Exploring the Impact of Lean Manufacturing on Flexibility in SMEs

Purpose: This paper describes the use of simulation and case-study research to assess flexibility gains induced by the adoption of three Lean Manufacturing practices. Design/methodology/approach: We gather useful material and information about the manufacturing process of a selected Small-Medium Enterprise by adopting a case-research approach. The Value Stream Mapping is the method used for visualizing flows of products and information along the production system. Starting from the current arrangement of the company, computer simulation is used to assess the benefits arising from Cellular Manufacturing, Just-in-Time Delivery by Suppliers, and Single Minute Exchange of Dies. Findings: To investigate the flexibility improvements coming from the introduction of Lean Manufacturing, we present a simulation model of the described company on which we performed our analysis. We quantify the flexibility of different configurations according to the new 5-step approach in order to segregate the contribution of different lean techniques. Originality/value: We extend the combined use of Case Research and Computer Simulation to the research on Manufacturing Flexibility within Small-Medium Enterprises. We enhance the knowledge on this under investigated context collecting quantitative field data. Moreover, building on the factorial Design of Experiment, we introduce a new 5-step method to appraise the costbenefit ratio of lean techniques for flexibility. The managerial implication of this research is


Introduction
Previous research acknowledges that manufacturing firms can achieve significant improvements from the introduction of lean practices (Singh, Garg & Deshmukh, 2008;Thomas, Barton & Chuke-Okafor, 2008).Lean approaches are recommended to mitigate uncertainty, reduce costs and improve productivity (Boyle & Scherrer-Rathje, 2009).Despite this relevance, there is paucity of studies specifically focused on lean applications to Small-Medium Enterprises (hereafter SMEs), and some authors claim for further research (Bakås, Govaert & van Landeghem, 2011;Shah & Ward, 2003).This paper aims at filling this gap.In particular, we address how the introduction of lean management practices can enhance manufacturing flexibility in SMEs.
By using computer simulation and a case-study we develop a staged approach to quantitatively assess the flexibility improvements induced by the introduction of three Lean Manufacturing practices.This complements existing studies that discuss the importance of Lean Manufacturing in SMEs solely from a qualitative perspective (Achanga, Shehab, Roy & Nelder, 2006;Rose, Deros, Rahman & Nordin, 2011;Zhou, 2016).
The paper is structured as follows.Section 2 presents the literature review, Section 3 shows the research methodology, Section 4 describes the case study, Section 5 shows the results of the investigation and the last section draws conclusions, limitations and point out avenues of further research.

Manufacturing Flexibility (MF)
Manufacturing Flexibility (hereafter MF) has been the subject of theoretical and empirical characterization (Vokurka & O'Leary-Kelly, 2000).It is usually defined as the capacity of a production system to handle variability in its operating conditions (Buzacott & Mandelbaum, 1985), and face uncertainties due to exogenous and endogenous factors (Mascarenhas, 1981).Gupta and Goyal (1989) remark that manufacturing systems should be flexible as this affects the overall performances, allows customization, and reduces costs.In line with this, Carpinetti, Gerólamo, and Dorta (2000) argue that firms' competitiveness can greatly depend on their capability to rapidly adapt to fulfill unpredictable market demands.This relevance in the western industry has been enforced by the effects of globalization, as mass customization in combination with shorter times of delivery can better compete against cheaper goods from low cost countries (Gerwin, 1993).These considerations apply also to small and medium businesses.In fact, it is said that the more SMEs are flexible, the more they can promptly respond to any customer requests (Singh et al., 2008), e.g.produce and deliver a limited amount of highly customized products in a feasible way ( Cagliano, Blackmon & Voss, 2001;Vassell, 1999).Although relevant for most businesses, it is claimed that flexibility is not a 'free good' (Bengtsson, 2001;Slagmulder & Bruggeman, 1992).Despite soft actions such as a higher cooperation between siloed departments can improve flexibility with little investment ( Hayes & Pisano, 1994), usually firms have to invest money in redundant and multipurpose production capacities (Abernethy & Lillis, 1995).This is critical in those companies -typically SMEs -which allocate few budgets to renew production facilities (Nemetz & Fry, 1988).In this respect, LM practices can be extremely functional, as their introduction brings manifold improvements with relatively low costs (Womack, Jones & Roos, 1990).This is the reason why this paper explores to what extent the adoption of Lean Manufacturing in manufacturing systems bring flexibility gains.

