Development of a Digital Twin of an Oil-Flooded Screw Compressor Using Measurement Data and Numerical Simulations

Abstract

The paper describes the development of a digital twin of an industrial compressor with an integrated compressed air adsorption dryer. Digital twin development is enabled by prototyping an oil-flooded screw compressor with an integrated dryer and numerical modelling of partial phenomena and the whole system. The results of the numerical modelling and measurements of the prototype are then used to train and adapt neural networks to develop the digital twin software. A sufficient amount of input data is essential for the development and subsequent application of the digital twin of the compressor including the integrated dryer. These data must correspond to both standard operating conditions and off-design conditions with respect to, for example, ambient temperature, intake air humidity, compressor load or regeneration air temperature and quantity, which are taken into account in the measurements and numerical simulations. Taking into account the different operating conditions allows the neural networks to find the optimal design parameters for the entire compressor station with respect to the specific requirements and conditions of the end users, which is an important aspect of the added value of the digital twin.