Virtual Tests
Subjecting digital materials to rigorous simulated environments and testing conditions.
Once a virtual sample is constructed, we evaluate its physical behavior using virtual tests: algorithms that dictate how temperature, boundary deformations, and external forces act on the simulation box over time. These tests are the direct computational counterparts to physical laboratory instruments such as capillary rheometers, dynamic mechanical analyzers (DMA), and tensile grips.
Sample Packing & Box Initialization
The testing lifecycle begins by generating the 3D spatial coordinates and molecular bond topologies for all components in the blend. Polymers, comonomer branches, and additives are assembled inside a 3D unit cell with Periodic Boundary Conditions (PBC), a mathematical boundary that wraps opposite faces of the box so particles leaving one side immediately re-enter from the other, effectively simulating an infinite bulk polymer without artificial surface edges.
Initial particle velocities are assigned from a Maxwell-Boltzmann distribution, matching the kinetic energy of the atoms to the target processing melt temperature. At this initial stage, polymer chains are packed incrementally into the cell before thermodynamic relaxation begins.
Figure: Chains and side-groups are procedurally constructed and introduced incrementally into the periodic unit cell boundaries.
Thermodynamic Equilibration (Annealing)
Before applying any external dynamic load, the digital sample must be relaxed to its lowest-energy, un-stressed thermodynamic state. We equilibrate the sample using two standard statistical mechanics ensembles:
- NPT Ensemble (Isobaric–Isothermal): Maintains constant particle count (N), constant pressure (P = 1 atm), and constant temperature (T). The simulation cell volume contracts and adjusts dynamically, allowing the material to consolidate to its true physical melt density.
- NVT Ensemble (Canonical): Maintains constant particle count (N), fixed volume (V), and constant temperature (T) to stabilize thermal kinetic velocities.
This annealing stage eliminates unphysical steric overlaps (atoms placed unnaturally close during initial packing) and relieves internal pre-stresses, mirroring how physical compounded test specimens are annealed before laboratory characterization.
Figure: The simulation cell volume contracts under NPT conditions, consolidating to target density and relaxing packing stresses.
Applying Dynamic Virtual Tests
Once equilibrated, we subject the sample to Non-Equilibrium Molecular Dynamics (NEMD) tests. By actively deforming the cell boundaries or imposing external momentum over time, we evaluate material response under real-world processing conditions:
- Rheology (Planar Shear Flow / SLLOD): The SLLOD algorithm imposes continuous affine shear deformation on the cell, causing polymer chains to tilt, orient, and slide past one another. This evaluates non-Newtonian flow behavior, shear-thinning indices, and first normal stress differences across shear rates from 10² s⁻¹ to 10⁶ s⁻¹.
- Transport Properties (RNEMD): Reverse Non-Equilibrium Molecular Dynamics imposes a controlled momentum flux across cell slabs to measure velocity gradients, zero-shear viscosity, and thermal conductivity.
- Mechanical (Tensile Elongation): Applies constant-rate boundary elongation along the primary axis while contracting transverse directions (Poisson effect) to measure tensile yield, modulus, and strain hardening.
Figure: Dynamic virtual testing: toggle between Rheology (SLLOD shear deformation) and Mechanical (tensile elongation) to observe real-time chain orientation.
On-The-Fly Calculation (OTFC Engine)
Simpal does not dump heavy, multi-gigabyte trajectory files for delayed post-processing. Instead, our proprietary OTFC property engine evaluates physical observables continuously in memory during the simulation run.
At every integration time-step, the engine computes and tracks:
- Virial Stress Tensor (σ_ij): The microscopic statistical equivalent of macroscopic Cauchy stress, capturing instantaneous shear and normal stresses.
- Radius of Gyration (Rg) & End-to-End Distance: Metrics tracking the spatial coil dimensions and uncoiling orientation of polymer chains.
- Stress Autocorrelation & Relaxation Spectra:Time-correlation functions that directly yield viscoelastic storage ($G'$) and loss ($G''$) moduli.
These extracted descriptors are immediately formatted and passed downstream to our AI property models and surrogate estimators.