A Max-Plus Algebra Approach to Study Time Disturbance Propagation within a Robustness Improvement Context

Martínez-Olvera, César; Mora-Vargas, Jaime

doi:https://doi.org/10.1155/2018/1932361

Mathematical Problems in Engineering

On this page

Abstract Introduction Literature Review Conclusions Appendix Data Availability Conflicts of Interest References Copyright Related Articles

Research Article | Open Access

Volume 2018 | Article ID 1932361 | https://doi.org/10.1155/2018/1932361

A Max-Plus Algebra Approach to Study Time Disturbance Propagation within a Robustness Improvement Context

César Martínez-Olvera¹and Jaime Mora-Vargas²

Academic Editor: Alberto Campagnolo

Received14 Sept 2018

Revised03 Nov 2018

Accepted29 Nov 2018

Published20 Dec 2018

Abstract

Industry 4.0 aims to ensure the future competitiveness of the manufacturing industry, where one of the major challenges faced by its implementation is the manufacturing/production system robustness (that is, able to perform in the presence of noise), as they may not be able to absorb input disruptions without bending or breaking. In this paper we propose to use the Max-Plus algebra approach to study the propagation of manufacturing disturbances (i.e., processing time variations), presenting a case study and performing a sensitivity analysis, with the idea of understanding under which conditions disturbance propagation takes place. Findings show that the impact propagation depends on where the variation source is located within the manufacturing system. Two are the main original contributions of this paper: the use of Max-Plus algebra to study the impact propagation of processing time variations and a four-step methodology to derive the equations representing the deterministic manufacturing system.

1. Introduction

As e-commerce has changed the distribution process, i.e., delivery distances are decreasing, the product variety is increasing, and more products are being sold in smaller quantities, the current supply chain needs to be rethought [1]. According to [2, 3], Industry 4.0 technologies will contribute to the more robust functioning of supply chains, by allowing a quicker and more reliable recognition of the potential external and internal disruptions and disturbances and the minimization or avoidance of their negative consequences. Different authors, i.e., [2–12], identify these value creation enabler technologies of Industry 4.0 as the Internet of Things, Big Data Analytics, and Cyber-Physical Systems. The real time Internet-networking of a growing number of physical objects with intelligent sensor/actuator technology is triggering the next stage of the industrial revolution known as Industry 4.0 [13]. Industry 4.0 aims to radically combine manufacturing, automation, and information and communications technology (ICT), in order to ensure the future competitiveness of the manufacturing industry [11, 14]. This requires the horizontal integration of the intelligent digital networking from different companies, with vertical integration of the autonomous, rules-based decision-making, and performance management within a single company, based on Big Data analysis [8], to create organizational transparency and seamless production [13]. More specifically, Industry 4.0 involves the networking and computerization of industrial production areas, in order to organize and optimize their operations, via the real time coordination of the production stages within value creation network, so they can boost customer satisfaction by producing customized products at serial production price [12, 15].

2. Literature Review

2.1. Industry 4.0 Reliability, Robustness, and Resilience

On the other hand, one of the major challenges faced by an Industry 4.0 implementation is the manufacturing/production system reliability, robustness, and resilience, that is, that seldom fails, able to perform in the presence of noise, that remains elastic under changing load, and that autonomously recovers from fault or failure situations [14]. Many manufacturing/production systems are not resilient to disruptions of their normal operations, as plenty of operations management techniques depend on assumption of stable and predictable environment, so their transforming processes are tuned in such way that they may not be robust enough to absorb input disruptions without bending or breaking [16–18]. References [19, 20] state that manufacturing/production systems need to be planned to be robust and resilient enough to maintain their basic properties and ensure execution and be able to adapt their behavior in the case of disturbances in order to achieve planned performance with the help of actions for recovery. Reference [21] summarizes this last as being resistant to different kinds of disturbances and able to react properly to them, while [22, 23] state that, in order to assure the effectiveness and efficiency necessary to keep a permanent high performance level, a manufacturing/production system must exhibit well behaved dynamics, that is, responsiveness and robustness. Now, robustness and resilience do not necessarily coexist, but both of them provide manufacturing/production systems with a sustainable competitive advantage in the global marketplace [24]. Even though robustness has been used interchangeably with resilience, they present some basic differences [25–27]:(i)Robustness, defined as the ability of a system to perform its intended function and retaining its original level of performance while sustaining some damage (reasonable variations/disturbances) without failure. Robustness is a crucial feature of a reliable manufacturing/production system [21].(ii)Resilience, defined as the ability of a system to adapt to major variations/disturbances and gradually return to its original level of performance (or migrate to desired different one).(iii)Reliability, defined as the probability that a system will be able to perform its function without failing for a specific time period, under certain operating conditions.

According to [28, 29], robustness is a static concept (the system can resist disruptions and retain its previous stable state), while resilience is a dynamic concept (the system can survive a disruption and return to a new different stable state). Finally, Table 1 shows the relationship between these robustness, resilience, and reliability, based on the work presented by [30–32].

