Learning Objectives
By the end of this section, students will be able to:
- Describe the types of hardware-accepted testing that are employed in industry.
- Differentiate between lot and part acceptance testing.
- Describe statistical process control and conceptually connect it to lot and part acceptance.
Everything thus far addressed in this chapter has gotten the AM part to the verge of production. Once a part has been qualified, and the system on which part is installed is certified, production may now begin. It should be noted that in some instances where certification can take a year or more, such as aircraft, parts may go into production before the aircraft is fully certified for delivery and use. While parts are in production, they will undergo both lot and part acceptance tests to not only ensure that the parts meet requirements, but often that the feedstock and processes used to make them are in control.
Lot Acceptance Testing
Lot acceptance testing refers to testing that is performed on a group (or lot) of parts to show that the processing the parts received meets specification. There is a range of definitions for a lot in AM, such as the following:
- All of the parts in a single build in any process that go through key post-processing operations, such as thermal treatment, together.
- A series of parts in DED that use the same lot of feedstock, the chamber atmosphere is never broken, and go through key post-processing operations together.
- A series of parts in BJP that made in multiple consecutive builds, and undergo de-binding, sintering, and key post-processing operations together
- A series of parts in VP that use the same lot of resin and are made consecutively
Note that each lot could theoretically consist of identical, opposite-hand, similar, or even different parts. What is being checked in lot acceptance is that the process met requirements. So, one type of lot acceptance would be a check of the build file and process monitoring that there were no unacceptable process events, such as energy source failure, recoater failure, contaminated atmosphere, and so on. While another type of lot acceptance could be a check of witness material in the build to verify that nominally acceptable material was produced. These types of tests consist of mechanical (mainly tensile) tests, micrographic (mainly in metals) chemistry tests, etc. Since AM builds parts in layers, careful thought must be put to the location of the witness coupons to ensure they are representative of the build process.
As occasionally practiced in other industries, destructive lot acceptance testing is sometimes performed in AM. This would be driven by a combination of part criticality, and overall maturity of the process. Destructive lot acceptance testing would generally consist of selecting one or more parts from the lot and performing destructive testing on it, in a similar or less extensive manner than destructive part testing for qualification. Different schemes may be used to select the part to be tested, with some requiring the same location in the lot, and others requiring a random location.
Failure of one or more of these tests can result in the entire lot being rejected, require the need for additional testing to demonstrate lot or part acceptability, or being analyzed by engineering authority to determine if any of the parts are acceptable either as-is or with rework.
Part Acceptance Testing
Part acceptance testing verifies that the part meets key requirements. It is generally performed on each and every part, although schemes for sampling (where some parts might not be qualified) do exist. Some examples of part acceptance testing include the following:
- Dimensional tests of key dimensions
- Nondestructive testing of both the part volume and/or the part surfaces
- Hardness testing to ensure proper response to thermal treatments
Failure of one or more of these tests can result in the part being rejected, indicate a need for additional testing, or require analysis by an engineering authority to determine if the part is acceptable as-is or with rework.
Statistical Process Control
Modern quality management systems will often utilize statistical process control (SPC), sometimes referred to as statistical quality control (SQC). SPC consist of recording data from either the process or the product and checking it against limits to determine if a process is in control. The goal is to use this tracking to get advance warning that a process is getting out of control before it happens and begins making unacceptable hardware. As a digital and data-rich technology, AM is ideally suited for this.
Examples of process variables and outputs that could be tracked include the following:
- Tensile properties of the witness coupons
- Key dimensions of the part
- Atmosphere gas consumption during build
- Laser output power
- Peak tip temperature in ME or peak current in an arc DED process
Having an in-depth understanding of the key variables in the process can help a product team decide what inputs and outputs to monitor, with the default option being to monitor all and let advanced data analytics software to make that determination. The former is preferred, as the latter can generate numerous false warnings (crying wolf) that distract the product team from other tasks.