Work Package 1 Update
During the reporting period of this newsletter (M12-M18), Task 1.5 Machine Learning training techniques for flexibility determination” was completed. iBECOME partners worked together to develop calculation models for the flexibility metrics and demand response the source of each metric was defined.
In this task, a literature review on the effectiveness of calibrated models for informing flexibility services was conducted, and a methodology for the development of machine learning algorithms is then proposed, with the aim of optimising a building for load shifting and load shedding, in response to market energy prices.
Based on the information collected via the literature review, it was decided to implement a three-step load shifting proof of concept (PoC) consisting of (1) stochastically generated demand and appliance use, (2) appliance load shifting algorithm and (3) a plant-level optimisation. With that, the developed algorithm will be used to inform load shifting and load shedding activities. To better enable data collection on the iBECOME demo sites, methods for load disaggregation were also investigated in order to determine effective methods of obtaining timeseries data on flexible end uses to inform the flexibility algorithms.
Load disaggregation
Convolution Neural Networks (CONV1D) algorithm was chosen for load disaggregation for a number of reasons found in the literature, comparing to other ML methods. The results of the load disaggregation algorithm showed promising, with accuracies between 65-92% at predicting time of use of air handling units. It was found that accuracy typically increased with increasing contribution to total floor area. This accuracy could be further improved with additional data. In addition to this, high fidelity data is required to effectively train the load disaggregation algorithms, ideally with frequencies of less than 30 seconds.
Load Shifting
A genetic algorithm was used to determine the optimum load shifting of flexible appliances, in response to changing energy prices. The recommendations carried out by the team result in an 28% reduction in price for the given day. The same algorithm will be demonstrated on the iBECOME demo sites, following a survey and data collection to determine the demand profiles of the flexible end uses.
Load Shedding
The load shedding algorithm is applied to a heating system for the initial implementation. In this case, the indoor air temperature is constraint to a thermal comfort range, defined by the algorithms in D1.1. Following investigation. air temperature is maintained within the acceptable thermal comfort range most of the time, with the optimised solution reducing energy costs by a further 25%. It is worth noting that the level of improvement will largely depend on the appliance time of use and the cost of the electricity.
Work Package 2 Update
In Work Package 2 the virtual BMS (vBMS) architecture has been defined and designed in order to integrate existing buildings’ automation systems or other monitoring solutions and to exchange data to the various services through APIs.
Four main scenarios have been identified when dealing with existing buildings and the methodology to integrate them in the vBMS: buildings without an automation/monitoring system, buildings with a Schneider Electric PLC, buildings with 3rd parties PLC with open communication protocols and buildings with 3rd parties PLC without open communication protocols.
To ensure there is a high level of security when dealing with data exchange from the buildings to the vBMS, it has been engineered a VPN to the cloud hosting the data aggregation software.
A specific focus was made on how to establish bidirectional integration and interoperability with on-site BMS and IoT sensors and controllers needed to run the services part of the vBMS itself.
The vBMS architecture has been developed and three different layers has been defined:
- Filed level layer: where the data needed for the vBMS services are collected
- Data aggregation layer: where the data are stored and exchanged
- Services layer: where the data are exchanged, and how the services performed
The iBECOME vBMS will incorporate services (Core services: Continuous commissioning services, Optimised renovation, Demand Response service) that will aim to improve the energy use, condition and the smart readiness of a building. Will also offer the option to any software service provider to utilize energy data (including raw, processed, simulated or predicted data) and enable energy or non-energy services to its users, in this document are described the Healthcare Management services and the Car Sharing platform.
To engage the user the data will be presented using dashboards allowing them to interact with the vBMS services and functionalities.
Work Package 3 Update
For WP3, the specific work package objectives included:
- developing the control algorithmic routines for (i) energy optimisation, (ii) comfort optimisation, (iii) flexibility optimisation, (iv) predictive maintenance and FDD, and (v) M&V
- developing the communication and controllability functions to accommodate additional energy and non-energy services from 3rd parties
- incorporating the services functions into the iBECOME vBMS
During the first 18 months of the project, WP3 has been focussed on R&D algorithmic development and evaluation for: fault detection and diagnosis (FDD) and predictive maintenance (Task 3.1); measurement & verification (M&V, Task 3.2); predictive control (Task 3.3) and demand response (Task 3.4).
The various algorithms developed for each WP3 subtask have undergone substantial development and evaluation: the initial selection of which was based on the literature review undertaken for each subtask. Using the iBECOME modelling tools and test models from WP1 and WP2 to generate virtual datasets, the evaluated algorithms were then tested as proof of concepts, with final recommendations and selections being made for use in iBecome. These final selections will be, as part of the upcoming Task 3.6, incorporated into the virtual BMS as production code.
Aside from the algorithmic integration work outlined above, the next area of focus for WP3 will be on the additional energy and non-energy services, namely Healthcare Management and Commuting Optimisation.
All the necessary infrastructure required to enable all the above services is in the process of being developed and integrated, as part of Task 3.6. This is scheduled to be completed in mid 2022
Work Package 4 Update
In the last 6 month period, activities for demonstrating the operational efficiency of iBECOME vBMS were initiated in two project demo sites of Country Crest (Ireland) and Helix building (UK). Currently, tasks 4.1- ‘Customer engagement, audits and hardware installation’ and 4.2- ‘Development and calibration of VE models’ are progressing.