Lean Manufacturing (LM)
As well known, Lean Manufacturing (hereafter LM) was initially set up in the Japanese automotive industry as a new concept, in reaction to the serious circumstance of the post-world war II economy.Womack et al. (1990), in his seminal book, suggests a five-step approach aimed to ban waste and maximize the performances of production flows.LM techniques conceived for these purposes are widely discussed in the operations management literature (Feld, 2000;Monden, 2011;Nahmias, 2001).Hereafter, we review those applications that in our opinion can likely influence the MF performances.
• Cellular Manufacturing (hereafter CELLMFG).In CELLMFG the company's production is structured into cells; these are defined as single working units, enclosing equipment and resources necessary to produce the highest number of similar products.
• Just-in-time Delivery by Suppliers (hereafter JITds).JITds ensures that suppliers deliver the right quantity at the right time in the right place (Shah & Ward, 2007).Ansari and Modarress in their work (Ansari & Modarress, 1988) state that the base of this technique is a partnership between the Supplier and the Company.Their study also confirms that JITds contributes to the improvement of product quality and productivity of each kind of company.
• Single Minute Exchange of Dies (hereafter SMED).Changing the production sequence from one product to another requires usually significant time for machines setup.SMED arises from the need to have a Quick Changeover.This procedure consists in converting as much as possible the IED (Inside Exchange of Die) in OED (Outside Exchange of Die), practically minimizing setup activities that require downtimes.

Value Stream Mapping (VSM)
Value Stream Mapping (hereafter VSM) is a method for visualizing flows of products and information along the production process, from raw materials to semi-finished and finished products.Icons and rules for flow representation are self-explicable, and the symbols set is currently unified in the managerial literature (Rother & Shook, 2003).Through VSM, managers can focus on activities not generating addedvalue to the finished products, in order to eliminate wastes.A VSM analysis starts with the drawing of the Current State Map that is the baseline for representing how operations are currently performed.The second step consists in the development of the Future State Map that shows the identified improvements.Generally, these improvements entail reshaping of the facility layouts, reduction of WIPs, and restructuring of the overall production process.Once completed, the method can start again as part of the systematic and continuous improvement process.

Simulation as Support to Value Stream Mapping
Since it is almost impossible to quantify the achievable gains in terms of KPI (i.e. the Work-in-process inventory) with a future state map only, simulation constitutes an appropriate complementary tool (McDonald, van Aken & Rentes, 2002).This solution to evaluate the profitability of an investment, that constitutes a base requirement for the financing of lean introduction (Sullivan, McDonald & van Aken, 2002), is usually cost effective and generally cheaper than a practical on the field test.Many examples of this approach combining VSM and simulations are available in literature (Abdulmalek & Rajgopal, 2007;Bernards, van Engelen, Schrauwen, Cramer & Luitjens, 1990;Gurumurthy & Kodali, 2011;Lian & van Landeghem, 2007;McDonald et al., 2002;Narasimhan, Parthasarathy & Narayan, 2007;Wang, Guinet, Belaidi & Besombes, 2009).One of the most popular simulation tools available on the internet is Arena Simulation (Detty & Yingling, 2000;Hammann & Markovitch, 1995;Kelton, 2002).This software has been selected for the present work considering that its diffusion is wide and that its features have been successfully proven in similar studies (Detty & Yingling, 2000;Lian & van Landeghem, 2002).

Research Methodology
A foundational case research provides the base data for the accomplishments of this paper.Then, computer simulation leverages that material to apprise the connections between LM and MF.Section 3 presents the main methods and tools adopted during this study.

