What CEOs Need To Know About M2M – Part 3 (How Machine Usage and Behavior Data Can Provide Business Insights)
Today’s posting will focus on machine data analytics. In my previous posting on what CEOs need to know about M2M, I referenced the Axeda value curve and discussed level 3, remote service and monitoring. Today, we’ll discuss Level 4… machine usage and behavior analysis… how collecting the right machine data and using the right tools to analyze the data can drive better products and services.
In a recent survey of our customers, we discovered that 87% of our customers are storing historical machine data.
What can machine data analysis tell us?
What are the benefits of building a data mart or big data system for machine data?
Here are the top 3 business drivers we’ve seen for machine data analytics:
- Predictive Maintenance – To decrease the cost of maintaining machines and to improve up-time
- Improved Product Design – To understand end-user behavior and usage patterns to design better products and prioritize new features
- Identification of Quality Issues – To understand what’s causing down-time; To identify issues with design, embedded software, part suppliers and manufacturing processes
Let’s look at each driver starting with Predictive Maintenance. If your company has equipment that needs to be maintained by you, your partners or your customers, then there are some basic questions to ask your IT, Product Management and R&D organizations:
- What machine data can enable us to proactively service machines to drive greater uptime and shorten Mean Time Between Failure (MTBFs)?
- What sensors do we have or what sensors can we add to help us detect failures before they happen?
- What patterns in machine data readings are we seeing that are early indicators of failures?
Answers to these questions will provide Engineering, Product Management and IT the requirements for capturing the right raw data on the machine. It will help field service proactive schedule maintenance or parts replacements, rather than over servicing equipment or waiting to be reactive to failures.
Product Design can be improved with machine data. Behavior data from your machines can give you insights into how end-users are using your machines. This data can be used as input into your next generation product requirements. It can help you design a product that responds to real world use cases.
Lastly, quality issues can be identified by analyzing machine data. Today, analysis of trouble tickets and field visits can provide indicators of issues in your machines. But rather than waiting for calls and tickets to accumulate, connected machine data can be collected and analyzed to detect patterns and identify issues much sooner. The patterns observed can document normal operation and identify exception patterns. These exception patterns can be used to detect glitches in software or hardware design or correlate issues with parts suppliers.
Machines can also have issues resulting from flaws in the manufacturing process. Understanding the relationships between problems and specific batches or production runs can identify a bad batch early and streamline the recall process. It is also possible the problems are more serious and still in in the current manufacturing process. In that case, the data may trigger the need to change the current manufacturing process. Net/net... If you can understand what’s causing down-time, you can identify any flaws in the manufacturing design or process of your machines.
In my next post, we’ll continue the business case discussion and move on to level 5, integration of machine data with CRM, ERP and PLM. M2M Application Platforms of today are becoming very good at turning raw machine data into standard IT formats and web services that other systems can consume. I’ll give several examples of how machine data is working its way into our customers back office systems to improve efficiency or enable new services.