LAMQC Case Studies

This article introduces the Missing Weld anomaly type and highlights its usefulness in delta qual and AM production applications. The Missing Weld anomaly is only available in LAMQC Frozen Process.

LAMQC Frozen Process is tailored for additive manufacturing (AM) production environments where a “correct” or “qualified build” can be defined.

We will examine components from two different builds and highlight how the Missing Weld  anomaly type calls attention to anomalies that LAMQC can not detect.

 

Delta qual efforts will find solid models of builds useful. This a solid model of an AM calibration block with a section of missing weld. The missing weld is not a separate anomaly type and is easy to miss.

Welded Region contains missing corner in LAMQC

Solid model of additively manufactured part with "missing weld" anomaly. Useful for AM delta qual.

Missing Weld anomaly type in LAMQC Frozen Process

 

LAMQC:
Calibration Block

This calibration block has both Bad Weld or Soot (green)  and Bright Recoat Off Part (purple) anomalies. The most striking part of this calibration block is the section of missing weld at the top left corner.

LAMQC does not see the region of missing weld as an anomaly and an operator or qualification team could easily overlook this issue.

Solid model of AM calibration block with a section of missing weld. The missing weld is not a separate anomaly type and is easy to miss.

LAMQC: Weld (grey)

Solid model of AM calibration block with a section of missing weld along with Bad Weld or Soot and Bright Recoat anomaly types. Created by LAMQC Prototype.

Missing Weld is not an anomaly type

LAMQC Frozen Process:
Calibration Block

LAMQC Frozen Process addresses limitations of LAMQC with the Missing Weld anomaly type by requiring the a “correct” or “qualified” build.

The weld solid model for the “correct”  calibration block includes the upper left corner as shown on the left image.

The image on the right illustrates the orange Missing Weld anomaly from the “current” build superimposed on the “correct” weld.

Solid model of "correct" or "qualified" AM calibration block. This calibration block does not contain missing weld.

Frozen Process: Weld (grey)

Solid model of AM calibration block with a model of the missing weld from LAMQC Production of the "current" build superimposed. LAMQC Production helps detect hidden build anomalies for AM production and delta quals.

Frozen Process: Missing Weld (orange)

LAMQC:
Turbine Blade

This turbine blade tip-shroud has Recoat Dark  (teal) and Recoat Bright (blue) anomalies as shown in the right image. These anomalies often negatively impact part quality.

Unlike the calibration block example, the missing weld in the turbine blade tip-cap is difficult to see in LAMQC, increasing the risk that an operator or quality team may overlook the problem.

 

"Current" build turbine blade with hidden missing weld.

LAMQC: Weld (grey)
Anomalies interior to weld

Solid model of turbine blade LAMQC Prototype anomalies including Recoat Bright and Recoat Dark. Missing weld is not an anomaly type in LAMQC Prototype.

Recoat Dark and Recoat Bright anomalies at the tip shroud.

LAMQC Frozen Process:
Turbine Blade

The Missing Weld anomaly type (right) allow easy identification of regions that lack weld. Missing Weld regions often have severely debited tensile and LCF and the surfaces often don’t clean up in machining, making the Missing Weld anomaly critical for delta quals and new machine certification.

If the AM Purchaser has stringent requirements, the part should be scrapped or be submitted on SDR (Supplier Deviation Request), NCN (Non-Conformance Notification) or for Engineering Review Board approval.

If AM Purchaser review process does not scrap the part, it should be dispositioned for tip-shroud inspection to determine the severity of the anomaly.

Solid model of turbine blade from the "correct" or "qualified" build. At first glance, this model appears to be very similar to the "current" turbine blade welded region, making it difficult to use for delta quals alone.

Unknown Part: All Anomalies

Non-continuous weld visible.

Solid model of the weld missing from the "current" build but present on the "correct" or "qualified" build. Useful for delta quals and serial production. The missing weld anomaly type allows operators to easily detect hidden build problems.

Bad Weld or Soot (green) removed.

Non-continuous weld visible.

Our second published case study illustrates LAMQC’s ability to find a hidden build failure and unsupported weld with in-situ inspection.

As before, the goal of the case study is to show:

  1. What an operator would have known.
  2. When they would have known it.
  3. What they could have done about it. 

In particular, this case study will show the use of LAMQC solid models to inspect the quality of individual parts from layer camera data.

