Same mould. Same material. Same process parameters. One part survives its service load without incident. The other cracks along a line nobody predicted, on a part that, by every spec sheet in the file, should have behaved identically.
This isn't a quality control story about a bad batch or an out-of-spec supplier. It's a story about what a material datasheet actually promises, and what it quietly doesn't.
A datasheet describes an ingredient, not a part
Specifying "30% glass-fibre-reinforced PP" specifies a formulation, not a performance envelope. Fibre-reinforced thermoplastics do not arrive at their final mechanical properties during compounding. They arrive there during flow, as the melt moves through the mould cavity and the fibres reorient in response to shear.
This matters because mechanical performance in a fibre-filled part is direction-dependent in a way solid metals simply aren't. A coupon cut and tested along the fibre orientation direction can show meaningfully different stiffness and strength than the same material tested transverse to it. The relevant question isn't "what does 30% glass-fibre PP do," but "what does 30% glass-fibre PP do at this specific location, in this specific orientation, given this specific flow history." Two parts can share a datasheet and still not share an answer to that question.
How flow actually organizes fibres
As melt enters a mould cavity, the classic skin-core structure forms. Near the cavity walls, high shear rates align fibres roughly parallel to the flow direction, forming a well-oriented "skin" layer. In the core, shear rates drop and fibre orientation relaxes toward something closer to random, or aligned transverse to flow depending on local flow kinematics. This isn't a defect. It's the normal, expected outcome of how polymer melts and suspended fibres respond to velocity gradients, and it's present in essentially every fibre-filled injection-moulded part.
The complication arrives where flow paths split and rejoin, which happens constantly around ribs, bosses, and any feature that forces the melt front to divide. Where two flow fronts meet again downstream, they form a weld line: a region where fibre alignment is disrupted, often with fibres oriented perpendicular to what would be the load-bearing direction elsewhere in the part. Mechanical properties at a weld line can drop sharply relative to the surrounding material, not because anything went wrong in processing, but because two melt fronts meeting head-on simply can't preserve the fibre alignment either front had on its own.
Why gate placement decides the outcome
Here is where two "identical" parts diverge. Gate location determines the entire flow path: where the melt front travels first, where it has to split around internal features, and critically, where the resulting weld lines land. Move a gate by even a few millimetres, or change from a single gate to a family tool with a slightly different runner balance, and the weld line locations can shift meaningfully, sometimes landing in a low-stress region in one case and directly across a high-stress load path in another.
This is exactly how two parts, moulded from the same material on the same press with the same cycle parameters, end up with their weakest point in entirely different locations. Nothing about the failure is random. It is fully determined by a flow history that most conventional design and QC processes never explicitly examine, because a structural stress analysis run against a single, spatially averaged material property has no way to see it.
Why this stays invisible until something cracks
A standard structural simulation workflow takes a material datasheet, assigns a single stiffness and strength value across the whole part, and solves for stress under load. That workflow will converge, produce a clean safety factor, and look entirely defensible, right up until a part fails at a location the model gave no special attention to, because the model had no representation of weld lines, skin-core variation, or local fibre orientation to begin with. The simulation isn't wrong on its own terms. It's answering a narrower question than the one the part actually needs answered.
This is precisely why coupling a mould-fill and fibre-orientation simulation to the structural solver is worth the added complexity and cost, particularly for parts with load paths that cross ribs, bosses, or multi-gate weld regions. A fill simulation predicts fibre orientation tensors throughout the geometry: where the skin is well-aligned, where the core has relaxed, and critically, exactly where weld lines will form given a proposed gate layout. Feeding that orientation data into the structural model, rather than a single bulk material card, makes the true anisotropic property distribution across the part visible before a single unit is moulded. Weak zones that would otherwise only surface in physical failure testing, or worse, in the field, become visible on screen during the design phase, when moving a gate or adding a rib costs nothing more than rerunning a simulation.
What this means in practice
None of this is an argument against fibre-reinforced thermoplastics, which remain an excellent choice for a huge range of structural applications precisely because of the stiffness and strength they add relative to unfilled resin. It's an argument for treating the material datasheet as a starting condition rather than a guarantee. The properties on that sheet were measured on a coupon with its own specific flow history, almost certainly not the flow history the actual part will experience.
What happens between the hopper and the finished part — the shear history, the weld line locations, the skin-core ratio at every wall thickness — is what actually determines whether the part holds up under load. Two identical-looking parts can carry two very different mechanical realities inside them, and the only way to know which reality has been built is to model the flow, not just the formulation.