Every year, the Material Handling Institute (MHI) partners with Deloitte to release their annual report detailing key challenges and trends in the supply chain. There were several interesting observations particularly regarding the rate of adoption of innovative technology. In the industrial space, technology adoption can be slow as manufacturers are hesitant to disrupt existing production processes to implement upgrades, even when future efficiencies can be realized. This paradigm is getting turned on its head regarding the supply chain of raw materials and finished goods. Adoption rates are accelerating, and many innovative and disruptive technologies are expected to be implemented throughout 70% or more of the supply chain within the next five years. Roughly 80% of survey respondents indicate that digital supply chains will be the primary model within the next five years.
The top challenge facing the supply chain is the increasing number of demands that customers are placing on suppliers and that is propagating through every link in the supply chain from end customers to suppliers of raw materials to manufacturers. With the digital interconnection of all parties and supply chains extended globally, the demand is for 24/7 availability. Manufacturers are looking to limit raw material inventories to reduce costs and are increasingly leaning on suppliers for faster service in support of this objective. Wholesalers and retailers are leaning towards fast delivery and high availability as differentiators in an increasingly competitive marketplace and are leaning on finished goods manufacturers to support them.
To meet these demands will require a combination of automation in production processes and automation in intelligence. Automating tasks and processes will be needed to deliver the productivity improvements required to meet the demands of the supply chain while automating intelligence and information flow will be critical to monitoring demand indicators and adjusting business decisions accordingly (ex. order material, increase capacity).
The key to maximizing the benefits from automating both production processes and business intelligence will be providing inter-connectivity between these systems. This can provide its own set of challenges as the network technologies used in each have different data models and performance characteristics.
In response to the increasing demands on the supply chain, several technology solutions are being pursued that will reinvent the traditional business models in this market space. According to the MHI Annual Report, the top technologies that are expected to be disruptive influences and provide competitive advantages include Automation and Robotics, Predictive Analytics, Sensors/Internet of Things, Artificial Intelligence and Autonomous Vehicles. These technologies satisfy the two core requirements for optimizing supply chains, productivity increases (Automation and Robotics, Autonomous Vehicles) and improved business intelligence (Predictive Analytics, Sensors/IoT, Artificial Intelligence).
It is important to keep in mind that none of these technologies are mutually exclusive. The greatest benefit will be realized in a combination of several or all the technologies shown. For example, one can realize better predictive analytics facilitated by internet-connected sensors installed into automated systems.
These technologies are different in how information needs to be presented. The needs of automated machinery (state measurement, task execution) are uniquely different from the contextually-based information needs of business intelligence applications. This results in a unique communication network needs which we will discuss separately.
Automation is expected to grow from the current adoption rate of 34% to 73% within the next five years (23% average annual growth rate). This is being driven by the need to improve productivity in the transportation of raw and finished goods within a facility, optimize existing capacity and provide more efficient order fulfillment. There are many common systems now used to aid this process.
Automated Storage and Retrieval Systems (AS/RS: Figure 2) are important in optimizing capacity as the system of elevators and shuttles allow for more of the vertical capacity in an existing building to be used for storage in addition to faster infeed and outfeed of material. Infeed and outfeed are often connected using systems of conveyors, or Automated Guided Vehicles in modular operations, for delivery of material to shipping or production cells. This improves the output of these facilities and reduces the need, and associated cost, to expand horizontally to achieve greater capacity.
Industrial robots have been able to assume many of the tasks once performed by people such as picking, sorting and packing of orders, loading, unloading, and other repetitive production tasks while human resources can be deployed towards tasks better associated with their problem solving and adaptive reasoning skills. Robots also have the key benefit of being able to run 24/7 like other automated machinery bringing tremendous productivity increases. Supply chains can now exist in an “always on” state.
As mentioned earlier, the network needs of automated machinery are unique in comparison to other networks which focus on delivery of information.
In machine automation, deterministic control is critical to performance. Determinism when referring to machine control means that the system is both predictable and repeatable when executing its tasks. The outcomes of the system may change based on changes in inputs, but the process that the machine control executes is uniform and performed within a repeatable period.
The networks that are used with automated machinery need to fit with this model of determinism. These networks typically involved small chunks of process data (bits and bytes) being sent in fast, repeatable intervals. Early in the days of automation controllers when vendors started to use networks to interface between controllers and machine peripherals, these networks were typically proprietary to the control vendor which allowed them to tightly control performance.
Starting in the late 1980s into the early 1990s, open fieldbus networks began to evolve. These network technologies were standardized to address the needs of machine control but allowed multiple vendors to participate if they were compliant to the standard. One of the major challenges is that these fieldbus networks are not natively compatible with enterprise applications so machines using these technologies are limited in what they can contribute.
Eventually, in the late 1990s into the early 2000s, many of the organizations that developed open fieldbus technologies began working on ways to do machine control networks on Ethernet. There were challenges that needed to be addressed with this approach. The standard TCP/IP stack structure used in most the applications was not inherently deterministic and since most information-level networks were also based on Ethernet, there was significant risk to machine performance if these networks were combined into a single network. This has been addressed by defining strict rules for network segregation and the different organizations have developed alternative approaches to TCP/IP for critical data transmission (ex. UDP/IP, Standardized Open Protocols built on IEEE 802.3). These networks are referred to collectively as Industrial Ethernet.
With increasing adoption of Industrial Ethernet, network architectures are flattening, and this will make it easier to ensure the level of connectivity needed to facilitate the flow of information from automated machinery to enterprise applications, and potentially to customers or other external stakeholders. The last challenge to overcome will be interpreting data from industrial networks so it can be presented in information-centric applications.
When discussing automated machine control and the networks they employ, this is generally referred to as OT (Operational Technology) and as discussed previously its data needs are unique in comparison to IT (Information Technology). From an IT perspective, OT data often lacks context, which is necessary for it to be useful to humans when making business decisions. To be beneficial, it will be necessary to apply devices or systems that can reconcile these two worlds.
Solving this challenge often requires a multi-layer approach. The first step is to ensure that the data from machinery is available and this can be challenging as many machine networks are not optimized for sharing process data with different owners. This can often be solved by using some form of two-sided communications bridge/gateway that is inserted as a native device on the machine network. The machine-facing side of the gateway serves as a repository for data sent to it from the machine controller and maintains segregation of its network from other systems. The other side of the gateway may interface with some other OT network such as Industrial Ethernet that may be integrating multiple machines, and these are commonly referred to as automation gateways. The other side of the gateway may also provide connectivity directly into IT-centric data models or application frameworks such as OPC UA, SQL, MQTT, .NET, etc. These devices that directly connect machine data into information-centric applications are often referred to as edge gateways.
An edge gateway can potentially have multiple levels of functionality. In its simplest form, it can serve as a method for making raw data available to information-centric applications, essentially an IT/OT bridge. This requires the external IT application to be configured to derive context from the supplied data and present it in a consumable form. In some cases, the edge gateway may support the functionality to run specific algorithms that can precompute and add context to data before it is made available to external applications. Finally, as cloud computing continues to expand in its use, edge gateways will need to interface to various cloud computing platforms, provide robust security and support connectivity to internet-connected networks such as Ethernet, 802.11 WiFi and Cellular.
Creating networks that are secure and offer high availability of data in an easily understandable form will most often require a combination of edge and automation gateways.