In this build, only 130 layers of the 800 layer build contain anomalies that a trained operator would have noticed. It is likely that these build anomalies would not have been caught in many AM facilities.

This 800 layer build started properly and ended properly. Significant build anomalies occurred on only 130 layers in the middle of the build.

LAMQC insitu inspection anomaly graph with out unsupported weld.

Build Anomalies Excluding Unsupported Weld

The Build Anomalies Graph illustrates significant Bad Weld or Soot anomalies from layers 220 to 350. 

The Bad Weld or Soot event is transient and, from this graph, it appears that the build has recovered.

LAMQC in-situ inspection including unsupported weld

Build Anomalies Including Unsupported Weld

The addition of Unsupported Weld anomalies to the Build Anomaly graph illustrates that the second half of the build did not behave as expected and instead produced unacceptably large Unsupported Weld anomalies.

Unsupported Weld  anomalies are weld that starts in the middle of the build without weld in the prior layer. Unsupported Weld is the single most serious build anomaly as layer adhesion, surface roughness, LCF and tensile results are often negatively impacted in these regions.

Build Anomalies – Solid Model

The anomaly solid models for this build indicate significant build issues on the left half of the build including Unsupported Weld (red), but less on the right side of the build. An AM operator would be reasonably tempted to assume the parts on the right side of the build plate are acceptable.

In the next section we will use LAMQC solid models to illustrate why that would have been a costly mistake.

Note that the solid models appear “rough” due to the low (700×700) resolution images. Higher resolution images improve solid model fidelity.

Part Details: Unknown

This part is one of the 6 parts on the left of the build plate. The parts have failed so completely that the part geometry is unrecognizable. It is likely that these 6 parts would have been immediately scrapped after the build.

The solid models show that there is a large region of Bad Weld or Soot (green)  in the center of the part (left image).

Additionally, LAMQC shows that there are significant gaps in the weld (grey), accompanied by Unsupported Weld anomalies in red.

Unknown Part: All Anomalies

Non-continuous weld visible.

Bad Weld or Soot (green) removed.

Non-continuous weld visible.

Part Details: Overhang Block

This overhang calibration block illustrates the ability of LAMQC to detect anomalies on low angle overhangs that present problems on many LPBF AM machines. 

The left image shows the weld (grey), along with the build anomalies, Bad Weld or Soot (green), Bright Recoat (blue) and Bright Recoat Off-Part (purple).

As expected, the lowest angle overhang contains the most anomalies, followed by the second-lowest angle overhang.

Overhang Block: All Anomalies

Significant anomalies at lowest angle.

Part Details: Impeller

The LAMQC solid models clearly show that the top impeller vanes experienced significant Bad Weld or Soot and many of the vanes experienced Bright Recoat anomalies.

A close inspection of the weld region (grey) on the left image shows that the vanes on the left side of the impeller failed to print properly. LAMQC does not consider these vanes to have printed with “weld”, resulting in the vanes failure to appear as “weld” in the solid models.

Impellers operate at a high rotational speed and high stress. Any build anomalies or lack of weld should result in a scrapped part.

Scrapping the part during AM by stopping the build saves AM costs associated with powder and machine time as well as post-AM manufacturing and inspection costs.

Scrapping the part after AM savings time and costs associated with post-AM manufacturing operations and inspection.

Impeller: All Anomalies
Weld: Grey

Bright Recoat:  Blue

Bad Weld or Soot: Green

Part Details: Turbine Blade

This turbine blade has both Bright Recoat (blue) and Unsupported Weld (red) build anomalies.

Unsupported Weld is present in the turbine blade at the perimeter of the blade. The weld cross-sectional area in the layer camera images grows and shrinks over the build height, not the design intent of a constant taper. Significant Unsupported Weld negatively impacts surface roughness, tensile properties and LCF life.

Turbine blades operate at a high rotational speed and high stress. Any build anomalies should result in a scrapped part when they become apparent.

Scrapping the part during AM by stopping the build saves AM costs associated with powder and machine time as well as post-AM manufacturing and inspection costs.

Scrapping the part after AM savings time and costs associated with post-AM manufacturing operations and inspection.

Turbine Blade: All Anomalies
Welded Region: Grey

Bright Recoat: Blue

Unsupported Weld:  Red

Part Details: Calibration Block

This calibration block has both Bad Weld or Soot (green) build anomalies as well was Bright Recoat of Part (purple) but the most striking part of this calibration block is a section of missing weld. 