2.2. Robustness, Resilience, and Production Disturbances Propagation

Robustness of a manufacturing/production system should aim for both a target value of the process outcome and a stable or consistent performance with minimum deviation or variation [33], while its resilience, to cope with disturbances and maintain their original state [28, 29, 34–36]. Now, robustness is related to the propagation by a given disturbance [37], so a realistic manufacturing/production system’s robustness study requires identifying each disturbance’s specific cause (unexpected/unwanted event) and effect (substantial deviation from a planned quality, quantity, time, cost value deviation), in terms of quantity of events and manufacturing downtimes they cause [21]. Internal/external industrial disturbances interfere not only with efficiency but also with the lead time of the manufacturing/production system [38], as their occurrence often leads to undesired perturbations degrading the system’s performance [21, 31, 32, 39–41], so understanding their dynamics is of paramount importance [22, 23]. According to [37], manufacturing disturbances can be classified as upstream, internal, and downstream disturbances, and more likely to propagate in highly networked manufacturing environments [38, 42]. Reference [43] refers to this propagation as ripple effect, that is, the failed nodes transfer the failure to their downstream neighbor nodes through connectivity links, with this process continuing through the entire supply network. This propagation of manufacturing disturbances affect the behavior of the production network [23], and their impact on the shop floor, the most sensitive part of a manufacturing system and where customer value is created [38], depends on the associated consequences derived from their propagation [31, 32, 39]. For this reason, authors like [44–47] have proposed models that try to predict the effects of manufacturing disturbance propagation (in this case, dimensional variation propagation), representing the disturbance propagation in terms of state-time equations (of some sort). If we take into account that (i) robustness and resilience are necessary to respond to disruptions [24], (ii) the topology of the manufacturing system affects its robustness and resilience [48], and (iii) in order to increase its robustness, a manufacturing system must be modelled and the disturbances’ negative effects studied [21], it becomes evident the need of a tool that facilitates the modelling of the propagation of a disturbance, from the manufacturing system structure point of view. For this reason, in this paper we propose to investigate the feasibility of using state-time equations to study, the propagation of manufacturing disturbances (i.e., processing lead time variation) downstream the production network. With this idea on mind, next section presents the literature review of relevant topics. Section three presents a case study and the evolution equations derived from the Max-Plus algebra approach are put to the test in Section four. Section five presents the conclusions derived from the research findings and establishes future research venues

2.3. Max-Plus Algebra and Modelling of Manufacturing Systems

Manufacturing/production systems can be modelled from a qualitative or quantitative point of view [49]: qualitative models (i.e., Automata, finite state machines, and Petri Nets) capture logical aspects of a system, while quantitative models (i.e., discrete-event simulation and Max-Plus algebra) highlight the quantitative system performance. Regarding these last, according to [50], discrete-event simulation models are dynamic in nature and are often easier to apply than analytic models, while Max-Plus algebra models are static in nature and express the event timing dynamics in terms of a set of algebraic linear equations analogous to conventional state-space linear equations [51]. However, even though discrete-event computer simulation is a widely used technique to analyze a manufacturing system [52], an important drawback is that it does not supply equations needed to analyze system’s behavior, making the understanding of how parameter changes affect important system properties such as stability, robustness and optimality of system performance very difficult [53, 54]. On the other hand, state-time dynamic models are an effective method to make quantitative analysis and evaluation of manufacturing systems, as it allows analyzing the influence of random disturbance factors on the system’s outputs [22, 23], so regarding this last issue, Max-Plus algebra can provide a much better insight than discrete-event simulation [52]. Max-plus algebra is an effective tool for modelling the event timing dynamics of a deterministic discrete-event system like a manufacturing line or a job shop manufacturing system [55–57]. According to [58], research related to Max-Plus algebra has been done in both the areas of developing the tool and the applications of the tool itself. A complete detailed description and analysis of the mathematics behind Max-Plus algebra can be found in [59, 60], a review of the basic concepts in [51–54, 61, 62]. A review of the Max-Plus algebra tool and its applications can be found in [58] and some of the work done on the modelling of manufacturing systems in [53, 54, 63, 64]. In summary, in Max-Plus algebra, the maximum operator (denoted by ⊕) replaces the addition operator, and the addition operator (denoted by ⊗) replaces the multiplication operator. The ⊕ called operation has a neutral element ϵ called zero; the ⊗ operation has a neutral element e, called unity. For all a, b ∈ real numbers R ∪ −∞ the Max-Plus operators are defined according to the following [49]:If we take into account the fact that state-space models are an effective method to quantitatively analyze the effect of manufacturing disturbances [23] and that Max-Plus algebra provides linear state-space-like equations to model manufacturing systems [51], then it becomes natural to follow a Max-Plus algebra approach to study the effects of manufacturing disturbances propagating downstream the production network. In fact, [58] mentions that an useful application of Max-Plus algebra equations is in not only evaluating the effect of variation in processing times on the total line idle time, but also possible performance measures as throughput, throughput time, and stations utilization. Next section presents a brief tutorial of the basic concepts of Max-Plus algebra and introduces the manufacturing system used as case study.

3. Case Study

In order to introduce the use of Max-Plus algebra approach for studying the propagation of disturbances in a manufacturing system environment, we will use the scenario presented in Figure 1, based on the one mentioned by [59]. Following is the notation used in further analysis made throughout the rest of the document:(i): variable denoting a certain machine type. In example, machine type 3 is denoted as M3.(ii): variable denoting a certain part type. In example, part type 1 is denoted as P1.(iii); variable denoting a part’s routing, that is using for a certain amount of time. In example, Part P2 routing is denoted as P₂M₁ (1 min), P₂M₂ (2 min), P₂M₃ (3 min).(iv)p.a.s.: variable denoting a part’s availability sequence, that is, the sequence of machines a part will make use. In example, the p.a.s. for P1 is M1 – M3 – M2 – M1 – M3.(v)m.a.s.: variable denoting a machine’s availability sequence, that is, the sequence of parts a machine will process. In example, the m.a.s. for M3 is P2 – P1 – P2 – P1.(vi)X_i: variable denoting the position of each has on its corresponding p.a.s.(vii) (k): variable denoting the starting time of X_i for the first unit of to be processed.(viii) (k+1): variable denoting the starting time of X_i after a previous unit of has been processed.(ix) (k+1): variable denoting the initial arriving time of part to the manufacturing system.(x): variable denoting the cycle time of part , that is, the amount of time it takes from its arrival until its departure from the manufacturing system.(xi): variable denoting the utilization of machine , that is, the ratio between the actual time machine being busy processing units of part and the actual time machine being available for processing units of part .(xii)w.t.: variable denoting the waiting time a unit of part , that is, the amount of time it takes from its arrival to machine until it starts being processed by machine .(xiii)m.a.t.: variable denoting the machine availability time, that is, the time at which certain machine becomes available to process the next unit of part of its machine’s availability sequence (m.a.s.).(xiv)p.a.t.: variable denoting the part availability time, that is, the time at which a certain part becomes available to be processed by the next machine on its part’s availability sequence (p.a.s.).

