Short answer up front — this is a common Abaqus/Explicit behavior: adding realistic plasticity + JC damage to a deformable projectile often prevents “automatic” element deletion unless the damage evolution / deletion flags are set correctly (or you explicitly kill elements in a user subroutine). Below I list likely causes, step-by-step checks you can run, and practical workarounds that have worked for others.

Likely causes

No effective damage evolution → element not eroded.

In Abaqus/Explicit element deletion only happens when the damage model both initiates and evolves to remove the element’s load carrying stiffness (D → 1 or fracture energy consumed). If you only specify an initiation value but not an appropriate damage-evolution (or give it an unrealistically large fracture energy), elements will get highly distorted but not deleted. (VT Software Docs, ceae-server.colorado.edu)

Built-in failure vs. progressive damage difference.

Some built-in failure options immediately remove elements (simple “fail” shear options), while the progressive damage route (damage initiation + evolution) requires correct evolution parameters to reach removal. Check whether you used a “shear failure” type or the progressive JC damage sequence. (Eng-Tips, VT Software Docs)

User material (UMAT/VUMAT) or custom models need explicit deletion flags.

If you implement JC plasticity/damage in a UMAT/VUMAT you must drive the Abaqus element-deletion mechanism yourself — e.g. set the designated state variable / STATUS to indicate a deleted material point (DEPVAR with DELETE option). If you don’t, Abaqus won’t remove the element. (WashU Engineering Courses, ResearchGate)

Constraints / contact / node sharing keeps elements “attached”.

If projectile elements are tied, constrained, or heavily loaded by contact that the solver treats as still active, you can see severe distortion rather than deletion. Contact formulations and tied interfaces can prevent node removal even if the element is flagged. (YouTube)

Numerical issues (hourglassing, severe distortion) obscure deletion.

With C3D8R you can get excessive hourglass or negative volumes; Abaqus may leave a distorted element rather than deleting it if the damage scalar SDEG never reaches the removal threshold. Output SDEG / status to confirm. (CAE Assistant)

Practical checklist / troubleshooting (do these in order)

Confirm solver: make sure you run Abaqus/Explicit (element deletion is explicit-oriented unless you use UMAT with delete option). (WashU Engineering Courses)

Output internal damage variables: request SDEG, STATUS, and element/point level variables for the projectile. Inspect if damage ever reaches 1 (or STATUS becomes 0). If SDEG stays small → evolution is wrong. (Reddit)

Check Damage setup:

Use Damage Initiation (ductile) with JC damage parameters.

Add Damage Evolution and choose a physically reasonable evolution (equivalent plastic displacement or fracture energy) so that the element stiffness goes to zero within a plausible strain/displacement. If fracture energy >> required value, elements won’t erode. (VT Software Docs, ceae-server.colorado.edu)

If using built-in JC failure, check the “failure option” type. If you used a “shear failure” quick option that should remove elements when criterion met; if you used progressive approach make sure evolution leads to D→1. (Eng-Tips)

If using VUMAT/UMAT: implement element deletion correctly. In VUMAT/UMAT use the documented DEPVAR, DELETE flag or set STATUS=0 for deleted integration points. Many forum answers and the manual show that you must set the deletion state variable explicitly. After doing so, request STATUS in output to verify deletion. (Eng-Tips, ResearchGate)

Check contact and constraints: temporarily disable contact or simplify to frictionless contact to test deletion behaviour. If elements delete with contact off, you need to modify contact definitions (allow separation, use surface erosion options, avoid tied constraints across parts you expect to fail). (YouTube)

Check element formulation / hourglass control: enable appropriate hourglass stabilization for C3D8R; refine mesh locally so elements deform less before reaching damage. Severely distorted elements may mis-report damage. (CAE Assistant)

Run a single-element / small block test: model a small sample of projectile material with JC plasticity + damage under your expected strain rate to verify the chosen parameters actually cause element deletion at the desired strains. This isolates the material law from contact/geometry complexity. (scholar.afit.edu)

Workarounds & alternatives

Explicit element killing in VUMAT/VUSDFLD: If built-in damage still fails to remove elements for your deformable projectile, use a VUMAT/VUSDFLD that sets the element deletion flag (STATUS or DEPVAR DELETE). This is the most robust way to ensure elements are actually removed when your composite criterion is met. (WashU Engineering Courses, ResearchGate)

Use a rigid projectile (if acceptable): you already noted rigid projectiles work — if projectile deformation is not critical to your failure mechanism, keep it rigid to avoid the complication.

Two-step approach: run impact with a rigid projectile to capture target response; if projectile deformation is necessary, run a separate deformable projectile-only test or use submodeling.

Coarse-grained erosion (user node deletion): some users delete elements/nodes via a subroutine and remesh or use adaptive techniques externally. These are more invasive but workable.

Quick diagnostics you can run now

Turn on SDEG and STATUS for the projectile and inspect output at a time when you expect failure. Do elements reach SDEG≈1?

Run a block tensile test on projectile material (same element type, mesh, and JC+damage settings) to confirm damage evolution → deletion.

If deletion fails in the simple test, increase damage evolution aggressiveness (lower fracture energy / smaller displacement) until deletion occurs — then port that setting back to the full model.

If deletion occurs in the simple test but not in full model, simplify contact/ties and reintroduce them gradually to find what prevents deletion.

References & reading (Abaqus docs + community notes)

Abaqus/Explicit: element deletion via user subroutines and built-in damage evolution. (WashU Engineering Courses, VT Software Docs)

Johnson-Cook plasticity & damage in Abaqus/Explicit — documentation and examples. (VT Software Docs)

Community threads & tips (Eng-Tips, ResearchGate) on using STATUS/DEPVAR DELETE in VUMAT/VUMAT to force deletion and on common pitfalls.

Here are several troubleshooting steps and potential solutions for the element deletion problem you're experiencing with a deformable projectile in ABAQUS. The key difference between your rigid and deformable projectile simulations is the introduction of a complete material model (plasticity and damage) to the projectile, which can introduce new complexities.

Most likely causes and solutions

Step-by-step diagnostic checklist

Advanced alternatives

If the built-in damage models and contact settings continue to be problematic, you may need a more direct approach:

• Implement failure in a VUMAT subroutine: This involves writing a User Material Subroutine (VUMAT) where you can explicitly set the STATUS variable to 0 for any element integration point that meets your failure criteria. This gives you full control over when and how elements are deleted.

Community references

• Forums on ResearchGate and Eng-Tips often discuss this exact problem. Many users have found that adjusting damage evolution or implementing VUMAT provides the needed control for complex interactions.
• Academic papers comparing JC and MMC models for specific materials can also be a valuable resource. Some studies even note that the MMC model can be more conservative in predicting failure.