Running this build in LAMQC Frozen Process would have highlighted the missing weld as a separate anomaly, allowing easier detection.

Look for this calibration block example in the near future as part of the LAMQC Frozen Process demonstrations.

Calibration Block: All Anomalies
Welded Region: Grey

Bright Recoat Off Part:  Purple
Bad Weld or Soot: Green

Unsupported Weld: Case Study Summary

1

What an Operator would have known.

The operator would been made aware of the large Bad Weld or Soot  anomaly on layer 220 as well as the Unsupported Weld anomalies. The operator can then review relevant layer camera images, the anomaly graph and solid models of the build anomalies.

 

2

When they would have known it.

The operator would have been notified of the Bad Weld or Soot anomaly immediately with an email alert, and if the build continued, additional alerts regarding Unsupported Weld. The other anomalies would have been visible in the LAMQC graph and solid models.

3

What they could have done about it.

Shut down build. This build has significant and persistent build anomalies. It should be shut down for all AM part use-cases, even for prototype parts.

 

1. General Information

Unique report identifier, material, and build name

2. Inspection Summary

Tabular anomaly sizes with respect to build limits

3. Build Anomaly Graph

Graphical anomaly sizes with respect to build limits

4. Layer Camera Images

Limited, relevant layer camera images.

General Information

The first section of the LAMQC inspection report is General Information. This section contains build identification details such as:

  • Report Number: Auto-generated unique Inspection Report identifier
  • Build Date: Date of AM build (when used Live)
  • Material: AM powder
  • Build Name: User generated build name
  • Machine Name: Company specific AM machine name or number
  • Layers Evaluated: The evaluated layers included in the inspection report.

LAMQC Build Report General Information Section and Pass/Fail

Inspection Summary

The Inspection Summary presents build quality information for each anomaly type in a table, including:

  • Area Limit: The maximum allowable anomaly area
  • Max Area: The maximum recorded build anomaly area
  • Status: PASS / FAIL. If Max Area exceeds Area Limit status is FAIL
  • Max Layer Number: The layer number of the maximum anomaly area

 

In this build, only Bright Recoat has failed inspection requirements.

The anomaly limit is 100 mm2 and LAMQC found a Bright Recoat anomaly with an area of 7,415 mm2 at layer 1595, resulting in a “FAILED” build.

LAMQC Inspection Report: Inspection Summary

Build Anomalies Graph

The Build Anomalies Graph section of the Inspection Report illustrates how the size of each anomaly progresses over the build.

While this build failed to meet Bright Recoat requirements, the graph shows that the build quickly recovered from each anomaly.

 

If the parts in this build will not be used in a critical application, the build may be acceptable. This inspection report and the solid models should be provided to the AM Purchasers to disposition this build.

However, if the parts in this build will be used in a critical application with well defined acceptability limits, it should be scrapped and not be allowed progress down the manufacturing router.

 

Its important to note that the Build Anomalies Graph presents anomaly size at a give build layer, but does not present the location of the anomaly relative the AM part(s). Understanding the location of the anomaly with respect to the AM part(s) requires the use of LAMQC anomaly solid models.

LAMQC Inspection Report: Build Anomaly Graph

Layer Camera Images

LAMQC understands that reviewing 200 images per build is just as impractical as reviewing 2000.

Therefore LAMQC Inspection Reports only contain layer camera images  of anomalies that most significantly exceed build limits. 

In other words, a LAMQC Inspection Report will include no more than 12 images, comprised of an exposure and recoater image for each of the six anomaly types.

In this example, only Bright Recoat anomalies exceeds limits. Consequently, the LAMQC Inspection Report includes only Bright Recoat layer camera images of layer 1595, the layer with the largest Recoat Bright anomaly. 

If the AM Purchaser considers this anomaly acceptable, we recommend that the AM Purchaser investigates increasing Recoat Bright anomaly limits.

LAMQC Build Report Layer camera images

Our first published case study illustrates LAMQC’s abilities to find a hidden short feed.

The goal of the case studies is to show:

  1. What an operator would have known.
  2. When they would have known it.
  3. What they could have done about it. 

This build is from an EOS M290, contains 2266 build layers (4532 images) and has been released by Oak Ridge National Labs. 

Nearly all build layers recoated properly. A well trained operator watching this build closely would likely have not have noticed any build issues.

This build is an excellent example of LAMQC’s ability to highlight stochastic (randomly occurring) AM build anomalies.

