Three pillars for digitalization success
The path to a digital production facility can be incremental, capturing benefits along the way through the proposed approach of tying lean practices to data collection so users throughout the facility have access to visualization and analytics.
- Learn an approach to developing and implementing industrial Internet of Things in a production facility.
- Tie lean practices to actual data collection enhances decision-making.
- Understand the importance and opportunity of ongoing communication for all activities.
- Companies incorporating digitalization and digital transformation into their facilities can glean more insights and improve efficiency in their production facility.
- Using a lean approach to this process and using short sprints can help companies make small, but incremental, gains where they are needed and use them as a proof of concept when expanding its scope.
Phoenix Contact’s facility in Bad Pyrmont, Germany, is an electronics production facility that implemented factory digitalization. This location has been operating as a smart production facility for several years. While digitizing, the plant operators identified and implemented projects that improved efficiency by more than 10% per year. They also found they could go from idea to execution up to five times faster. Wrapping digital production information around existing lean and agile processes helps make this data easily accessible to employees. This creates a feedback loop that can increase production.
Phoenix Contact’s U.S. production facility, located near Harrisburg, Pennsylvania, has 70,000 square feet of production space divided into electronic and electromechanical production. U.S. manufacturing operations formally started in 2003, with custom products for the U.S. market. Development capabilities were added incrementally and production grew to include standard catalog products for global distribution.
In 2021, Phoenix Contact started a five-year, $19 million investment program to better serve organic growth in the North American market, reduce supply chain risk and better use the facility’s overall assets.
The program consists of nine projects divided into 39 subprojects. Six subprojects are focused on creating a fully digital factory. The effort has been planned as a part of daily work, with existing production staff implementing the activities. The same team is also responsible for deploying the new machines in the expansion projects. This makes prioritization of floor activities critical. While it slows the speed of the journey, it is a practical approach. Short “sprints” on pilots or events allow experimentation, proof of concepts and quick feedback about new activities.
The Bad Pyrmont electronics production facility became a template for potential success. The U.S. facility is different, which meant Bad Pyrmont’s approach couldn’t be copied. In addition to electronic and electromechanical production, Harrisburg has a legacy manufacturing execution system (MES) on many workstations and most automated machinery. The different production areas run almost independently and all information systems and key performance indicator (KPI) reporting were established in the same way. While Bad Pyrmont’s approach could be followed, the project team needed to adjust for the equipment and reporting tools already in place.
The digitalization approach consisted of three pillars: Shop floor management, data collection and analytics and visualization.
1. Shop floor management
Shop floor management is the institutionalization of the lean process. The team has been practicing lean techniques for more than a decade. Phoenix Contact typically held Kaizen events when identifying issues or installing new product lines. The shop floor management program makes these activities daily habits.
This subproject began in mid-2021 when the project team expanded the daily Gemba walks to all production areas and shop floor management meetings using near real-time data from the process (Figure 1). The walks and the meetings focused on addressing issues immediately with the personnel as close to the events as possible. Regular 5S events (sort, straighten, sweep, standardize and sustain) occur, with results being tracked. Process improvement activities run continuously.
2. Data collection
Data collection activities began in June 2022. Data is required from three unique categories of equipment: The MES, equipment that is connection-enabled primarily through Open Platform Communications (OPC) or an application programming interface (API) and stand-alone legacy machines and workstations.
The MES was initially deployed for shop floor scheduling in one of the electromechanical production areas. All automated machinery built internally also includes connectivity to the MES. It measures operational equipment effectiveness (OEE) and benchmarks operations with similar machines globally. These two applications will be merged to expand shop floor scheduling to automated machinery and measurement of the entire process value flow. All data is available only through centralized enterprise applications.
It also is possible to make this data available to visualization tools for KPI reporting and analytical tools selected by the end user. Standardized APIs provide access to user-specific applications and visualization.
Connection-enabled equipment provides access through a standard interface, primarily OPC UA. Most of this equipment is purchased from machine manufacturers that provide connectivity as standard. Examples of this equipment include surface-mount line modules, automatic test equipment and semiautomated packaging. New custom or semi-custom locally sourced machines have an open-source controller, which also acts as an message queuing telemetry transport (MQTT) server. This provides the same connectivity as the stand-alone machines and workstations (Figure 2).
Stand-alone machines and workstations make up a large portion of existing assets in production and include manual, semi-automatic and fully automatic machines or work areas. These are controlled by industrial computers or earlier-generation programmable logic controllers (PLCs). The machines are part of daily production, so the project team needed a low-risk, noninvasive approach to connect the equipment. The inputs are often parallel wired in the control cabinet, to the sensors with the desired data. These are wired back to an open-source controller, which acts as an MQTT gateway to the production network.
