What CloudBMS is
A single, white-label platform that watches your energy assets, predictstheir behaviour, optimizes their cost, safely controls them, and reports it all — across mixed fleets and multiple sites, with strict per-tenant isolation.
See everything
Live power, energy, state-of-charge and status from every inverter, meter, genset and battery — one screen per portfolio.
Trust the numbers
A physics digital twin gives battery health (SOH) and remaining life (RUL) with honest confidence bands — not vendor marketing.
Save money, provably
The real bill (incl. demand charge & power-factor penalty) is reconstructed, then a cost-optimal plan shows exactly how each rupee is saved.
Control safely
Peak-shaving and load-shedding run shadow → operator-approved → enforce, always read-back verified, never a blind write.
Prove the investment
Model any source mix → solar+storage with payback, NPV, IRR, LCOE and CO₂ avoided — a board-ready business case.
Stay compliant
ISO 50001 / ESG / CBAM reports on schedule, from aggregated metered data only.
How it works, at a glance
A small, cheap edge client on your site's network reads your equipment and talks outbound-only to the multi-tenant cloud. Heavy analytics run in the cloud; your site needs little or no hardware.
On your site
An edge client reads Modbus/CAN/SunSpec from the inverter, meter and genset, buffers through outages, and (optionally) drives relays. No inbound ports.
In the cloud
Ingestion → TimescaleDB (isolated per tenant by row-level security) → digital twin, forecaster, MILP optimizer, savings engine and compliance — served to one branded web app.
Connect your system
Pick the tier that matches the site. Monitoring can need zero hardware; control needs a tiny gateway.
E0 · Cloud poll
The inverter's own cloud (GoodWe SEMS, Huawei) is polled for monitoring. Fastest to onboard where a vendor cloud exists.
E1 · ESP32 micro-gateway
A microcontroller reads Modbus locally and drives load-shed relays. Replaces the mini-PC on most sites; closed-loop survives internet outages.
E2 · Mini-PC / Pi gateway
For many protocols at once or a long local historian. Runs the same Docker edge images.
The four steps to live data
Register the site & assets
Add a site and tap the equipment installed (inverter, PV, battery, genset, meter). Each asset gets a nameplate — its ratings feed the twin and the optimizer.
Point the edge client at your equipment
Give it the inverter's LAN address (Modbus-TCP) or RS485 details, plus your cloud URL and site credentials. Connectivity is outbound-only — no port-forwarding.
Confirm the register map
Match each metric to the right Modbus register from your device's manual. Readings should now appear live on the Site console within a minute.
Set your tariff
Enter your Time-of-Use tariff (and demand charge / power-factor penalty if any). Everything financial — the bill, savings, payback — is priced from this.
Deep guides: connectors & vendor clouds, the ESP32 firmware build/commissioning, and any-inverter Modbus mapping are documented for your installer.
Calibration & accuracy — be certain
Before you trust a single derived number, verify the inputs. CloudBMS is built so every figure traces back to metered data — here's how to check each link in the chain.
| What to verify | How to check it | You're calibrated when… |
|---|---|---|
| Live readings | Compare the Site console's live kW / SOC / voltage against the inverter display or a clamp meter. | values match within your meter's tolerance |
| Register map | Read each register back and confirm it matches the device app before enabling any write. | no metric reads an implausible value or wrong sign |
| Energy totals | The twin integrates power to kWh, discarding gaps >15 min. Compare a day's kWh to the meter's daily total. | daily kWh agree within a few % |
| The bill | The reconstructed bill prices load against your ToU tariff. Compare a month to your utility invoice. | reconstructed cost ≈ invoiced cost |
| Forecast | Run forecast accuracy — it back-tests on held-out days and reports MAPE / MAE. | MAPE is stable and in a sane band for your site |
| Twin health | SOH/RUL come with confidence bands that tighten as data accumulates. | bands narrow over weeks; no sudden unexplained jumps |
| Control writes | Every write is dry-run first and read-back verified — the log shows would-write vs actual. | dry-run values are correct before you enable enforce |
Overview dashboard
Your portfolio in one glance: how many sites and assets, what needs attention, and quick jumps into the work.
- 1KPI tiles. Sites, assets, active advisories and open work orders. Each is a button — click to drill straight into that view.
- 2Live recommendations. The savings/health engine surfaces the highest-value actions across the fleet, newest first, colour-coded by severity.
- 3Quick actions. One-tap into Fleet, Optimize and Compliance — the three things operators open most.
How to read it
Advisories are ranked by value and urgency, not chronology — the top item is usually the biggest quick win. A rising work-order count with flat advisories means execution is lagging detection; the reverse means you're keeping up.
Site console & energy flow
The beating heart of a site: where power is coming from and going to, right now.
- 1Energy-flow diagram. Animated power paths between PV, battery, grid, generator and load. Line thickness ≈ magnitude; direction shows import vs export.
- 2State-of-charge ring. Battery SOC at a glance. The ring colour warns as you approach the reserve floor the policies protect.
- 3Live tiles. Instantaneous kW per source, self-sufficiency %, and grid import/export — the numbers you calibrate against your meter.
How to read it
When the sun is up, healthy sites show solar covering load with surplus charging the battery or exporting. If the grid arrow is importing at midday with PV available, something is capping solar — a good calibration check. At night, discharge should track the tariff: hold through cheap hours, cover the expensive peak.
Assets & the digital twin
Click any asset to see live behaviour beside its physics twin — health, remaining life, and performance vs expectation.
- 1State of health (SOH). Usable capacity vs nameplate, from an electrochemical + coulomb-counting model — with a confidence band, not a single fragile number.