As stated before, Figure 1 shows the sequence of machines each part follows, denoted as part routing (that is, using ), where columns represent and rows represent . The processing times per part type are as follows:(i)Part P1: M2 (3 min), M3 (4 min).(ii)Part P2: M1 (1 min), M2 (2 min), M3 (3 min).(iii)Part P3: M1 (5 min), M2 (3 min).

where transportation and set-up times are negligible, and all processing times are deterministic in minutes.

Now, according to [56] there are four methods to derive Max-Plus algebra models for a manufacturing system:(i)Timed event graph or a directed graph representation of the system [49, 55].(ii)Max-plus linear queuing networks [57].(iii)The system specifications.(iv)Block diagrams [53, 54, 56].

Reference [63] claims that obtaining the evolution equations from a timed event graph can be cumbersome, and, for that reason, he presents the following recursive formulation:whereNow, in this paper we are more interested in the practical application of this modelling approach, so the industrial practitioner can truly exploit the benefits of it, independently the level of knowledge of the mathematics behind Max-Plus algebra. With this idea on mind, we present an easy-to-follow four-step algorithm to derive the evolution equations of the manufacturing system.

3.1. Proposed Algorithm to Derive the Evolution Equations

In order to exemplify the benefits of the proposed algorithm, we will use the following reentrant manufacturing system (next section will highlight the features of such type of system):(i)Part P₁: M₁ (t1), M₃ (t3), M₂ (t2), M₁ (t1), M₃ (t3).(ii)Part P₂: M₃ (t3), M₂ (t2), M₁ (t1), M₃ (t3), M₁ (t1).

Step 1. For each , establish its sequence (in a column format) and label each (from left to right, top to bottom) in terms of X_i. This will be known as the part availability sequence, p.a.s. (Figure 2).

Step 2. Based on the previous step, for each , identify the sequence (from left to right, top to bottom) in terms of its corresponding X_i. This will be known as the machine availability sequence, m.a.s. (Figure 3).

Step 3. Based on Steps 1 and 2, build the m.a.s./p.a.s. sequencing matrix (Table 2). For example,(i)P1 p.a.s. has X1 as the element in the sequence (corresponding to M1), followed by X3 as the (corresponding to M3), X5 as the (corresponding to M2), X7 as the (corresponding to M1), and X9 as the (corresponding to M3)(ii)P1 m.a.s. has X1 as the element in the sequence, followed by X6 as the , X7 as the , and X10 as the , all of them corresponding to M1

Step 4. Put together the evolution equations for each corresponding X_i (see (5)–(14)), where the first element of each X_i evolution equation corresponds to a m.a.s. relationship sequence, and the second element to a p.a.s. relationship sequence. It must be highlighted that any n+1 => n relationship is denoted as (k), and any n => n+1 is denoted as (k+1).

3.2. Benefits of the Proposed Algorithm

According to [58], the main difficulty in modelling reentrant manufacturing systems using Max-Plus algebra is the main underlying assumption that the system (to be modelled) must be representable by a decision free, Timed Event Graph. This means, in Petri nets terms, that there is only one upstream and one downstream transition for each place. This is something a reentrant manufacturing system clearly does not comply with. Moreover, the same author presents the following reentrant manufacturing system:(i)Part P₁: M₁ (operation A), M₂ (operation B), M1 (operation C).

And the corresponding recursive formulation is whereAs we can see, for a very simple reentrant manufacturing system (one product – one machine used twice), the Max-Plus algebra involved can be cumbersome to the common average industrial practitioner, where, on the other hand, the proposed four-step methodology allows the depiction of fairly more complex reentrant manufacturing systems (two products – two machines used twice). For this reason, we consider the proposed methodology as a value-adding original contribution of this paper.

3.3. Simulation Study

As we are interested in studying the impact of varying the processing time (at each ,) has on the rest of the manufacturing system, we apply the recursive formulation (see (3)) or use the four-step algorithm proposed in the previous section to obtain evolution equations of the manufacturing system as follows:Table 3 shows the starting time of each product at each machine , for a total processing of fifteen units per (that is, fifteen units of P1, P2, and P3, respectively). Table 4 shows the seven different tested scenarios, consisting of varying the processing time of each , one at the time, while maintaining the original processing times of rest of the . For example, Scenario 1 consists of varying t1 from 1 to 5, while maintaining unchanged the original processing times of t2 through t7. Moreover, in order to test the validity of (17)–(23), a discrete-event simulation model of the case study was developed using the simulation software ARENA [65], validated, and used to collect sample data, following an approach similar to [66]. The running conditions of the seven different tested scenarios were the following: fifteen units per were simulated; a new unit of can enter the manufacturing system, only after the previous unit of departed already; the running time was long enough to allow all the fifteen units per to be processed and depart from the manufacturing system.

