BUS606: Operations and Supply Chain Management

In this section, the recent advances and the challenges presented during the life cycle of a manufacturing network are discussed. A typical modern manufacturing network is composed of cooperating original equipment manufacturer (OEM) plants, suppliers, distribution centres, and dealers that produce and deliver final products to the market. The topics discussed include supplier selection, supply chain coordination, initial network configuration, manufacturing network complexity, inventory management, capacity planning, warehousing, lot sizing, ICT support tools, and dynamic process planning, monitoring, and control. These topics are in line with the life cycle phases of a manufacturing network (Fig. 5).

Fig. 5 Manufacturing network life cycle

Supplier selection

The building blocks of any manufacturing network are the cooperating companies. The significance of the selection of these stakeholders (supplier, vendors) has been indicated as early as in 1966 is known as the supplier selection problem. This decision-making problem is highly challenging since it goes beyond simple comparison of component prices from different suppliers. It is often decomposed into sub-problems of manageable complexity, such as formulation of criteria for the selection, qualification of partners, final selection, and feedback verification. In Fig. 6, the decomposition of the supplier selection problem into small more manageable problems is presented, together with indicative methods for solving these sub-problems.

Fig. 6 Supplier selection problem, its decomposition into small more manageable problems, and indicative methods for solving them 

The supplier selection problem becomes even more complicated in the era of MC since a certain level of adaptability and robustness is necessary when operating within a volatile and rapidly changing environment. The most commonly used criteria in supplier selection studies include quality and performance. However, when having to deal with unpredictability and fluctuating demand, which are common in MC, additional factors need to be considered such as management compatibility, transparency of operations, strategic direction, reliability, and agility. While trying to adhere to eco-friendliness directives, frameworks incorporate environmental footprint criteria to green supply chain design. Moreover, several other criteria may be relevant according to the design and planning objectives of a niche supply chain, which could be identified using data mining methods.

The Internet and web-based platforms are used in recent years in order to counterbalance uncertainty, monitor altering parameters (e.g. weather in supply routes), and proactively adapt to changes. Moreover, several proposed supplier selection models incorporate the relative importance of the supplier selection factors depending on the types of targeted MC implementation, e.g. for the component-sharing modularity type of MC, the requirements for selecting suppliers would not be the same as the component-sweeping modularity implementation type. Like in the case of a stable low variety production, the analytic hierarchy process (AHP) is commonly used as a means to solve the multi-criteria decision-making problem of supplier selection. Incorporating uncertain information about the real world, essentially extending the Dempster–Shafer theory, the D-AHP method for solving the supplier selection problem. The suggested D numbers preference relation encapsulates the advantages of fuzziness and handles possible incomplete and imprecise information, which is common in human-driven systems such as supply chains. Similarly, a combined analytic hierarchy process - quality function deployment (AHP–QFD) framework that handles uncertain information, selects suppliers, and allocates orders to them. A multi-criteria decision-making method to support the identification of business-to-business (B2B) collaboration schemes, especially for supplier selection.

Supply chain coordination

The literature on organisational knowledge creation points out that "coordination" plays an important role in combining knowledge from stakeholders, while it also mediates the relationship between product modularity and MC. A report on coordination mechanisms for supply chains.

Concerning coordination in supply chains, in general, two topologies are studied, namely the centralised and the decentralised one (Fig. 7). In the first, the coordination decisions are taken by a central body, often the leading supply chain OEM, whereas in the second, each member independently makes its own operational decisions. The decentralised topology has been proven to improve the performance in the context of MC. A supply chain that is commissioned to provide a variety of customised products requires a total systems approach to managing the entire flow of information, materials, and services in fulfilling customer demand. Further incentives have to be provided to the members, so as to entice their cooperation through the distribution of the benefits of the coordination for instance.

Fig. 7 Centralised and decentralised supply chain topologies. In a centralised topology, material and information move only downstream. In the decentralised one, material/information can be transferred both upstream and downstream to better serve customisation, personalisation, and/or regionalisation

The need for adaptation to the new MC requirements has led to the definition of a novel framework for autonomous logistics processes. The concept of autonomous control "describes processes of decentralised decision-making in heterarchical structures, and it presumes interacting elements in non-deterministic systems, and possess the capability and possibility to render decisions independently". However, regardless the topology, the alignment of the objectives of the different collaborating organisations in order to successfully carry out projects, optimise system performance, and achieve mutual profits is indispensable. While an action plan suffices for the coordination of a centralised supply chain, it is inadequate with a decentralised one since entities tend to exhibit opportunistic behaviour. Nevertheless, in terms of overall network performance, decentralised topologies have shown great benefits for serving the mass customisation paradigm.

