--- title: 'Automation ROI' description: Executive summary of metering-fleet automation impact — coverage, data quality, savings across ten directions, ROI, TCO and payback. section: AI Analytics weight: 12 related: - ai-analytics/fleet-reports/decarbonization - ai-analytics/fleet-reports/top-problem-nodes --- import Alert from '@/components/docs/Alert.astro'; A summary report for the head of a utility (resource-supplying) company or an investor. It answers five questions in a single document: **what** in the metering fleet is automated, **how well** it works, **how much money** it brings, **what gets in the way**, and **what to do next**. Written for people without a technical background — the headline numbers are visible in 30 seconds, the details and caveats follow below. This is a management assessment for decision-making (whether to expand the automation program, and by how much), not an accounting statement. Every figure is honestly tagged with a confidence level — see below. ## What it shows The report consists of seven blocks, top to bottom: | Block | What it contains and why | |---|---| | 30-second summary | five headline numbers: fleet coverage, connection reliability, annual savings, return on investment (ROI), payback period | | Infrastructure | which types of correctors are deployed in the fleet and in what condition — shows how homogeneous the fleet is | | Telemetry quality | how reliably nodes come online and how clean their data is — the foundation of trust in all the figures | | Economic effect | the main block: savings across 10 directions, grouped by confidence level (see below) | | Return and payback | ROI, payback period, total cost of ownership (TCO) over 5 years | | Risk map | nodes and groups that need attention first | | Problems and recommendations | what is critical, what is under observation, what can be improved | ## Three confidence levels for savings This is the report's key principle — it guards against inflated promises: - **Confirmed** (base) — a measurable effect on the fleet's real data. Only this is used to build the payback and ROI calculation. - **Scenario** — a model estimate based on industry norms, requiring verification against a specific company's accounting policy. Shown separately, not included in payback. - **Possible** (optional) — potential revenue if a condition is met, for example selling CO₂ credits. Counted only with a confirmed contract; disabled by default. ## Ten directions of savings | Direction | Confidence | How it is calculated | |---|---|---| | Reading-collection visits | confirmed | remote reading removes regular crew visits: number of active nodes × visit cost | | Prevented emergency visits | confirmed | early signals prevent some incidents: ≈0.4 incidents/node/year × cost of an urgent visit (×1.5) | | Gas losses | confirmed | automation reduces losses by ≈2% of the fleet's real annual volume × tariff | | Office labor | confirmed | less manual entry and reconciliation: hours saved × engineer's rate | | Battery optimization | confirmed | scheduled replacement upon actual discharge instead of "just in case": savings per node × active nodes | | Reaction time | scenario | problems are found and fixed faster: hours saved × rate | | Billing accuracy | scenario | less undercounting in bills (the same physical gas as in "Gas losses" — not double-counted) | | Metrology | scenario | fewer unplanned verifications/calibrations thanks to data-quality control | | Compliance | scenario | readiness for regulator inspections: estimate per node | | CO₂ credits | possible | monetizing methane reduction — only with a carbon-market contract | Directions with a single physical source (gas losses, billing, CO₂ credits all sit on the same ≈2% of gas) are tagged as alternatives and enter the total only once — this eliminates triple-counting of the same money. ## How it is calculated **First, "empty" nodes are filtered out** — otherwise the total is overstated. A node is excluded if it has been offline for a long time, consumes almost nothing over the year (a dummy device), or does not provide a monthly consumption archive at all. Everything remaining is the live fleet, and the economics are calculated on it; the report shows how many nodes were filtered out and why. **Gas losses and billing are not calculated as "average consumption × number of nodes"** — that is misleading, one plant equals a hundred apartments. The actual sum of real annual consumption from each node's monthly archive is taken. If there is no data, the item is honestly tagged "no data" rather than filled in with an estimate. **Payback and ROI are calculated only from confirmed savings.** The logic is simple: annual net benefit = confirmed savings minus annual maintenance; payback = investment ÷ net benefit × 12 months; ROI = net benefit ÷ investment. Scenario and possible effects do not participate in these numbers. Connection quality is measured on a sample of nodes (200 by default) — for a fleet of a thousand nodes this is sufficient for a reliable estimate. Optionally, a short management summary written by AI can be added (it does not affect the calculations). ## Run parameters | Parameter | Default | What it means | |---|---|---| | Period from / to | last 30 days | analysis window; for annual return, take a year | | Nodes for connection assessment | 200 | how many nodes to poll for connection quality (0 = all) | | Parallel fetch workers | 4 | lower to 2-4 on a download error (HTTP 429) | | Cost per field crew visit | 5000 | price of one crew visit | | Engineer hourly rate | 500 | for estimating labor savings | | Gas tariff | 6.95 | gas price for losses and billing (per m³) | | Cost per automation device (CAPEX) | 30000 | cost of one automation device (investment) | | Subscription / OPEX per device | 6000 | annual maintenance of one device | | Exclude long-silent, days | 90 | offline longer than N days → node not counted | | Exclude micro-consumption, m³/year | 50 | consumption below N over the year → dummy device | | Exclude without monthly archive | yes | nodes without a monthly archive are not counted | | Display currency | RUB | RUB / USD / EUR / UZS / KZT / CNY | | Exchange rates | default | can be overridden manually (JSON format) | | CO₂ contract | no | enables the CO₂ credits block in the economics | | AI executive summary | no | a short management summary written by AI | | Utility company | from context | filter the fleet by resource-supplying company | ## How to read the result Start with the 30-second summary: coverage, reliability, annual savings, return, payback. Then — the economic effect by direction. What counts as good: ROI from 50% and a payback period up to 24 months is a strong argument for expanding the program. A negative ROI is a reason to reconsider maintenance costs or the fleet's composition. What to pay attention to: - **Coverage decides everything.** If less than 30% of the fleet is covered by data, the report is flagged as premature: first grow coverage, then count the money. - **Look at which savings the conclusion rests on.** Payback is calculated only from the confirmed part; scenario and possible are given for reference and require separate validation. - **CO₂ credits are disabled by default** and do not affect payback until a carbon-market contract is confirmed. Typical questions to be ready for when defending before an investor: "why exactly this much savings" (the answer is in each row's formula), "isn't the same gas counted twice" (no — alternative rows are tagged and not summed), "what happens as the fleet grows" (the 5-year TCO shows recurring costs). The report assesses the effect for prioritizing management decisions; it does not replace a financial audit. Confirmed savings rely on measured fleet data; scenario and possible rely on models and conditions that the company must verify for itself.