A little background on cloud services
Before diving into the manufacturing system specific topic, it is necessary to clarify the term “cloud
services”. There are multiple cloud platform models in use today but this article will focus on the
opportunities provided by leveraging new commercial cloud component services in a mix with on-premise software versus looking at moving the entire information system to the “cloud”.
Types of cloud services:
• Infrastructure-as-a-Service (IaaS) – Virtual machine services accessed over the network providing
computational and storage capabilities
• Software-as-a-Service (SaaS) – Software applications (such as PLM, ERP, Inventory Control, or CRM)
provided as a service from the cloud
• Platform-as-a-Service (PaaS) - Platform software services (including application server, database
server, web user interface framework, workflow orchestration services, and enterprise service bus
middleware) on which to build custom web service-based applications for the enterprise
• Component-as-a-Service (CaaS) - Commercial cloud delivered technology, data and connectivity services (such as identity verification, app marketplace, supply chain data exchanges, analytical engines and document storage) that can be incorporated into on-premise or cloud-based custom enterprise applications. In contrast to regular web sites focused on interacting with human users, these cloud component services provide APIs to exchange data directly with applications.
From a financial point of view these cloud services can be very exciting because they are provided
under term and usage pricing models. But in addition to the new pricing models that are making new technology adoption more affordable, we should also get excited about the new functionality and integration options enabled.
The benefits of using SaaS and IaaS have been covered in many general information technology articles, but I haven’t seen much discussion on the potential for PaaS and CaaS for manufacturing systems. I also want to propose that a mixed model leveraging a combination of distributed on-premise applications and cloud component services should provide the best framework for future manufacturing systems.
Cloud services and new standards evolving around the Industrial Internet of Things (IIoT) are rendering the old division between automation, IT and business obsolete. The availability of quality cloud component services that are easy to assemble into a custom application will change the future of the Information Technology (IT) department. Other enterprise departments will be able to assemble applications they need themselves on a UI, workflow and integration framework. The IT department will change from focus on developing, installing and supporting software applications, to focus on providing guidelines and a framework for assembling integrated enterprise applications. The framework provided by the IT department will include security schemes, brokering of the mix of commercial IaaS, SaaS, PaaS and CaaS, and integration standards that the enterprise will use.
The question should not be whether or not our manufacturing system will be on the cloud, but instead figuring out how much of our manufacturing system will be on the cloud to maximize the benefits to our organizations and customers. Cloud computing is here to stay providing new software delivery models and information technology options. A mix of on-premise plus on-cloud solutions will most likely provide the best landscape for future manufacturing systems, and in fact some software vendors are already moving in that direction. Business critical data, and speed/availability sensitive applications can be maintained on-premise. Integrated enterprise systems can leverage a layered and distributed landscape of on-premise applications and on-cloud components integrated via SOA (service-oriented architecture) interfaces. The new IT options enabled by cloud services will become more affordable and easier to maintain which will be of great interest to smaller organizations with a smaller IT staff.
A hybrid on-premise and on-cloud model for manufacturing systems
What would this mixed on-premise and on-cloud model look like? I could see it having: (a) an enterprise level business process management (BPM) and workflow orchestration engine, (b) integration mechanisms to enable receiving and sending information between the BPM layer and other layers from machine to plant to supply chain, (c) several persistent layers of data with different precision and time horizons for data needed at the controller level versus business level, and (d) a combination of role specific and enterprise system applications designed to optimize the job of each user role and function in the overall integrated enterprise system.
Manufacturing systems orchestrate interdepartmental processes
The Manufacturing Operations department cannot effectively be managed as an island isolated from other enterprise departments including Engineering, Supplier Management, Quality Management, Human Resources, Facilities Management and Financial Management. We need effective ways to create information threads for complete business processes across departments that do not depend on manual translation of information. We currently run many interdepartmental business process via paper, email, and with many manual interpretations and translations of data inputs to outputs along the way. These manual interdepartmental business processes are error prone and cannot scale to handle higher volume of transactions.
Business Process Management (BPM) and Workflow tools can be used to string together business processes that span across enterprise departments. For example, managing the release of a revision to a product design from engineering to supply chain to revise component parts, to quality to revise inspection requirements, to fabrication to revise NC programs, and to production to revise assembly procedures. All these processes need to be orchestrated so that they cut in at the right time so the right version of the components are available for production with the right machine and inspection instructions.
An ideal manufacturing system platform would facilitate (a) integration throughout different functional layers including automation control, operations management, equipment maintenance, supply logistics, and business management layers, and (b) integration throughout parallel management systems including physical, human, information, sustainability, economic, social, and regulatory systems.
Cloud component services for manufacturing systems
Types of cloud component services that could be leveraged in a manufacturing system include:
• Data Services including Data Storage, Data Format Translation, Document Repository
• Analysis Services including statistical correlation or predictive analysis that does not have to run
real-time along with the run of a machine process
• Specialized Apps including Document Authoring and File Viewers for specific CAD formats
• Industry Market Places to acquire parts and materials more effectively
• Supply Chain Data Exchanges for controlled communications tied to contracts between multiple tiers of suppliers in the product value chain
• Weather Services to optimize energy utilization settings at the factory
• Product Transportation and Distribution Hubs to get the product to the distribution channel and
customer in the most efficient manner
• Identity Verification Services to make sure the person or machine trying to exchange information is authorized to access the information requested.
• Printing Services in case you must still print something at the corner printing store : )
Distributed layered manufacturing system between business and machines
Could we just move the whole manufacturing system to the cloud? Perhaps for some cases, however at some plants more intimate controller-machine interactions should probably remain on-premise near the machine. For a more distributed and layered system, the general MOM (Manufacturing Operations Management) functions of Work Order Management, WIP Management, Quality, Maintenance, etc. could be provided by applications hosted in a cloud service that would interface to multiple lower level “Smart Machine Controller Appliances” at the shop floor.