Initial manufacturing network configuration

The initial manufacturing network configuration must consider the long-term needs of cooperation and often determines its success. In a constantly changing environment, the configuration of the manufacturing network must be, therefore, flexible and adaptable to external forces. The problem has been extensively addressed in the literature using approaches classified in two main categories, namely approximation (artificial intelligence, evolutionary computation, genetic algorithms, tabu search, ant colony optimisation, simulated annealing, heuristics, etc.) and optimisation techniques (enumerative methods, Lagrangian relaxation, linear/nonlinear integer programming, decomposition methods, etc.) and their hybrids (Fig. 8). Focusing on agile supply chains, a hybrid analytic network process mixed-integer programming model with uttermost aim the fast reaction to customer demands. Fuzzy mathematical programming techniques have been employed to address the planning problems for multi-period, multi-product supply chains. A coloured Petri Nets approach for providing modelling support to the supply chain configuration issue. A dynamic optimisation mathematical model for multi-objective decision-making for manufacturing networks that operated in a MC environment.

Fig. 8 Issues to be considered during the initial manufacturing network configuration and indicative methods used

Still, the accuracy of planning ahead in longer horizons is restricted. The incorporation of unpredictable parameters in the configuration through a projection of the possible setting of the network in the future may lead to unsafe results.

Inventory management/capacity planning/lot sizing

Inventories are used by most companies as a buffer between supply chain stages to handle uncertainty and volatile demand. Prior to the 1990s, where the main supply chain phases, namely procurement, production, and distribution, were regarded in isolation, companies maintained buffers of large inventories due to the lack of regulatory mechanisms and feedback. The basis for manufacturing and inventory planning was relatively safe forecasts. However, in the era of customisation the basis is actual orders and the pursuit is minimisation of inventories. These requirements constitute inventory management and capacity planning functions very important for a profitable MC implementation.

In complex distributed systems such as modern manufacturing companies with a global presence, the question of optimal dimensioning and positioning of inventory emerges as a challenging research question. Various strategies for inventory planning have been reported based on how the underlying demand and return processes are modelled over time, thus making a distinction between constant, continuous time-varying, and discrete time-varying demand and return models. Integrated capacity planning methods encompassing stochastic dynamic optimisation models over volatile planning horizons exhibit high performance in the context of MC and personalisation. The DEWIP (decentralised WIP) control mechanism, focusing on establishing control loops between work centres for adjusting the WIP levels dynamically. Its performance was assessed against other well-accepted systems such as LOOR, Conwip, and Polca. Methods used for solving the capacitated lot sizing problem are indicatively shown in Fig. 9.

Fig. 9 Methods (indicative) used for solving the capacitated lot sizing problem

In particular, in just-in-time (JIT) environments, MC impacts the amount of inventory that needs to be carried by firms that supply many part variants to a JIT assembly line. In addition, the supply of parts is performed either on constant order cycles or more commonly under non-constant cycles. The goal chasing heuristic, pioneered within the Toyota production system, seeks to minimise the variance between the actual number of units of a part required by the assembly line and the average demand rate on a product-unit-by-product-unit basis, while applying penalties for observed shortages or overages. Of course, information sharing and partner coordination systems are a prerequisite for JIT procurements. For instance, DELL, which achieved a highly coordinated supply chain to respond to MC, communicated its inventory levels and replenishment needs on an hourly basis with its key suppliers and required from the latter to locate their facilities within a 15-min distance from DELL facilities. Another consideration during inventory management is the type of postponement applied in a company. Studies have shown that postponement structures allow firms to meet the increased customisation demands with lower inventory levels in the case of time postponement (make-to-order), or with shorter lead times in the case of form postponement. Also, an assemble-to-order process, a variation of form postponement, does not hold inventory of the finished product, while in form postponement, finished goods inventory for each distinct product at the product's respective point of customisation is kept. An indicative example is given in the case of Hewlett Packard, where using form postponement, the company achieved the postponement of the final assembly of their DeskJet printers to their local distribution centres.

Logistics management

Logistics can play a crucial role in optimising the position of the customer order decoupling point and balance between demand satisfaction flexibility and productivity. In a customer-centric environment, the supply chain logistics must be organised and operated in a responsive and at the same time cost-effective manner. Customisation of the bundle of product/services is often pushed downstream the supply chain logistics, and postponement strategies are utilised as an enabler for customisation. Maintaining the product in a neutral and non-committed form for as long as possible, however, implicates the logistics process. Traditional logistics management systems and strategies need to be revisited in the context of customisation, since distribution activities play a key role in achieving high product variety, while remaining competitive. Most OEMs form strategic alliances with third-party logistic (TPLs) companies. The introduction of TPLs in the supply chain serves two purposes. First, it acts as a means of reducing the complexity of management for an OEM through shifting the responsibilities of transportation, and in many times customisation, to the TPLs. Second, it extends the customisation capabilities as TPLs can actively implement postponement strategies. Postponement strategies with logistics as an enabler are located at the bottom of Fig. 10 and can serve all types of customisation, from plain shipment to order up to extremes of engineer-to-order or personalisation.

