Home >Open Calls >Experiments > ARCTUS

AI Prediction Models for Commercial Refrigeration Condensing Unit Systems.

Energy, Skills
ARCTUS image


ARCTUS is an experiment designed to spark the digital transformation of a traditional HVAC SME. Its modular nature will enable support and assistance on two real-life challenges. Firstly, it will introduce established data-driven methodologies to the HVAC engineers, in order to enhance the skillset of its employees, increasing the innovation potential, especially when new products or services can emerge from data-driven applications. Consecutively, the engineers will be trained with methods that can unveil the potential that underlies in data, and aim to build novel, commercial services. Secondly, as a direct outcome of the training activities, ARCTUS will support the specific development of AI models aiming to predict upcoming malfunctions and forecast energy consumption in RCUs that operate in commercial environments. The development of such models, will be used both for research purposes, as well as for the development of a commercial service, offered to clients and stakeholders.


I-Space Coach


I-Spaces Involved

Main Objectives
  • Train HVAC experts
  • Create AI models
Main innovations

ARCTUS supports two innovations: the first is targeted on the training of HVAC experts on technologies and applications of data-driven methods, and the development of predictive models.
The second innovation focuses on the development of two predictive models for commercial refrigeration units, which will lead to a marketable service. Currently, such application is commercially non-existing, therefore it has the potential to create an innovative product, as well as new knowledge.


IMPACT 1: Reduce downtime in customer operations
A customer can timely prepare for replacing the problematic components, thus reducing the delivery time, as well as plan the best time for maintenance.

IMPACT 2: Reduce economic impact on unforeseen events
Energy consumption forecasting allow the optimization of energy sources and minimizing costs, and the fault prognostics prevent from unfortunate events like the damaging of products that are stored in cooling facilities.


Key Exploitable Results (KERS)

KER1: AI model for fault prognostics
Offering early warnings on upcoming potential faults.

KER2: Energy consumption forecasting model
Providing energy consumption forecasts for adjusted operation.

KER3: Identification of opportunities
Seminars on AI applications in the HVAC domain, sparking innovation.

KER4: Increase competence
Leading to the development of products which will generate revenues, attract new clients and bring public attention to Klimamichaniki as a frontrunner SME.

Technical KPIS
  • Introduce data-driven methodologies
  • Train engineers on HVAC applications
  • Create forecasting model
  • Create fault detection model


Klimamichaniki was founded in 1984 in Thessaloniki, Greece, as an engineering consultancy and construction company which dealt with air-conditioning installations. The expertise expanded through the years and presently covers any type of heating, ventilation, and air-conditioning (HVAC) systems.