ASHRAE Journal - August 2023 - 52

FIRST PLACE 2023 ASHRAE TECHNOLOGY AWARDS
areas while maintaining thermal comfort (for details
refer to the Innovation section).
Cloud-Based Smart Energy Management Platform (CBSEMP)
RCx is labor-intensive work and is thus only conducted
periodically. Therefore, potential energy savings
between studies would not be identified and achieved
in a timely manner. To deliver higher energy efficiency
consistently, a digital solution was sought. A centralized
cloud platform was built, namely the cloud-based smart
energy management platform (CBSEMP), to collect, analyze
and visualize real-time energy and operation data.
It enables real-time data collection, big data analytics
based on artificial intelligence rule-based expert system
and machine learning and the ability to integrate with a
digital twin solution.
Five individual systems from Cityplaza are connected to
CBSEMP, namely the HVAC air-side system, the CP-North
chiller plant system, the CP-South chiller plant system,
the seawater pump house system and the power meter
system. In total, 1,646 unique points are connected to
the CBSEMP, with data collected at 15 minute intervals,
equivalent to 158,000 data points per day. The team
developed a standardized naming guideline for the cloud
platform to incorporate the siloed BMS datapoints.
Through the analytics, the platform is able to identify
actionable insights to improve operational efficiency,
reduce energy consumption and enable continuous
monitoring of the building's performance to help
achieve corporate objectives of 1.5°C (2.7°F) Science
Based Targets and the net zero vision of 2050.
Innovation
Digital Twin Study
We also explored the use of a digital twin in combination
with our CBSEMP to identify further energy savings.
A digital twin provides a virtual 3D model reflecting
actual physical space to perform prediction, optimization
and evaluation of potential changes to a building's
operation and maximize its performance
We combined on-field data (CBSEMP real-time data
and on-site measurement) and modeling (3D CFD simulation
for temperature and airflow) to analyze our building
energy performance, thermal comfort and IAQ.
AHU Optimization
Connecting the north and south wing of the mall is
52
ASHRAE JOURNAL ashrae.o rg
A U G UST 2023
CASE STUDY
a bridge with an area of 1,553 m2 (16 716 ft2), a floor to
ceiling height of over 6 m (20 ft) and a skylight area
of 563 m2 (6060 ft2). The area acts as an exhibition
area across different seasons. We received complaints
of thermal discomfort from different exhibitors
throughout the year. To address the thermal discomfort
issue, the 10 AHUs serving the bridge were fully operated
to cool the environment.
However, the CBSEMP identified that AHU return
air temperatures were relatively low. On-site
measurements found that the bridge was overcooled,
with space temperatures around 20ºC (68°F). However,
the open space of the bridge meant there were no
physical locations to install temperature sensors to
measure the occupancy zone's temperature to accurately
control the 10 AHUs.
A CFD study was conducted to better understand the
airflow distribution of the environment under different
conditions and AHU operation combinations. It was
concluded some AHUs had a larger impact on the thermal
comfort than others; thus those AHUs were classified
as essential and will maintain operation to serve the
area over different seasons. The study also concluded
that six AHUs were sufficient to maintain a comfortable
indoor environment and ventilation at the bridge.
Through CFD and on-site measurement, it was found
that the space temperature is now around 24ºC (75°F).
Thus, a relationship can be built between the return
air temperature of the AHU and the space temperature
for better control. Through reducing AHU operation,
energy savings was achieved while maintaining a good
indoor environment.
Cooling Load Prediction
Another major innovation is the development of a
cooling load prediction module, which would help
identify the optimal chiller sequencing for the day.
The prediction uses machine learning (LightGBM
model) to predict Cityplaza's future 24 hours' cooling
load. The algorithm is built using the previous year's
data as training data. During operation, the model
takes in the past 7 days' input data to predict the future
24 hours' cooling load at an hourly timestep. Input
parameters include building cooling load, environmental
data and major systems' electricity consumption.
The model is retrained bi-weekly with newly acquired
data to have an updated reflection of recent operation
https://ashrae.org/
ASHRAE Journal - August 2023

Table of Contents for the Digital Edition of ASHRAE Journal - August 2023
