Mold stress analysis FEA simulation

• 1
  What mold stress analysis FEA simulation actually solves
• 2
  The four load cases that decide mold life
• 3
  How to set boundary conditions without fooling yourself
• 4
  Reading the results: stress, deflection, and fatigue life
• 5
  Why FEA results depend on how the mold is cut
• 6
  What simulation cannot tell you
• 7
  When simulation is worth the time, and when it is not

A mold is a pressure vessel that opens and closes. Steel that looks solid at rest becomes a spring under cavity pressure, clamp force, and thermal growth. Mold stress analysis FEA simulation takes the tool geometry, splits it into elements, applies those loads, and returns displacement and stress fields you can compare against material limits.

The output is not a single number. You get a contour map of von Mises stress, a deflection plot across the parting line, and a safety factor per region. The useful question is never 'is the mold strong enough' but 'which 20 mm of this mold carries the risk, and what happens there after 200,000 cycles'.

FEA also exposes problems that no amount of machining accuracy can fix. A core pin seat can be machined to ±0.005 mm and still crack, because the failure is a stress concentration, not a dimension. Simulation moves that discovery to the screen instead of the press.

The method has limits. Linear elastic FEA assumes the steel returns to its original shape, which holds below yield but not in a crash or an overloaded clamp. Treat the results as a map of where to look, not a certificate.

Cavity pressure is the first case and usually the largest. Injection pressures for engineering plastics commonly sit between 30 MPa and 120 MPa at the gate, and die casting pushes far higher. The pressure acts normal to every cavity surface, so thin cores and long core pins see bending, not simple compression.

Clamp force is the second case and it is easy to underestimate. A 200-tonne press closes the platens with that force whether the parting line is flat or not. If the support plate deflects, the cavity flashes and the operator adds clamp tonnage, which increases the deflection. Simulation of the support plate under clamp load usually explains flashing better than cavity pressure does.

Thermal load is the third case and the one most often skipped. A hot runner manifold at 250 °C bolted to a plate at 60 °C grows by roughly 3 mm per meter of steel. Constrain that growth and the manifold bows or the bolts yield. Transient thermal FEA shows the gradient during the first ten shots, when distortion is worst.

Residual stress is the fourth case and it comes from the steel itself. Rough machining removes material unevenly and leaves a stressed skin. If you machine both sides symmetrically and stress-relieve between roughing and finishing, the finished cavity stays where the cutter left it. Skip that step and a flat plate can bow 0.05 mm while it sits on the bench.

Boundary conditions decide the answer more than mesh density does. Fixing the entire bottom face of a support plate makes it infinitely stiff and hides the deflection you are trying to measure. Constrain only the bolt footprints and let the rest of the plate bend.

Contact between plates is nonlinear and worth the extra solve time. If you model the parting line as bonded, the cavity cannot open and the flash prediction disappears. Frictionless or frictional contact with a 0.1 to 0.2 coefficient gives a more honest picture of how the stack behaves.

Mesh quality matters most at corners, fillets, and pin seats. A second-order tetrahedral mesh with three or more elements across a fillet radius is a reasonable starting point for a typical mold insert. Run a mesh convergence study on the critical region only; refining the whole model wastes hours.

Material data is the weakest link. Use temperature-dependent yield strength and thermal expansion for the actual grade, not generic tool steel. For a part that runs hot, a room-temperature modulus and a hot yield strength in the same model will mislead you.

Separate the two failure modes before you look at any contour. Deflection is a stiffness problem and it shows up as flash, short shots, or dimensional drift. Stress is a strength problem and it shows up as cracks at fillets, pin seats, and sharp internal corners. Different fixes apply to each.

A safety factor against yield is a reasonable first screen. Below 1.5 on a high-cycle mold, we look at the geometry again. Peak stress at a sharp corner is often a mesh artifact, so read the value one or two elements away from the singularity and check whether the stress field is smooth.

Fatigue is where the honest numbers live. Compare the alternating stress at the critical node against the endurance limit of the steel at operating temperature, then apply a knockdown for surface finish and any stress concentration. Shot count matters: 100,000 cycles is a different design problem than 2,000.

Deflection limits are set by the part, not by the steel. For an automotive connector with a ±0.05 mm tolerance, a 0.03 mm parting-line deflection already eats most of the budget. For a large, non-critical housing, 0.2 mm may be irrelevant.

A simulation is only as good as the geometry it is given, and the geometry is only as good as the machining process that produces it. A radius modeled at R3 that comes off the machine as R1.5 raises the stress concentration factor at that corner. The same applies to a pin seat modeled as a press fit but machined with a clearance.

This is why we run FEA and CAM from the same model. At GreatLight we machine mold inserts, cores, and support plates on 127 high-precision CNC machines, including 16 simultaneous 5-axis machining centers with a Ø400 mm rotary table. Complex cores with undercuts and deep ribs are cut in fewer setups, which keeps the modeled fillet and the machined fillet the same size.

Tolerance and finish also feed back into the model. Holding ±0.005 mm on a pin seat and Ra 0.8–1.6 μm on a sliding surface changes both the fit and the fatigue behavior. Surface roughness acts as a stress raiser, so a polished fillet at Ra 0.2–0.8 μm performs better than a turned one at Ra 1.6–3.2 μm.

When the analysis says a seat needs more support, the fix is a design change, not a tighter tolerance. We would rather move a boss or add an insert than promise a dimension the press will destroy in the first thousand shots.

FEA will not predict a cracked water line from mineral scale, a stuck sprue from a worn bushing, or a warped part from a cooling imbalance you never modeled. It answers structural and thermal questions within the assumptions you gave it.

It also will not replace a first-article inspection. Simulation predicts, the CMM confirms. For molds where the cavity dimensions carry a ±0.05 mm tolerance, we inspect the cut geometry, not the mesh, before the tool goes into the press.

Use the model to rank risks and pick the design that survives the press. Use measurement to prove the tool matches the model. The two together are cheaper than either one alone.