Technology & compliance
Safety is not a feature of the product. It is the process.
Batteries and power systems are inherently dangerous, and second-life packs arrive with a history somebody else wrote. Everything below exists to make that history known, testable and documented before a system reaches a customer.
From salvage yard to switchboard
Acquisition
Packs are sourced from Australian salvage auctions and wreckers, overwhelmingly from statutory write-off vehicles. Make, model, VIN, pack serial, supplier and extraction date are recorded at intake.
Dismantling & recovery
High-voltage work by qualified personnel with appropriate training and PPE. Open-circuit voltage, state of charge, BMS-reported State of Health, lifetime throughput and physical condition are logged — the pack's second-life birth certificate.
Test & assembly
Capacity, internal resistance and temperature stability are measured per module. Modules are assembled with an Upcycle battery management system and active balancer, then cycle-tested to validate capacity and thermal behaviour before enclosure.
Install & commission
Installed by a licensed electrical contractor with DC contactors, isolation, current-limiting protection and dangerous-goods labelling. Commissioned, instrumented and remotely observable from day one.
Battery Provenance Log
Every pack has a file.
Each battery carries a QR-coded unique identifier and a record that follows it for life: where it came from, what condition it arrived in, every transport and storage movement, every safety inspection, every configuration change, and its full operational history once deployed.
For a corporate customer this is the difference between an interesting idea and an auditable asset. It is also how we improve — a fleet with known provenance is a dataset for predicting how second-life cells actually age in Australian conditions.
What the record holds
- Vehicle make / model / year / VIN
- Manufacturer pack serial, extraction date
- Supplier and acquisition channel
- OCV, SoC and BMS-reported SoH at extraction
- Lifetime charge / discharge throughput
- Photographic condition log, fault notes
- Transport, packing and DG documentation
- Storage location and duration
- Safety inspection and configuration history
- Deployed site, BMS configuration, operating logs
Cooperative control
Adding a battery without a fight.
Two batteries behind one meter is a control problem before it is an electrical one. Each system's charging looks like load to the other, so both chase a target that keeps moving — energy circulates, cycles are spent for nothing, and neither system reaches the state of charge it was aiming for.
Our controller resolves this by treating the incumbent battery as a known participant rather than as noise. It observes the existing system's behaviour, schedules Upcycle charge and discharge around it, and holds within limits that keep both systems inside their own safe operating envelopes. The incumbent is not modified, reconfigured or degraded — and the customer keeps the asset they already paid for while adding capacity around it.
- Trained on
- 14 live installations
- Longest record
- Approaching 2 years, continuous
- Method
- Non-intrusive load monitoring
- Tuning
- Per deployment, per household
The site model
Fourteen homes have been teaching this algorithm for two years.
The hardware is what people notice. The reason our systems earn their keep is the software, and the reason the software works is that we have been collecting operating data from every installation since the first one went in — half of the fleet now approaching two continuous years.
That fleet is not a marketing asset. It is an instrument. Every system reports at high resolution, and every one of them is a different house: a different roof, a different tariff, a different family, a different set of machines switching on and off. No two energy profiles are alike, and a generic control algorithm treats them as if they were.
It learns the house, not a house
Non-intrusive load monitoring reads the whole-of-home power signal and separates it into the machines behind it — hot water, pool pump, air conditioning, cooking, EV charging — without a sensor on any of them. Over months, the model learns not just what is running but when, and how that changes with weather, season and the rhythm of the household.
Then it is tuned for that deployment
Each system ships with a control algorithm parameterised for its own site: its measured load shape, its solar yield, its tariff, its battery capacity, and the behaviour of any other battery already there. Commissioning starts the tuning; it keeps refining as the record lengthens.
And every install improves the rest
Each new site widens the training set, so the model arrives at the next house already knowing more than it did at the last. This is the part a competitor cannot buy: two years of high-resolution second-life storage data across a real, diverse fleet, and it compounds.
The battery fills through the midday price trough and carries the house through the evening peak — sized against the evening the model expects, not an average one.
Illustrative day, representative of measured residential profiles. Actual behaviour is site-specific.