The Internet of Things (IoT) interconnects “smart” devices via standard Internet Protocol Suite
(TCP/IP). The Industrial Internet of Things (IIoT) has to strive for the same goal for a new generation of “smart” machines. New IIoT standards should not maintain the old networking standards (Profibus, Fieldbus, etc) used for older machine automation controls. To achieve this goal we might need an intermediary application layer that will translate “smart” communications to older “not as smart” controllers, machines and sensors that do not speak the new integration language standards directly.
In the IIoT, the proliferation of connected smart machines, devices, and sensors can result in an
explosion of data. A distributed layered system is a way to contend with this explosion and organize
people-to-people connections, people-to-machine, and machine-to-machine communications. If
organizations embrace this opportunity to organize these automated communications and leverage all the resulting data at different layers, the visibility and analysis enabled will lead to new levels of
visibility, capabilities, accuracy, and control.
A Smart Machine Controller Appliance would be placed on-premise closer to the machines it controls for any of the following reasons:
(a) The controller function needs high-availability and high-speed connectivity
(b) The controller has large bandwidth requirements for the amount of data that needs to be analyzed and aggregated for immediate feedback back to the running machine process
(c) The controller appliance needs to aggregate data locally during the processing of a machine job
because the higher level systems do not need to know each little data transaction generated by the
machine during the job. The higher level MOM functions only need summary data like job start, job end, quantity complete and quantity scrapped. This approach minimizes the requirements on bandwidth due to unnecessary levels of details traveling into the cloud systems.
The requirement for real-time control of machine processes is a very good reason for a local controller appliance. If connectivity is lost between the MOM user and the MOM application for a few minutes, it might be a bit frustrating for the user, but it will not cause a machine process to fail causing loss of invested labor and material for in-process product. The controller appliance would have a cache memory to store a full machine program before it starts a machine run or similar to how video streaming is handled for movies on demand. If connectivity is lost between the MOM application and the controller appliance for a few minutes, the appliance should be able to continue and start transmitting again when the connection is regained.
For example, in industries where the same product is made in large quantities every day for a long
period of time and only summarized information is required at the MOM level, it would be easy to see the MOM functionality hosted in a cloud service. If we only need to know at the MOM level that the machines produced 100,000 units in the current run, it would be wasted network bandwidth and application processing to send 100,000 transactions to the MOM level.
The appliance approach also provides a way to support older machines and controllers in the new Smart Manufacturing systems landscape as newer smarter machines are introduced to the factories. The Smart Machine Controller Appliance would be able to talk “smart” up to the central MOM layer using standards like OAGIS, MIMOSA, ISA88, and ISA95, and talk down to the older machines using protocol standards like OPC UA, MTConnect, EthernetIP CIP, Profinet, Modbus, MQ Telemetry Transport (MQTT).
The distributed layered model also extends into higher systems levels that operate between global
locations, into the supply chain, or into customer systems. There can be a layer of systems on top of
the MOM layer that only needs data aggregated at higher levels. At higher levels of the organizations
and business processes, we might only need work order, or project, or daily numbers. We might not need to know the details of each machine job when we are looking at performance metrics for the
organization.
Smarter machines are coming
As the standards evolve for the IIoT, newer “smarter” machines will enter the factories with the
capabilities of the Smart Machine Controller Appliance already built in. The detailed machine
transactions will be gathered in real-time from intelligent sensors and other monitoring technology
that will provide current machine conditions, the state of the production process in the work cell,
state of automated material handling, and local analysis of data for SPC control, alerts, and
programmed automated self-adjustments. Smart machines might also be equipped with sensors that verify product quality during the production run instead of waiting until the entire run is complete to do manual sampled inspection.
When machines become smarter with on-board controllers and diagnostics, the traditional functions of the SCADA system are handled by a combination of functions at the smart machine and at the MOM layer. For example, a MOM function can provide a view of the manufacturing process status or manufacturing assets’ status on a plant layout diagram.
Apps become smaller and focused
Today’s enterprise systems landscape is full of monolithic software applications that have been loosely interconnected at the database level, operate inconsistently, and are not optimized for any one function. If interoperability standards are embraced for integration of future smart manufacturing systems, not only inside the plant, but among global locations and the supply chain, I can see a future where smaller specialized apps (short for applications) are easy to incorporate into the manufacturing systems landscape. Smaller apps would be optimized for specific functions and specific users and specific hardware platforms (phone, tablet, work station or wall mounted monitor).
References:
- “Lessons Learned From Cloud in Manufacturing Industries”, Hagemeyer, Koslowski, Halpern, Scheibenreif, and Shanler, Gartner, 2014
- “The Evolution of Manufacturing Software Platofrms: Past, Present, Future”, Mark Davidson, LNS Research, 2013
- “The White Book of Cloud Adoption”, Ian Mitchell, Stephen Isherwood, and Marc Silvester, Fujitsu, 2011
Will companies move data to the cloud? It is hard to talk about data as one thing. There are different types of data with different detail, persistence horizon and sensitivity requirements. I don't believe an organization has to persist all or any data in the cloud to use cloud services but that should be an option explored. Some uses of cloud services might require data sent as part of communication exchange. Cloud services will provide encryption protection schemes. Organizations can also persist sensitive data on-premise and join to data on-cloud. There are many options for different applications. And by the way , many organizations are trusting their customer and sales information on cloud apps like Salesforce today. That trend is growing.
Posted by: Conrad | March 13, 2015 at 09:22 AM