🎓 Lesson 12
D5
NOx Sensor Cross-Sensitivity Compensation Algorithms
A NOx sensor cross-sensitivity compensation algorithm is a smart correction method that helps the sensor tell NOx apart from other gases (like CO or hydrocarbons) that can trick it into giving wrong readings.
🎯 Learning Objectives
- ✓ Explain how cross-sensitivity to CO and HCs degrades NOx sensor accuracy using sensor response curves
- ✓ Apply polynomial compensation models to correct raw sensor voltage for known interferent concentrations
- ✓ Analyze ECU diagnostic trouble codes (DTCs) linked to uncompensated NOx sensor drift (e.g., P229F, P2201)
- ✓ Design a two-point linear compensation lookup table for NOx sensor output at 200°C and 400°C exhaust temperatures
📖 Why This Matters
In modern diesel aftertreatment systems, inaccurate NOx readings due to cross-sensitivity can trigger false DTCs, cause unnecessary DEF dosing (increasing operating costs), or—worse—mask real NOx exceedances, leading to regulatory noncompliance and costly field recalls. For mining fleet operators, uncorrected NOx sensor errors directly impact SCR system reliability in underground mines where ventilation constraints make NOx accumulation especially hazardous.
📘 Core Principles
NOx electrochemical sensors (e.g., amperometric dual-cell designs) generate current proportional to NOx partial pressure—but also respond to CO (≈3–8% of NOx sensitivity), unburned HCs (≈5–12%), and O₂ fluctuations. Cross-sensitivity arises from overlapping redox potentials and catalytic surface reactions on the sensing electrode. Compensation requires either: (1) multi-parameter calibration using reference gas blends under controlled temperature/flow, (2) model-based inversion using Arrhenius-type temperature dependence and interferent-specific response coefficients, or (3) adaptive estimation via Kalman filtering fused with upstream lambda and temperature sensors. Industry practice favors hybrid look-up + linear interpolation due to ECU memory and computational constraints.
📐 Polynomial Interference Compensation Model
The most widely deployed compensation uses a second-order polynomial correction applied to raw sensor output (V_raw) based on measured CO concentration and exhaust temperature. This model balances accuracy and embedded-system feasibility.
CO-Compensated NOx Voltage
V_comp = V_raw − (k₁·[CO] + k₂·[CO]²) − k₃·(T − T_ref)Corrects raw sensor voltage for CO interference and temperature deviation from reference point.
Variables:
| Symbol | Name | Unit | Description |
|---|---|---|---|
| V_comp | Compensated sensor voltage | V | Voltage output after cross-sensitivity correction |
| V_raw | Raw sensor voltage | V | Uncorrected analog output from NOx sensor |
| [CO] | Carbon monoxide concentration | ppm | Measured upstream CO level (typically from broadband lambda or dedicated CO sensor) |
| T | Exhaust gas temperature | °C | Temperature at NOx sensor mounting location |
| T_ref | Reference temperature | °C | Calibration reference point (usually 300°C) |
Typical Ranges:
Post-turbine, pre-SCR: 50–200 ppm CO
Underground mine operation: 200–600 ppm CO during transient load
💡 Worked Example
Problem: Given: Raw sensor voltage V_raw = 1.42 V, measured CO = 85 ppm, exhaust temperature T = 320°C. Use: V_comp = V_raw − [0.0012·CO + 0.00004·CO²] − [0.002·(T−300)].
1.
Step 1: Compute CO interference term = 0.0012 × 85 + 0.00004 × (85)² = 0.102 + 0.289 = 0.391 V
2.
Step 2: Compute temperature offset = 0.002 × (320 − 300) = 0.040 V
3.
Step 3: Apply correction: V_comp = 1.42 − 0.391 − 0.040 = 0.989 V
Answer:
The compensated voltage is 0.989 V, which corresponds to ~182 ppm NOx (using factory calibration slope of 0.0054 V/ppm), falling within the typical post-SCR range of 150–250 ppm.
🏗️ Real-World Application
In Komatsu’s 930E-4 haul trucks (2022+), a field-reported NOx sensor drift was traced to high CO levels (>120 ppm) during cold-load acceleration in low-ventilation underground ramps. The ECU’s default 1st-order CO compensation failed; engineers reprogrammed the lookup table using NIST-traceable gas mixtures at 250°C, 350°C, and 450°C, reducing reported NOx error from ±47 ppm to ±9 ppm—restoring SCR urea dosing fidelity and cutting DEF consumption by 11% over 6 months.