CUSTOMER CASE STUDY
Performance Optimisation of Chillers
& HVAC Equipment
Project Highlights
- Increased productivity.
- Made Building Energy Management software available to personnel who is not energy managers, such as maintainers and building managers.
- Reduced operation costs by evaluating the drop in performance and enabling predictive maintenance.
- Enabled smart investment in new assets.
10%
Reliability Increase
10x
Productivity Raise
15x
False Alert Reduction
CASE OVERVIEW
A world leader, international energy company (top 3 worldwide in Energy management, water management, waste management, transport), approached Sensewaves to help them in the energy management of chiller machinery. These chillers were used for the ventilation of large size commercial centres in Dubai.
The company had already installed their own software system for the energy management of the chillers and made it available to the energy managers on site. The goal was to monitor the performance of the chiller, detect potential failures and ensure it operates at maximum efficiency, while ensuring comfort for the people visiting the building.
CHALLENGE:
Raise the productivity of energy managers
The insights provided to the energy managers were limited. Whilst operational parameter information was displayed, no further analysis was done within the software. In order to interpret the operation parameter information, an energy manager needed to spend a fair amount of time. This was particularly difficult, given the amount of sites she/he was supposed to monitor daily. Furthermore, the software was accessible only to skilled energy managers, preventing building managers and maintenance personnel from getting substantial value from it. Finally, built-in alarm system was threshold based and did not take into account the weather forecast and other contextual parameters, thus generating a lot of false positives.
The reason for deploying Adaptix was to use AI to produce smart alerts and advanced insights to help energy managers raise their productivity by enabling multi-site parallel supervision. Furthermore, an other objective was to produce insights that make sense to operating personnel other than energy managers, such as the building managers and maintainers.
The ultimate goal of the pilot was to reduce the maintenance cost due to the high amount of specialised personnel involved with the old version of the system (energy managers and maintainers), make maintenance operation more efficient and evaluate the impact of the wear of the chillers and the decline in performance with respect to the different models in order to improve the process of investment in new equipment and assets.
SOLUTION
Sensewaves have used Adaptix.AI together with their dedicated smart building software package applied on machine data (operational temperature), environmental variables (internal and external temperature and humidity, degree days) as well as the energy consumption for each unit of the chiller (primary pump, ventilation, cooling tower, chill water).
Among the tools that are deployed:
- Smart alerts: based on self-learning, the system automatically identifies over/under consumptions as well as failure of equipment
- Smart tagging: an online learning tool allowing the user to tag special events and create his own alerts for similar events in the future
- Prediction tool: hourly consumption forecast for the next day, 1-4 hourly forecast for the next week.
TEST BEFORE DEPLOYMENT
The results were evaluated by energy managers. The evaluation protocol consisted in having site energy managers that were familiar with the data to manually search for anomalies such as over/under consumptions & failure of equipment and then compare them with the results from Adaptix.
PERFORMANCE INDICATORS
SCIENTIFIC KPIs
- RECALL:
99% - PRECISION:
55% for unsupervised mode and increased up to 89% in the semi-supervised mode
COMPARISON OF ADAPTIX SMART ALERTS WITH STANDARD ALERTS OF BMS:
- RELIABILITY:
Recall score of engineer was 90%. Adaptix reduced false negatives by 10% - FALSE ALERT REDUCTION:
Adaptix scored 30% higher in recall and around 1500% higher regarding precision, vastly reducing false alerts. - SPEED OF DETECTION:
Adaptix was able to spot anomalies in 2 years of data for 1 chiller installation in 1 second time. In order to analyse the same amount of data using the original BMS, an engineer would need 1-3 working days.
TECHNOLOGY STACK, INTEGRATION AND DEPLOYMENT
Next step is to use Sensewaves’ multi-tenant, cloud-based deployment and make available Adaptix functionalities through its API. A pilot will run in 6 months and will lead to full scale deployment for 100 chiller units. The purpose is to train site managers and focus on integrating the software into the actual workflow of energy managers and maintenance engineers.