A digital toolkit to model energy consumption of manufacturing operations within the industry 4.0 factory

The fourth industrial revolution (known as Industry 4.0) introduces technologies that offer for the first time the opportunity to build a "Digital Factory" that complements the real one. As environmental legislations now penalise energy waste, energy efficiency of production operations is becoming critical. This research considers using one of the "Digital Factory" concepts to simulate manufacturing operations in order to predict their consumption so that it can be later optimised. This concept is the virtualisation of production operations and this thesis considers the special case of machining. In particular, an end-milling case study is investigated.

Other researchers have attempted to develop solutions in this area but hardly any can provide a realistic simulation to a physical machining process. The published literature shows a lack of solutions that relate the numerical control instructions to the energy consumption. Even the ones that do, they demonstrate limitations such as incompatibility with any workpiece geometry, absence of visualisation of materialcutterinteraction or lack of integration of changes in the depth and width of cut into the consumption estimation. Several of them also lack validation. Additionally, most of the proposed implementations are not bundled in a software easy to use and integrate into a broader system.

Moreover, a literature review has shown that the experiment design and development process required to collect the data for the energy model development, a time-consuming and prone to human errors procedure, is often overlooked. The same applies to the post-processing of the collected data.

In order to provide solutions to these limitations, this thesis introduces:1. A software for the automation of the design of experiment for end-milling machining.2. A novel method for the automation of the feature extraction from experimental end-milling machining data.3. A software for the estimation of the energy consumption of end-milling operations through virtual machining.All three solutions that are introduced by this thesis were validated and the developed software was verified against specifications that were defined before it was implemented. The validation of the design of experiments software produced the data that was used to validate the automated feature extraction methods and to build the models for the virtual machining toolkit.

In summary, this thesis developed a novel virtual machining toolkit for the prediction of the energy consumption of end-milling processes that can estimate the material cutting energy consumption with an average error of 3:27%. The results for the spindle acceleration energy prediction were mixed, as the expected performance was achieved only when the initial spindle speed was zero.

Complementary to this, a software that can automate the design of end-milling experiments and a novel automated feature extraction method were developed, with the latter identifying on average 94.91% of the material cutting states within the power signal and reporting 94.08% detection F1-Score for the material cutting states and 78.18% for the spindle acceleration identification.