Fig. 10 Postponement strategies for different supply chain structures and logistics

Moreover, the management of logistics is a process inherently based on communication and collaboration. Developments of either function-specific or all-in-one ICT solutions targeted on logistics. Tools for warehouse and transportation management, ERP, supply chain management (SCM), and information sharing are reported under the umbrella of e-logistics. The concept of virtual logistics is also proposed for separating the physical and digital aspects of logistics operations, having Internet as an enabling means to handle ownership and control of resources.

Supply chain control

The information transferred from one supply chain tier to the next in the form of orders is often distorted, a phenomenon known as the bullwhip effect. In particular, when customer demand is volatile such as the case is in MP, the bullwhip effect misguides upstream members of the supply chain in their inventory and production decisions. Nevertheless, the performance of the supply chain is highly sensitive to the control laws used for its operation. The application of the wrong control policy may have as a result the amplification of variance instead on its minimisation. Dynamic modelling approaches have been proposed to manage supply chains, accounting for the flow of information and material, to capture the system dynamics. Multi-agent approaches for modelling supply chain dynamics. Software components known as agents represent supply chain entities (supplier, dealers, etc.), their constituent control elements (e.g. inventory policy), and their interaction protocols (e.g. message types). The agent framework utilises a library of supply chain modelling components that have derived after analysis of several diversified supply chains. For instance, a novel oscillator analogy in presented for modelling the manufacturing systems dynamics. The proposed analogy considers a single degree of freedom mass vibrator and a production system, where the oscillation model has as input forces, while the manufacturing system has demand as excitation. The purpose is to use this simple oscillator analogy to predict demand fluctuations and take actions towards alignment.

Another necessity in supply chain control is the traceability of goods. Traceability methods, essential for perishable products and high-value shipments, exploit the radio frequency identification (RFID) technology during the last years. A traceability system that traces lots and activities is proposed by Bechini et al. The study examines the problem from a communication perspective, stressing the need to use neutral file formats and protocols such as XML (extended markup language) and SOAP (simple object access protocol) in such applications. The emerging technology of IoT can provide ubiquitous traceability solutions. Combining data collection methods based on wireless sensor network (WSN) with the IoT principles, the method can support the traceability of goods in the food industry. In a similar concept, the role of an IoT infrastructure for order fulfilment in a collaborative warehousing environment. The IoT infrastructure is based on RFID, ambient intelligence, and multi-agent system, and it integrates a bottom-up approach with decision support mechanisms such as self-organisation and negotiation protocols between agents based on a cooperation concept.

Supply chains formed for servicing customisation are more complex as structures and less predictable in their dynamic behaviour than stable traditional supply chains. Recent complexity studies deal with the emerging aspects of increasing complexity of manufacturing activities and the dynamic nature of supply chains. The importance of managing the complexity in supply chains is evident, as recent studies depict that lower manufacturing network complexity is associated with reduced costs and overall network performance. A complete and comprehensive review of complexity in engineering design and manufacturing.
Simulation and ICT support systems for manufacturing networks life cycle

Robust and flexible ICT mechanisms are rendered necessary for improving performance in each of the previous life cycle aspects of supply chains and for bridging inter- and intra-enterprise collaboration environments. Digital enterprise technologies in general represent an established, new synthesis of technologies and systems for product and process development and life cycle management on a global basis that brings many benefits to companies. For instance, the benefits offered by the adoption of virtual engineering through the life cycle of production are shown in Fig. 11. To manage the huge portfolio of products and variety, as well as tracking the expanding customer base, ERP and CRM suites are necessary tools. Additionally, cloud technology is already revolutionising core manufacturing aspects and provides ample benefits for supply chain and manufacturing network life cycle. Cloud technology and the IoT are major ICT trends that will reshape the way enterprises function in the years to come.

Fig. 11 Increased efficiency through virtual engineering approaches

Simulation for manufacturing network design

Literature on ICT-based systems for improving manufacturing networks is abundant and highlights the need for increased penetration of ICT systems in design, planning, and operation phases. A simulation-based method to model and optimise supply chain operations by taking into consideration their end-of-life operations is used to evaluate the capability of OEMs to achieve quantitative performance targets defined by environmental impacts and life cycle costs. A discrete event simulation model of a capacitated supply chain is developed and a procedure to dynamically adjust the replenishment parameters based on re-optimisation during different parts of the seasonal demand cycle is explained. A model is implemented in the form of Internet-enabled software framework, offering a set of characteristics, including virtual organisation, scheduling, and monitoring, in order to support cooperation and flexible planning and monitoring across extended manufacturing enterprise. Furthermore, the evaluation of the performance of automotive manufacturing networks under highly diversified product demand is succeeded through discrete event simulation models with the use of multiple conflicting user-defined criteria such as lead time, final product cost, flexibility, annual production volume, and environmental impact due to product transportation. Finally, the application of the mesoscopic simulation approach to a real-world supply chain example is illustrated utilising the MesoSim simulation software.

