Solar installation SMS alert system for under $15

TL;DR: Twilio SMS API + Python + Node-RED. Cost: $0.04/SMS. If the installation produces < 20% of the seasonal norm for 2 hours — SMS. Twilio Free Trial = $15 credit to start — enough for over a year of alerts at sensible usage levels.

A photovoltaic installation normally runs unattended. You find out about problems when the electricity bills return to pre-solar levels, or when you happen to check the app. An SMS alert system notifies you when something’s wrong — a cable came loose, the inverter is showing an error, a new tree’s shadow is blocking panels.

Twilio setup

Registering an account at twilio.com is free. The Free Trial gives you $15 credit — at $0.04/SMS that’s 375 messages, enough for over a year of normal use.

Buy a Twilio number (US: $1/month, or use the Free Trial number with limitations). Download from the dashboard: Account SID and Auth Token.

pip install twilio

Sending an SMS in Python:

from twilio.rest import Client

account_sid = "ACxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
auth_token  = "your_auth_token"
client = Client(account_sid, auth_token)

def send_alert(message: str, to_number: str = "+48600123456"):
    client.messages.create(
        body=message,
        from_="+12025551234",  # your Twilio number
        to=to_number,
    )
    print(f"SMS sent: {message}")

Five lines of code. There’s no simpler SMS API.

Data from the installation

We assume solar installation data is already in InfluxDB (MQTT from the inverter or MPPT controller → Telegraf → InfluxDB v2). Flux query fetching the average production over the last 2 hours:

from(bucket: "solar")
  |> range(start: -2h)
  |> filter(fn: (r) => r._measurement == "mppt" and r._field == "panel_power_w")
  |> mean()
  |> yield(name: "avg_production_2h")

The result is average power in watts. For a 3 kWp installation in July at noon you’d expect ~2000–2500W. If it returns 300W at 12:00 on a sunny day — something is wrong.

Alert logic in Node-RED

The Node-RED flow consists of five nodes:

1. Inject (every hour): Trigger that runs the flow every 60 minutes throughout the day.

2. InfluxDB In: Executes the Flux query for the last 2 hours, returns average power as msg.payload.

3. Function node — alert logic:

const avgPower = msg.payload[0]?._value ?? 0;
const month = new Date().getMonth() + 1; // 1-12

// Seasonal norms for a 3 kWp installation [W]
const norms = {
    1: 150, 2: 250, 3: 450, 4: 700, 5: 900, 6: 1000,
    7: 950, 8: 850, 9: 600, 10: 350, 11: 180, 12: 120
};

const threshold = norms[month] * 0.20; // 20% of norm = alert threshold
const hour = new Date().getHours();

// Alert only during daylight hours (9:00–17:00)
if (hour < 9 || hour > 17) {
    return null; // don't send alerts at night
}

if (avgPower < threshold) {
    msg.alertMessage = `SOLAR ALERT: production ${Math.round(avgPower)}W < threshold ${Math.round(threshold)}W (norm ${norms[month]}W). Check the installation.`;
    return msg;
}

return null; // all good

4. HTTP Request to Twilio API (or call Python script):

Twilio endpoint via HTTP POST with Basic Auth (Account SID + Auth Token). Body: Body=...&From=...&To=... (form-urlencoded).

5. InfluxDB Out: Write the alert to InfluxDB with tag type=sms_alert for history and dashboard display.

Anti-spam

Without throttling, the system could send 8 alerts per day (8 hours × 1/h). Over a week of problems — 56 SMS, cost $2.24. Unacceptable.

Solution: Node-RED Context Storage to remember when the last alert was sent:

const lastAlert = context.get('lastAlertTimestamp') ?? 0;
const now = Date.now();
const cooldownMs = 6 * 60 * 60 * 1000; // 6 hour cooldown

if (now - lastAlert < cooldownMs) {
    return null; // too soon for the next alert
}

// If we're sending an alert — save the timestamp
context.set('lastAlertTimestamp', now);
return msg;

Maximum 2 alerts per day (6h cooldown × 2 = 12h coverage during the day). Alternatively, save lastAlertTimestamp to InfluxDB instead of Node-RED context — it’ll survive a Node-RED restart.

Seasonal norms

Norms for a 3 kWp installation on a 35° south-facing roof, central Poland location (52°N). Empirical values averaged from PVGIS data:

Month	Daily norm [kWh]	Peak hourly norm [W]
January	1.2	150
February	2.0	250
March	3.6	450
April	5.6	700
May	7.2	900
June	8.0	1000
July	7.6	950
August	6.8	850
September	4.8	600
October	2.8	350
November	1.5	180
December	1.0	120

Scaling to your installation size: multiply values by your_peak_kWp / 3. For a 6 kWp installation — multiply by 2.

Summary

An SMS alert system for a solar installation is an afternoon project: Twilio (5 minutes setup), Node-RED flow with logic (2 hours), calibrating seasonal norms (1 hour). Operating cost: $0.04 per SMS × maximum 2 per day × 365 days = $29.20 per year — and that assumes a year-long problem. Under normal conditions, a few alerts per year costs $1–2. Peace of mind for the price of a cup of coffee a month.