Data Collection in the Factory

One of the toughest challenges that a manufacturing black-belt faces is institutionalizing data collection systems that yield project-critical information. Whether the needed data is attribute or variable, it is still a tough challenge to implement the system.Once the GR&R issues are resolved, it doesn’t get any easier. At this point, the black-belt has to make it happen on the floor. Many of us have been there, standing on the line for hours training the personnel on the proper measurement procedures required.

If you are an integrated black-belt, you have even less time to dedicate to the system implementation, yet the need for the system still exists. I’ve found that in this case, it helps to obtain maximum leverage from resources at your disposal.

Some of the best sources of leverage are the front-line data collectors. These people are typically team leaders or production team members that actually make the products on the shop floor. They have a wealth of practical knowledge of the production process, and are, in most cases, willing to help improve the process in any way possible. However, I’ve seen cases where data collection sheets are pushed onto the data collectors, without any input from them, without giving them any insight as to where the data is going, or how it will be used. Naturally, the response from the data collector is less than enthusiastic, and since the system depends heavily on him/her, the system is likely to fail.

Handpicked Content:   The Psychology of Awards

Here are some ways that I’ve found to leverage these very important team members to ensure data collection success:

1) Involve the people that build the product at the very start of your implementation. In this way, they are involved and realize why the need for the data is important. Remember, physically taking the data can be more of a burden to them, so it is absolutely important that they realize how important it really is.

2) Give the team members a chance to make the data collection system successful before going to their supervisor. It may be very tempting when activities seem to stall to go to the supervisor for support. Support the team members, and they will support you in most cases. Of course, there will be cases where you will need supervision to help.

3) Let the team member collecting the data be a part of the data report-out. Eventually, the data has to be reported on or used in an analysis. Giving the team member the opportunity to see how it is used and analyzed is a learning opportunity for him/her, and will also give a feeling of being a part of the business.

I’d really like to hear how your organizations have handled data collection at the front-line. I look forward to hearing about your experiences.

Comments 2

  1. Mike LaChapelle

    One of our objectives in Six Sigma is to collect process improvement data with as little effort as possible. Automated data collection systems can be a big help.

    In many industries, particularly process industries like the plastics, chemical, or food industries, key systems are continuously monitored and data on dozens of operating parameters is recorded and stored. In these situations, the ability to understand and sort through the mountains of data to find useful information for process improvement is critical.

    In other industries, you can ease the burden of data collection by planning ahead with process improvement in mind. Ask yourself, “What pieces of data do I think are important quality indicators or drivers”, and develop a plan to collect that data.

    If you can’t collect the data automatically, then the next best thing is to make it easy for the operator to measure and record the data. Simplify the process as much as possible and provide the tools needed to make the job easy.

    I remember one temperature measurement that required climbing up on top of a silo, leaning over to catch a sample of material coming out of the pipe entering the silo, pouring the sample into a thermos bottle, and then measuring the temperature with a thermometer.

    We tried a number of different methods to simplify the process, while maintaining accuracy. We finally ended up using an infrared thermometer, which we validated as being both accurate and fast.

  2. Kosta Chingas

    Thanks for the comment, Mike.

    I totally agree with making the data collection process as simple as possible with minimum effort and maximum effectiveness. Automatic (properly implemented) is definitely the way to go. The critical path is then analysis rather than data collection.

    One of the drivers of this blog is a system that was in need of development which required data collection on multiple outputs (15 or so) that were correlated to each other over the short-term. Over time however, with process adjustments and long term effects, the outputs would not correlate, thus requiring data collection on all the outputs over the longer term until the team understood the system behavior. It wasn’t really a matter of over-scoping, since we knew there was an underlying system relationship based on the outputs over the short-term. A failure in any of the outputs would have constituted a defective unit with respect to the overall project "Y" that we were working on.

    To make matters more challenging, an automatic data collection system for this system would have cost upwards of $500K, which was out of the scope of the project, since if we understood the system behavior, we could either predict or control the variables causing the break in the correlation, and thus reduce the measured variables to a very manageable level.

Leave a Reply