Real-Time NO₃⁻ Monitoring and Feedforward Carbon Dosing

Control of Denitrification in High-Strength Industrial Wastewater

1.   Background and Challenge

This case study examines the optimisation of biological denitrification at a hydrogen peroxide production facility operating the anthraquinone process. The wastewater generated at this site is characterised by an unusually complex matrix, presenting significant challenges for both accurate nitrogen monitoring and effective carbon dosing control.

1.1  Wastewater Characterisation

The influent wastewater to the denitrification stage contains a combination of organic solvents, nitric acid (HNO₃) from acid washing operations, and various process additives and stabilisers. This results in nitrate (NO₃⁻) concentrations ranging from 50 to 100 mg/L, with strong intra-day variability driven by production scheduling and batch operations.

 Table 1: Influent wastewater characterisation and monitoring implications

1.2  Identified Process Challenges

Two fundamental challenges were identified that prevented effective denitrification control:

• Severe instrument interference: The combination of residual hydrogen peroxide (a strong oxidant), high salinity, and organic solvent by-products creates a matrix in which conventional UV/optical NO₃⁻ analysers produce measurement errors exceeding 200%. This rendered real-time nitrate data unreliable for process control purposes.
• Lagging dosing strategy: In the absence of reliable real-time monitoring, the plant operated a fixed-rate carbon dosing regime supplemented by manual operator adjustments. Dosing corrections were made reactively, based on delayed laboratory total nitrogen (TN) results from grab samples, introducing a lag of several hours between load changes and dosing.

Together, these factors created a control environment characterised by chronic under-dosing during high-load periods (leading to TN consent exceedances) and systematic over-dosing during low-load periods (resulting in unnecessary chemical expenditure and elevated operational costs).

2.   Technical Solution

2.1  Analyser Selection — Ion Chromatography with Optical Detection

The Aquamonitrix online analyser was selected to address the measurement interference problem. Unlike conventional single-pass optical or UV absorption methods, the Aquamonitrix platform separates NO₃⁻ from interfering matrix components using ion chromatography prior to quantification. This two-stage approach — chromatographic separation followed by optical detection — effectively eliminates cross-interference from oxidants, salts, and organic species.

Under the complex wastewater matrix described above, this instrument consistently achieved measurement accuracy exceeding 95%, compared to errors greater than 200% observed with conventional optical analysers. This represented a step-change in data reliability sufficient to underpin closed-loop process control.

2.2  Observed Load Dynamics

Continuous monitoring with the Aquamonitrix analyser revealed the true extent of influent NO₃⁻

variability, which had previously been masked by inaccurate or infrequent measurement:

• Peak NO₃⁻ concentration: ~95 mg/L
• Average NO₃⁻ concentration: ~39.6 mg/L
• Significant intra-day fluctuations driven by batch production cycles

This variability — a ratio of approximately 2.4:1 between peak and average loads — confirmed that any fixed dosing strategy would inevitably produce significant periods of both under- and over-dosing, regardless of the set-point chosen.

2.3  Feedforward Control Strategy

The reliable real-time NO₃⁻ data enabled the project team to develop a quantitative carbon dosing model based on influent nitrate concentration. The control logic exploits favourable characteristics of this specific wastewater:

• NO₃⁻ is the dominant nitrogen species, simplifying the mass balance
• Low influent COD eliminates endogenous carbon as a significant variable
• Low ammonia concentrations confirm that nitrification-derived nitrate is negligible

These conditions made the system ideal for feedforward control, in which the carbon dose is calculated directly and prospectively from the measured influent NO₃⁻ load, rather than relying on feedback from effluent quality. The dosing pump setpoint is updated continuously in near real-time as analyser data is received, maintaining the stoichiometric carbon-to-nitrate ratio required for complete denitrification across all load conditions.

3.   Results and Performance

3.1  Operational Comparison

The following table summarises the key operational differences between the previous fixed dosing regime and the implemented feedforward control strategy:

 Table 2: Comparison of carbon dosing strategies — fixed vs feedforward control

3.2  Key Performance Outcome

The ~40% reduction in carbon source consumption represents a direct and recurring operational cost saving. The magnitude of this saving reflects the extent to which the previous fixed-dosing regime was over-supplying carbon during periods of low nitrate load — a systematic inefficiency that real-time feedforward control eliminates by design.

Effluent TN performance also improved materially. By maintaining proportional carbon dosing during high-load periods, the system eliminates the under-dosing events that had previously caused TN spikes and potential consent breaches.

4.   Discussion

This case study illustrates a principle that applies broadly to industrial denitrification: the quality of the control outcome is bounded by the quality of the measurement data. In wastewater matrices where conventional sensors fail, advanced analytical techniques — such as ion chromatography-based separation — are a prerequisite for effective process control, not merely an enhancement.

The feedforward control architecture adopted here is particularly well-suited to wastewater streams in which influent nitrogen speciation is predictable and COD variability is low. Where these conditions are met, influent NO₃⁻ concentration becomes a highly reliable predictor of instantaneous carbon demand, enabling a proportional and prospective dosing response that feedback-only systems cannot achieve.

The broader implication for process engineers is that investment in reliable online analysers in challenging matrices should be evaluated not merely as a monitoring cost, but as an enabler of process optimisation with quantifiable return on investment through chemical savings and improved effluent compliance.

5.   Conclusions

• Conventional optical and UV-based NO₃⁻ analysers are unsuitable for use in hydrogen peroxide production wastewater, producing errors in excess of 200% due to matrix interference from oxidants, salinity and organic solvents.
• Ion chromatography-based online analysis (Aquamonitrix) achieved >95% measurement accuracy in this matrix, providing data of sufficient quality and frequency to underpin closed-loop process.
• Real-time monitoring revealed peak-to-average NO₃⁻ variability of approximately 2.4:1, confirming that fixed carbon dosing strategies are fundamentally unsuitable for this application.
• Feedforward carbon dosing control, driven by continuous influent NO₃⁻ data, achieved a ~40% reduction in external carbon source consumption compared to the previous fixed-dosing.
• Effluent TN compliance was maintained consistently under the feedforward control strategy, eliminating the reactive exceedance events associated with lagging manual.
• This case demonstrates that measurement reliability is the critical enabling factor for advanced denitrification process control in complex industrial wastewater matrices.