Getting a Handle on Optimizing Day-to-Day Process Manufacturing Operations - Mastering JMP

Video was recorded in January 2025 using JMP 18.

Are you a manufacturing engineer who supports day-to-day operations?  Do you need to launch an effort to understand and optimize your process, but don’t know where to start? Would you like a straightforward way to visually summarize and explore your manufacturing data and then dig into process capability and root-cause analysis to help improve the process?

In this session, we will demonstrate how to use key JMP capabilities that support day-to-day operations, provide process insight, help you solve problems, proactively address how to avoid small process excursions, and improve results. We will use the following steps to analyze and compare typical manufacturing processes for which we have yields, environmental conditions, financial information, manufacturing location, equipment and monitors.

• Use Graph Builder to evaluate and compare processes interactively using Box and Trellis Plots, Yield Graphs, Geographic Maps and Distributions.
• Understand process capability using Goal Plots and Process Capability Plots and create graphs that are easy to drill into.
• Use Control Chart builder to examine graphs and related statistics for processes in control or out of control.
• Locate and drill into sources of variability using Variability Gauge Charts and Fit Model graphs, summary statistics, and trade space analysis to find optimal settings.
• Find root cause of process deviations by applying interactive machine learning using Partition.

 Suggested Prerequisites:

• Some experience evaluating process data.

After you use the attached materials to try the techniques in the video, consider using the companion Data Mining for Solutions in Root Cause Investigations Hands-On Activity to use data mining to find root cause. 

Questions answered by @HydeMiller , @scott_allen and @ChristianStopp at the live webinar: 

Q: How do you group the process monitor variables into an overall group in the columns section?

A: Select those process columns in the column panel on the left side of the table, right click them, you'll see an option for Group Columns. 

Q: How does the analysis know the USL and LSL?

A: They can be stored and saved as metadata as column properties.

Q: Can I have multiple triangles in the Goal Plot, like Ppk 1.3 + 1.0?

A: Yes. You can set the triangle to a target Ppk and then use the slider to look at the sensitivity as the triangle gets larger and smaller.

Q:  Can you run the Capability Analysis using Cp (instead of Cpk)?

A: The Goal Plot only displays Ppk index for the triangle zone. You can view the Cpk values for the individual processes by toggling them on/off in the Red Triangle.

Q: How do you add the spec limits to each monitor parameter?

A: There are several way. One easy way is to right click the Column header > Column Properties > Spec Limits.

Q: Can you add warnings? 

A: Yes, from the Control Chart Builder red triangle.

Q: How can I set an alarm for points that are above the Spec Limits?

A: There is no way to set a rule for that. You can do it graphically by selecting points on your Distribution Chart. You could also circle them within the control chart if you wanted to.

Q: Would the beyond limits test in the drop down under warnings allow you to be alerted when outside the spec limit?

A: The distinction is between the control limits, which JMP is calculating for you (process defined), and the spec limits (user defined). 

Q: Is the "subgroup" on the x axis just the order the values appear?

A: Initially, yes, it is the row order (like components coming off a line), although you can define it in other ways.

Q: is there any way with the PHASE on a Control chart to push the control limits to a pooled limit ?

A: When you use a Phase, the Control Limits will be calculated for each category in the Phase column. If you want to over-ride the calculated Control Limit, you can specify Control Limits and Targets in as Column Properties. These Control Limits will be displayed in the Control Chart, even if you have different phases.

Q: Is there a way to create a limits legend next to the Control Chart? I'm only aware of how to do a row legend of the columns, but not the limits.

A: By default, you will get a limits summary table next to a control chart in control chart builder. You can also place control limit values directly on the Control Chart under the Red Triangle: Limits > Show Limit Labels.

 Q: Can I  exclude those parameters with lower p-values?

A: Yes. You can remove them from the model by selecting them and clicking Remove.

Q: What R^2 value should we look for?

A:  There is no rule for this because it depends on many things and is sometimes  industry dependent. The general idea is the closer to 100%, the more your predictors are explaining about the response.

Q: How could i document, for example,  all monitors vs process variability charts, and send the journal/other JMP documentation file to my peers/vendors for review and their data processing? I tried to put multiple charts horizontally, but JMP default produces charts in vertical way. Or is there an option to edit the chart grid?

A: There are a few ways to address this depending on the reports you want to communicate.  One way is it to consider Process Screening ( Analyze > Quality and Process > Process Screening ). This will calculate summary stats for all your process variables and summarize them in a table. You can then select the charts for which  you want to display graphs. If you want to report the results out into a custom report, you can use a journal by copy/pasting different charts into the journal or you could look into creating a custom report using the JMP scripting language.

Comments from attendees at the live webinar:

• When you build an IMR chart, and have Spec Limits applied to a variable, with the "Show As Graphic Reference Lines" clicked, they will show on the chart.  They'll also show on the distribution plots.  And it's appropriate to view the plot that way (you wouldn't want to add spec limits to XbarR).