Case Research (CR)
The setting of a case-study is considered extremely crucial for gathering useful material and information for the implementation of LM (Ohno, 1988).It is commonly used to fill the need for experiential evidences about lean introduction into small companies (Bakås et al., 2011;Moeuf, Tamayo, Lamouri, Pellerin & Lelievre, 2016).Therefore, in the following we analyze the manufacturing department of a SME producing decorated glass items, which we purposively select for this study.Hereafter, this company is referred to as ALFA, since part of the disclosed information are confidential.According to Voss et al. (Voss, Tsikriktsis & Frohlich, 2002), theory building can greatly benefit from case research (hereafter CR) in its early stages.In particular, validation of operations management practices accounts for manifold field evidences (Lewis, 1998).According to Yin (2013), CR is an empirical exploration that aims to investigate a contemporary phenomenon in its real context.This happens especially when the boundaries between phenomenon and context are not clearly evident, and when multiple sources of evidence -such as field observations and interviews -are used.Thus, this strategy is recommended when the following criteria are fulfilled: • Linkages between the phenomenon under examination and the context are not evident; • Events cannot be reproduced in a laboratory; • Events observation is possible; • Typical research questions are of 'how' and 'why' types.
These criteria fit well with the aims of this work, as we investigate the correlations between LM and MF through field analysis of a case study.
Literature discriminate case studies on the basis of two main parameters: • The number of cases examined: single or multiple case-study.
• The aim of the study: holistic (single units of analysis) or embedded (multiple units of analysis).
Figure 1.Case Studies (Yin, 2013) We implement a single-case study, considering this the best option to provide managers with a detailed example of cost-benefit analysis (Burgess, 1993;Crotty, 1998;Voss et al., 2002).Even if the evidences are based on a single case, general considerations may nevertheless be formulated about cause-effect relationships and operational modes.Furthermore, we select a holistic approach to focus a manufacturing facility that cannot be molded into embedded subunits.On the base of the previous considerations, we set the outline of the research strategy used in this paper as presented in Table 1.The paper addresses the following research question:

Parameter
How LM and MF are connected in the SME context.
We clarify the flexibility gains that are specifically due to the introduction of lean methods.We apply VSM in combination with the use of computer simulation to measure flexibility objectively.To this concern, we develop a set of metrics for the quantitative evaluation of performance improvement.

Computer Simulation (CS)
This study builds on Computer Simulation (hereafter CS) to assess the benefits arising from the above mentioned LM techniques.Even in presence of widespread knowledge on the benefits that LM can bring, it is essential to quantitatively estimate these benefits in order to compare them to the costs induced by any modifications in the manufacturing system.As anticipated before, simulation analysis can notably complement qualitative approaches such as VSM (Jeong & Phillips, 2011;Marvel & Standridge, 2009;McDonald et al., 2002;Xia & Sun, 2013), to get approval from the company direction.Therefore, CS can be central to favoring the adoption of LM (Detty & Yingling, 2000).

Case Study
As said, ALFA was purposively selected as this firm well fits in the paper's scope due to the following reasons: a) It is a small manufacturing company in a niche market, competing with medium-and big-sized rivals; b) Managers indicate that flexibility is relevant for ALFA, but not easy to achieve since departments that are in charge of each stage of the manufacturing process often take their decisions in isolation; c) Managers consider the capacity to adapt to the market a critical asset to be enhanced; d) Managers declared they would have supported this study providing us with manifold, even confidential data.

Case Study Description
Since the '50s, the primary business of ALFA is the decoration of glassware by screen printing.In the beginning, ALFA mostly focused on the local market, but time after time it gained a good reputation for its crafting abilities that, in combination with cutting-edge technologies, make the difference in this kind of business.This is the reason why around 50% of revenues are currently from sales to foreign countries.
In the last years, however, ALFA is facing a downtrend due to the unfavorable economic situation in the Italian consumer market.In addition, it is suffering the growing competition, especially on foreign markets, on which its shares are slightly but continuously reducing.As the overall sales decrease, the demand for customized products tends to increase.In sum, ALFA was typically looking for a way to obtain flexibility gains without raising its baseline costs (Madrid-Guijarro, García & van Auken, 2009;Rose et al., 2011).Managers agreed that LM could be the straightforward way towards this result.

Data Collection Methodologies
To increase triangulation (Voss et al., 2002), we collected empirical data from several sources.We interviewed different managers of the production department, carried out field surveys, analyzed production databases and datasheets.We also directly observed how the production process was run.
Having little data on production and setup times, we also collected field data and validated them with the managers.
Table 2 summarizes the time we were engaged in field activities and provides an overview of the efforts required for future analysis of similar small-medium manufacturing facilities.The next section describes how ALFA produces its products.