The method chosen to link all of this together is a microservice architecture. Each data source or the analytical program collecting data resides in containers accessed through APIs. Applications, or individuals with access rights through the API, have near real-time data without other preprocessing or storage. Open-source tools support this effort.
The benefits of this architecture are flexibility, speed and expanded use of the existing workforce. It also does not tie anyone to a single supplier. Applications can be created as required and piloted. If successful, internal specialists then harden and release them to the larger community.
Developers create applications using programming environments they are familiar with. The structure is not supplier-specific, so tools from different suppliers can be used and mixed within the architecture.
The MES is an enterprisewide application; however, through the API, data can be accessed to allow other user-defined applications. For example, the digital twin obtains machine status from the MES to display in a separate visualization tool (Figure 3). This architecture fit into the IT security infrastructure while allowing personnel to work within production network guidelines.
3. Analytics and visualization
The analytics and visualization were in place at those workstations with automated machinery or MES. Operators use these tools to call up, open, track and close work orders. Machine OEE and statistical process control tools are also accessible. Additional access to the enterprise resource planning, MES and product life cycle management provides production KPI data at the direct production, supervisory and manufacturing engineer levels or at the machine or workstation level. The intent is expanding information access the enterprise.
The digital twin was the first step toward this, providing situational awareness. Here is one way to explain it: A user finds something on the operator interface in the same place located on the production floor. If a particular machine is located outside of the production entrance and to the right, the user will find it there on the digital twin graphic. The machine graphic allows access to basic information: whether on or off, the number of units produced or rejected and more. It also includes any current process improvement projects, with status in the process or production area. The KPI board locations in production can be accessed electronically through the digital twin, as well.
Employee engagement generates new ideas
Direct and continuous communication has empowered employees to share ideas for process improvement. All employees in the Phoenix Contact U.S. organization are updated on high-level activities at least quarterly. The leadership team is updated monthly and all production and project personnel receive weekly updates.
This communication and engagement allows greater participation. For example, a mechatronics-technician apprentice involved in one of these updates expressed interest in programming some of the apps outside of PLCs. He was paired with a manufacturing engineer with programming experience to deploy the first MQTT gateway on-site.
In another example, a second-shift mechatronics technician made a hack using a QR code taped to a machine to help him remember the frequently used spare part numbers. He shared this with his supervisor, who recognized the benefits and the risks. They worked with the IT department to build a custom augmented reality (AR) solution.
They built an application to scan the machine using an authorized cellphone or tablet so they could identify and immediately access all machine documentation, including the parts lists (Figure 4). This pilot was successful and is now being scaled throughout the department. Having the data server across the enterprise allows more user access to individuals who are not reliant on others to create reports or access data. This is a soft gain, but it provided real results.
Initial results and lessons learned
The shop floor management program has provided significant gains. Despite supply chain issues, production output grew by 23% in 2022. The Kaizen events improved efficiency between 7% and 33%. The production floor’s new layout has optimized floor space, creating room for the addition of two new product lines. The effort has also resulted in updated value-stream maps, which will help prioritize the data collection activities.
Data collection is moving forward. To date, 30% of all machines and workstations have been connected. The MES API integration has allowed a pilot project to collect and display machine status through the digital twin. That success qualified the concept and has kicked off a project to automatically display real-time KPIs overall. From there, the project team can drill down to individual machines for daily production management. The company has piloted a new pick-and-place machine with their standard interface and developed a schedule for the remaining equipment. Two stand-alone machines have piloted the MQTT data collection concept and scheduled the remaining workstations.
The digital twin is active and it’s been kept updated with new equipment installations and plant relayout activities. The current digital machine status is available. Several machines’ data can be displayed through open-source software. A digitalization steering committee comprising operations and IT personnel has been formed to lead the qualification, deployment and standardization of applications involving analytics and visualization. The first programs involved the expansion of the existing MES applications across all areas of production. This is almost complete.
Two more pilot applications have been rolled out. The first involves labor management and the second involving process quality. Each is in a different production area, but multiple projects can be run simultaneously to help expand knowledge across the site. If successful, both pilots will be deployed throughout the facility.
Two pilots involving AR have been completed and are beginning to scale these to other areas. One is a vendor-specific solution to help onboard new employees. The other is the machine troubleshooting application.
Best practices for future projects
All production facilities and their specific activities are unique, but other organizations can use this approach if they follow a few best practices.
First, the facility needs to define a clear direction or overall purpose of the production digitalization and communicate that to the organization regularly.
Tying lean to the digital effort is important. Actual machine data allows better and faster decision making. Collecting data along a process value stream provides a starting point and immediate feedback following a change. While there are different ways to collect data, making that data available to all users is important.
Run pilots first and don’t try to be perfect. If the pilot will get 80% of the initial scope, it’s good enough, which will help prevent companies from getting stuck. Build on those successes and communicate them to the entire organization.