01
Begin with the use case and reporting boundary
Energy data becomes useful when a named team can act on it. Define whether the project supports abnormal-usage alerts, tenant or branch comparison, retrofit verification, operational control, management reporting or a combination.
Agree the reporting boundary: whole site, distribution board, floor, branch, equipment group or critical load. Record what is intentionally excluded so that later dashboards are not mistaken for a complete building account.
Decision owner
Name who reviews the data, how often, and what action follows an exception.
Baseline and context
Record operating hours, occupancy, floor area, weather or production context where relevant to comparison.
Portfolio consistency
Use a common naming, timestamp, unit and hierarchy model across branches or floors.
02
Turn the electrical scope into a data-point map
The point map connects physical distribution to digital reporting. It should identify the board or circuit, meter, current-transformer arrangement where applicable, measured parameters, polling interval, gateway, protocol, destination and dashboard name.
Electrical verification
Confirm safe installation access, isolation strategy, space, rating, phase arrangement and the responsible competent parties before work.
Data architecture
Define protocol, addressing, time synchronisation, buffering, communications resilience and where raw data is retained.
Security and ownership
Agree network zones, outbound connections, credentials, vendor access, data controller, retention and export rights.
Dashboard semantics
Make every chart traceable to a named physical point, unit, aggregation rule and quality status.
03
Commission the data chain end to end
A green status light is not commissioning. Test from the physical load through sensing, metering, gateway, network, platform and dashboard. Challenge the system with known changes and compare values against an agreed reference method.
Installation evidence
Capture labels, orientation, ratios, wiring checks, serial numbers and mapped locations.
Data-quality tests
Check units, sign, phase, timestamp, gaps, duplicates, scaling and aggregation over an agreed observation period.
Operational tests
Prove alerts, user permissions, export, offline buffering, reconnection and issue escalation.
Acceptance record
List passed points, open defects, accepted limitations, owners and completion dates.
04
Keep statutory roles and performance claims explicit
Hong Kong’s Buildings Energy Efficiency Ordinance has defined compliance and energy-audit requirements. EMSD states that mandatory audits are conducted under the Energy Audit Code, with Registered Energy Assessors serving a regulated role. Requirements are changing in stages, including provisions taking effect on 20 September 2026.
UPNX energy-monitoring delivery should therefore be described as an engineering and data-visibility scope unless the proposal separately names the applicable statutory requirement, qualified party and deliverable. Savings should be measured against an agreed method; they should not be guaranteed by installing meters.
Sources
Primary references
These primary sources were used to check public requirements, the purpose of standards or recognised practice. Project-specific applicability must be confirmed.
FAQ
Frequently asked questions
How many meters does a workplace need?
There is no useful universal number. Start from the decisions and load hierarchy, then select enough measurement points to isolate material, controllable or business-relevant loads without creating an unmaintainable system.
Is an energy-monitoring installation the same as a statutory energy audit?
No. Monitoring provides measured data. A statutory audit under Hong Kong’s BEEO follows the applicable Energy Audit Code and regulated assessor requirements. Confirm current applicability with EMSD and a qualified advisor.
Can a dashboard prove energy savings?
A dashboard can provide evidence, but savings require a defined baseline, adjustment method, comparison period, data-quality controls and an agreed verification approach. Installation alone does not prove causation.