The semi-explicit nonsmooth Newmark time integrator for robust unilateral contact in dynamic fragmentation simulations

This paper introduces and validates a semi-explicit Nonsmooth Newmark (NSN) time-integration scheme that robustly handles unilateral contact in dynamic fragmentation simulations by strongly enforcing constraints, thereby achieving superior stability and accuracy over penalty-based methods while revealing that contact dissipation can paradoxically increase fragment counts by improving damage localization.

The Gist

Imagine you are trying to simulate what happens when a solid object, like a ceramic plate or a rock, gets hit so hard it shatters into thousands of tiny pieces. This isn't just a simple break; it's a chaotic explosion where pieces fly apart, smash into each other, bounce off walls, and grind together.

The paper introduces a new computer program (a "time integrator") designed to simulate this chaos without the computer crashing or giving nonsense results. Here is how it works, broken down into simple concepts:

1. The Problem: The "Spring" Trap

To simulate breaking, scientists usually use a method where they pretend the material is made of tiny springs. When the material breaks, the springs snap. When pieces hit each other, they use "penalty springs" to push them apart so they don't pass through one another.

The Analogy: Imagine trying to stop a bowling ball with a rubber band.

• If the rubber band is too loose (low stiffness), the ball passes right through it (unphysical).
• If the rubber band is super tight (high stiffness) to stop the ball perfectly, it acts like a rigid wall. But if you make it too tight, the computer has to take tiny, tiny steps to calculate the bounce, making the simulation take forever.
• The Paper's Claim: The old way (using these tight springs) is unstable. It causes the computer to drift, lose energy, or crash, especially when there are millions of collisions happening at once.

2. The Solution: The "Traffic Cop" (Nonsmooth Newmark)

The authors created a new method called Nonsmooth Newmark (NSN). Instead of using rubber bands to push pieces apart, this method acts like a strict traffic cop at a busy intersection.

The Analogy:

• The Bulk (The Car): The main body of the object moves freely and smoothly. The computer predicts where the car would go if there were no obstacles. This part is calculated very quickly (explicitly).
• The Contact (The Intersection): If the car hits a wall or another car, the "traffic cop" steps in. Instead of pushing the car back with a spring, the cop instantly says, "Stop! You cannot go there." It enforces a hard rule: No passing through.
• The Magic: This method treats the "no passing" rule as a hard law of physics rather than a soft spring. It allows the computer to take much bigger steps in time because it doesn't have to worry about the rubber band getting too tight.

3. The "Split Personality" Approach

The paper describes this method as "semi-explicit." Think of it as a two-step dance:

1. Step A (The Prediction): The computer guesses where everything will be in the next moment, ignoring collisions.
2. Step B (The Correction): If the guess shows two pieces overlapping, the computer instantly corrects their speed and position to fix the overlap, just like a billiard ball hitting another and instantly changing direction.

This allows the simulation to be fast (like the prediction) but accurate and stable (like the correction).

4. What They Found (The Experiments)

The authors tested this new "traffic cop" method against the old "rubber band" methods using three scenarios:

• The Bouncing Ball: A simple ball bouncing on the floor. The new method was just as accurate as the best existing methods but handled the bounces without losing energy or getting jittery.
• The Striking Bar: A metal bar hitting a wall. The old methods struggled with the speed of the impact, but the new method handled the "crunch" perfectly, keeping the energy calculations correct.
• The Shattering Bar: A bar that is already cracked and then hits a wall. The old methods required such tiny time steps to stay stable that they were incredibly slow. The new method could take huge steps, running 27 times faster while being more accurate.

5. The Surprising Discovery: Confined Shattering

The most interesting part of the paper involves a "confined" experiment. Imagine a bar shattering inside a small box instead of in open space.

• The Old Intuition: You might think that if pieces bounce off the walls and lose energy (dissipation), there would be less energy left to break the material, resulting in fewer, bigger chunks.
• The Paper's Finding: The opposite happened. When the pieces bounced off the walls and lost a little energy (contact dissipation), the material actually broke into more, smaller pieces.
• Why? The authors explain that the "bouncing" acts like a filter. In a perfectly elastic (bouncy) world, stress waves bounce around wildly, causing the material to get "confused" and develop many tiny, weak cracks that don't fully separate. When the walls absorb some of that energy, the waves calm down. This allows the stress to focus on specific spots, driving cracks all the way through to create clean, separate fragments.

Summary

The paper presents a new mathematical tool that simulates breaking objects by treating collisions as hard, instant rules rather than soft springs. This makes the computer simulation:

1. More Stable: It doesn't crash or drift.
2. Faster: It can take bigger time steps.
3. More Accurate: It correctly predicts how many pieces an object will break into.

The authors conclude that this tool is ready to be used for complex 3D simulations, such as understanding how spacecraft debris breaks apart or how rocks shatter in avalanches, by providing a robust way to handle the chaotic dance of millions of colliding fragments.

