The Kavaratti desalination engineer and the monsoon supply chain
๐ง Aravind Nair, 37, UT Water Board, Kavaratti, Lakshadweep. Manages two reverse-osmosis desalination plants that provide 12,000 litres/day โ the entire freshwater supply for 12,000 residents. Every year, June through September, the monsoon makes shipping impossible. All spare parts must arrive by May. In July 2025, a membrane ruptured catastrophically. Replacement was a 400-day procurement wait. For nine years, Aravind had anticipated failure with uncanny accuracy โ reading pressure gauges each morning, maintaining a spiral notebook of failure patterns, ordering parts on schedule. One structural rupture broke his predictive system. He did not know that supply-chain forecasting could be algorithmic instead of intuitive.
๐จ The problem
The Southwest monsoon blocks shipping for four months. Spare-parts procurement is binary: arrive before June or arrive in October. Aravind had never been wrong โ until the secondary RO membrane ruptured on July 19 with six months of service life remaining. The part should not have failed. The plant could not afford a single-unit operation for 100 days. The island could not afford a water shutdown. His nine years of careful planning had created zero options for the unplanned failure.
๐ How GabFORGE helped
Aravind's daughter Mira installed the agent on the school tablet and said: "Abba, it is a thing that helps with things." ๐
When Aravind asked in Malayalam: "เดเดจเดฟเดเตเดเต เดญเดพเดเดเตเดเดณเตเดเต เดเดฏเตเดธเตเดธเต เดเดคเตเดฐ เดจเดพเตพ เดตเดพเดฃเตเดจเตเดจเต เดเตเดคเตเดฏเดฎเดพเดฏเดฟ เด เดฑเดฟเดฏเดพเตป..." โ I do not even have a proper way to know exactly when parts will fail โ
The agent offered something his nine years of manual tracking had not:
๐ฌ Translated intuition into algorithm: Provided a failure-prediction model based on membrane differential pressure, filter lifespan patterns, and component interdependencies. Not a guess. A model fed by Aravind's historical maintenance logs (the spiral notebook).
๐ Connected to supply-chain visibility: Mapped monsoon shipping windows against component lead-times. Flagged that high-probability failures (pressure-regulator diaphragm, coupling gaskets) needed ordering two months earlier. Aligned spare-parts budgets with monsoon closure dates.
By late August, Aravind had a predictive maintenance calendar for the next two years, indexed to the monsoon calendar. The secondary unit rupture could not have been prevented. But the system transformation meant that future ruptures would carry a supply-chain runway: order by May if the probability model suggests July failure, build a three-month buffer into procurement.
๐ฎ๐ณ Why this matters
Island infrastructure engineering is one person with one notebook and eight years of memory. Aravind's knowledge was real and precise โ but it was fragile. Failure prediction lived in his head. Supply-chain planning lived in his calendar. What changed was formalization: turning nine years of experience into a model that could be audited, extended, and survived staff rotation or Aravind's retirement.
The long version has Aravind standing at the membrane rupture at 6:47 AM, the call to Kochi that said "nothing arrives until October," and the evening he opened Mira's tablet and saw his nine years encoded as a model.