Data analysis in manufacturing, plant reliability and developing your workforce.

Welcome back, Troubleshooters! It’s that time of the week again… Last week in Part 2 of this series on how to leverage big data in manufacturing, we talked about the overwhelming volume of data that is now available to manufacturers, and how it can create so much noise that you can miss the key signals. Today we’re talking about data analysis in manufacturing.

Hoping for an “Aha!” moment

Not that long ago, when there wasn’t all that much data coming in from their various production processes, manufacturers could look at the few data streams they did have, and mine downwards, waiting for an “Aha!” moment that would reveal an opportunity for increased efficiencies.

We have reached the point now, or at least soon will, when this approach will no longer work. There’s too much data to sift through, and simply hoping to stumble on the relevant signals by sheer luck is not a viable strategy.

What is needed is a more intentional approach to data analysis in manufacturing. Experts are now beginning to recommend a goals-first data management strategy that will help weed out the irrelevant data and focus on the important digital information coming from the factory floor.

Using your data intentionally

Leveraging manufacturing data is all about knowing what you need beforehand. It’s a critical first step before you begin data analysis in manufacturing to take the time to reflect on your business’s needs and decide what information you need to improve operations or processes, and to map out an intentional data plan. Following are some suggested steps for working out your company’s data strategy.

1. Identify your data goals.

The first step is identifying your overall goals for using your data (e.g., increasing profits, decreasing waste, eliminating duplication of effort, etc.). What can data do to help you achieve them?

2. Identify your business objectives.

Identify your business objectives so you can build a data strategy around them. Objectives have specific, measurable targets that will help achieve overall goals (e.g., increase production numbers by 5%, or decrease factory rejects by 10%). Filter out the data streams that don’t speak to your priority objectives.

3. Identify your key data streams.

Streams of data are flowing into your enterprise from many different directions. Which are most relevant to your goals and objectives? Which can be used to impact operations or processes directly? Take inventory of areas where you’re losing money, and consider the data available to you that affects those areas.

For example, if warranty claims are a big drain on revenue (as they are in the auto industry—up to 3% of annual revenue goes right back out the door to handle warranty claims), predictive maintenance can help save money on claims by improving quality control. All of the data streams that come from vibration sensors, temperature sensors, etc., are therefore relevant to catching potential equipment failures early, before they cause quality problems.

Or, if production-line downtime is cutting into your profits (it can cost thousands of dollars for every minute the line is down), find the key streams of data that can be used to reduce it, e.g., analytics from your maintenance professionals’ training program that can tell you which people are the quickest at solving electrical faults in production equipment so they can be deployed to the factory floor first.

It’s critical to identify the essential metrics that lead directly to solving problems, and then screen out everything else that’s irrelevant.

4. Turn data into information.

Recall from last week that data and information are not the same thing. Data needs to be transformed into information if it’s going to be useful. It may be that you need to integrate and analyze several different data sources together in order to have valuable information. This might require the expertise of outside data management consultants.

5. Recognize that human analysis has limitations.

As we mentioned last week, at some point, human beings will no longer be able to monitor and analyze big data effectively on their own. According to data entrepreneur Sundeep Sanghavi, “Monitoring data in isolation limits its utility. Large-scale interconnected processes require a heterogeneous approach to monitoring in collaboration with the entire universe of the system.” Your data must be accessible across national and international operations and not isolated in silos, and this may require Big Analytics to integrate and sort through your data and make real time, money-saving decisions on the floor. We’ll take a look at that in greater depth in the coming weeks.

And that’s it for today, Troubleshooters! Join us next week for Part 4 of Cutting Through the Noise.

Follow Troubleshooting Thursdays

Tune in to Troubleshooting Thursdays for reliable tips, general troubleshooting process, and industry insights. Stay up to date with Simutech Multimedia:

Have a subject you would like Troubleshooting Thursdays to cover? Send Simutech Multimedia an email at [email protected].

Looking to give simulation learning a try? Get started with our award-winning first solution here: Get Demo.

Get started with your demo

Share This