- 2Remaining useful life (RUL). Days to end-of-life, extrapolated from the health trend. The band widens when data is thin and tightens as it accumulates.
- 3PV performance ratio. Actual vs weather-expected yield. A persistent gap flags soiling, shading or a failing string — before it costs a season.
How to read it
SOH tells you what the battery can still do today; RUL tells you how long you have to plan a replacement. Trust the band, not the point — a 94% ± 2% is solid; a 94% ± 12% means keep gathering data. A PV ratio drifting down week over week is the earliest, cheapest maintenance signal you'll get.
Analytics & the savings roadmap
The real bill, decomposed — and a plain-language plan of exactly how you cut it.
- 1Reconstructed bill. Your actual cost rebuilt from metered load × your ToU tariff, including demand charge and power-factor penalty. This is the baseline everything is measured against.
- 2Savings headline. Quick wins (operational, free) vs capital upside (investment), annualised — so you know what's a switch-flip and what's a project.
- 3Strategy table. Each lever with its evidence, expected annual saving, confidence and payback — sorted by value.
How to read it
Start at the top of the table — highest value, highest confidence. Quick wins are operational (schedule changes the controllers can automate); capital upside needs spend and is where the Feasibilitystudy takes over. The demand-charge and PF-penalty tiles are often the fastest money on a C&I bill.
Optimize — the cost-optimal plan
A mixed-integer optimizer computes the least-cost hourly mix of solar, storage, generators and grid — and ranks your gensets by true delivered cost.
- 1Recommended strategy. When to charge, when to discharge, when to export — in plain hours you can act on or let the controller run.
- 2Projected saving. Optimized cost vs unmanaged, with CO₂ avoided. The bar compares the two so the gap is obvious.
- 3Battery schedule. The 24-hour dispatch: green charges (cheap/solar hours), amber discharges (expensive peak).
How to read it · merit order
The optimizer also ranks generators by delivered cost per kWh. A genset sitting at 0% utilisation is telling you it's uneconomic to run against grid/solar — hold it as backup, don't dispatch it for energy. That single insight often reframes a diesel-heavy site's whole operating pattern.
Feasibility — the solar business case
Model any current supply (grid / diesel / gas, any combination) → any solar+storage addition, and get an investor-grade case: payback, NPV, IRR, LCOE, CO₂ and fuel avoided.
- 1Recommended scenario. The winning option with saving/yr, payback, IRR and NPV, plus a source → solar flow.
- 2Scenario cards. Add-Solar vs Add-Solar+Battery side by side — investment, saving, payback and CO₂ avoided, the best one ringed.
- 3Live sensitivity. Drag solar capex, cost of capital and energy price — payback, NPV, IRR and LCOE update instantly.
How to read it
Payback is the gut check; NPV is whether it creates value at your cost of capital; IRR is the return to compare against other investments; LCOE is your solar's cost per kWh to compare against the tariff it displaces. Off-grid diesel sites usually show the fastest payback — displaced diesel is the most expensive energy there is.
Policies & control
Turn insight into action — safely. Control climbs a deliberate ladder: shadow (logs only) → backtest (must pass) → operator-approved enforce (read-back verified).
- 1A policy. e.g. peak-shaving with a demand cap and a reserve floor, on a chosen battery. Mode shows shadow vs enforce; a badge shows if its backtest passed.
- 2Shadow log. Every ~30 s the runner logs what each closed-loop policy would do — the charge/discharge decision and its reason — writing nothing.
- 3The gates. Enforce is blocked until a backtest clears a minimum peak-reduction; live writes are dry-run, guardrailed and read-back verified.
Compliance & maintenance
Board-ready reports and predictive upkeep — the paperwork and the wrench-time, handled.
How to read it
Reports draw on aggregated metered data only — raw per-device telemetry never leaves the platform — and every figure is reproducible from the record, which is exactly what an auditor or financier wants. Maintenance is driven by the twin: tasks are prioritised by predicted failure, not a fixed calendar, so you spend wrench-time where it pays.
Your learning path
The same platform serves a first-day facility manager and a fleet-scale energy trader. Pick your level.
Log in and open your site
Watch the energy flow for a day. Learn what 'normal' looks like — when solar peaks, when the battery works.
Read the Overview advisories
Act on the top green quick-win. It's free and reversible; you'll see the effect on the bill.
Download a compliance report
See your month decomposed into energy, demand and carbon. This is the language you'll use with stakeholders.
Check one calibration
Compare a day's kWh to your meter. Confidence starts with one verified number.
Estimate your savings
A back-of-envelope feel for the opportunity. Drag the sliders — this is illustrative; the in-app Feasibility study computes the real case from your metered load.
Frequently asked
Do I need to buy hardware?
Not necessarily. If your inverter has a vendor cloud (e.g. GoodWe SEMS), monitoring needs zero on-site hardware. Control needs a tiny gateway — from a ~$15–40 ESP32 upward.
Will you control my equipment without asking?
No. Control is off by default. Writes are dry-run first, refuse any unconfirmed register, are read-back verified, and enforce requires an operator to approve after a backtest passes.
How do I know the savings are real?
Every figure traces to metered data. The bill is reconstructed from your load and tariff and can be checked against your invoice; the baseline for savings is your plant optimally dispatched today, not a strawman.
Does my raw data leave the site?
Only aggregates are exported for analytics and reports; raw per-device telemetry stays on the platform, and tenants are isolated at the database by row-level security.
Can I white-label it for my clients?
Yes — branding, colours and subdomain are per-tenant, so you can resell CloudBMS as your own product from day one.
See it on your own site
Bring one site's data in and we'll walk you through the numbers — live monitoring, the savings roadmap, and a solar business case built from your real load.
This guide is public — no login needed. Share the link with anyone.