Article Sidebar
Main Article Content
Abstract
Irrigation controllers must switch reliably and respond safely to faults before they are connected to pumps and valves. This study tests an ESP32 irrigation controller in the Wokwi simulator, with remote monitoring through the ThingSpeak cloud service. A potentiometer provided an uncalibrated input index, a slide switch represented rain, and two LEDs showed the pump command and a safety lockout. The control rules started irrigation below 30%, stopped it at or above 40%, blocked it during rain, and enforced a latched 90-second limit on continuous operation. Separate simulator traces recorded a threshold stop at 40.76%, a rain-triggered shutdown, and a timeout exactly 90.000 seconds after start. After the input validator was corrected, an injected non-finite value stopped the pump command at 10.006 seconds, and the command restarted at 15.016 seconds once a valid value returned. In a 604-second cloud trace, 30 snapshots were scheduled, 25 uploads were confirmed and 25 records were stored. Four intermediate snapshots never reached the server, consistent with the latest-only queue design, and the final snapshot was still pending when logging ended. The stored records cover 86.21% of the 29 eligible sampling instants. Because these results come from separate development-stage runs rather than one regression suite on a frozen release, they support only the tested command transitions and cloud reporting. They do not demonstrate field performance, water savings or resilience to network outages. Machine-learning forecasting is identified as future work; no predictive model was trained or evaluated.