Batch Traceability: Tracing Materials Through Production

From one incoming batch to three finished batches

In this simplified example, 100 kg of RM-410 is processed. Of that quantity, 60 kg is combined with 20 kg from a second incoming batch, RM-411, to produce mixture MX-610. The remaining 40 kg of RM-410 is processed separately. All identifiers and quantities are fictional; the table describes documented actual operations, not a planned recipe.

Documented paths therefore lead from the single incoming batch RM-410 to FP-810, FP-811, and FP-812. RM-411 is connected only to FP-810 and FP-811. Processing the materials together in MX-610 is the decisive transition.

EPCIS uses the TransformationEvent event type to connect consumed input objects to produced output objects. Its semantics allow each input to have contributed to each output within the recorded transformation context. Such a relationship does not specify the exact share of an input material in an individual finished product.^[1]

For this example, a concern about RM-411 means that both outputs of MX-610 need to be investigated. Without further evidence, an analysis must not assume that the material went exclusively into FP-810. Conversely, the shared use of RM-410 alone does not establish a material path from RM-411 to the separately produced FP-812.

The required process relationships must therefore express whether material was actually used, merely planned for an order, or transported together with other containers. These relationships lead to different results.

Trace origins backward and material use forward

Starting with FP-811, the backward search leads through MX-610 to RM-410 and RM-411. It identifies the documented material inputs for this finished batch. An Electronic Batch Record can additionally bring together the steps performed, inspections, and decisions for FP-811. To trace across multiple batches, the material relationships must also be preserved between these records – including a cross-company review in a digital supplier audit.

Starting instead with RM-410, the forward search follows its uses through to all three finished batches. Their subsequent movements are also relevant. At the point in time considered, the example shows the following situation:

For traceability in production, these quantities must refer to the same point in time. If goods are shipped after that point, they still belong to warehouse stock in this snapshot; a later update must account for them as shipped. Combining an older stock extract with a newer shipment list can instead count the same quantity twice.

Thus, 48 kg in the organization's warehouse and 68 kg in documented shipments are connected to RM-410. These 116 kg initially define the investigation scope derived from the material paths. The table does not establish that every unit is actually defective or that the same action is required for every quantity.

Preparing for a recall would now involve compiling the affected delivery items, recipients, and available contact persons, among other information. The organization's own shipping record cannot establish whether a recipient still holds the goods or has already processed or forwarded them. The investigation must continue there using the available information. A subsequent response from a recipient should record both the time to which it refers and the time it was received. Pending responses remain visible as open items; they must not be interpreted as confirmation that no goods remain affected. This keeps traceability in production understandable as later information arrives, without silently replacing the original snapshot.

The query time also belongs to the result. If FP-811 is shipped after the initial analysis, the earlier stock balance no longer describes the current situation. A saved result list therefore needs a traceable data cutoff and an assigned responsibility for keeping it up to date.

Establish the connection when material is actually used

For traceability, GS1 distinguishes between significant events in an object's lifecycle and the data describing each event. Examples include receiving, packing, shipping, and transporting.^[2] Applied to the production example, the critical capture point is where RM-410 is actually used in MX-610.

The documentation must distinguish that use from a mere reservation. An order specifying 60 kg of RM-410 does not establish that this exact material was used. Nor does a label on an empty container prove which material was previously taken from it.

For the investigation shown here, at least the following questions must be answerable: Which unambiguously identified incoming batch was used? In which operation and in what quantity? Which outputs were produced? When did this happen, and which record supports the assignment? This selection of fields is a design inference from the example, not a universal mandatory field list.

When splitting a container's contents, container identifiers must also be distinguished from batch identity. Two new containers may still hold the same batch. Changing their identifiers must not break the material's origin references. Conversely, placing containers on the same pallet does not establish that their batches were mixed.

Rework can also create another material path. If a remainder from an earlier production run is reused, that input must retain its actual origin. A generic entry labeled “rework” would not identify which earlier batch was involved.

If an incorrect material assignment is discovered later, a correction event must connect the original entry to its correction. Subsequent analysis must be able to account for the corrected relationship. It must also be assessed which previously generated result lists or decisions based on them were affected by the incorrect assignment.

A complete material chain is not yet a mass balance

In the example, total inputs of 120 kg correspond to 116 kg of finished goods and 4 kg of recorded process loss. The arithmetic balances. It does not establish that the quantities were measured correctly or that the amount booked as loss actually arose in that way.

Traceability and mass balance answer different questions. The material chain connects inputs and outputs. The balance additionally checks whether the quantities reconcile within a defined scope. Units, measurement bases, and periods must be comparable. For example, a net quantity must not be reconciled unnoticed against a weight including packaging.

If only the two finished batches totaling 78 kg were documented for MX-610, 2 kg would initially remain unexplained against the 80 kg input. The connections to RM-410 and RM-411 could nevertheless be fully recorded. They would not replace the missing explanation of the quantity.

Likewise, the 116 kg of finished goods found when investigating RM-410 does not contradict its original 100 kg. The investigation scope includes entire connected finished batches, including the additional material mixed in. It is not a calculation of the share of RM-410 contained in them.

An unexplained difference should remain visible as such. Automatically calculating a residual item called “loss” would balance the figures without explaining actual use.

Assess the identified scope against a specific test case

Whether a solution represents the material flow adequately can be tested directly with this example. When selecting compliance software, the demonstration should start with RM-410 and show the three connected finished batches, including stock and shipments. The same investigation is then performed for RM-411. FP-812 must not appear as materially connected solely because RM-410 was used in both processing paths.

In another run, one assignment is deliberately missing. The demonstration must show how comparison with independently documented processing or weighing operations reveals the gap; the tester knowing which assignment was removed is not sufficient. An unexplained quantity such as the 2 kg in the MX-610 counterexample can indicate a need for investigation, but does not prove that a particular assignment is missing. Conversely, a reconciled mass balance does not establish that all material relationships are complete. A technically successful query returning an empty list must not be presented without further checking as proof that no material was used downstream.

This test concerns physical material history. The data lineage of the analysis instead concerns the path from the underlying data to the resulting list. Data provenance makes the data's origins and creation context traceable. Both can help verify the result; neither replaces a missing record of actual material use.

Even a fully documented material chain has limits: shared equipment, mix-ups, or possible environmental transfer may matter to an investigation without appearing in that chain as material inputs. Whether they implicate additional batches requires a separate assessment of the specific incident. The material search alone determines neither product safety nor the final recall scope.

The configured process must explicitly represent the splits, merges, and downstream relationships needed for a specific traceability task.