The problems that serve as a starting point for determining project objectives are:
(1) There is no developed methodological framework in the research area of energy efficiency and energy consumption
(2) In the supply chain of natural gas, as one of the importan energy sources in public sector, there are losses that could be reduced by adequate simulation modeling.
To identify characteristic groups (clusters) of public buildings for which data on energetic characteristics, measures for improving energetic characteristics, and energy consumption are available in existing information systems.
To develop a methodological framework for analyzing the influence of specific characteristis of buildings and types of implemented measures on predicted and real savings in energy consumption, CO2 emission and cost, by using intelligent data analytics.
To assess financial compensation and long-term financial effects of specific measures of increasing energy efficiency by advanced analytical methods.
To suggest improvements in supply chain of natural gas as one of the important energy sources in building sector by simulation modelling that will lead to reduce the total energy cost.
The objectives will be achieved by several steps:
Collecting data from the Information system of energy management (ISGE) and from other sources of state institutions.
Developing prediction and classification models by testing more machine learning methods that have shown their success in prescriptive and predicting analytics (cluster analysis, neural networks, decision trees, association rules, and others).
Suggesting guidelines for integration of methods in predicting energy efficiency, consumption and cost through unique methodological framework.
Cluster analysis of buildings in public sector according to common characteristics and energy consumption.
Comparing the accuracy of methods in modelling in order to propose the most successful model for each group of consumers and to provide an integrative approach.
Implementing guidelines into the model of supply chain of natural gas, that will enable more efficient management of this energy source.
Some of the hypotheses that will be tested in the project research are:
Buildings in the public sector of Croatia could be grouped into clusters according to their characteristics and behavior in energy consumption (for example according to age of buildings, types of termal protection, geometrical characteristics, type of thermo-technical system, location etc.).
Machine learning methods will provide more accurate prognoses of payback period than the standard method of simple payback period (JPP) which is currently used by experts for determining compensation of suggested measures for increasing energy efficiency.
Some of the machine learning methods will produce more accurate prognoses from others, therefore it will be possible to select them as more successful for certain problems or clusters of buildings.
Simulation model of supply chain will reveal some of the possibilities for cost reductions.