The impact of varying the processing time (at each ) has on the rest of the manufacturing system was evaluated in terms of he following:(i)Cycle time () of each : Table 5 shows the calculation of the (for the case of Scenario 1 with t1 = 1), based on Table 3. For example, Ct₁ = starting time of P₁M₃ + processing time t6 – starting time of P₁M₂; Ct₂ = starting time of P₂M₃ + processing time t7 – starting time of P₂M₁; Ct₃ = starting time of P₃M₂ + processing time t5 – starting time of P₃M₁, and so on.(ii)Utilization () of each : Table 5 shows the calculation of the (for the case of Scenario 1 with t1 = 1), based on Table 3. For all three , the utilization was calculated by adding the processing time of each using the same machine, multiplied by the fifteen processed units, and divided by the highest departure time of the processed unit. For example, in Scenario 1 with t1 = 1, M₁ consumed 6 minutes (t1 = 1 + t2 = 5), for a total of 90 minutes (15 6), and was divided by 143 (starting time of X7 = 140 plus t7 = 3), for a utilization of 90/143 = 0.629.(iii)Waiting time (w.t.) before each : Table 6 shows the calculation of the w.t. (for the case of Scenario 1 with t1 = 1) before each , according the following rules:(a)XI; IF t7 X7 (K+1) ≥ t2 X2(K+1) THEN w.t. = 0 ELSE w.t. = t2 X2(K+1) - t7 X7 (K+1).(b)X2; IF t5 X5 (K+1) ≥ t1 X1(K+2) THEN w.t. = 0 ELSE w.t. = t1 X1(K+2) - t5 X5 (K+1).(c)X3; IF t6 X6 (K+1) ≥ t5 X5(K+1) THEN w.t. = 0 ELSE w.t. = t5 X5(K+1) - t6 X6 (K+1).(d)X4; IF t1 X1 (K+1) ≥ t3 X3(K+1) THEN w.t. = 0 ELSE w.t. = t3 X3(K+1) - t1 X1 (K+1).(e)X5; IF t2 X2 (K+1) ≥ t4 X4(K+1) THEN w.t. = 0 ELSE w.t. = t4 X4(K+1) - t2 X2 (K+1).(f)X6; IF t3 X3 (K+2) ≥ t7 X7(K+1) THEN w.t. = 0 ELSE w.t. = t7 X7(K+1) - t3 X3 (K+2).(g)X7; IF t4 X4 (K+1) ≥ t6 X6(K+1) THEN w.t. = 0 ELSE w.t. = t6 X6(K+1) - t4 X4 (K+1).

Regarding Table 5, the row with the calculation presents two extra values, i.e., for Ct₂, the 9.4 and 9.53 values. The first value refers to the average Ct₂ value obtained from Table 5 and the second value from the simulation run output. For both cases, the results are practically the same. The same applies to the calculation. Regarding Table 6, the row with the w.t. calculation presents two extra values, i.e., for X2 and t2 = 5, the 1.5 and 1.466 values. The first value refers to the average w.t. value obtained from Table 6 and the second value from the simulation run output. For both cases, the results are practically the same.

3.4. Experimental Results

Figure 4 shows the impact of varying the processing times of P₁M₁, from t1 = 1 to t1 = 5, have on the waiting times right in front of P₃M₁. Figures 5(a)–5(g) show a similar analysis for the tested scenarios presented in Table 4.

(a) Scenario 1

(b) Scenario 2

(c) Scenario 3

(d) Scenario 4

(e) Scenario 5

(f) Scenario 6

(g) Scenario 7

Three things must be noted regarding Figures 5(a)–5(g):(i)Each is located according to the layout presented in Figure 1.(ii)The is presented at the end of each column, and the U at the end of each row.(iii)All the m.a.t. relationships are always presented on the horizontal direction, as each row represents a certain machine .(iv)All the p.a.t. relationships are always on the vertical direction, as each column represents a certain part .

Figure 6 shows an analysis of the impact of varying the processing time of P₂M₁ has on the rest of the manufacturing system (Scenario 1 of Table 4, Figure 5(a)), where such impact is typified in terms of degree link types and INC/CON effects:(i), , , and degree link type refers to the degree of closeness between the affecting (that is, the source of processing time variation) and the affected . In this sense, a degree link type means the affecting and the affected s are adjacent, where, in a link type, there is a in between the affecting and the affected s. This idea of using , , , and connectivity degree links is taken from the supply network risk propagation model proposed by [43].(ii)INC (increase) and CON (constant) effect refers to the effect the affecting has on the m.a.t/p.a.t. of the affected and the value at which these last start increasing, or the value at which these last remain constant.

The analysis of Figure 6 shows the following:(i)X1 (P₂M₁): its m.a.t increases due to a degree link from P₃M₁; its p.a.t. remains constant (and then starts increasing) due to a degree link from P₂M₃. This results in the pattern A of Figure 7.(ii)X2 (P₃M₁): both its m.a.t and its p.a.t. increase due to a degree link from P₂M₁ and a degree link from P₃M₂. This results in pattern B of Figure 7.(iii)X3 (P₁M₂): its m.a.t. increases due to a degree link from P₃M₂; its p.a.t. remains always constant due to a degree link from P₁M₃. This results in the pattern C of Figure 7.(iv)X4 (P₂M₂): its m.a.t. remains constant due to a degree link from P₁M₂; its p.a.t. increases due to a degree link from P₂M₁. This results in the pattern D of Figure 7.(v)X5 (P₃M₂): its m.a.t. remains constant (and then starts increasing) due to a degree link from P₂M₂; its p.a.t. increases due to a degree link from P₃M₁. This results in the pattern E of Figure 7.(vi)X6 (P₁M₃): its m.a.t. remains constant (and then starts increasing) due to a degree link from P₂M₃; its p.a.t. increases due to a degree link from P₁M₂. This results in the pattern F of Figure 7.(vii)X7 (P₂M₃): its m.a.t. remains constant due to a degree link from P₁M₃: its p.a.t. remains constant due to a degree link from P₂M₂. This results in the pattern G of Figure 7.

