Challenges – the many

Metro North Health (MNH) has a diverse set of buildings with large variability in age and systems contained within. Like many hospital buildings, they are in a constant state of flux and receive continual and ongoing upgrades. While telemetry and sensors have the ability to generate large datasets, most of them are inaccessible due to on-premise data silos coupled with inconsistent naming conventions, while electricity and gas invoices arrive as PDF or paper documents. Modelling data becomes a per-building activity, making scalable application development an impossibility. Without a standard metadata schema, application development cannot happen across the entire sector. This project will allow MNH to trial the CSIRO cloud-based platform the Smart Building Data Clearing House (DCH).

1. The current implementation of DCH was found to be complex requiring bespoke knowledge for integration. Upgrades are required to enable development houses a reasonable time to interface with a database system, as a user permissions issues consumed more time than anticipated from the project budget.
2. In addition to the complexity of getting data to DCH was the complexity of getting it out of the existing system, the main requirement of this project. There were two main issues with this. The first is some complex naming conventions used across the hospital building portfolio to create “meta information.” These were overcome but required several testing iterations to generate custom code to handle ‘the edge cases’ for generating the brick model to ingest into the DCH.
3. The second was Niagara and JAVA’s lack of good memory management. This resulted in multiple stoppages and manipulation of data queries to slowly get the required data instead of getting large chunks as would be the standard approach. Time for integration of other sites would be greatly improved from the knowledge gained from this project.
4. The successful integration of these hospital buildings into the DCH will enable MNH to achieve strategic goals optimising energy performance against REETSEF KPIs. It will also inform requirements for future enterprise BMS implementations, and enable greater collaboration with research and industry partners. In addition to achieving these strategic goals, MNH would be better able to achieve operational goals related to energy performance optimisations, including balancing energy usage against achieving indoor air quality (IEQ) targets, maximising asset effective-life spans through targeted upgrades of infrastructure that are no longer performing efficiently, and determining improved HVAC load requirements for new projects.
5. For the wider iHub project, the onboarding and integration methodology, lessons learnt and documentation functions to support future channel partners such as Bar-tech in onboarding buildings to DCH, and deployment of applications and services.

The KPIs have been demonstrated at the Royal Brisbane and Women’s Hospital using a limited data set. Slow changing asset data from buildings such as electrical meters (NMI) to building relationships was unverified. Information on HVAC systems from the building management control system was unavailable. Likewise, weather and telemetry data at the hospital site were unavailable. Sample data sets were used to substitute for accurate contextual information. Thus, proper analysis at the hospital level and confidence in the KPI output was limited. Greater data availability and validation of assets and their operation will improve the results. More buildings and accurate data will demonstrate the effectiveness and maturity of KPIs for validation, feedback and improvements.

Regarding energy intensity, more contextual information will improve the quality and effectiveness of indicators. For example, m² is not sufficient to raise awareness of poor performance at a site compared to a benchmark without considering factors such as; climate, building and/or equipment age, building orientation, degree of medical specialisation etc. It is only when these factors are taken into consideration that a benchmark can be made and buildings can be effectively normalised for comparison against the benchmark. Therefore, energy intensity using a single normalisation factor is a shallow dive or precursor for further investigation. However, in many instances, as was the case in this project, more information is not readily available.

Once effective normalisation occurs, poor performance can be understood, benchmarks established and more meaningful comparisons between sites can occur. Once poor-performing sites are identified a deeper dive into operational technology, and data from HVAC systems and/or the BMCS should be sought to determine causation.

Regarding KPI metrics, aggregations of time-series data, such as consumption divided by max consumption are simple measures to implement and useful to understand as they align with network charges and indicate the potential for improvement. Likewise for renewable energy generation, however, the building typology of hospitals has less potential for generation as they are generally multi-level buildings with less roof space.

In summary, KPIs are a measure of performance, what happens once owners and operators are aware of performance issues? It is likely they face the following challenges:

• Having access to timely, accurate data; from billing, time-of-use (time-series), and integration of BMCS, to slow-changing asset data.
• Digitalising and automating systems pro-grammatically, removing the need for manual tasks and eliminating the potential for error.
• Realising operational to information technology integration, real-time management and programmatic control from device-to-cloud and cloud-to-device signals, alerts and alarms, fault detection, and algorithms such as MPC to match demand with supply for a balanced network with improved security at a reduced cost.

Advanced modelling – applied

REETSEF reporting functional requirements are derived from the relevant iHub report by Queensland University of Technology (QUT)116, referred to as ‘the REETSEF report’. 10 KPIs from this report were produced using sample energy and asset-related data sets for the beneficiary, Queensland Health. The resultant reports were produced in Microsoft Power BI demonstrating how the KPIs can be useful in helping to understand, manage, and improve the energy productivity of healthcare facilities and the value of renewable energy utilisation in these facilities.

Buildings Evolved have developed a sophisticated modelling tool that provides users with an interface or web form enabling them to upload, transform and write data from the following sources to a database:

• electricity meter data agent (NEM12);
• electricity retailer invoices (ERM & Origin);
• Bureau of Meteorology (BoM);
• Australian Energy Market Operator (AEMO);
• solar irradiance data (Solcast);
• GHG gas emissions factors;
• battery management system; and
• solar PV inverters.

The technical overview of the data architecture from ingestion via the web form, to processing and transformation, to producing KPIs and reports for analysis.

The basic architecture involves the following processes:

• ingest data via ETL;
• lay down data into a database;
• allow scripts to run across datasets to calculate required outputs;
• allow an array of scenarios and assumptions to represent different what-if scenarios for modelling;
• automatically calculate costs on an interval-by-interval basis;
• produce report outputs; and
• easily allow database querying to be performed by any standardised BI reporting tools, such as Tibco Spotfire, Tableau, Power BI, Grafana et. al.

Various open-source coding languages and software frameworks were used to develop the solution including:

1. Python,
2. JavaScript,
3. PostgreSQL, and
4. React.JS, amongst others.

This architecture has user access control, logs and validation processes to ensure data completeness and accuracy. The dashboards presented below, designed from KPIs in the REETSEF report used data generated from the modelling tool.

Users can access the modelling tool via a web form built on JavaScript and React JS framework. The modelling tool provides users with ‘one-click’ data ingestion of multiple data file formats relating to building energy management to produce the following outcomes:

1. Validate data quality, format and type.
2. Transform and normalise data in an appropriate schema for report outputs, and for energy management applications, such as:

• Billing, and financial modelling of energy systems.
• Time-series analysis of energy, and
• Carbon reporting.

Being able to save time on energy analysis whilst ensuring information quality and completeness is crucial for organisations seeking to take ownership of building energy management initiatives, to save money, increase efficiency and reduce emissions.

You can access the KPI reports from the Github repo here. Note that you will need to have Microsoft Power BI desktop version (free download) to open the reports. More information on the modelling tool available here.