(+55) 254. 254. 254
Helios Tower 75 Tam Trinh Hoang Mai - Ha Noi - Viet Nam
How SapienX helped a National Bank to transform operational efficiency
1 solution for 100+ sites
With this portfolio, each site is unique.
Different sizes. Some site operate with 50+ staffs and some only have 4/5 staffs. The needs are different and the energy and water usage is unique to each site.
Different locations and contexts. Bank branches what to ease the access to a physical branches. Branches can be found in shopping centers, in office building, main streets, etc. These set up impact differently the energy and water usage.
Inconsistent operating hours
Operating hours heavily impact the energy and water usage on site.
As any B2C or retail service, knowing the business hours is crucial to perform an accurate usage analysis. In big cities some sites are open have extensive opening hours (10 hours/day) and limited opening hours (6 hours / day) is smaller towns.
The branch Director can change staff schedules depending on the business context and without notice.
The fact that opening hours are constantly changing and not uniform across the board.
Wide range of equipment loads
The bank run branches with staff and clients constantly coming and going. Therefore they use naturally HVAC systems to control and offer a comfortable temperature for everyone. The staff also uses large range of computers that can run 24/7 or only within the operating hours.
The bank also manages private Data Centers and Computers Centers to store and process transactions. These data centers are made of servers, routers that are well-known to be energy-intensive. Most of the time, no downtime is off the cards therefore a allowed so
All these equipment loads and different are critical for the company but there are also a a clear challenge in terms of energy and water monitoring.
1. AI-powered anomaly detection
It is where the magic happens.
AI models use operating hours, temperature, weather, HDD, CDD, water usage and power usage.
The AI-powered models are able to detect and diagnose anomalies with a high level of precision. The models can detect newly commissioned equipment loads or pick up any major change on-site with better precision and accuracy without any manual intervention. The real-time processing also enables the client to pick up any anomalies with greater speed.
The high level of accuracy allows false-positive reduction which leads to reducing the time spent on triaging and troubleshooting falsely detected incidents (OPEX reduction). The team can focus on high-value issues.
2. IoT & data collection
Several meters and sensors are commissioned on each site. Meters and sensors collect the actual energy usage (kWh) for various equipment loads. These meters also capture contextual data such as the actual temperature (°C) which is crucial for an HVAC system.
Data is collected with a 15-mins interval. The shorter interval the better as it allows the system to zoom-in for a thorough and detailed analysis.
Other contextual data is collected such as weather data to collect external temperature, humidity, heating degree days (HDD) and cooling degree days (CDD).
3. Cloud-ready Big Data infrastructure
IoT, meters and sensors enable high-quality and real-time data collection. However, it is only feasible with a solid real-time-ready cloud infrastructure. In our case, AWS (Amazon web Services) is the cloud provider used.
The real-time infrastructure includes the data collection, real-time processing of unstructured data, secured data storage in Data Warehouses.
This infrastructure enables unbounded stream of data, which make this infrastructure scalable by-design.
The end goal is to enables a faster response time to incident.
4. AI-powered pattern recognition
Some sites operate 24/7, and some operate with regular business hours (9 am – 5 pm), some are closed on Mondays and others on Sundays. With 100+ sites to monitor, it is not the best use of one’s time to manually set and update manually operating hours. Missing, erroneous or outdated information can lead to inaccurate anomaly detection.
Using usage data, the AI model can derive business hours and shift changes without any manual inputs. This technology enables to seamlessly detect the shift change and adjust the analysis accordingly across the board.