Optimizing Water Safety through Data Intelligence.

A machine learning-based chlorine demand forecasting system designed for the Ambatale Water Treatment Plant. Preventing shortages, eliminating waste, and ensuring public health.

hydro_admin_dashboard.php

Chlorine Demand Forecast

DECEMBER 2025

Predicted Total Demand

0 kg

+4.2% vs Last Year

Model Confidence

Random Forest / LSTM

Daily Demand Projection

Safe Levels

Peak: 930kg

Why Modernize Now?

Moving away from manual estimation to data-driven precision.

Current Manual Process

Overstocking: Creates chemical wastage and expiration risks.
Understocking: Severe public health risks and water quality degradation.
Static Estimation: Fails to account for seasonal and climatic fluctuations.

The Machine Learning Solution

Dynamic Forecasting: Analyzes historical data + climate patterns.
Operational Efficiency: Reduces manual workload and calculation errors.
Cost Reduction: Optimizes procurement to match exact demand.

System Capabilities

Built for the National Water Supply and Drainage Board.

Smart Dashboard

Visualizes daily demands, seasonal trends, and safety thresholds in real-time.

Prediction Engine

Uses Random Forest & LSTM models to predict total monthly demand with high accuracy.

Data Portal

Secure entry interface for hourly operational data (pH, turbidity, conductivity).

Auto-Reporting

Generates procurement documents instantly for inventory audits and planning.

Operational Workflow

How ChloroCast transforms raw data into actionable insight.

1. Data Collection

System gathers historical chlorine usage and climatic data (rainfall, temperature) from NWSDB records.

2. Preprocessing

Raw data is cleaned, normalized, and missing values are handled to ensure model accuracy.

3. ML Analysis

The Random Forest & LSTM models analyze patterns to generate demand predictions for the upcoming cycle.

4. Visualization

Results are displayed on the interactive dashboard for immediate procurement planning.

Financial Impact

Minimized Wastage

Public Health

Consistent Quality

Efficiency

Data-Driven Ops