Arbitrage that actually captures the spread
Knowing when a household will draw power is what makes price arbitrage work. The algorithm schedules charge and discharge against the tariff the customer is on — a fixed time-of-use structure, or exposure to wholesale spot pricing where their retailer offers it — while holding back enough capacity to cover the evening the model expects, not an average one. Buying at the midday floor and supplying the house through the evening peak is the whole game, and it is won or lost on how well you predicted the evening.
It works with the house, not around it
The controller is built on open, widely supported home-automation protocols, so it integrates with what a customer already runs rather than demanding its own walled garden. Loads that can be moved — hot water, pool pump, EV charging, air conditioning — can be handed to the algorithm to schedule automatically, or left entirely under manual control. Customers who like tinkering can tinker; customers who want it to just work can ignore it.
A mobile app that tells you something
Most energy apps show a number that goes up. Ours shows where the energy went, appliance by appliance, because the model already knows: what the pool pump costs a month, what the last cold snap did to the heating, how much of the evening ran off the battery rather than the grid. Alongside that sits live state of charge, backup runtime available, and direct control of connected devices.
The same technique, at industrial scale
The load-decomposition method that identifies a pool pump in a suburban meter is the same one that separates rectifier baseload from traffic-driven radio power at a telecommunications site, and flags plant beginning to fail. The residential fleet is where the technique was proven; the industrial monitoring service is where it is sold.
- Grid compliance
- Via CEC-registered PCE
- Battery system
- Designed and constructed to the standards listed
- Installation
- Licensed electrical contractor
Standards & approvals
The ordinary approval path, deliberately.
We do not ask a network operator or a corporate compliance team to evaluate anything unfamiliar on the grid side. Power conversion equipment is commercial off-the-shelf and already registered with the Clean Energy Council; the connection application, the inverter standards and the installation standards are the ones every other battery installer uses. What is new is inside the battery enclosure, and that is designed and built against the list below.
| Standard | Scope | Applies to |
|---|---|---|
| IEC 62619:2022 | Safety requirements for secondary lithium cells and batteries in industrial applications | Battery system |
| AS/NZS 5139 | Electrical installations — safety of battery systems for use with power conversion equipment | Battery system, installation |
| AS/NZS 3000:2018 | Electrical installations (the Wiring Rules) | Site installation |
| AS/NZS 4777.1:2024 | Grid connection of energy systems via inverters — installation requirements | Installation |
| AS/NZS 4777.2:2020 | Grid connection of energy systems via inverters — inverter requirements | Power conversion equipment |
| IEC 62053-21 | Electricity metering equipment — static meters for active energy, Class A | Monitoring & metering |
| SA TS 5398:2025 | Electrical energy storage equipment — safety requirements | Battery system |
| AS/NZS 62368.1 | Audio/video, information and communication technology equipment — safety | Telecom-site equipment |
| UN 38.3 | Transport testing of lithium batteries | Cells and modules |
| ADG Code Ed. 7.9 | Australian Dangerous Goods Code — UN 3480 and UN 3536 consignments | Transport |
| IEC 60947-2 | Low-voltage switchgear — circuit breakers | DC protection |
Standards list describes the design and construction basis of Upcycle Energy systems. Product certification status is confirmed per project.
- Platform
- Ruggedised x86, DC input
- Runtime
- Containerised micro-services on Linux
- Comms
- Dual-SIM mobile, secure remote access
Autonomous controller
The system runs itself, and tells you how.
Every deployment includes a controller built on open-source, containerised components running on ruggedised industrial hardware. It does the work that makes an unattended battery plant trustworthy rather than merely functional.
Energy management
Per-tenant power balancing, battery management integration over CAN bus, inverter control over Modbus, automatic transfer switch operation, and coordination with any battery already on site.
Environmental management
Temperature, humidity and air quality held in range with combined passive and active cooling, tuned by a learning algorithm to hold 15–35 °C at the lowest energy cost.
Data pipeline
Streaming collection, local storage and external backup, so an audit or settlement question can be answered from the record rather than from an estimate.
Fault & alarm management
Threshold and signature-based alarming with SMS, email and in-app notification, plus a maintained failure-mode watchlist per site.
Security & access
Secure remote access over an aggregated dual-SIM tunnel, with access control and release management for the active equipment on site.
Observability
Dashboards for load profile, peak-shaving potential, backup runtime available, prioritised opportunities and anomaly detection — per site and across a portfolio.