3.5. Results Analysis and Derived Conclusions

The previous analysis seems to suggest that the further a certain affected is from the affecting P₂M₁, the less impact its varying processing time has on it. As the ultimate objective of this research effort is to study the conditions under which the manufacturing disturbance propagation takes place, we performed a similar analysis on the seven tested scenarios of Table 4 (Table 7 presents a summary of the results obtained from this analysis, while the Appendix presents the full details of the results obtained). For this matter, we defined the type of impact the affecting has on the m.a.t. and p.a.t. of the affected : primary (TOT), the m.a.t./p.a.t. of the affected increases; secondary (PAR), the m.a.t./p.a.t. of the affected remains constant and then increases; or null (NON), the m.a.t./p.a.t. of the affected remains constant. Also, we classify each affected in terms of its location: level I, for X1 (P₂M₁) and X2 (P₃M₁); level II, for X3 (P₁M₂), X4 (P₂M₂), and X5 (P₃M₂); level III for X6 (P₁M₃) and X7 (P₂M₃). An analysis of the results presented in Table 7 reveals the following facts:(1)Independently at which level the affecting is located (i.e., I, II, and III), it has a primary impact on the m.a.t. of its immediate affected s ( degree link), 85.71% of the times. From here, its frequency starts decreasing: 21.42% for a degree link, 11.76% for a degree link, and 9.09% for a degree link. Moreover, 50% of the times there is a primary impact on an affected s, is due to a degree link affecting , and only 25% of the times from degree link, 16.66% from a degree link, and 8.33% from a degree link. A similar situation presents the p.a.t. case. These results lead us to conclude that the primary impact from the affecting is mostly contained by the immediate, degree link affected s.(2)Independently at which level the affecting is located (i.e., I, II, and III), it has a secondary impact on the m.a.t. of its immediate affected s ( degree link), 14.28% of the times. From here, its frequency increases (42.85% for a degree link) and then decreases again (23.52% for a degree link). Moreover, 14.28% of the times there is a secondary impact on an affected s, is due to a degree link affecting , 42.85% of the times from degree link, and 23.52% from a degree link. A similar situation presents the p.a.t. case. These results lead us to think that there must be some residues of the primary impact (absorbed by the degree link affected s) that keeps propagating further. This issue is reflected on the fact that the secondary impact affecting the degree link affected s is almost three times higher than the frequency of the secondary impact affecting the degree link affected s (42.85% versus 14.28%).(3)Independently at which level the affecting is located (i.e., I, II, and III), it has a null impact on the m.a.t. of its immediate affected s ( degree link), 0% of the times. From here, its frequency starts increasing: 35.71% for a degree link, 64.70% for a degree link, and 90.9% for a degree link. Moreover, 0% of the times there is a null impact on an affected s, due to a degree link affecting , 19.23% of the times from degree link, 42.30% from a degree link affecting, and 38.46% for a degree link. A similar situation presents the p.a.t. case. As these results go on the opposite direction of those regarding the frequency of primary impact, and the fact that a null impact is the complement of a primary impact, we arrive to the same conclusion already stated: the primary impact from the affecting is mostly contained by the immediate degree link affected s.

Figure 8 shows the same results of Table 7, where Series1 refer to the TOT impact type, Series2 to NON impact type, an Series3 to the PAR impact type; , , , and refer to the connectivity link type.

3.6. Implications for Robustness Improvement

According to [1, 19, 20, 37], the improvement of robustness in a production system is provided by the presence of inventory (to make up production short-falls) and capacity buffers (slack available at resources). On the other hand, [67] mentions that, in order to improve the robustness of a production/supply chain network, its topology needs to be modified. Author in [68] expands this idea when he states that a dynamic network reconfiguration bring robustness, as new machines can be plugged to provide redundancy. Reference [37] claims that the robustness of the system depends on the resource with the minimum ability to recover. Reference [40] proposes a dynamic network of the manufacturing system, where this critical resource is the one with the highest correlation between performance and system disturbance. For [43], there is the node that provokes two or more downstream nodes to be disrupted. It is the authors belief that, within a context of a dynamic reconfiguration of the network topology, typical of an Industry 4.0 environment, where the production system reconfiguration takes place through the easy integration of plug-and-produce, fully automated, digitized, highly cost-efficient, smart new manufacturing units [69], following an approach like the proposed in this paper would allow(i)identifying under which conditions a resource will become critical(ii)establishing the appropriate stock/inventory and time/capacity buffers countermeasures to minimize it

4. Conclusions and Future Research

One of the major challenges faced by the implementation of the Industry 4.0 concept is the robustness of the manufacturing/production system, that is, the ability to absorb manufacturing disruptions without failing or breaking. In this paper, we proposed the use of a Max-Plus algebra approach to study the impact of manufacturing disturbances (i.e., processing time variations) on the performance of a manufacturing system, as they propagate through it. For this purpose, an algorithm to derive the evolution equations (of a manufacturing system) was proposed and applied to a case study. Seven different scenarios were analyzed and the findings show that, independently where the processing time variation source is, the primary disturbance effect is contained mostly by the immediate elements and the secondary disturbance effect initially increases (as it propagates) and then vanishes. Future research will consider using the Max-Plus modelling approach proposed in this paper, to identify, within a context of a dynamic reconfiguration of the network topology, under which conditions a resource will become critical, so the appropriate robustness improvement strategies can be taken.

Appendix

See Tables 8 and 9.