As a part of task 4.1, energy audits were initiated for both the demo sites to gather the data required in developing the simulation models such as building façade, HVAC and equipment details, lighting and utility bills. Once the required information is gathered completely, it will be used to finalise the two demo site simulation models in Task 4.2.
Due to Covid-19 restrictions on-site, site audit for Country Crest is currently being conducted virtually through online meetings, whereas the data collection is completed for Helix building. A baseline survey is also developed to engage with the customers of both demo sites, to determine the qualitative aspects and features of indoor built environment in them. The overall survey is focused mainly on three environmental attributes: Indoor air quality, thermal comfort and lighting. The survey has dedicated subsections with relevant questions for each of these attributes.
The output from the survey will be utilized to determine the baseline against how the iBECOME vBMS impact will be compared in the later stages of the project. The survey will be carried out in coming months on both the sites combined with training for the occupants explaining the benefits and operational features of iBECOME vBMS.
During the next 6 months, development and calibration of dynamic simulation model will be completed for the demo sites using the latest information available from site. Moreover, these calibrated simulation models will facilitate the deployment of iBECOME services as their main modelling tool in task 4.3. ICT hardware installation will also be deployed in the sites, after conducting gap analysis to finalise additional sensors and hardware required to deploy site specific iBECOME services developed in WP3.
Work Package 5 Update
Task 5.1 / Customer engagement and training
The engagement of the end-users includes both a survey and training sessions. Within this task meetings have been organized between CIVIESCO, CEA and EOLYA in order to work on the elaboration of a survey that will be used locally to collect occupants’ feedbacks on Indoor Environment Quality conditions. The survey will be used to gauge the qualitative aspects and features of the indoor environment and it represents a baseline for the calculation of the iBECOME vBMS impact. The survey will be carried out before and after the deployment of the iBECOME vBMS.
During the coming months the survey will be shared to the tenants/occupants to collect feedbacks for different outdoor conditions in order to better qualify IEQ conditions for different seasons. Training sessions will present the functions, operational features and interfaces of the iBECOME vBMS along with its usability characteristics. Partners in charge of each service will be asked to collaborate in order to prepare the training sessions. Training materials have been developed in collaboration with WP7.
Task 5.2 / Building audits and modelling
Within this task, local partners performed energy audits and gathered building’s information so IES could develop, using its technology, dynamic building simulation models for the two demo sites that are retrofitted: World Trade Center (WTC) in France and ASP Della Carnia (ASP) in Italy. The energy audits gather all quantitative information (energy bills, data from BMS and smart sensors and meters, building data and HVAC systems). Once the models are created, they are calibrated over their IEQ and their energy consumption according to the methodology determined in T2.3. The calibrated model will be used in T5.4 as the main modelling tool for the deployment of the iBECOME vBMS but also from the building owners and operators to explore what-if scenarios for future energy upgrades of their premises.
For both demo sites, some additional calibration work is needed to improve the representativeness and the quality of the models as more measured data and information will come in the future.
Task 5.3 / Retrofit implementation supervision
Within this task retrofit work has been overseen to ensure the coordination with iBECOME activities.
Also, physical visits of the two demo sites and technical online meetings have been organized with the objective to analyse the IT/network architecture downstream of the Building Management System in terms of communication protocols, existing devices (sensors/meters/actuators) and associated measurements (inputs) and setpoints (outputs).
Moving forward, additional hardware will be deployed in order to be able to deploy iBECOME vBMS and to host innovative services developed in WP3.
Work Package 6 Update
In WP6 two tasks have been completed in the reporting period (MX-MY). In T6.1 the market research and analysis for vBMS platforms have been performed and summarised in the deliverable D6.1. It concluded that in Europe the two key broad markets, the “standard” BMS and Smart Buildings markets are projected to grow significantly in the coming years due to the rise in energy prices and the evolution of the regulatory framework. For different areas worldwide the Smart Buildings market will grow between 2020 and 2025 according to Figure 1.
Figure 1: Smart buildings market value break-down per world region (2020-2025) reflecting smart cities 2020’s global spending distribution. Source: RINA elaboration of IDC data
The study also presents a section devoted to the already existing vBMS platform and the investigation of different market segments from commercial to industrial buildings, from ESCOs to demand response applications. The results can be considered a usable basis for further business modelling foreseen in iBECOME.
Task 6.2 has also been concluded by achieving a review of the regulatory framework described in D6.2 that includes the main European standards and regulations for tertiary buildings that are mainly addressed in iBECOME. Moreover, the regulatory context of ESCOs in Italy and France has been analysed and a review of the available building management system (BMS) communication protocols and standards has been presented. Finally, the last section has been devoted to data privacy and cybersecurity that are relevant topics for the digitalization of the building stock since data management is at the heart of iBECOME vBMS platform.
Task 6.3 is at its initial phase and in the next few months aims at creating a template of a collaborative contractual framework that allows establishing a working arrangement between the stakeholders involved in the installation and the provision of services through the iBECOME platform. 2022 started with a site visit on the 4th of January at the nursery house ASP della Carnia, the Italian demo case of iBECOME. The scope of this visit was to discuss with the building director how to mitigate the pressure that the pandemic is creating on the organization, workers and occupants in view of the survey and training sessions that will be organised in the coming months.