Resilience by Teaming in Supply Chains and Networks

Resilience research and practice require encompassing principles, which will enable the convergence of concepts and approaches across disciplines. Furthermore, sustainable resilience requires design and operation methodologies which acknowledge that supply network participants are fault-prone; they can, and will, fail at some point in time. Supply networks must be capable of maintaining operations despite these failures, with minimal protection from under-utilized resources. Acknowledging the above facts, this book introduces a discussion on central resilience concepts and presents a novel approach to designing and operating resilient supply networks based on teaming. The conceptual discussion of the initial chapters distills key fundamentals from seemingly dissimilar disciplines in supply network research and provides the foundation for future resilience research. The latter chapters of this book present a comprehensive framework, Resilience by Teaming, developed based on the notion that a team of weaker agents can outperform a single flawless agent, under the right conditions. Combined, fundamentals and framework have the power to shape future complex supply networks to provide higher service levels with minimal disruption by enabling smart collaboration among fault-prone agents.

Real-world systems are inherently complex, formed by a large number of self-governing interconnected agents which dynamically update their behavior rules and connections based on context and environment changes. In order to model and understand the basic properties of these systems, it is necessary to start from a common definition of their components and main evolution processes. In this chapter, the foundations introduced by Reyes Levalle and Nof (2017) to model supply networks as complex adaptive systems are presented. The traditional (linear) supply chain model is discussed and contrasted with a more advanced formalism, i.e., supply networks. Supply network components, agents and links, as well as basic network structure dynamics, formation and reconfiguration, are defined based on the notion of autonomous agents and graph theory. The concepts introduced in the following sections will provide the theoretical basis for supply network modelling and analysis, and for the discussion of supply network resilience, throughout this book and in future research efforts in the field.

Several authors recognize the work of (Holling 1973) in ecosystems’ resilience as the seminal work in systems resilience. Based on the observed behavior and evolution of various ecological systems, Holling (1973) defined resilience as a property thereof, responsible for the persistence of relationships within the ecosystem in the face of changes to system variables or parameters. Over the next 40 years, research in systems resilience spread over a wide range of disciplines, from psychology to supply network management; nevertheless, resilience in the context of supply networks still remains largely unaddressed (Bhamra et al. 2011). In this chapter, the definition and components of resilience in the context of supply networks introduced in Reyes Levalle and Nof (2017) are analyzed. Five fundamental principles of supply network resilience are introduced, in line with resilience definitions from several supply network domains. These core concepts will provide a platform to unify resilience understanding across domains in supply network research. Furthermore, in order for supply network principles to materialize, resilience must be addressed in two dimensions, structure and control protocols, and at two levels, agent-level and network-level. These resilience components are introduced in the subsequent sections and discussed within the supply network formalism presented in Chap. 2.

The supply network formalism and resilience fundamentals introduced in Chaps. 2 and 3 provide the theoretical foundation to understand resilience in SNs. Practitioners, however, require actionable insights and strategies to create and operate resilient supply networks. Literature from physical, digital, and service supply networks contains numerous techniques to achieve these objectives. Nevertheless, the field lacks a clear mapping of these techniques to core resilience strategies to design and operate SNs. In this chapter, the three main strategies applied in the design of resilient structures in SNs—redundancy, excess resources, communication network efficiency—are presented. The aim is to provide practitioners and researchers with a high-level map of design strategies which encompass many existing approaches, and, possibly, several future ones, to guide their efforts to create resilient SNs. Strategies are discussed in connection with existing examples found in literature to provide a wide overview of techniques used within each strategy to achieve resilience at agent and network level.

Resilience is a dynamic property of SNs. The ability of a SN to minimize the effect of disruptions on its performance depends on the interaction between its structural properties and the set of flow control protocols used to dynamically assign and re-assign flow and resources. Hence, SN design should not only be concerned with the structural or control aspects in isolation but rather with the integrated design of these and their seamless interaction. Flow control protocols manage digital, physical, and service flows, and communication exchanges, among and within SN agents in order to conduct operations under normal and disrupted conditions, without altering existing SN structure. In this chapter, the three main types of flow control in SNs—sourcing control, internal resource control, and distribution control—are characterized and their interrelation with SN structural characteristics is discussed within the SN formalism introduced in Chap. 2. Several examples from literature illustrate applications that belong to each of the three protocol types.

