PhD research project · Mechanical engineering technology

Material-aware G-code, with the reason behind every number

AdvisorAI turns a 3D part model and a workpiece material into a complete CNC process plan. It assigns cutting speed, feed and depth of cut from physical models, checks them against machine, thermal and stability limits, and records which limit governs each value.

The problem

A CNC control programme stores the spindle speed and the feed, but not why they have those values. The workpiece material influences the programme only through the technologist's judgement or a lookup table in CAM software, and that reasoning is lost once the programme is written.

When a regime has to be changed, nobody can tell whether it was limited by the machine's power, by the cutting temperature, or by vibration, and each calls for a different remedy.

Aim of the project

To make the transition from material properties to programme parameters explicit, reproducible and verifiable. It is formulated as four principles, implemented as a constrained selection algorithm, and delivered as working software.

About the research →

Objectives

The research tasks of the dissertation, as implemented in this system.

Formalise the principles

State how speed, feed, depth of cut and tool changes follow from measurable material properties, as explicit equations rather than tables.

Build consistent models

Force, temperature, tool life, deflection, stability, surface and energy models that share inputs, units and assumptions.

Select with constraints

Assign each regime subject to spindle power, thermal, stability and roughness limits, and report the one that governs.

Generate and check G-code

Produce an ISO 6983 programme and simulate it along the toolpath before it reaches a machine.

Plan the calibration

Define an ISO 8688-1 test programme that determines every model coefficient experimentally.

Compare across materials

Quantify how six materials from five ISO 513 groups change the assigned regime on one machine and cutter.

Key features

Everything below runs in the browser from a single STL file.

Guided eight-step workflow

From model upload through features, process plan, physics analysis and G-code to validation and export.

Constraint trace

Every regime is reported with the limitation that fixed it: machinability band, spindle power, temperature or chatter.

Thermal assessment

Two-zone cutting temperature and a homologous-temperature criterion referred to the alloy's solidus.

Optimization

A grid search over speed, feed and depth, ranked by Pareto dominance, with a TOPSIS compromise regime.

Check your own G-code

Upload your program: AdvisorAI simulates it on the stock, finds gouges, collisions and overloads, re-tunes feeds and speeds, and says whether a better version exists.

Energy and CO2

Cutting power, spindle load and energy per part computed from force and the actual toolpath length.

See all features →

How it works

  1. Upload

    Load a binary or ASCII STL exported from any CAD system, or use the sample block.

  2. Configure

    Choose the material and machine; the system detects surfaces, pockets, holes and thin walls.

  3. Analyse

    Principles P1–P4 assign the regime; the physics models predict force, temperature and tool life.

  4. Export

    Review the simulation, then download the G-code, the report and the data as JSON.

Project in numbers

4
formalised principles (P1–P4)
8
coupled physical models
6
materials in the case study
3
interface languages

Research status. The quantitative results are model predictions with provisional coefficients. The experimental calibration programme (ISO 8688-1) is specified but has not yet been carried out. See Limitations.

Try it on your own part

No account is needed. Uploaded files are processed in memory and are not stored.