Research-Stack/3-Mathematical-Models/AMMR/WAVE_OVERHANGS_HEAT2D.md

4.9 KiB

Wave Overhangs + Heat-2D Integration

Why this belongs here

PrusaSlicer-WaveOverhangs introduces wave-generated toolpaths for unsupported FDM overhangs. The slicer fills unsupported bottom-surface regions with recursively generated wave paths instead of ordinary support material. The algorithm is described as wave-propagation inspired: waves continue until they fill available space and can diffract around corners and holes.

This makes it a natural companion to heat-2D:

wavefront toolpath geometry
    -> deposited polymer strand
    -> cooling / reheating / contraction
    -> warping or stable overhang

heat-2D should not try to reproduce the slicer. Its role is narrower:

Given a wave-overhang toolpath source pattern, estimate the thermal residue field and failure-risk envelope.

Core PDE

The native heat-2D equation is:

\partial_t u = \nabla\cdot(\alpha\nabla u)+f

For wave overhangs:

Term Interpretation
u(x,y,t) local thermal state / cooling residue of the overhang layer
alpha(x,y) effective thermal diffusivity of polymer, air gap, perimeter anchor, or fiber-filled material
f(x,y,t) moving deposition source from nozzle path
boundary conditions cooling from air/fan, anchor conduction into perimeter, or imposed bed/chamber temperature

Toolpath-as-source model

Represent the wave path as a moving heat source:

f(x,y,t)=Q_n\exp\left(-\frac{\|\mathbf{x}-\mathbf{x}_{nozzle}(t)\|^2}{2\sigma_n^2}\right)

where:

Symbol Meaning
Q_n deposited thermal source strength
x_nozzle(t) time-parametrized wave toolpath
sigma_n effective bead/nozzle heat radius

Cooling sink:

f_{cool}(x,y,t)=-h_f(u-u_{air})

Combined:

\partial_t u=\nabla\cdot(\alpha\nabla u)+f_{nozzle}+f_{reheat}+f_{cool}

Warping risk proxies

WaveOverhangs documentation identifies warping as a coupled thermal, mechanical, and process-control problem. This adapter only models the thermal part directly, but it can output risk proxies.

1. Temperature-gradient stress proxy

G_T=\|\nabla u\|

Higher local thermal gradients imply higher differential contraction risk.

2. Reheat activation proxy

R_{reheat}(x,y)=\max_t \mathbf{1}_{u(x,y,t)>T_g}

This estimates where previous strands may re-enter a mobile polymer state.

3. Curl-risk proxy

C_{curl}=w_1\|\nabla u\|+w_2R_{reheat}+w_3t_{hot}-w_4A_{anchor}

where A_anchor measures proximity/connection to perimeter or previously stabilized strand.

4. Span-size risk

C_{span}\propto L_{unsupported}^2

Large unsupported spans amplify nozzle-pressure and contraction risks.


Parameter bridge to slicer settings

WaveOverhangs setting Heat-2D proxy
line spacing source-path spacing / wavelength
line width source radius and deposited amount
flow ratio source amplitude Q_n
print speed source dwell time
fan speed cooling coefficient h_f
perimeter overlap anchor conduction / boundary coupling
minimum wave width geometry mask pruning threshold
monotonic / zig-zag / smart source ordering and local reheat history

Test cases

Test A: single wave stripe

Goal: validate moving-source thermal trail.

Expected:

  • smooth trail behind nozzle,
  • peak temperature decays after source passes,
  • stronger fan coefficient reduces hot lifetime.

Test B: adjacent wave lines

Goal: test line spacing / line width / flow ratio coupling.

Expected:

  • tighter spacing increases overlap,
  • higher flow ratio increases heat accumulation,
  • slower print speed increases local dwell and bonding but may reheat earlier lines.

Test C: monotonic vs zig-zag ordering

Goal: quantify local heat buildup from path ordering.

Expected:

  • monotonic gives neighboring lines more cooling time,
  • zig-zag lowers travel but can increase local heat accumulation,
  • smart ordering should reduce unsupported-start risk, though geometry support is outside pure heat diffusion.

Test D: large unsupported span

Goal: identify when thermal gradients and hot lifetime become too large.

Expected:

  • larger spans increase curl-risk proxy,
  • uniform two-sided cooling reduces vertical-gradient proxy if modeled with layered extension,
  • fiber-filled material proxy should use higher conductivity, lower expansion risk, and higher stiffness in downstream mechanical model.

Scope warning

This is not a full FDM mechanics model. Pure heat-2D does not solve:

  • bead sag,
  • viscoelastic shape memory,
  • nozzle pressure deformation,
  • polymer crystallization,
  • 3D strand geometry,
  • mechanical stress equilibrium.

Correct use:

thermal afterimage and risk proxy generator for wave-overhang toolpaths

Downstream mechanical models should consume the heat-field output rather than be hidden inside this solver.