Data Availability

The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

References

V. Gružauskas and M. Vilkas, “Managing capabilities for supply chain resilience through it integration,” Economics and Business, vol. 31, no. 1, pp. 30–43, 2017.
View at: Publisher Site | Google Scholar
J. Monostori, “Robustness- and complexity-oriented characterization of supply networks’ structures,” Procedia CIRP, vol. 57, pp. 67–72, 2016.
View at: Publisher Site | Google Scholar
J. Monostori, “Robustness and complexity of supply chains,” in Proceedings of the INDUSTRY 4.0, LOGISTICS 4.0: Challenges and opportunities, Budapest, Hungary, 2017.
View at: Google Scholar
P. M. Swafford, S. Ghosh, and N. Murthy, “Achieving supply chain agility through IT integration and flexibility,” International Journal of Production Economics, vol. 116, no. 2, pp. 288–297, 2008.
View at: Publisher Site | Google Scholar
J. C. Chen, C. Cheng, and P. B. Huang, “Supply chain management with lean production and RFID application: A case study,” Expert Systems with Applications, vol. 40, no. 9, pp. 3389–3397, 2013.
View at: Publisher Site | Google Scholar
C. Klotzer and A. Pflaum, “Cyber-Physical Systems (CPS) in Supply Chain Management – A definitional approach,” in NOFOMA 2015 Post Conference Proceedings : Towards Sustainable Logistics and Supply Chain Management, 27th Conference of the Nordic Logistics Research Network, pp. 12–19, Molde, Norway, October 2015.
View at: Publisher Site | Google Scholar
M. Digital, How to Navigate Digitization of The Manufacturing Sector, McKinsey & Company, 2015.
R. Sauter, M. Bode, and D. Kittelberger, “How Industry 4.0 Is Changing How We Manage Value Creation,” https://www.horvath-partners.com/, 2015.
View at: Google Scholar
V. Navickas and V. Gružauskas, “Big data concept in the food supply chain: Small markets case,” Scientific Annals of Economics and Business, vol. 63, no. 1, pp. 15–28, 2016.
View at: Publisher Site | Google Scholar
R. Zhao, Y. Liu, N. Zhang, and T. Huang, “An optimization model for green supply chain management by using a big data analytic approach,” Journal of Cleaner Production, vol. 142, pp. 1085–1097, 2017.
View at: Publisher Site | Google Scholar
J. M. Müller, L. Maier, J. Veile, and K. I. Voigt, “Cooperation strategies among SMEs for implementing industry 4.0,” in Proceedings of the Hamburg International Conference of Logistics (HICL) – 23. Digitalization in Supply Chain Management and Logistics, W. Kersten, T. Blecker, and C. M. Ringle, Eds., 2017.
View at: Google Scholar
C. Schröder, “The challenges of industry 4.0 for small and medium-sized enterprises,” Expertise for the Friedrich-Ebert-Stiftung on the Project A Good Society - Social Democracy, 2017.
View at: Google Scholar
E. V. Bitkom, E. V. Vdma, and E. V. Zvei, “Implementation Strategy Industrie 4.0,” Report on the Results of the Industrie 4.0 Platform, 2016.
View at: Google Scholar
D. O. Chukwuekwe, P. Schjølberg, H. Rødseth, and A. Stuber, “Reliable, robust and resilient systems: towards development of a predictive maintenance concept within the industry 4.0 environment,” in Proceedings of the Euromaintenance 2016 Conference, vol. 24, Athens, 2016.
View at: Google Scholar
D. Ivanov, A. Dolgui, and B. Sokolov, “A dynamic approach to multi-stage job shop scheduling in an industry 4.0-based flexible assembly system,” in Advances in Production Management Systems. The Path to Intelligent, Collaborative and Sustainable Manufacturing, vol. 513 of IFIP Advances in Information and Communication Technology, pp. 475–482, Springer International Publishing, Switzerland, 2017.
View at: Publisher Site | Google Scholar
F. Rydzak, P. Magnuszewski, J. Sendzimir, and E. Chlebus, “A concept of resilience in production systems,” in Proceedings of the 24th International Conference of the System Dynamics Society, 2006.
View at: Google Scholar
F. Rydzak and E. Chlebus, “Application of resilience analysis in production systems - bombardier transportation case study,” System Dynamics Society, vol. 6, pp. 4076–4093, 2007.
View at: Google Scholar
T. J. Pettit, J. Fiksel, and K. L. Croxton, “Ensuring supply chain resilience: development of a conceptual framework,” Journal of Business Logistics, vol. 31, no. 1, pp. 1–21, 2010.
View at: Publisher Site | Google Scholar
D. Ivanov and B. Sokolov, “Control and system-theoretic identification of the supply chain dynamics domain for planning, analysis and adaptation of performance under uncertainty,” European Journal of Operational Research, vol. 224, no. 2, pp. 313–323, 2013.
View at: Publisher Site | Google Scholar | MathSciNet
D. Ivanov, A. Dolgui, and B. Sokolov, “Supply chain design with disruption considerations: review of research streams on the ripple effect in the supply chain,” Preprints of the 15th IFAC Symposium on Information Control Problems in Manufacturing, 2015.
View at: Publisher Site | Google Scholar
N. Stricker and G. Lanza, “An approach towards improving the robustness of production systems,” Advanced Materials Research, vol. 1018, pp. 461–468, 2014.
View at: Publisher Site | Google Scholar
N. A. Duffie and L. Shi, “Maintaining constant WIP-regulation dynamics in production networks with autonomous work systems,” CIRP Annals - Manufacturing Technology, vol. 58, no. 1, pp. 399–402, 2009.
View at: Publisher Site | Google Scholar
D. Tang, H. Zhang, T. Huang, and Q. Cai, “Dynamic analysis of production network from perspective of order flows,” Procedia CIRP, vol. 56, pp. 215–219, 2016.
View at: Google Scholar