Manufacturing Process
The company sells its products by catalog, which includes a significant number of raw glass items.
These objects can be variously decorated, taking into account manifold request for customized designs.Starting from a few dozens of base materials, the adoption of several different decorations produces an overall product line of more than 200 finished products.On average, ALFA produces +10,000 items per month, 30% having customized decorations.This study focuses on the set of product families in the ALFA catalog that delivers most sales and production volumes.The standard production process develops along different phases.Production starts with quality control of raw materials.Once raw items have been visually controlled, they are grouped into lots that are moved through fork-lifts trucks to intermediate buffers, located near the production cell.As soon as the cell capacity is available, the assigned operator setups the printing machines, first making adjustments on the basis of the item's sizes, then installing the serigraphic frame requested by the first printing layer.
Product decoration requires 2 or more printing layers.Printed goods are then transferred to the heating units (i.e.warehouses and ovens for respectively pre-heating and heating).The 6 conventional ovens available in the factory are sized to process one lot of products per time.The heating parameters -temperature and time -are adjusted according to the item's weight and shape.As soon as the treatment is completed, the batch is stocked in another buffer, waiting for the polishing unit.
After polishing, items are ready for additional silk-screen printings, quality control, and final packaging.Currently, the production facility works on a single 8-hour shift, 5 days per week.
Production batches are scheduled on a daily basis, following customer's orders as well as the need for replenishing inventories of the factory outlet, which is close to the production facility.The described production flow has been mapped according to Rother and Shook (2003) and represented in Figure 2, that also shows the departments of the production facility.The current manufacturing performances of ALFA are summarized in

Simulations Analysis
To estimate the flexibility gains coming from the introduction of LM, we developed a simulation model of the described process, on which we performed our analysis.Assumptions and objectives of this simulation study are illustrated in the following section.

Predicted Performances
In this study, we simplify the model from Suarez, Cusumano, and Fine (Suarez, Cusumano & Fine, 1996), and focus on Volume Flexibility (hereafter VOLF) and Mix Flexibility (hereafter MIXF), that drive most of the overall MF (Hallgren & Olhager, 2009;Metternich, Böllhoff, Seifermann & Beck, 2013).To estimate and compare the flexibility performance of different manufacturing systems, we developed two process parameters that ease the characterization of Manufacturing Volume and Manufacturing Mix: Volume Indicator (hereafter VI, Equation 1), Mix Indicator (hereafter MI, Equation 2).To provide a proper definition of these indicators, the following parameters must be defined: CPL is the Current Production Level that the system is required to produce, namely the number of batches produced each day; MPC is the Maximum Productive Capacity that a given system arrangement can achieve assuming a random productive sequence; QIP is the average Quantity of Identical Products' batches manufactured in series during the process. (1) (2) In the current situation, the minimum acceptable level of production for ALFA is about the 50% of maximum capacity (VI = 50%) and the best optimization of the sequence is 2 identical lots in series (MI = 2).
The vertices of the operating window, which are the extreme conditions of process parameters, are presented in Table 6.Building on these parameters, we measured the flexibility of different configurations according to the following 5-step approach shown in Figure 3 and described below.
Figure 3. 5-step Approach Step 1. Calculation of maximum productive capacity In the first step, we calculate the Maximum Productive Capacity (hereafter MPC) of the system via software.It represents the simulated capacity of the system reacting to the following inputs: immediate request of 30 batches of each product in the catalog; MI = 1.This value is strongly dependent upon the system configuration and defines the production levels for the next simulations.
In facts, two levels of volume rate are tested with the software model: 100% and 50% of maximum capacity.
Step 2. Generation of the production sequences The combination of different levels of VI and MI generates the 4 operating points declared in Table 6.
We generate 4 production sequences based on each of these points to be used as input of the simulation models.The production sequences consist of 5 days of warm-up, actual simulation of 600 days (30 samples of 20 days).
Step 3. Simulation of the system The Arena Simulation is the software used for this study.We test a specific system configuration considering the four simulations, each one completed considering the four sequences previously generated in Step 2. The adoption of the factors in Table 4 is commented below: The SMED technique is interpreted in the context of this research as a method to reduce the setup time of the print stations, the setup time of the frames for the screen printing, and the required handlings for the controls of raw materials.The adoption of this technique is linked to the investments needed to modernize tools and equipment, currently used in the operations.Figure 4 shows the VSM updated considering a 50% reduction of times related to changeover and control activities.This value is in line with the expectation of ALFA's management.The performances of ALFA operating in the SMED configuration are summarized in The data presented in Table 5 show a 7.9% decrease in WIP.The beneficial effect of SMED on inventory has been also evaluated with a different assumption: 75% decrease in changeover and control activities.In this case, the decrease in WIP is 11.7%.The adoption of JITds is based on the principle that raw materials must be received at the exact moment in which they are required, unlike the current situation whereby they are received once a day.
The use of such a system entails that the supply chain is able to satisfy this requirement.Otherwise, the possible reduction in the inventory would be simply transferred from a buffer inside the production system to an external warehouse.Figure 5 shows the VSM considering the introduction of JITds: The performances of ALFA operating in the JITds configuration are summarized in The restructuring of the layout according to the concept of CELLMFG involves a major reorganization of functions.The factory is no more divided into departments but organized into individual, self-sufficient production units including a dedicated printing station and an oven.The transition requires investments for the adjustment of the layout, the training of the personnel, who will need to play multiple roles, and a period in which the facility must operate at a reduced capacity to allow the restructuring.
Figure 6 shows the VSM considering the introduction of CELLMFG:

