How many edge devices does your Industrial Cloud Analytics Solution require?

How many edge devices does your Industrial Cloud Analytics Solution Require?

Steven Garbrecht - Chief Digital Architect
Steven Garbrecht - Chief Digital Architect

I have worked with a few cloud analytics solutions over the last few years and they all seem to have one thing in common. If approached in isolation, the cloud architecture may specify several IOT edge devices; as many as one for each machine in your plant. Furthermore, you may utilize multiple cloud vendors, where each may require a different IOT edge device to connect. An example may be one vendor for asset performance management and another for manufacturing operations management analytics. This may be reduced if you go with a single vendor but this is not always possible as it limits choice. These devices are typically inexpensive, anywhere from $100 to $1,000, but do you really need that many connection points to accomplish the job?

For each cloud edge device that you have, there’s a technical debt that is incurred in the form of a new connection point to be secured and also the ongoing need to send updates to the device periodically as cloud software revisions are released. This is called Continuous Integration and Continuous Deployment or CICD and is something that DevOps-based systems have been taking advantage of for many years, behind the scenes for ERP systems for example. This works well in the business world, but there are special considerations for real-time operations management and manufacturing environments in the area of availability, network performance, security and systems management where an architecture that requires a myriad of connections can be problematic.

There may be an even more fundamental problem in that many cloud software vendors have a great deal of expertise in IT-based solutions, but when it comes to organizing information at the shop floor level, with a myriad of PLCs, control systems, historians, databases and sensors, they lack the specific knowledge on how to approach things in a more holistic fashion. Fortunately, there are solutions today that scale very well and take into account security, communication performance, common namespace as well as development productivity. By using a Data Bridge configuration, the need for multiple IOT gateways to enable a cloud solution is dramatically reduced. One such product that can enable a Data Bridge solution is Siemens WinCC Open Architecture (OA) software which is a SCADA platform and a whole lot more. In many cases, a single gateway can be employed to accomplish the transfer of real-time and historical information to the cloud. Furthermore, the results of analytics and number crunching that occurs at the cloud level can be sent back down to the shop floor via the Data Bridge to adjust setpoints automatically based on analytical results. This provides a secure, bidirectional interface with enterprise applications and can close the loop on supply chain optimization.

Be an Army of One when it comes to Digital Transformation

Be an Army of One when it comes to Digital Transformation

Steven Garbrecht - Chief Digital Architect
Steven Garbrecht - Chief Digital Architect

It’s often said that if you want to change the world, start by changing yourself first. For the last few years, I’ve been working with customers on large digital transformation projects. Becoming a digital enterprise is a goal for many large industrial organizations. At the same time, Analysts Firms like LNS Research identified back in 2016 that “around 80% of manufacturers have no data historians or manufacturing execution systems (MES) and many with little automation at all, especially in discrete manufacturing industries”. Furthermore, Gartner research mentions that “70% of captured manufacturing data goes unused”. We need to start at the plant level to make improvements that will benefit the entire Manufacturing Corporation.

So where do you start to make improvements? Let’s say that today you have a few Programmable Logic Controllers (PLCs) controlling the more repetitive parts of your operation and the rest is being done with clipboards, Excel and poster boards where the paper-based daily reports are displayed.

Start with an assessment of the outcomes you would like to have:

- Ability to generate the daily and weekly reports automatically
- Ability to automatically monitor OEE by manufacturing line or cell and identify bottlenecks
- Ability to automatically count inventory at each process step to understand material in process
- Ability to feed work instructions through a central repository to operators and machines

Next move from the data that you have to the data that you need

- Assess the PLC information you have already been logging
- What additional sensors and data loggers will be needed to capture the difference?
- Don’t forget the smartest sensor you have in the plant, which is your operator. How best to capture their manual information in the system?

Look at the capabilities that you need:

- Visualization of process parameters
- Ability to enter reason codes for downtime
- Collection of time series data and events in a high-speed database
- Ability to generate reports on demand and at a scheduled basis
- Provisions for mobile access to information and operator rounds data
- Ability to send setpoint changes to controllers based on operator actions
- Notifications to the right people when help is needed
- Ability for remote specialists to look at the operation and help to troubleshoot problems

Finally consider the technology that’s available today to do this:

- A flexible SCADA System that provides visibility, control and monitoring for both operators and supervisors
- A high-speed data historian with reporting, trending and data analysis capabilities
- Workflow software that can be employed to consistently execute procedures
- An alarm and event subsystem that can notify people when things go out of bounds or problems arise
- Mobile applications that can be used on iPads, cell phones and other mobile devices
- KPI dashboard software that can be displayed on monitors strategically positioned in the plant to give feedback on production progress and whether people are ahead or behind goals
- Batching recipe management software that can automatically download operating parameters without inducing human error as part of the transfer
- Temperature, current, pressure, inductive and proximity sensors to convert the physical world into meaningful digital information

These are things that can be done well ahead of any large digital transformation initiative and will help to generate and organize your data and information at the plant level to be leveraged in future cloud-based big data and analytics projects. At the same time, providing immediate value to the local operation in worker productivity, quality, throughput, inventory reduction, asset management, machine availability, energy reduction and operations agility.