Y. Kristianto, A. Gunasekaran, and P. Helo, “Building the “Triple R” in global manufacturing,” International Journal of Production Economics, vol. 183, pp. 607–619, 2017.
View at: Publisher Site | Google Scholar
P. Seferlis, D. Vlachos, E. Iakovou, and A. Xanthopoulos, “An optimal control theory-based framework for supply chain resilience,” in Proceedings of the 3rd International Conference on Manufacturing Engineering (ICMEN), Chalkidiki, Greece, 2008.
View at: Google Scholar
V. L. M. Spiegler, M. M. Naim, and J. Wikner, “A control engineering approach to the assessment of supply chain resilience,” International Journal of Production Research, vol. 50, no. 21, pp. 6162–6187, 2012.
View at: Publisher Site | Google Scholar
V. L. M. Spiegler, Designing Supply Chains Resilient to Nonlinear System Dynamics, Doctoral Dissertation, Cardiff University, 2013.
View at: Publisher Site
J. Fiksel, “Designing resilient, sustainable systems,” Environmental Science & Technology, vol. 37, no. 23, pp. 5330–5339, 2003.
View at: Publisher Site | Google Scholar
J. Fiksel, “Sustainability and resilience: toward a systems approach,” IEEE Engineering Management Review, vol. 35, no. 3, p. 5, 2007.
View at: Publisher Site | Google Scholar
J. Lee, M. Ghaffari, and S. Elmeligy, “Self-maintenance and engineering immune systems: Towards smarter machines and manufacturing systems,” Annual Reviews in Control, vol. 35, no. 1, pp. 111–122, 2011.
View at: Publisher Site | Google Scholar
O. Gluchshenko, “Definitions of Disturbance, Resilience and Robustness in ATM Context,” DLR Report IB 112-2012/28, DLR, Institute of Flight Guidance, Germany, 2012.
View at: Google Scholar
O. Gluchshenko and P. Foerster, “Performance based approach to investigate resilience and robustness of an ATM System,” in Proceedings of the 10th USA/Europe Air Traffic Management Research and Development Seminar, ATM 2013, USA, June 2013.
View at: Google Scholar
C. Giannetti and R. S. Ransing, “Risk based uncertainty quantification to improve robustness of manufacturing operations,” Computers & Industrial Engineering, vol. 101, pp. 70–80, 2016.
View at: Publisher Site | Google Scholar
C. W. Zobel and D. Cook, “A decision support framework to assess supply chain resilience framework,” in Proceedings of the 5th International Conference on Information Systems for Crisis Response and Management, ISCRAM 2008, pp. 596–605, USA, May 2008.
View at: Google Scholar
K. L. Croxton, “Ensuring supply chain resilience: development of a conceptual framework,” Journal Of Business Logistics, vol. 31, no. 1, pp. 1–21, 2010.
View at: Google Scholar
C. A. Lengnick-Hall, T. E. Beck, and M. L. Lengnick-Hall, “Developing a capacity for organizational resilience through strategic human resource management,” Human Resource Management Review, vol. 21, no. 3, pp. 243–255, 2011.
View at: Publisher Site | Google Scholar
G. Frizelle, D. McFarlane, and L. Bongaerts, “Disturbance Measurement in Manufacturing Production Systems,” in Proceedings of the Advanced Summer Institute of the Esprit Network of Excellence in Intelligent Control and Integrated Manufacturing Systems, pp. 159–162, 1998.
View at: Google Scholar
K. Knüppel and I. Nikitin, “Target-based evaluation of disturbances in production systems,” Advanced Materials Research, vol. 1018, pp. 589–596, 2014.
View at: Publisher Site | Google Scholar
P. Leitão and F. Restivo, “Predictive disturbance management in manufacturing control systems,” in Proceedings of the IMS International Forum, pp. 589–596, 2004.
View at: Google Scholar
T. Becker, M. Meyer, and K. Windt, “A network theory approach for robustness measurement in dynamic manufacturing systems,” in Proceedings of the Annual Cambridge International Manufacturing Symposium - Disruptive Supply Network Models in Future Industrial Systems: Configuring for Resilience and Sustainability, Cambridge, UK, 2013.
View at: Google Scholar
P. Garcia-Herreros, J. M. Wassick, and I. E. Grossmann, “Design of resilient supply chains with risk of facility disruptions,” Industrial & Engineering Chemistry Research, vol. 53, no. 44, pp. 17240–17251, 2014.
View at: Publisher Site | Google Scholar
H. Carvalho, V. Cruz-Machado, and J. G. Tavares, “A mapping framework for assessing Supply Chain resilience,” International Journal of Logistics Systems and Management, vol. 12, no. 3, pp. 354–373, 2012.
View at: Publisher Site | Google Scholar
L. Tang, K. Jing, J. He, and H. E. Stanley, “Robustness of assembly supply chain networks by considering risk propagation and cascading failure,” Physica A: Statistical Mechanics and its Applications, vol. 459, pp. 129–139, 2016.
View at: Publisher Site | Google Scholar | MathSciNet
R. Suri and O. Kevin, “System-level robustness through integrated modeling,” in Proceedings of the 11th International Conference on Design Theory and Methodology. ASME Design Engineering Technical Conferences, Las Vegas, Nevada, USA, 1999.
View at: Google Scholar
J. Camelio, S. J. Hu, and D. Ceglarek, “Modeling variation propagation of multi-station assembly systems with compliant parts,” Journal of Mechanical Design, vol. 125, no. 4, pp. 673–681, 2003.
View at: Publisher Site | Google Scholar
Y. Ding, J. Jin, D. Ceglarek, and J. Shi, “Process-oriented tolerancing for multi-station assembly systems,” Institute of Industrial Engineers (IIE). IIE Transactions, vol. 37, no. 6, pp. 493–508, 2005.
View at: Google Scholar
A. Das, P. Franciosa, and P. K. S. Prakash, “Transfer function of assembly process with compliant non-ideal parts,” Procedia CIRP, vol. 21, pp. 177–182, 2014.
View at: Google Scholar