Case Study Results
This section discusses the findings from our simulation experiments with respect to the research question presented in section 3.1.
How LM and MF are connected in the SME context.
We find that LM techniques have different impacts on the operational performance of the plant.These effects are quantified in Table 9.Looking at the first columns in Table 5 the effect on CR, is evident for all techniques -CR Avg.values are all negatives and, in absolute value, 1.6% for SMED, 0,1% for JITds, and 1.0% for CELLMFG -.
SMED, as expected, is the most useful.In fact, the setup times are the necessary No Added Value operations, which mostly penalize the flexibility in this context.
With reference to the effect on IR, the analysis on differences between the WIP at 100% and at 50% capacity is statistically complex because of the high fluctuation of the inventory itself (very high standard deviations are associated with the metrics).Due to the high reductions of the parameters, it is anyway clear that the contribution of JITds and CELLMFG to the maintaining of an inventory level, adequate to the production level is positive -IR Avg.values are all negatives and, in absolute value, 31.19% for JITds, and 58.60% for CELLMFG -.In particular, JITds seems to be the most effective technique, able also to reduce the variability in WIP.
Observing the coefficients related to UR, it is possible to conclude that the CELLMFG is the methodology to be used for minimizing unproductivity.Being the highest value in absolute value -UR value is 33.1% for CELLMFG -, in fact, the division of the system into independent sub-units grant high performances, both at high production levels and when the market demand is low.
-937-Journal of Industrial Engineering and Management -https://doi.org/10.3926/jiem.2119 According to the obtained responses data, LM can be considered a good strategy to increase MF as it contributes to equalize the system performances under varying boundary conditions (production mix or volume).
The present study shows that the benefits achieved by the use of SMED, JITds, and CELLMFG have strong implications also on the flexibility of production systems.
For instance, reducing the setup time by using SMED, the ratio between the value-added operations and the overall cycle time increases.Consequently, the general improvements obtainable in this case study through the Single Minute Exchange of Die are a 7% increase in production.The decrease in tooling is particularly important when a high Mix Flexibility is required.Moreover, the simulation shows that the difference in productivity between a low-mix and a high-mix scenario drops from the 37% of the basic arrangement to the 35% of the SMED configuration.Based on this, the SMED can be considered a good solution for manufacturing systems where the production of batches with variable size is required.
The higher is the investment on these techniques, the higher are the percentage benefits.
The Just-in-Time is commonly used for a general reduction of the inventory.The analysis of the company ALFA shows that a daily supply of raw materials produces a high level of WIP.Simulating an advanced replenishing logic by JITds, the stock levels drop significantly -about 30% -.The reduction in inventory is more pronounced when the level of production, and therefore the daily storage of raw materials, is high (Perceptual decrease is almost constant).This phenomenon contributes to flatten the WIP stock, and, thus, increases the Volume Flexibility.
Lastly, CELLMFG is a production strategy that consists in a division of the plant into self-sufficient units in order to facilitate operations.This method, which is often associated with an increased sharing of tasks and responsibilities, allows for a better use of resources.The simulations indicate that the reorganization of production in cell would allow almost the same level of production with less human resources (one operator less).The use of CELLMFG has the advantage to permit the disabling of production units without affecting the utilization of the resources.This feature is particularly useful in the case of high variation in production Volumes.The results of the simulations clearly show an almost constant utilization of employees by varying production levels -about 83% for an optimized sequence of products and 66% for random lists -, with a great effect on the volume flexibility.

Table 2 .
Efforts and references of field activities

Table 7 .
Manufacturing Performances of JITds configuration

Table 8 .
Manufacturing Performances of CELLMFG configuration

Table 9 .
Summary of Effects