Technical Summary

Technical Summary: The Semi-Explicit Nonsmooth Newmark Time Integrator for Robust Unilateral Contact in Dynamic Fragmentation Simulations

Problem Statement
Numerical simulations of dynamic fragmentation involve a dual challenge: capturing the rapid topological evolution of a solid continuum shattering into discrete fragments and resolving the dense, simultaneous contacts that occur between these fragments and crack faces. While the Finite Element Method (FEM) with extrinsic cohesive zone models (CZM) effectively handles fracture, conventional contact treatments pose significant difficulties. Penalty-based methods, often used for their compatibility with explicit time integration, suffer from a fundamental trade-off: low stiffness permits unphysical interpenetration, while high stiffness restricts the stable time step to computationally prohibitive levels. Furthermore, the interaction between fracture and regularized contact introduces nonsmoothness that leads to persistent numerical energy drift and instability, masking the underlying physics.

Methodology
The authors propose and validate a novel semi-explicit time-integration scheme: the Nonsmooth Newmark (NSN) method. This approach adapts the Nonsmooth Contact Dynamics (NSCD) framework to dynamic fragmentation by decoupling the treatment of bulk dynamics and contact interactions:

1. Semi-Explicit Formulation: The scheme splits the dynamics into a "smooth" part (bulk deformation, external loading, and cohesive fracture) and a "nonsmooth" part (contact reactions).
   • The smooth dynamics are integrated explicitly using a second-order Newmark-$\beta$ scheme ($\beta=0, \gamma=1/2$), maintaining high accuracy during free-flight phases.
   • The nonsmooth dynamics (unilateral contact) are treated implicitly at the velocity level using a first-order implicit Euler approach. This strongly enforces Signorini gap constraints without artificial compliance.
2. Algorithmic Implementation: The contact problem is formulated as a Linear Complementarity Problem (LCP), which is solved as a convex Quadratic Programming (QP) problem restricted to the active contact set. This allows for efficient, parallelizable solutions that scale with the number of active contacts rather than the total system size.
3. Modified Traction-Separation Law (TSL): To address the numerical challenge of infinite cohesive stiffness as damage approaches zero in explicit schemes, the authors introduce a stiffness cap ($\tilde{k}$). This regularization prevents the time step from vanishing while preserving the convexity of the contact problem.
4. Validation and Application: The framework is implemented within the open-source Akantu FEM library. It is validated against analytical solutions and established nonsmooth methods (Moreau-Jean and CD-Lagrange) using benchmarks such as the bouncing ball and impacting bar. It is subsequently applied to 1D dynamic fragmentation under both free and confined expansion.

Key Contributions and Results

• Accuracy and Stability: Benchmark tests demonstrate that the NSN scheme achieves accuracy comparable to established nonsmooth methods (Moreau-Jean and CD-Lagrange) and significantly outperforms penalty-based approaches. In the impacting bar test, NSN maintains second-order accuracy for displacement and superior velocity accuracy compared to CD-Lagrange, particularly in elastic regimes.
• Computational Efficiency: Although the NSN method incurs a higher computational cost per time step (approximately 5 times that of a purely explicit penalty scheme due to the QP solve), its robust stability allows for significantly larger time steps. For 1D benchmarks, this results in a total time-to-solution that is comparable to or better than purely explicit approaches, with energy conservation errors reduced by orders of magnitude (near machine precision).
• Physical Insights in Confined Fragmentation:
  • Energy Budget Shift: In confined expansion scenarios (where fragments collide with rigid walls), the energy budget for fracture shifts from the local kinetic energy of individual fragments to the global kinetic energy of the entire system. This leads to a drastic reduction in mean fragment size and an increase in total fracture energy dissipation compared to free expansion.
  • Counterintuitive Role of Contact Dissipation: The study reveals that introducing contact dissipation (inelastic collisions, $e < 1$) reduces the total energy dissipated by fracture yet increases the final fragment count. This occurs because contact dissipation acts as a low-pass filter, suppressing high-frequency stress wave oscillations. This filtering promotes cleaner stress-wave propagation and better damage localization, driving cracks to full separation rather than allowing damage to remain distributed in partially damaged zones ($d < 1$).

Significance
The paper establishes the NSN scheme as a robust tool for generating high-fidelity fragmentation statistics in dense-contact scenarios. By rigorously treating contact as nonsmooth while retaining explicit efficiency for bulk dynamics, the method overcomes the numerical instabilities inherent in penalty-based formulations. The authors claim that this framework is essential for simulating complex multi-body interactions, such as those found in orbital debris scenarios, where frequent fragment-fragment interactions drive secondary fragmentation. The work provides a critical step toward extending high-fidelity fragmentation analysis to higher dimensions, leveraging the method's ability to preserve physical fidelity without the prohibitive cost of global implicit solves.