Oracle Fusion WIP Sampling: How Inspection Works

Oracle Fusion Quality Management lets you attach an inspection sampling plan to a work order operation so that only a statistically chosen subset of units is inspected, rather than every unit. The decision you must make first is not the sample size, but the sampling basis and the inspection specification, because those two choices determine which units get checked, how many, and against what criteria.

Formula

The sample size is calculated as:

Sample size = ⌈Lot size × Inspection level factor⌉

where the inspection level factor is 1.0 for General II, 0.5 for General I, and 0.25 for Reduced. The accept number is:

Accept number = round(AQL% × Sample size / 100)

The reject number is simply the accept number plus one. In this simplified model, the lot is accepted if the number of defective units found in the sample is less than or equal to the accept number; if it exceeds the accept number, the lot is rejected.

Note that Oracle Fusion uses the actual ANSI/ASQ Z1.4 tables, not a simple formula. The formula above gives a close approximation for planning purposes, but the system uses the exact tabular values for the chosen inspection level and AQL.

Worked Example

Using the defaults: lot size = 100 units, inspection level = General II (factor 1.0), AQL = 1.0%.

1. Sample size = ⌈100 × 1.0⌉ = 100 units (so 100% inspection).
2. Accept number = round(1.0% × 100 / 100) = 1.
3. Reject number = 2.

So you would inspect all 100 units, and if you find 2 or more defects, the lot is rejected. This matches the ANSI table for a lot of 100 at General II, AQL 1.0%.

How Inspection Sampling Works in Oracle Fusion

In Oracle Fusion Quality, sampling is tied to the quality specification attached to a work order operation. When you report completion of an operation, the system checks the sampling plan and decides whether the current transaction requires inspection. The plan is defined with a sampling type (e.g., fixed, lot-based, or time-based) and a specification that lists the characteristics to be checked.

The key steps in the process are:

• Define a sampling plan in the Quality Management work area, specifying the sampling type and the inspection level.
• Attach the plan to a quality specification, which is then associated with a work order operation or a resource.
• When the operation is completed, the system calculates the sample size based on the lot quantity and the plan rules.
• If the sample size is less than the lot size, the system randomly selects the required number of units for inspection.
• You record results for each characteristic on the sample units. The system aggregates the results and applies the accept/reject criteria to determine the disposition of the lot.

If the lot is rejected, you can create a quality issue, initiate a corrective action, or route the lot to rework. If accepted, the lot moves to the next operation or to inventory.

What to Decide First: Sampling Basis and Specification

Before you set up sampling, you must decide two things: the sampling basis and the inspection specification. The sampling basis determines whether you sample per lot, per batch, per time interval, or per transaction. The inspection specification defines the characteristics (dimensions, tolerances, visual criteria) that will be checked.

Choosing the wrong basis leads to over- or under-inspection. For example, if you sample per transaction but your work order is split into many small transactions, you could end up inspecting far more units than intended. Conversely, if you sample per lot but the lot is huge, you might miss defects that appear later in the lot.

The inspection specification must be aligned with the actual quality requirements. If the specification is too loose, defects slip through; if too tight, you reject good product. Start with the customer’s requirements and regulatory standards, then define the characteristics and tolerances.

Common Problems and Misconceptions

A frequent misconception is that the sample size is always a fixed percentage of the lot. In reality, the ANSI tables give sample sizes that increase with lot size but not proportionally. For a lot of 500, the sample might be 50, but for a lot of 1000, it might be 80, not 100.

Another issue is ignoring the inspection level. Using General I instead of General II reduces the sample size, but also increases the risk of accepting a bad lot. The choice of level should reflect the criticality of the product and the supplier’s history.

Also, many users forget that the accept number is not a percentage of the sample. It is a discrete count. For a sample of 50 with AQL 1.0%, the accept number is 1, meaning you can have at most 1 defect. If you find 2, the lot is rejected, even though 2/50 = 4% is above the AQL.

Comparing Sampling Types: Fixed vs. Lot-Based vs. Time-Based

Oracle Fusion offers several sampling types. The most common are:

• Fixed sampling: Always inspect a fixed number of units, regardless of lot size. Useful when the lot size is stable and the cost of inspection is high.
• Lot-based sampling: Sample size depends on the lot quantity, following a plan like ANSI Z1.4. This is the standard for most manufacturing.
• Time-based sampling: Inspect units at regular time intervals (e.g., every hour). Good for continuous processes where defects may occur over time.

For discrete manufacturing, lot-based is usually the right choice. For process industries or high-volume production, time-based may be more appropriate. Fixed sampling is rare and only used when the lot size is constant and the risk is well understood.

Cost and Risk Considerations

Sampling reduces inspection cost but increases the risk of accepting defective lots. The AQL (Acceptable Quality Limit) is the maximum percent defective that, for the purposes of sampling inspection, is considered acceptable as a process average. It is not a guarantee that every lot will have that quality level; it is a contractual quality level.

Choosing a lower AQL (e.g., 0.1% instead of 1.0%) increases the sample size and the cost of inspection. You must balance the cost of inspection against the cost of defects reaching the customer, which includes warranty claims, rework, and reputation damage.

Alternatives to Traditional Sampling

If sampling is not giving you the confidence you need, consider alternatives:

• 100% inspection: Inspect every unit. This is only feasible when the cost of a defect is extremely high, such as in medical devices or aerospace.
• Statistical Process Control (SPC): Monitor the process in real time and take corrective action before defects occur, reducing the need for end-of-line inspection.
• Automated inspection: Use vision systems or sensors to inspect every unit at high speed, eliminating the need for sampling.

These alternatives are not always practical or cost-effective, but they can be considered when the risk of defects is unacceptable.

Frequently Asked Questions

Q: What is the difference between AQL and LQ?
AQL is the quality level that is acceptable for the process average. LQ (Limiting Quality) is the worst quality level that the sampling plan will accept with a low probability. In practice, AQL is used to define the plan, and LQ is used to assess the risk of accepting a bad lot.

Q: Can I use sampling for non-conformance reporting?
Yes. When a sample unit fails a characteristic, you can record the nonconformance in Oracle Fusion Quality. The system can automatically create a quality issue and route it for disposition.

Q: How do I change the inspection level after the plan is created?
You can edit the sampling plan in the Quality Management work area. However, changing the level affects the sample size for future lots. Existing lots that have already been inspected are not affected.

Q: What happens if the lot size is smaller than the sample size?
In that case, the sample size is capped at the lot size, meaning you inspect 100% of the lot. This is standard in ANSI tables, where the sample size never exceeds the lot size.

Q: Is the sampling plan applied at the operation level or the work order level?
It is applied at the operation level. You attach the quality specification to a specific operation in the routing, so different operations can have different sampling plans.

Related

• Turn a Product Walkthrough into a Polished Tutorial Video
• Recover Deleted TIFF Files: Full Guide
• Enterprise AI Security: Full Lifecycle Guide

in lot_size | Lot size (units) | units | 100 | 1 | 100000
sel insp_level | Inspection level | General II=1, General I=0.5, Reduced=0.25 | 1
sel aql | AQL (%) | 1.0%=1, 0.65%=2, 0.40%=3 | 1
out sample_size | Sample size (units) | units | ceil(lot_size * insp_level) | 0
out accept_num | Accept number | units | max(0, round(aql * sample_size / 100, 0)) | 0
out reject_num | Reject number | units | accept_num + 1 | 0