Existing simulation-based approaches do not tackle the numerous issues of manufacturing network design in a holistic integrated manner. The results of individual modules used for tackling network design sub-problems often contradict each other because they refer to not directly related manufacturing information and context (e.g. long-term strategic scheduling vs. short-term operational scheduling), while harmonising the context among these modules is challenging. This shortcoming hinders the applicability of tools to real manufacturing systems as it reduces the trustworthiness of results to the eyes of the planner among other reasons.

Enterprise resource planning

An ERP system is a suite of integrated software applications used to manage transactions through company-wide business processes, by using a common database, standard procedures, and data sharing between and within functional areas. Such ICT systems entail major investments and involve extensive efforts and organisational changes in companies that decide to employ them. ERP systems are becoming more and more prevalent throughout the international business world. Nowadays, in most production distribution companies, ERP systems are used to support production and distribution activities and they are designed to integrate and partially automate financial, resource management, commercial, after-sale, manufacturing, and other business functions into one system around a database.

A trend, especially in the mid-market, is to provide specific ERP modules as services. Such need generates the challenge for ERP system providers to offer mobile-capable ERP solutions. Another issue is the reporting and data analysis, which grows with the information needs of users. Research in big data analytics and business intelligence (BI) should become more tightly integrated with research and applications of ERP.

Customer relationship management

In Internet-based retailing, which is the preferred business model followed in MC, customer information management is a necessity. In particular, exploiting consumer data, such as purchase history, purchasing habits, and regional purchasing patterns, are the cornerstone of success for any company active in MC. In business-to-business and business-to-customer, CRM suites are thus indispensable. According to Strauss and Frost, CRM involves, as a first step, research to gain insight so as to identify potential and current customers. In a second step, customer information is used to differentiate the customer base according to specific criteria. Finally, the third step involves customised offerings for those customers that are identified as "superior" from the previous phase, enabling thus, the targeted offering of customised products. During the first step of identification of customers, market research and consumer behaviour models are used. In a second phase, for establishing differentiation techniques, data mining and KPIs assessment are used. Finally, for fine-tuning customisation options, information such as price, variants, promotions, and regions are examined.

As Internet becomes ubiquitous in business, CRM has been acknowledged as an enabler for better customisation since it offers management of the new market model less disruptively. Internet-enabled CRM tools also bring the customer closer to the enterprise and allow highly responsive customer-centred systems without significant increase in costs. e-CRM implementations have been assessed in the study. Noticeably, most major CRM suite vendors have started providing cloud-based services, a business model that suits SMEs that cannot afford huge ICT investments. Based on the balanced scorecard method, the study assessed e-CRM performance using 42 criteria in a number of companies. The results show that a successful CRM implementation is associated with tangible outcomes, such as improvements in financial indicators, customer value, brand image, and innovation. Finally, the latest generation of CRM tools, referred to as social CRM, exploit social networking technology to harness information about customer insights and engagement.

Cloud computing and manufacturing

A comprehensive definition of cloud computing is provided by the National Institute of Standards and Technology: "a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources (e.g. networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction". Several applications have been reported in recent years where a cloud infrastructure is used to host and expose services related to manufacturing, such as: machine availability monitoring, collaborative and adaptive process planning, online tool-path programming based on real-time machine monitoring, manufacturing collaboration and data integration based on the STEP standard, and collaborative design.

The benefits of cloud for improving manufacturing network performance are numerous (Table 1). Cloud can offer increased mobility and ubiquitous information to an enterprise since the solutions it offers are independent of device and location. Moreover, computational resources are virtualised, scalable, and available at the time of demand. Therefore, the intensive costs for deploying high-performance computing resources are avoided. In addition to that, purchasing the application using the model software as a service is advantageous for SMEs who cannot afford the huge investments that commercial software suites entail. However, there are some considerations also (Table 1). A main challenge for the adoption of cloud in manufacturing is the lack of awareness on security issues. This major issue can be addressed using security concepts and inherently safe architectures, such as privately deployed clouds. The security concept must include availability of ICT systems, network security, software application security, data security, and finally operational security. Considerable funding is spent by the global security software market, in order to alleviate security issues. Recent reports show that the expenditure on cloud security is expected to rise 13-fold by 2018. Moreover, there is the possibility of backlash from entrenched ideas, manufacturing processes and models caused by the hesitation for the adoption of innovative technology. Finally, the lack of standardisation and regulation around cloud hinders its acceptance by the industry.

Table 1 Benefits and drawbacks of cloud technology for manufacturing