Machine Data For Manufacturers - Key Metrics You Should Be Tracking

The increased adoption of the Internet of Things (IoT) has bombarded manufacturers with a flood of data that they can hardly handle with traditional information technology methods.

And as factories become more tech-intensive, it has become imperative for manufacturers to find ways of tracking and analyzing this machine data in order to remain competitive.

This means manufacturers need to zero in on the key metrics that affect their operations, allowing them to monitor and optimize production across their shop floor.

But before we can get into that, here’s a quick recap of the basics.

With the sheer automation of the mechanized sectors of a manufacturing operation, a trove of data leaks from the machines and devices installed in the factory.

In simpler terms, machine-generated data is just that—data originating from the devices on the production line and other factory areas. Every action and decision made by a machine without direct human input is logged in a system or database, creating machine-generated data.

This is a shift from the traditional methods that relied heavily on proprietary software to log and sort through different non-uniform data from factory devices. However, the current deluge of data requires more processing to provide actionable insights for teams controlling the operations.

Here are some of the different types of machine-generated data originating from a typical manufacturing plant.

As manufacturing operations become even more complex, so do the kinds of devices required on the factory floor. Consequently, these machines churn out different types of data in their various functions.

As such, machine-generated data can be loosely grouped into these types:

• Sensor data: Because machines run near-continuously during their scheduled run time, they go through different functions and processes that give off valuable data. Sensors in the machines pick up pressure, temperature, humidity, vibrations, acceleration, and even power levels. This kind of machine data collection plays a central role in keeping a finger on equipment health through overall equipment effectiveness (OEE), informing predictive maintenance and troubleshooting decisions.

• Log data: Some machines use databases, logging various kinds of data that is key to analyzing performance and other critical elements on the factory floor. Additionally, various applications, web servers, and file systems included in manufacturing processes can be ideal sources for this machine-generated data.

• Network data: A factory floor tooled with connected machines and smart devices generates plenty of network data as these instruments communicate with each other during the course of production. Not only do the machines interact wirelessly via edge devices, but they can also do so through wired connections over local networks. As such, analyzing machine data makes it easier to keep an eye on the integrity of the network over which the machine data travels.

Watch Our On-Demand Webinar: Holistically View Your Operations With Machine Monitoring →

Manufacturers that embrace digital technologies in their operations are better equipped to optimize processes, ensuring that production moves along smoothly and improves over time. The degree to which a business is able to drive continuous improvement depends on how and from where they implement machine data collection efforts.

Before embarking on this collection journey, it’s imperative to comb through your business processes to determine your current machine data collection capacity and identify potential gaps in data collection.

Subsequently, you can roll out various solutions—unique to your operation—to obtain the necessary data to drive improvement.

Here are some examples of key sources for machine data collection:

• On-floor machines: The very machines on the factory floor are a great source of valuable data. Modern machines are fitted with various sensors that relay different forms of data. This data provides updates on the status of the given machine, allowing operators to make informed decisions as necessary. Additionally, some plants have edge computing that analyzes this data, eliminating the need for human input in the machine optimization process.

• Connected systems: Machines on the floor don’t always give the complete picture. Therefore, using connected factory systems allow you to integrate external applications to gain a more holistic view of data produced by machines and industrial equipment.

• Human input: Machine tools and connected systems can only do so much to collect vital data. In several instances, you need operators on the floor to manually input some data to give the set more context, making it easier to make informed decisions.

With all the above in mind, here are some key manufacturing metrics you should be tracking:

• Production volume: This is the amount of product that your plant can get off the production line.

• Machine uptime and downtime: Also known as run time, this is the actual time a machine operates in a specific period. This metric highlights time wasted during stoppages, breakdowns, or shift changeovers.

• Throughput: The amount of product that a machine churns out during a specific period. This metric can also apply to the entire production line to check its efficiency.

• Overall Equipment Effectiveness (OEE): A measure of productivity, OEE outlines the share of time a machine works at peak performance. The metric is a product of machine availability, performance, and quality.

• First pass yield: This is the share of products coming off the line that have no defects and meet specifications without any rectification work required.

• Mean Time Between Failures: MTBF shows a manufacturing operation the operation time lost to equipment failure. As such, it’s also an indicator of a machine’s reliability.

• Mean Down Time: Referred to as MDT, this metric is a comprehensive indication of the time it takes for repair and maintenance. It includes any delays related to the time it takes for replacement parts to arrive and time lost due to a technician’s ability.

• Energy cost per unit: This is the cost of electricity, steam, oil, or gas required to produce a given product unit in the factory.

Once you have collected the key machine-generated data you need to track, the next crucial step is to turn such data into values. Having machine-generated data only is not enough. Research shows that while data lakes are becoming commonplace in the industry, scientists are still spending 80% of their time cleaning data on a spreadsheet instead of running analyses and refining algorithms. Inefficient data lakes are leading manufacturers to the trap of data-rich, information-poor.

Data-rich, information-poor (DRIP) is a syndrome in which organizations are rich with data but lack the processes to utilize such data and create competitive advantages. Unfortunately, DRIP has become the defining description of many manufacturers who have invested heavily in technology: their data overflows, yet they lack the bandwidth to do anything with that data beyond printing it out in PDFs and creating a few charts and graphs.

To get out of the DRIP trap, manufacturers need to enrich machine-generated data with human inputs and provide the where, how, and by whom of data collection. Data contextualization is the action of adding human inputs to machine-generated data to elevate data to information. It allows your business to gain meaningful Insights that are actionable and is a fundamental step in the journey to predictive and adaptable manufacturing systems.

By combining different data sources and types, data contextualization empowers frontline workers with real-time information and actionable insights to make timely and impactful decisions right on the shop floor. This means not only improvements in efficiency, quality and productivity but also a new level of agility that can help organizations scale fast and sustainably.

To learn how you can start contextualizing your machine-generated data and turn it to real values, listen to our webinar on data contextualization and implementation.