Tecnalia – OSARGI: Operation and Maintenance through Predictive Models for HOSPITAL COMPLEXES
Tecnalia
Sector: Construction
Business Case
Hospital management systems employ reactive control strategies that lack the flexibility to adjust to changing boundary conditions (usage, weather, etc.) and allow for optimal system operation. Preventive maintenance strategies, on the other hand, provide suitable levels of equipment reliability and availability, but do not make it possible to minimise maintenance costs. The developed platform enables the transition to predictive O&M strategies by exploiting the capabilities provided by predictive models, optimisation algorithms and IOT networks.
Objectives
Real-time, non-human intervention optimisation of operation strategies and process/service (HVAC) adjustments to minimise their operating costs (continuous commissioning, leveraging building flexibility and price signals, etc.) Maximise the availability, reliability and useful life of HVAC equipment by means of anomaly detection, diagnosis and prognosis functionalities, minimising the cost associated with maintenance work. Ensure environmental quality provided to users (thermal comfort, indoor air quality, etc.)Garantizar la calidad ambiental proporcionada a los usuarios (confort térmico, calidad de aire interior, etc)
Use case
The solution is based on a monitoring platform that is installed as a top layer of the BEMS, composed of specific modules to optimise the operation and maintenance of processes/services (HVAC). The modules have an ecosystem of optimisation algorithms and predictive models developed through the combined use of physical models (EnergyPlus and Modelica), Machine Learning techniques (regression and classification) and an incremental learning procedure. Based on the prediction of boundary conditions (weather, energy prices, usage), this ecosystem provides O&M optimisation functionalities (anomaly detection, diagnosis and prognosis).
Infrastructure
On Premise & Cloud
Technology
Automatic or Deep Learning
Data
Weather forecast service, prediction of energy price signals, HVAC system operating parameters, internal operating parameters of monitored equipment, environmental conditions, settings/schedules.
Resources
Weather forecast service, prediction of energy price signals, HVAC system operating parameters, internal operating parameters of monitored equipment, environmental conditions, settings/schedules.
Difficulties and learning
Generation of synthetic data (definition of physical models for operation and maintenance); availability of historical data for operation; availability of labelled historical data for the operating parameters of the monitored equipment (especially the corresponding anomalous operating states).
KPIs (business impact and metrics of the model)
5-25% reduction in the energy cost of air conditioning. 10-25% reduction of maintenance costs associated with air conditioning systems. Interoperability through standard communication protocols (BEMS, CMMS, IOT networks, etc.).
Funding
HAZITEK
Collaborators, Partners
Tecman, Sedical, Homsa