Over the last decade, researchers and practitioners have been increasingly focusing on developing strategies to create resilient supply networks. Most approaches focus on optimizing the trade-off between operational costs and resilience through increased protection, in the form of redundancy and excess resources, and/or the use of more reliable agents/resources. Nature and humans, however, show us that it is also possible to create resilient systems with fault-prone agent and resources through smart designs and distributed control protocols. The driving force behind these resilient systems is teaming. Teaming is a process by which a set of agents form a network and collaborate to achieve their individual goals and, perhaps, a common objective. Inspired in the Fault-tolerance by Teaming principle of Collaborative Control Theory (Nof 2007), the Resilience by Teaming framework (Reyes Levalle 2015; Reyes Levalle and Nof 2015) comprises a set of principles and protocols to design and operate resilient supply network on the basis of forming collaborative teams among fault-prone agents. Leveraging the mathematical formalism to describe and model supply networks from Chap. 2, the set of fundamentals for resilience in supply networks from Chap. 3, and the basic strategies to design and operate resilient SNs from Chaps. 4 and 5, the following sections introduce the main components of the Resilience by Teaming framework. Chapters 7 through 9 discuss each component of the Resilience by Teaming framework in detail.

Every SN agent a in \(A^{K} \cup A^{O}\) needs to receive input flow from a set of predecessors P _a in order to operate and provide output to its successors. Such agents face a medium-term strategic decision, i.e., which agents to select as P _a , and a short-term operational decision, i.e., which agents in P _a to source from at time t. These distributed strategic decisions of agents a ∈ A shape the dynamic topology of the supply network, and give rise to its structural resilience, as well as determine, at least in part, the ability of each SN agent to cope with sourcing disruptions. This chapter presents the theoretical aspects of two sourcing protocols inspired in the Fault-tolerance by Teaming principle, FTT, of Collaborative Control Theory (Nof 2007), CCT, to select a team of predecessors P _a and to adaptively manage flow from these agents. The Sourcing Team Formation/Re-configuration Protocol (STF/RP) and Sourcing Flow Control Protocol (SFCP) leverage inherent randomness in individual agents in P _a to provide a more stable and resilient input to agent a.

Supply network agents transform flow received from their predecessors into output flow by executing a series of internal processes. These agents require a set of internal resources \(r \in R_{a}\) which are oftentimes arranged in a network configuration under central control of agent \(a\). Internal resources are classified in processes, i.e., resources that transform local input flow into local output flow, and storages, responsible for temporal stowage of flow to balance flow rate fluctuations. Internal resource network resilience can be enabled by design and flow control protocol decisions. As found among reviewed articles in Sects. 4.​3.​1 and 4.​3.​2, resilience can be built-in by embedding excess capacity and storage. However, these strategies entail one-time costs, e.g., equipment purchase and installation, and operation costs, e.g., energy consumption and maintenance, and need to be utilized efficiently to increase resilience in a sustainable manner. To this end, the principle of Fault-tolerance by Teaming from CCT lays the foundation to leverage weaker, thus less costly, resources to achieve a performance equal to, or better than, that achievable by a single flawless, and costlier, resource. Team formation in internal resource networks (IRNs) is influenced by parameters selected at the design stage and decisions made in real-time based on available status information. Design choices define limits to the ability of flow control protocols to re-parameterize an IRN in order to protect resources at risk of disruptions. The following sections present design guidelines and a flow control protocol to achieve resilient operation of internal resource networks.

Agents in a supply network deliver flow to other agents in the network or to entities outside the scope of the SN. Flow distribution occurs through flow links \(fl_{i \to j}\) which are subject to disruptions that affect service levels, possibly extending far beyond the reach of a single link. Therefore, creating resilient links and managing flow to avoid and overcome disruptions without impact on the service level is of utmost importance. This chapter presents two protocols from the Resilience by Teaming framework: Distribution Network Formation/Re-configuration Protocol (DNF/RP) and Distribution Flow Control Protocol (DFCP). DNF/RP implements a fuzzy network formation and reconfiguration strategy designed to create routes that balance the cost of flow distribution with the ability to re-configure the route in the face of disruptions. DFCP performs real-time monitoring and route re-configuration to avoid disruptions and ensure flow delivery according to the predefined SLAs. The use of resilient distribution protocols in combination with the sourcing and internal control protocols introduced in preceding chapters provides the foundation for Resilience by Teaming in supply networks.

