Need direction on your lean manufacturing journey? Listen to the data

Know you know, don’t assume
By Jeff Sipes 

Listen. Can you hear the data speaking? Is it whispering … or screaming? Does it point you toward constructive improvement, or are you just putting out fires?

Improvement starts with having the appropriate data. All work happens in a process, and analysis helps us understand the work happening in that process.

“Don’t jump to conclusions” is an often used idiom for a reason. When you do it, you often find the conclusion was wrong. Why? More than likely, you jumped without data or analysis. Relying on feelings, intuition, and opinion can quickly lead you to a wrong conclusion, which can lead to rework, disbanding your previous work, and continued firefighting. This can negatively affect customer satisfaction and, even worse, the safety of employees.

Do You Really Know?

This all comes down to the distinction between thinking you know something and knowing you know something. Let’s say something happened to the welding process for a product in the shop. You know the welding department had to scrap 12 pieces, really just an outcome or lagging indicator, but otherwise you have no data.

No matter. You’ve always been action-oriented, so you quickly determine that John, the second-shift welder, must have made an error. John’s new to the company, after all, and he’s learning this particular weld process. Besides, he’s had attendance problems. It seems logical, so you immediately talk to John and his supervisor on second shift.

But is John really the problem? Do you really know or just think you know? What data would increase the probability that you really know what happened to those 12 pieces? For starters, you could confirm that John did indeed produce those 12 pieces. If he did, then look at the weld. Does it have excessive porosity, too much or too little weld bead, or was the bracket welded slightly off the target location? If you find one of these conditions, ask yourself, how do I know? Does your nonconformance reporting clearly define the root causes? If it doesn’t, well then, how do you know? Each of these three conditions—porosity, weld bead size, and bracket positioning—will take the analysis in different directions.

If the weld has a porosity problem, was the correct gas being used? Was air around the nozzle disrupted by unexpected airflow because of poor shielding? Was the material covered with moisture because it was sitting outside immediately prior to welding? If the weld size was wrong, was it because the welder could not read the print, or was the print unclear? If the bracket was mislocated, were there worn locators? Was the wrong bracket used? Or did someone misread a tape measure?

Data and relevant information will get you to the root causes. Jumping to a conclusion could take you to a completely irrelevant place, and the problem will persist. Make sure you know.

Types of Data and Analysis

Let’s take a look at takt time and cycle time analysis. Takt time is the pace at which you need to operate to meet customer demand, determined by dividing the available time by the units of demand. If the available time is 450 minutes (480 minutes in an eight-hour shift minus 30 minutes for breaks and lunch) and the demand is 45 pieces per day, then takt time equals 10 minutes per piece. To meet the demand for the weldment, you need to start a new piece every 10 minutes.

Now consider the cycle time—that is, the time it really takes to produce the product. In the case of the weldment, you need to know how long it takes to complete one piece. The cycle time is an important piece of data in this calculation. Do you know the cycle time, or do you think you know? Has someone observed the work, captured the cycle time with a watch, and looked for examples of waste affecting the cycle time?

If the cycle time is 12 minutes, you have a capacity problem. You cannot make enough weldments in the normal eight-hour shift to meet your customer demand. If the cycle time is 7 minutes, you have excess capacity. The takt time and cycle time data are speaking to you.

Another basic but powerful analysis tool is the run chart, which shows the pattern of occurrence over time. The data must be plotted on the run chart in the sequence in which an event occurs. This might be the quantity of units produced per shift on the assembly line or the number of quotes processed per day. Whatever it tracks, the run chart gives you a picture of the item you are measuring.

Sure, you might have the data in a spreadsheet, or perhaps individual copies of quotes in a physical folder. The data is there, but it is not user-friendly or easy to see. The run chart makes trends visible, showing spikes and valleys. For instance, if you’re tracking order quantities for a product over the year, a run chart can help make seasonality more apparent.

The Pareto chart, another basic analysis tool, shows the number of occurrences of variables or events by category over time. Consider a press brake operation that is not performing up to standard. You have too much unplanned downtime and are not sure how to improve the situation. Just charging forward with a gut feeling might be dangerous and counterproductive. The Pareto chart provides structure to help you critically understand the process. Once again, it makes the data speak.

Let’s suppose you gather data over a week and find the press brake had 22 occurrences of unplanned downtime. That by itself might get everyone’s attention (the machine was down how many times?)— but it provides little guidance for improvement. Go one level deeper in the data and you find that tools crashed two times, a fault sensor stoppage occurred six times, a hydraulic valve stoppage occurred two times, a hydraulic fitting leakage occurred three times, an electric motor reset occurred eight times, and a backgauge adjustment occurred one time.

This is better, but in narrative form this simply looks like a stew of things that went wrong. Put it in a Pareto chart with the item occurring most frequently first, and the rest in descending order, and you achieve greater clarity and focus for improvement. In this case, the Pareto chart shows that the electric motor reset and the fault sensor stoppage represent 64% of the occurrences. In short, the Pareto chart makes your data speak to you.

These are just some basic analysis tools. If you want to go deeper, then consider investing in Six Sigma skill development. Start with Green Belt-level skills to complement the lean manufacturing skills your team already has. If you want a few employees to become data specialists, you eventually might send them to Black Belt training.

Know You Know

You need to know you know, not just think you know. Thinking you know is simply too risky in modern business. As you progress in your lean journey, lead times will shorten, inventory will shrink, and work velocity will increase. You have little margin for error.

As you become more data-savvy, you’ll find you have many sources of data, sometimes too many. As you sort through it all, look for ways to make the data speak. Make it visible. Put the data and analysis in charts and graphs. Be creative and make it user-friendly while still keeping it correct and relevant.

Data is the oxygen for analysis and improvement. Sometimes your data will whisper and other times it will shout. The question is, are you listening?

Original Article: https://www.thefabricator.com/thefabricator/article/shopmanagement/need-direction-on-your-lean-manufacturing-journey-listen-to-the-data