings for the Cost value
Full size table
Table 6 Ratings for the Ease value
Full size table

The team of experts decided to select six improvement techniques to be implemented, considering improvements with an ECE value of at least 20 and in accordance with the chain of direct–indirect losses previously identified as being valid. The optical detection system was discarded due to its high cost, low efficacy, and ease of implementation. Once the improvement techniques had been selected, the team evaluated the \({\mathrm{SUE}}_{\mathrm{to}-\mathrm{be}}\) and other intermediate indicators.

$${\mathrm{SUE}}_{\mathrm{to}-\mathrm{be}}=\frac{1070-880}{1070-860}=\frac{190}{210}=90\mathrm{\%}$$
(12)
$$\mathrm{Standard}\;\mathrm{Set}-\mathrm{Up}\;{\mathrm{Efficiency}}_{\mathrm{to}-\mathrm{be}}=\frac{210}{210}=100\%$$
(13)
$$\mathrm{Design}\;\mathrm{Set}-\mathrm{Up}\;{\mathrm{Efficiency}}_{\mathrm{to}-\mathrm{be}}=\frac{190}{210}=90\%$$
(14)
$$\mathrm{Internal}\;\mathrm{Set}-\mathrm{Up}\;{\mathrm{Efficiency}}_{\mathrm{to}-\mathrm{be}}=\frac{190}{190}=100\%$$
(15)

It can be seen that the values of the indicators increased significantly. The Design Set-Up Efficiency rose from 26 to 90%, while the Standard and Internal Set-Up Efficiency reached its maximum attainable value. In fact, implementing the selected improvement techniques would eliminate all the process and internal losses and significantly reduce design losses. This resulted in an expected \({\mathrm{SUE}}_{\mathrm{to}-\mathrm{be}}\) of 90% with an enhancement of 72%. Overall, the total changeover time dropped from 1070 to 210 s. In addition, operators reported that the improvements implemented resulted in a significant simplification and standardization of activities.

5 Discussion

As emerged from the industrial application, SSD has achieved good results with a significant reduction in changeover times. Compared to traditional SMED implementation, SSD has several advantages. For starters, SSD is a step-by-step procedure that aims to harness the full potential of SMED since its implementation has proven to be challenging. L-Matrix and TC-Matrix identify which and where set-up losses occur and provide an assessment of achievable time savings.

In addition, by simultaneously applying this logical approach to several machines, fixtures, or equipment, it is possible to assess which unit requires specific attention. Indeed, SSD can pass the critical phase of identification of the pilot testing area for SMED projects. Specifically, the set of lean indicators allows the analysis team to determine whether a change-over process is executed efficiently and thus which SMED projects should be developed first. By comparing the different SUE values, it is possible to evaluate which unit requires special attention, and moreover, by comparing the “as-is” values with the “to-be” values it is possible to quantify the expected gains. Another important difference between traditional SMED and SSD is that the first investigates the correlations between savings to establish the implementation sequence of the most effective improvements. Finally, SSD constitutes a more detailed analysis as it considers critical factors beyond effectiveness, such as ease of implementation and cost.

To reach its full potential, the tool must be tested in different production contexts. This could allow us to identify the possible limitations that the previous application did not provide.

6 Conclusions

This paper presents a new lean tool called Set-up Saving Deployment, which improves set-up efficiency by classifying, analyzing, and removing set-up losses within a changeover process, and supports decision-making for SMED implementation. By simultaneously applying this logical approach to several machines, fixtures, or equipment, it is possible to assess which unit requires specific attention.

To systematically support the application of this new tool, three novel matrices were proposed. In particular, (i) the L-Matrix identifies the losses related to a specific set-up phase and provides an initial qualitative screening, (ii) the TC-Matrix investigates the possible relationship between the losses resolutions and quantitatively estimates the time savings achievable by removing the losses, and (iii) the ECE-Matrix clarifies which corrective actions are most suitable for each loss in relation to many aspects such as the time impact of the improvements on the specific loss (Efficiency), the costs that should be incurred to implement the action (Cost), and the feasibility required to implement the improvement activities (Ease).

The validity of the tool was confirmed during a case study of an industrial application. The site production manager reported that SSD was very effective in standardizing set-up activities, which were usually performed without a systematic approach. The results obtained demonstrate that the SSD achieves an optimal solution for a SMED project in accordance with any budget constraints within the factory. Using this tool, the team of analysts identified losses occurring during the set-up process, classified them, and implemented tailor-made corrective actions that improved set-up efficiency by 72%. This resulted in the total changeover time dropping from 1070 to 210 s.

As industrial digitization continues to grow, in accordance with the principles of Industry 4.0, we believe that SSD can be integrated into business software and thus become easier to implement and maintain. Indeed, its well-structured step-by-step procedure ensures ease of electronic implementation through interconnected worksheets. In addition, this integration could provide real-time visibility of all assets to determine which machines in the plant are not operating at peak efficiency.

Future activity can concern the develop tools and methods that can accurately quantify implementation costs while maintaining an operational view of set-up optimization. One way to do this would be to introduce a cost/benefit analysis to assess the appropriateness of a decision by weighing its potential costs and benefits. This will entail developing structured cost metrics to increase the accuracy of the analysis by modifying the construction of the ECE-Matrix. In addition, the implementation of corrective actions could consider not only economic aspects but also sustainable aspects such as work ergonomics and operator stress. Finally, because of the qualitative nature of the scores used to evaluate the ECE, a future analysis using fuzzy logic can be proposed to make the analysis more consistent.