Skip to content

Repository files navigation

Understanding Condition Monitoring

HPCL Internship · Visakh Refinery, Visakhapatnam

An integrated project linking predictive equipment-failure models to spare-parts inventory and procurement decisions for critical rotary equipment at HPCL's Visakh Refinery. Goal: use degradation forecasts to time procurement, cut holding cost, and reduce emergency expediting — validated on a simulated operational quarter.

Duration: [May 2026 – July 2026] Division: [Maintenance & Reliability / Supply-Distribution — Maintenance] Nature: Simulation / proof-of-concept (not deployed to production)


Problem

Maintenance procurement was reactive: emergency part expediting and either overstocked or unavailable spares for high-value rotary equipment. Driven by HPCL’s digital transformation goals to reduce overall maintenance costs by detecting equipment anomalies before catastrophic failure.

Approach

End-to-end pipeline:

  1. Failure prediction — built predictive failure models for critical mechanical components using equipment health index, sensor data and historical failure records with XGBoost algorithms.
  2. Degradation forecasting — forecasted component lifecycle degradation to anticipate replacement timing by XGBoost.
  3. Inventory linkage — aligned model outputs with spare-parts inventory data and vendor lead times.
  4. Reorder optimization — optimized reorder points for high-value spares and designed just-in-time (JIT) procurement timed to predicted degradation.

Results (simulated over one operational quarter)

  • ~17.5% lower holding cost on critical mechanical components via JIT procurement aligned to predicted failures.
  • ~24% fewer emergency part-expediting incidents from degradation-based procurement timing.
  • [Availability metric — VERIFY: see note below] for high-value rotary equipment spares during planned turnarounds, by cross-referencing equipment health indices with vendor lead times.

All figures are from a simulated scenario, not live operations. Baselines and assumptions are documented in [/results].


Tech Stack

  • Language: Python [3.x]
  • Data: pandas, numpy
  • Modeling: scikit-learn
  • Forecasting: XGBoost
  • Visualization: matplotlib, seaborn
  • Other: Excel, Power BI, SAP export

Repository Structure

.
├── data/            # synthetic / anonymized only — no confidential HPCL data
├── notebooks/       # failure modeling, degradation forecast, inventory optimization
├── src/
├── results/         # metrics, baselines, assumptions
└── README.md

Running

git clone https://github.com/KilkankB/HPCL--Understanding-Condition-Monitoring.git
cd HPCL--Understanding-Condition-Monitoring
pip install -r requirements.txt

Key Learnings

  • Gained hands-on expertise in processing real-time sensor data and deploying predictive analytics to detect early-stage equipment degradation.
  • Developed a deep understanding of rotating equipment failure modes and their critical impact on refinery throughput, safety, and turnaround schedules.

Note on Data

HPCL operational data is confidential. This repository uses available data only. No proprietary data is published.

About

Condition Monitoring (CM) is the process of monitoring measurable parameters of machinery, such as vibration, temperature, lubricant condition, and noise, in order to identify a significant change that is indicative of a developing fault.

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages