Shop floor data collection systems 101

When you think of all the data that can be collected off the shop floor or from the field, all in the name of Industry 4.0, company executives are both excited and nervous – excited about the prospects for automated monitoring and control of operations but overwhelmed by the sheer volume of data and options available. Shop floor data collection (SFDC) systems provide an accurate and timely audit trail for tracking a wealth of information such as labor efficiency, product quality, machine utilization, energy consumption, deviations from standard, and so on. It has, therefore, equal relevance to Operations and Maintenance of assets, products, processes, and the surrounding environment.

SFDC systems consist of four components: data input, data manipulation, data output, and system architecture.

Data input

There are three types of data input devices:

Measurement devices. These are used to determine the attributes or properties of things, or the state of the environment. For example, a thermocouple is used to determine the temperature of a product and/or its environment.

Detection devices. These are used to test for the presence of things, to discern changes to the characteristics of things or the environment, or to determine if things or the environment exhibit certain attributes. Examples are, respectively, a photocell detecting the presence of a carton, a motion detector sensing the movement of a piece of equipment from point A to point C, and a smoke detector which is triggered by some threshold density of smoke.

Recognition and identification devices. These are used to sense and differentiate the pre-determined attributes of things themselves; or code describing things, the environment, or the process. For example, suppose the weight of finished product has been previously determined and encoded in the form of a barcoded label. A unique serial number and the date of manufacture has also been encoded on the same label. A barcode reader in the shipping department scans and identifies the product in order to update inventory records. If the interval between readings exceeds a pre-determined value, downtime is assumed and maintenance may be necessary.

Some of the more popular technologies used in recognition systems include barcode, magnetic strip, optical character recognition (OCR), transponders and inductive tags, radio frequency identification (RFID), charged couple, data entry terminal, speech recognition, and machine vision.

Data manipulation

This is affectionately referred to as “number-crunching”. The brains behind SFDC systems are essentially instrumentation that varies in shape, size and functionality. Many assets today carry microprocessors and sensors capable of collecting and manipulating data directly from the asset. Other assets require programmable controllers or other intermediary equipment to monitor and control many input and output devices.

David Berger, P.Eng. (AB), MBA, is president of The Lamus Group Inc., a consulting firm that provides advice and training to extract maximum performance, quality and value from your physical assets, processes, information systems and organizational design. Based in Toronto, Berger has held senior positions in industry, including for two large manufacturers, and senior roles in consulting. He has written more than 450 articles on a variety of topics such as asset management, operations management, information technology, e-commerce, organizational design, and strategy. Contact him at [email protected].

Some of the more common applications are as follows:

Labor reporting. The simplest automated time and attendance systems require the input of information describing who is present in Operations and Maintenance. This can be accomplished using employee badges with barcode, magnetic strip, RFID or chip technology. More complex labor reporting systems use the same technology to input changes made by employees in terms of current location or job being worked on. This information is, in turn, used to determine production / maintenance labor efficiency.

Materials management. This includes tracking receipts, work in process, and finished goods produced, as well as shrinks and gains throughout the manufacturing process (e.g., yield, scrap, rework, refeed, pilferage). Material tracking systems require quantities to be input either automatically using sensors, or manually via a keypad or scanner. Actual quantities produced would be compared with expected outputs for variance analysis, work-in-process tracking and so on. The maintenance department must ensure that assets produce maximum expected output and quality.

Overhead cost control. Machine utilization, indirect material usage, energy consumption, and downtime are examples of overhead cost factors requiring control. As well, SFDC systems can be used for condition-based maintenance, for example, feeding asset meter readings directly into the CMMS for comparison to PM trigger points. This implies that the CMMS can be used to alarm maintenance management of out-of-control conditions, such as a machine cycling incorrectly.

Data output

Data is continuously collected, manipulated and presented in a useful format to workers and management of both Maintenance and Operations. Output devices include monitoring devices (e.g., display screens, large-character display boards, alarms and printers), control devices (e.g., relays, switching equipment, and regulators), and storage devices (e.g., the cloud, hard disk, RAM). Data collected may also be sent in raw or summary form, directly or indirectly via the cloud to Operations through an ERP system, and to Maintenance through the CMMS.

There are four key forms of output as follows:

Graphics display. One of the most basic SFDC functions is to graphically represent the entire manufacturing process. This might take the form of large, hard-wired panels that take up an entire wall, or multiple pages of graphics on a simple monitor. Regardless, operators can track all processes in a plant from the comfort of an environmentally-controlled room, on-site or from a remote location.

For example, an operator can monitor a graphical display showing a vessel filling with liquid. Graphical displays are hierarchical so operators can zoom in and out by double-clicking on marked hot spots. The condition of the process or product can be monitored using colors (e.g., a heat map) or actual readings. If the pressure or temperature goes beyond a given set point, the graphic vessel might turn red and flash. Real-time temperature and pressure readings can then be displayed adjacent the graphic vessel.

Some vendors offer simulation or what-if capability. This can be useful for a number of reasons, including as a scheduling aid, or to help operators practice procedures for handling anomalies. Furthermore, digital twins (i.e., virtual equivalents to the physical world) can be used to simulate and test new automation ideas and algorithms prior to an expensive implementation.

Alarming. This is usually displayed in tabular format showing individual events that triggered an alarm condition, using colors or other techniques to denote severity of the alarm. The event could be a SFDC device that has measured, detected or recognized a certain condition, trend or trigger. Each line on the table represents a given alarm event, usually cross-referenced to jump to the appropriate graphics display screen.

Trending. Graphs can be displayed showing how shop-floor data collected is trending. Multiple readings can be viewed on one graph with user-defined time periods showing actual data points, upper control limits, lower control limits, and statistical tendencies.

Management reporting. As with most computer systems, online or printed reports can be prepared for exception reporting, variance analysis, reliability analysis, and other analytics. Artificial intelligence can be used to supplement analytical tools.

System architecture

In today’s world of IIoT and Industry 4.0, there are many options for architecting the system, from a centralized, cloud-based solution to more decentralized strategies such as edge computing and low-power wide area networks (WANs). Localized computing is gaining popularity as the cost can be lower, response time faster, and less bandwidth required compared to the cloud. Cybersecurity may be both an advantage of edge computing in terms of the reduction in data travelling over the network, and a disadvantage in terms of opening back-door entry points if improperly designed.

This story originally appeared in the August 2021 issue of Plant Services. Subscribe to Plant Services here.