Production lines can be described as IRNs of a supply network agent, consisting of interconnected processes and storages that transform raw materials into finished products. In almost every case, their objective involves maintaining constant throughput to match demand with minimum WIP. Nonetheless, resources are subject to random failures, creating local downtimes (i.e., at resource level) that can spread through the IRN affecting other non-failed resources and, ultimately, the QoS an agent provides to its successors in the SN. As pointed out by Hudson et al. (2014), the problem of unbalanced production lines with unreliable equipment has received significant attention for over half a century. Nevertheless, despite advances in modeling and control of these systems, modern high speed production systems are still challenged by the succession of process failures that lead to line downtimes and, consequently, throughput reduction. This problem is becoming ubiquitous, as production systems become “leaner” and are increasingly required to process a more diverse mix of products with higher speeds and performance.

Case Study A analyzes the relative merits of a teaming-based flow control protocol, IFCP (Reyes Levalle et al. 2013; Reyes Levalle 2015), versus traditional production line control methods such as Kanban, CONWIP, DWIP, and Base-stock. Results show that a teaming approach to resilient production line control is capable of reducing work-in-process and throughput variability when compared to other traditional control approaches for the same line throughput performance.

Flow delivery problems in distribution networks have received significant attention from researchers, as these challenges are ubiquitous in modern physical supply networks. Initially, research focused mostly on vehicle routing problems (VRPs), a sub-set of flow delivery problems concerned with the selection of intermediaries and delivery paths to minimize the total cost of delivery. As QoS gained importance, approaches to the solution of VRPs began to incorporate pre-negotiated delivery lead times as constraints to ensure fulfillment of SLAs. However, in practice, dynamic network conditions oftentimes hinder static solutions if these do not incorporate enough slack to allow for unexpected delays in the delivery plan. Small parcel distribution networks have grown significantly over the last years, mainly fueled by the exponential growth of e-commerce driven by Amazon and store pickup and home delivery options offered by large retailers such as Walmart, Target, and Best Buy, among others. Over time, increasing pressure from these systems’ customers to obtain best-in-class QoS with shorter leadtimes and lower cost eroded the margin of static solutions, obtained through traditional VRP approaches, to cope with congestion and/or disruption in flow links and/or at intermediaries. Furthermore, some SN agents became increasingly unwilling to accept tardiness, even in the event of unforeseen events, often including penalties in SLAs for late deliveries. In this new scenario, resilience-enabling distribution protocols based on teaming (DNF/RP and DFCP), capable of adapting dynamically to changing distribution network conditions, outperform traditional approaches based on cost and/or leadtime minimization (Reyes Levalle 2015a, b).

Agent-level resilience mechanisms received significant attention over the last years. A wide variety of approaches have been developed based on a trade-off between increased resilience and higher redundancy and use of excess resources. Despite the capacity of some approaches to increase agents’ resilience, their relative benefits have seldom been analyzed from a global perspective. Research in complex systems has successfully linked network formation/re-configuration phenomena, driven by agent-level association/dissociation decisions, and the impact of topology on network performance and survivability (Barabási and Albert 1999; Albert et al. 2000; Albert and Barabási 2002; Tangmunarunkit et al. 2002; Thadakamalla et al. 2004; Brede and de Vries 2009). Nevertheless, decision mechanisms analyzed mostly rely on simple probabilistic association/dissociation rules and fail to account for more complex decision criteria involved in supply network formation and re-configuration processes. Regardless of the magnitude of the benefits obtained from any agent-level strategies, it is not possible to extrapolate these benefits to the performance of the entire SN. Furthermore, it is necessary to understand if local benefits are obtained at the expense of performance losses in other SN agents, and whether these approaches create vulnerabilities that hinder SNs ability to be resilient. This chapter analyzes and discusses the benefits of RBT-based decisions relative to traditional SN formation and re-configuration mechanisms based on the results of Reyes Levalle (2015) and Reyes Levalle and Nof (2015a). Results show how local teaming-driven decisions effectively modify SN topology and its ability to overcome disruptions with minimal impact on the SN objectives.

Resilience is becoming a central concern in modern complex systems. As such, supply networks need design and operation mechanisms capable of enabling multi-level resilient behavior. In order to guide researchers and practitioners in these tasks, Chaps. 2–5 introduce the basic elements required to understand resilience requirements and model SN behavior. These foundational concepts should provide the base for future research efforts in resilient systems. Inspired in the principles of Collaborative Control Theory, Chaps. 6–9 present a teaming-driven approach to design and operate supply networks: Resilience by Teaming. A suite of protocols and design guidelines are presented in detail to provide practitioners with a generalized approach to design structures and control flow in physical, digital, and service SNs. The protocols and design considerations discussed in RBT are analyzed in real-world case studies in Chaps. 10–12. In this final chapter, the major contributions of the book are revisited. Future research lines are discussed not only in relation to the extension of Resilience by Teaming, but also from a general perspective for complex resilient systems.