Y. Shuai, X. Wang, and L. Zhao, “Research on measuring method of supply chain resilience based on biological cell elasticity theory,” in Proceedings of the 2011 IEEE International Conference on Industrial Engineering and Engineering Management (IEEM), pp. 264–268, Singapore, Singapore, December 2011.
View at: Publisher Site | Google Scholar
I. Nasri, G. Habchi, and R. Boukezzoula, “Scheduling and control modeling of HVLV systems using max-plus algebra,” in Proceedings of the Fifth International Workshop on Verification and Evaluation of Computer and Communication Systems (VECoS 2011), pp. 62–71, Tunis, Tunisia, September 2011.
View at: Publisher Site | Google Scholar
H. Ullah, “Petri net versus queuing theory for evaluation of FMS,” Assembly Automation, vol. 31, no. 1, pp. 29–37, 2011.
View at: Publisher Site | Google Scholar
A. Seleim and H. ElMaraghy, “Max-plus modeling of manufacturing flow lines. Variety management in manufacturing,” in Proceedings of the 47th CIRP Conference on Manufacturing Systems, CMS 2014, vol. 17, pp. 71–75, Canada, April 2014.
View at: Google Scholar
J. Stanczyk, M. Eckart, and J. Raisch, “Modelling and performance evaluation of DES: a max-plus algebra toolbox for MATLAB,” in Proceedings of the International Conference on Informatics in Control, Automation and Robotics, ICINCO 2004, pp. 270–275, Setubal, Portugal, 2004.
View at: Google Scholar
A. Imaev and R. P. Judd, “Hierarchial modeling of manufacturing systems using max-plus algebra,” in Proceedings of the 2008 American Control Conference (ACC '08), pp. 471–476, Seattle, WA, USA, June 2008.
View at: Publisher Site | Google Scholar
M. Algebra, A. Imaev, and R. P. Judd, “Block diagram-based modeling of manufacturing systems using max-plus algebra,” in Proceedings of the 2009 American Control Conference, pp. 4711–4716, St. Louis, MO, USA, June 2009.
View at: Publisher Site | Google Scholar
S. Lahaye, L. Hardouin, and J.-L. Boimond, “Models combination in (max,+) algebra for the implementation of a simulation and analysis software,” Kybernetika, vol. 39, no. 2, pp. 143–154, 2003.
View at: Google Scholar | MathSciNet
M. Singh and R. P. Judd, “Efficient performance evaluation of a job shop using max-plus algebra,” in Proceedings of the 2012 Industrial and Systems Engineering Research Conference, G. Lim and J. W. Herrmann, Eds., Reno, Nevada, USA, 2012.
View at: Google Scholar | MathSciNet
S. E. Z. Soudjani, D. Adzkiya, and A. Abate, “Formal verification of stochastic max-plus-linear systems,” IEEE Transactions on Automatic Control, vol. 61, no. 10, pp. 2861–2876, 2016.
View at: Publisher Site | Google Scholar
A. S. S. E. Seleim, Modeling and Performance Analysis of Manufacturing Systems Using Max-Plus Algebra, Electronic Theses and Dissertations, University of Windsor, 2016.
F. L. Baccelli, G. Cohen, G. J. Olsder, and J.-P. Quadrat, Synchronization and Linearity, John Wiley & Sons, Ltd., Chichester, New York, NY, USA, 1992.
View at: MathSciNet
B. Heidergott, G. J. Oldser, and J. van der Woude, Max Plus at Work: Modeling and Analysis of Synchronized Systems: A Course on Max-Plus Algebra and Its Applications, Princeton University Press, 2006.
View at: MathSciNet
B. De Schutter and T. V. D. Boom, “Max-plus algebra and max-plus linear discrete event systems: An introduction,” in Proceedings of the 9th International Workshop on Discrete Event Systems (WODES '08), pp. 36–42, Goteborg, Sweden, May 2008.
View at: Publisher Site | Google Scholar
P. Harvim, S. A. Gyamerah, and P. K. Boateng, “Max-plus algebra and application to matrix operations,” British Journal of Mathematics and Computer Science, vol. 12, no. 3, pp. 1–14, 2016.
View at: Google Scholar
A. Doustmohammadi, E. W. Kamen, and R. Lumia, “Direct generation of event-timing equations for generalized flow shop systems,” in Proceedings of the Photonics East '95, pp. 50–62, International Society for Optics and Photonics, Philadelphia, PA, USA, 1995.
View at: Publisher Site | Google Scholar
H. Goto, K. Shoji, H. Nagai, and M. Onuma, “On the max-plus linear representation for event-varying systems with buffer and order constraints,” in Proceedings of the 2007 IEEE International Conference on Industrial Engineering and Engineering Management, IEEM 2007, pp. 1823–1827, Singapore, December 2007.
View at: Google Scholar
D. Kellton, R. P. Sadowski, and D. T. Sturrock, Simulation with ARENA, McGraw-Hill Higher Education, New York, NY, USA, 2004.
E. Shuiabi, V. Thomson, and N. Bhuiyan, “Entropy as a measure of operational flexibility,” European Journal of Operational Research, vol. 165, no. 3, pp. 696–707, 2005.
View at: Publisher Site | Google Scholar | MathSciNet
Z. Sha and J. H. Panchal, “Towards the design of complex evolving networks with high robustness and resilience,” Procedia Computer Science, vol. 16, pp. 522–531, 2013.
View at: Google Scholar
S. Wang, J. Wan, D. Li, and C. Zhang, “Implementing smart factory of industrie 4.0: An Outlook,” International Journal of Distributed Sensor Networks, vol. 12, no. 1, pp. 1–10, 2016.
View at: Publisher Site | Google Scholar
A. Raza, L. Haouari, M. Pero, and N. Absi, “Impacts of industry 4.0 on the specific case of mass customization through modeling and simulation approach,” in Customization 4.0, Hankammer et al., Ed., Springer Proceedings in Business and Economics, pp. 217–234, Springer International Publishing, Switzerland, 2018.
View at: Publisher Site | Google Scholar

Copyright

Copyright © 2018 César Martínez-Olvera and Jaime Mora-Vargas. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

PDF Download Citation

Download other formats

Order printed copies

Views

1059

Downloads

975

Citations