Digital Thread: How to Link DFMEA to the Shop Floor With an AI-Enabled NPD Platform

Ask most quality leaders who their biggest competitor is, and the honest answer isn’t another software vendor — it’s Excel.

DFMEAs, PFMEAs, and control plans still get built and maintained largely by hand. Statistical process control tools are often sophisticated, but they sit on their own, disconnected from the FMEA and control plan that defined why a characteristic mattered in the first place. The result is a silo between engineering and manufacturing: limited traceability from a design decision all the way to what actually happens at the workstation, and delayed detection of the variation already present in the product.

That silo is exactly what a digital thread is built to close — and it’s the subject of a recent Omnex webinar, presented by Chad Kymal, Antony John, and Greg Gruska, on digitalizing product development and the shop floor with an AI-enabled NPD platform. This is the first of a three-part series; we’ll walk through the digital thread itself here.

The Real Problem: Great FMEAs That Never Reach the Shop Floor

Talk to enough quality teams and the same gaps come up again and again:

• Excel is still the default. DFMEAs, PFMEAs, and control plans are frequently built and kept up to date manually, even alongside genuinely sophisticated SPC software.
• Engineering and manufacturing don’t share one system. The people who write the FMEA and the people running the process are working from different documents that don’t automatically stay in sync.
• Traceability breaks somewhere in the middle. A characteristic identified as critical at the design stage can lose that context by the time it reaches an inspection sheet.
• Detection lags behind reality. Variation that’s already present in the product isn’t caught until well after the fact, instead of while it’s happening on the line.

One large automotive OEM described its own version of the fix as a “connectivity of documents”: the print connects to the PFMEA, which connects to the control plan, and so on. Omnex’s own version of that connectivity starts one step earlier — at the DFMEA — and runs all the way through to the SPC chart, using AQuA Pro for the engineering side and Inspection Control for the shop floor.

Following One Characteristic From the Print to the SPC Chart

The clearest way to see a digital thread at work is to follow a single characteristic through every document it touches. In the webinar, that characteristic is called BP005 — and the same mechanics apply to every other characteristic on the same print.

It starts with the print

Every drawing has a characteristics table — what most engineers know as the ballooned print, with balloon numbers marking each dimension that matters. You can bring that table in from Excel, or build the balloon print directly and carry it straight into your PPAP. BP005 is the clearance between the auto adjuster and the hub. It’s specified as an inverted delta, a bilateral tolerance, which means SPC is going to be involved downstream.

The DFMEA works in functions, not characteristics

Here’s the part that trips a lot of teams up: a DFMEA doesn’t contain characteristics at all — it contains functions. To carry forward what’s critical, significant, or safety-related, you open the characteristics table inside the DFMEA and link it to the relevant function. In this case, the function is the auto adjuster automatically maintaining the proper amount of adjustment on the brake drum — a safety and regulatory requirement, which is why it carries a severity of 10. BP005, the characteristic tied to that function at assembly, inherits that same severity the moment it’s linked.

Process flow, PFMEA, and control plan inherit what came before

BP005 shows up next in the process flow, at the operation that assembles the drum brake skeleton components — marked with the special characteristic symbol you’ll see follow it through every remaining document. In the PFMEA, because that failure mode is linked back to the design failure, it automatically inherits the severity of 10 from the DFMEA. Nobody re-types it, and nobody re-judges the risk a second time. From there it lands in the control plan as a safety-compliance characteristic with its own process control chart, and in the work instructions at that same operation, with every control attached and visible.

Inspection sheets and SPC take it from there

From the control plan, Inspection Control builds the actual shop-floor inspection sheets — and it sorts them automatically, so operators see start-of-shift checks at the start of a shift and changeover controls at a changeover. For BP005, the operator records a 10-piece sample straight onto a tablet or mobile device, or the reading comes in automatically from a caliper or a CMM.

What the SPC Data Actually Shows You

Once that data starts flowing, the value shows up fast. A six-pack SPC chart generated straight from the Omnex software, built from 25 subgroups, made it possible to see that two different assembly machines were contributing to the same characteristic — a stratification in the data that would be easy to miss in a spreadsheet. In this case, the distribution was normal and the process was capable, but the point is that you can see it at all, and see it in one place.

This isn’t limited to special characteristics, either. You can run an attribute study on an attribute characteristic, or calculate Cp and Cpk for any characteristic on the control plan, at any time. If a lot starts trending toward defective, the system can warn the company, alert the supervisor, and escalate if the control plan still isn’t followed — instead of finding out after the part has already shipped.

Put together, that’s the full thread: the characteristic number on the ballooned print, traced through the DFMEA, into the process flow, the PFMEA, the control plan, the work instructions, the inspection sheets, and finally the SPC study — for every characteristic on the drawing, not just BP005.

Family FMEA: Build the Risk Analysis Once, Inherit It Everywhere

Once a product family exists in AQuA Pro, you don’t start the next derivative FMEA from a blank page. Omnex has called this a family FMEA since 1988 — the same idea some customers, including Ford, refer to as a foundation FMEA.

You build the parent FMEA once, at the platform level — a battery module, for example — and then create child parts underneath it. Every child inherits everything from the parent automatically. Change something in the parent, and every child updates with it. Change a child without touching the parent, and the parent and its siblings are unaffected. Children can have their own child parts underneath them too, so the inheritance chain keeps going as far as your product structure needs it to.

Getting There Without a Rip-and-Replace

None of this requires starting over. If you’re already on AQuA Pro, you don’t need to switch platforms — the digitalization capability comes from adding the Inspection Control module for the shop-floor half of the thread. AI validation, recommendations, and FMEA generation are a separate, optional add-on on top of that, not a requirement to get the digital thread itself.

If parts of your documentation still live in Excel or another tool, you can bring that into the thread too, without switching everything at once. The AI PPAP reviewer can review every PPAP element — DFMEAs, process flows, control plans, inspections, and SPC studies — regardless of which tool produced them. And the AI agent can generate new FMEAs from your existing knowledge bank in Excel or SharePoint, where connectors already exist; a genuinely separate legacy system may need a connector built for it.

If you’re extending any of this to your supplier base, suppliers get their own branded portal with their own logins, authenticated with multi-factor authentication. Supplier licenses are priced separately from — and less expensively than — internal user licenses.

What’s Next in This Series

This webinar is part one of a three-part digitalization series. Part two covers Omnex’s AI PPAP reviewer — validating an entire PPAP packet, element by element, before it ever reaches a customer or comes in from a supplier. Part three covers using Omnex’s AI agent to generate DFMEAs and PFMEAs directly, building functions and requirements first, then failure modes, and either exporting the result to Excel or XML or feeding it straight into the same digital thread described above.

Across all of it, the reported payoff is consistent: a real reduction in the cost of poor quality, faster launch cycles, fewer warranty and customer issues, and a lot less manual effort re-entering the same data more than once.

• 1

  Lower cost of poor quality — variation gets caught through Cp/Cpk and SPC monitoring instead of after the part has shipped

• 2

  Faster launch cycles — the family FMEA approach means each new derivative part inherits its risk analysis instead of starting from a blank template

• 3

  Fewer warranty and customer issues — because the control plan is actually executed and monitored, not just filed away

• 4

  Less manual effort, faster ROI — the same characteristic doesn’t need to be re-typed at every stage from print to SPC chart

Frequently Asked Questions