Nominal Layer camera images

This build started properly, ended properly and ALMOST all of the layers in between were acceptable.

LAMQC in-situ monitoring anomaly line chart

Build Anomalies – Graph

LAMQC generates a graph of anomaly areas at each build layer. Each anomaly type is represented by a different colored line on the graph.

The blue line shows Recoat Bright anomalies and the green line shows Bad Weld or Soot anomalies.

The graph for this build illustrates that there were three significant Recoat Bright events and one significant Bad Weld or Soot event.

Only one Recoat Bright event violated anomaly limits. LAMQC would have notified the operator via email of a Recoat Bright event at layer 1595.

Build Anomalies – Solid Model

Graphs provide an effective build quality summary, but do not communicate the location of anomalies relative to AM part(s). Anomaly locations are often required to properly disposition AM parts.

Solid models, combined with Part Purchaser acceptance criteria, allows Part Providers to determine if anomalies occur in acceptable or rejectable regions of the build.

LAMQC generates three solid models for this build:
Weld:                         AM parts, supports 
Bad Weld or Soot:   Bad weld, soot
Recoat Bright:          Shortfeeds, super elevation, peel ups

LAMQC finds more anomaly types, but only two anomaly types are present in this build.

The solid models clearly show the parts impacted by the Recoat Bright and Bad Weld or Soot anomalies. Clear communication of anomaly type, location and size enables the Part Purchasers to understand build quality and properly disposition AM part(s).

Build Report

Allowing any part to continue down the manufacturing router requires a PASS / FAIL decision. In LAMQC, the summary information required to make this decision is contained in the build report. 

LAMQC build reports contain:
-General build information,
-Inspection limits for each anomaly type
-PASS/FAIL for each anomaly type
-Overall PASS / FAIL
-Graph of anomaly sizes for each layer
-Layer camera images of the largest anomalies of each type

LAMQC build reports are simple to understand and should be combined with anomaly solid models to provide substantiation details on AM Supplier Deviation Requests (SDR’s), Non-Conformance Notification’s (NCN’s) or Engineering Review Boards (ERB’s).

LAMQC is designed to effectively communicate the size, location, orientation and type of AM build anomaly to the AM Purchaser before part(s) continue down the manufacturing router.

Page 1: PASS/FAIL, General Build Data and Anomaly Graph

LAMQC AM In-situ build report, page 2

Page 2: The largest failed anomalies of each type

Hidden Shortfeed: Case Study Summary

1

What an Operator would have known.

The operator would been made aware of the large Recoat Bright (short feed) anomaly on layer 1595 as well as the smaller Recoat Bright and Bad Weld or Soot anomalies. The operator review relevant layer camera images, the anomaly graph and solid models of the build anomalies.

 

2

When they would have known it.

The operator would have been notified of the Recoat Bright anomaly immediately with an email alert. The other anomalies would have been visible in the LAMQC graph and solid models.

3

What they could have done about it.

Prototype Part:
Review LAMQC alert and data.  Continue build.  Ensure anomaly/build recovery.

Critical Application:
Highlight anomaly regions for focused CT scan, cut-ups and/or effects of defects studies.
If the part is found to be acceptable, open LAMQC allowable limits.

Critical Application, Well Defined Acceptance Limits:
If the builds exceeds well defined acceptance limits, terminate build.

If you can see it on layer camera images, we can see it in LAMQC.

LAMQC uses your existing in-situ layer camera data to highlight build anomalies for AM production and prototyping applications.

Anomalies visible on in-situ layer camera images often negatively impact mechanical properties and build repeatability.

Using existing layer camera images limits LAMQC’s ability to directly detect anomalies that are not visible in layer camera images.

 

Detectable Build Anomalies

Unsupported weld, soot, defective weld, short-feeds, coater tears, peel-ups, super-elevation.

Each of these anomalies are visible in layer camera images and are visible in LAMQC. 

 

Non-Detectable Build Anomalies:
Energy density variations less than ~20% (material and parameter dependent),  spatter particle tracking,  individual pockets of porosity or key-holes, 1 vs 2 contours.

These anomalies are not visible in layer camera images and not visible in LAMQC. Minor laser power variations are visible with optical tomography (OT) data.

LAMQC anomaly detection matrix from in-situ Renishaw layer camera images. Detection of build anomalies in LAMQC is helpful for AM qualification, delta-quals and documentation.

LAMQC detects anomalies that appear in layer camera images