So, think globally and act locally in your digital transformation journey. 

The Three Rules of Industrial Operations Management and Industrial IoT Applications

The Three Rules of Industrial Operations Management and Industrial IoT Applications

Steven Garbrecht - Chief Digital Architect
Steven Garbrecht - Chief Digital Architect

I've been thinking about the work I've been doing over the last couple of decades and what the basic requirements are for any new product coming into the industrial operations management space. This is based upon my experience working in DCS, SCADA, Historians, MES, ERP, APM, Optimization, Simulation and other operations management application areas. The formula is simple and any startup company should consider these three aspects when developing a new software product for the industrial market. As with anything, these statements are directional rather than absolutes.

Rule 1 - It should add to what you have rather than replacing

In the world of industrial plants and manufacturing there are very few new or greenfield opportunities. Many of these plants have been around for decades and have seen several generations of computer and automation technology applied to their sites. There are literally thousands of hours of engineering work that's gone into the configuration of these control systems, databases and applications. The key is not to replace what is working. This is the rallying cry of all engineers. If it isn’t broken, don't try to fix it. Therefore, the real value comes in adding something to what they already have that gives new insights and provides operations optimization what tells them how to make improvements using the data that they already have. A great example of this was the historian technologies that came out in the late 1980s. They were placed on top of distributed control systems and provided information management for people outside of the plant floor environment. A whole series of applications spawned from the use of this historical information. It was truly revolutionary. Another great example is SCADA systems that were developed in parallel with the acceptance of Microsoft-based minicomputers and added value to existing PLCs. The next big series of applications will come via cloud and edge-based systems that leverage data from existing automation and other databases but do not attempt to replicate functionality that works today on premises. Tying in the supply chain with these type of applications will provide a revolutionary way to approach solving problems that have been plaguing industrial operators for years.

Rule 2 - It integrates with everything you have today

Ever since I started to work in this industry in 1991, every position I've come to has identified the problem of silos of information as plaguing industrial operations. Whether it was multiple control systems, or multiple plants that needed to be visible across the enterprise, or looking across disparate applications such as HR, Quality, ERP and MES, the story has remained the same. Integrate together what I have and hold back no capabilities in integrating to everything I have. One of the keys is to make sure that you have connectors for all the different data sources inside of the plant or industrial operations environment. This includes real-time information sources like control systems and PLCs, and also transaction-based systems. If you don't have the interfaces to the systems then partner with vendors that provide interfaces that can convert the information into a form that can be leveraged in your application. This can't be stressed enough as the very first project you will encounter will require that data source that you don't have an interface for today. It's also important that you provide easy mechanisms for making this connection that doesn't require hours of services work to pull off. The more automatic the connection, the better. Consider API toolkits to allow partners to create new interfaces and make it open source so the project community can help build your interfaces.

A great example of this are device integration servers that are used to connect to different types of PLCs and expose the data in a common structure that can be used by a supervisor control program or by an Historian to log the information in a format that can be used by many different programs. Another are data interfaces from ERP, LIMS, EAM or MES systems. It's surprising how many new applications coming into the market do not make it easy to interface the core application to other data sources without a lot of engineering effort.

Rule 3 - It can be customized to meet your specific business needs

Every manufacturing plant or industrial site is in fact a snowflake. It has specific configurations and ways of doing things that need to be accounted for. This isn't to say that an organization should not try to standardize their operational templates across multiple sites to drive standardization or common ways of doing things. This is needed to provide comparable measurements across facilities for things such as downtime, quality and production throughput KPIs. If you think of a new application project as a continuous improvement project, there are certain aspects that need to be configured to exactly match what the operation needs to do at that particular site. We talk about “leading with lean improvement techniques" and then adding digital capabilities to capture the data and provide insight that allows your organization to further lean out. The more configurable you can make your industrial solution, the more easily it will be accepted by the masses across many types of industries and organizations. The key is making this customization up-gradable and maintainable through a standard set of tools and processes. That way as the technology improves, the customized configuration can improve with it.

One good example of this is a SCADA system that can be easily customized to meet specific visualization requirements and operations procedures for a manufacturing plant. Also, an MES system that can be configured specifically for how the operations work within a plant. Or an EAM system that can be customized to monitor for certain events and failure mechanisms for asset performance management.

As a packaged application provider or SaaS solution developer, if you follow these three rules in developing new industrial operations management or IoT applications, the adoption of your products will be much faster and you won't have to make radical architectural changes to meet new customer requirements as the fundamentals will be built-in. Take this for what it's worth as one person's opinions having worked in the marketplace for a while.