# Research on the performance degradation and failure mechanism of exhaust Systems for explosion proof diesel engines in different mining applications

## TL;DR

This study examines how exhaust systems in explosion-proof diesel engines used in mining degrade over time and proposes solutions to improve their performance and safety.

## Contribution

The study introduces a multivariable regression model and machine learning for predicting exhaust system degradation in mining engines.

## Key findings

- Wet exhaust systems degrade fastest due to sludge accumulation, causing a 15.2% increase in backpressure and 12–15% power loss.
- Dry systems have lower emissions but suffer thermal fatigue in condensers after 1000 hours of operation.
- Combined systems exceed PM emission limits at 300 hours due to hybrid failures.

## Abstract

Mine explosion-proof diesel engines are critical for underground coal mining, yet the long-term performance degradation of their exhaust systems poses substantial risks to operational safety and environmental compliance—a gap not addressed in short-term performance-focused studies. This study investigates the degradation and failure mechanisms of dry, wet, and combined exhaust systems under simulated mining conditions using a JHP4105DZDFB-G engine. Experimental results show that the wet system exhibits the fastest degradation, with exhaust backpressure increasing by 15.2% within 500 hours of continuous operation under simulated mining conditions due to sludge accumulation, leading to a 12–15% power loss. In contrast, the dry system maintains lower emissions (NOₓ, HC, CO, and PM were reduced by 68.36%, 71.71%, 55.39%, and 82.28% compared to the wet system) but suffers thermal fatigue in condensers after 1000 hours. The combined system shows hybrid failures, with PM emissions exceeding regulatory limits (0.4 g/kWh) at 300 hours. Mechanistic analysis reveals that wet systems fail primarily due to mechanical blockage and corrosion, while dry systems succumb to thermal-mechanical fatigue. A multivariable regression model and machine-learning algorithms are developed to predict degradation thresholds, enabling proactive maintenance. Thermal management optimization for dry systems is proposed.