A pilot constructed treatment wetland for pulp and paper mill wastewater: performance, processes and implications for the Nzoia River, Kenya

Background
Sustainable water pollution control calls for effective enforcement of regulations and adoption of cleaner production technology as well as effective end-of-pipe treatment of effluents. The final effluent quality of many municipalities and industries in Kenya seldom comply with government-prescribed effluent discharge guidelines. There is, therefore, a need for a sustainable technology that can reliably achieve acceptable effluent quality for discharge into the environment at minimal cost. Natural and artificial wetland systems have been used as a cost-effective alternative to conventional wastewater treatment methods for improving final effluent quality. Data and information on pulp and paper mill wastewater treatment in constructed wetlands are few while performance data that can guide design and operation under tropical environment conditions are lacking.
This study was undertaken to explore the potential of a constructed wetland to improve the quality of the final effluent from Pan African paper mills (E.A.) Limited (PANPAPER) in western Kenya in order: 1) to be in compliance with national discharge regulations, and 2) to protect the receiving aquatic environment, the River Nzoia, and downstream riparian users. In this thesis the problematic wastewater components were characterised (Chapter 2). The data were used to evaluate the performance of the PANPAPER Mills wastewater treatment ponds and the wetland system with respect to removal of nutrients, organic matter (BOD, COD), suspended solids (TSS), and phenols (Chapters 4-6) under various operational conditions.
A pilot-scale constructed wetland covering a total area of 48.5 m2 was located in the tree nursery just below the final stabilization pond of PANPAPER Mills. It consisted of eight subsurface flow (SSF) cells each of dimensions 3.2 m (length) x 1.2 m (width) x 0.8 m (depth) cells and two cells of dimensions 6.2 m (length) x 1.5 m (width) x 0.8 m (depth). The latter were initially operated as free water surface flow and later as subsurface flow systems. The subsurface flow cells were planted in pairs with Cyperus immensus, Typha domingensis, Phragmites mauritianus and Cyperus papyrus. The Cyperus immensus did not establish well due to frequent attacks by vermin monkeys and were therefore removed after eight months and the cells left implanted. The larger cells were planted with Typha domingensis. All cells were filled with gravel to a depth of 30 cm.
The experimental systems operation was dynamic and ran for a total period of 3 years from 2002 to 2005. It involved different operation modes, hydraulic loading rates and retentions in order to optimise pollutant removal while maintaining good plant vitality. Initially the wetland was operated on a batch load-drain mode starting with 5-day retention time (batch phase 1). It was assumed that this format would enhance organic matter degradation. Plant vitality was relatively poor and was partly attributed to low nitrogen loading at a long retention time. A shorter retention time of 3 days (batch phase 2) was subsequently used. Although plant vitality increased, there was a reduction in treatment efficiency with respect to TSS. It was therefore decided to use a continuous flow operation mode. The results of batch phase 2 had shown better wetland performance at 3 days than at 5 days. A third phase of batch operation was undertaken as a repeat of the first phase when the wetland was considered to be mature. In all there were three phases of batch operation and two of continuous flow. In the first phase of continuous flow plant growth was at steady stage while in the second phase the plants were at an exponential growth stage. A tracer study using lithium chloride was conducted in the first phase.
Wetland performance
The study revealed that the PANPAPER Mills pond system was actually performing well as per its type and design. However, the concentration of pollutants in the final effluent (average 45±3,
394±340, 52±6, and 0.64±0.09 for BOD, COD, TSS and phenols, respectively) discharged into the Nzoia River does not comply with the national discharge limits. Mean total nitrogen and phosphorus were about 3 mg/1 and 0.7 mg/1, respectively giving a low N:P ratio.
Evapotranspiration (8-16 mm/day) was found to be an important component of outputs in the water budget of the wetland system making up to 15-32 %, depending on the system type. ET rates were different for the aquatic plant species studied. It was not possible to deduce the actual retention time and other hydraulic parameters (efficiency and number of "tanks in series") under continuous flow, as there was no discernable tracer concentration curve for all wetland cells. For this wastewater, which has high organic matter content, the study should be conducted with a different tracer. Alternatively, lithium chloride may still be used but with continuous feed instead of pulse feed, as was the case in this study.
Plant tissue nutrient concentrations were lower than in healthy natural wetland plants. Nitrogen concentrations based on dry weights in Phragmites, Typha and papyrus were 9.2±0.7, 7.4±0.5, and 6.1±0.2 mg/g, respectively while phosphorus concentrations were 1.7±0.12, 1.9±0.11, 1.6±0.14 mg/g, respectively. Despite this, Typha and Phragmites had satisfactory aboveground biomass production (10896 g/m and 3015 g/m2 dry weight, respectively) when compared to natural wetlands. The growth of papyrus was sub-optimal with an aerial biomass of 3075 g/m2. In general, plant vitality and growth was lower during batch mode wetland operation. Below ground root and rhizome growth was variable. Typha roots penetrated the entire bed depth (approx. 30 cm) while Phragmites and papyrus rooting depths were in the top 20 cm and 10 cm, respectively.
Plant uptake of nutrients exceeded inputs by influent in the exponential growth stage. Nutrient mass flows indicated that in this low loaded system mineralisation and cycling of nutrients in accumulated sediments and/or in senescing/decaying plant organs are important for sustaining plant growth.
Mean removal efficiency for total nitrogen was in the range of 49 - 75 % for planted cells and 42 - 49 % for unplanted ones in continuous flow. For phosphorus removal efficiencies were 30 - 60 % in planted cells and (minus) 4 - 38 % in unplanted ones. Removal efficiency of up to 25 % may be attributed to uptake into plant shoots. Harvesting of plant shoots should be appropriately timed to avoid depletion of the nutrient pool. The removal efficiencies were lower during batch operation modes.
The constructed wetland effectively removed BOD (up to 90 %) and TSS (up to 81 %) from the wastewater to concentrations below that prescribed by the regulating authority in Kenya. However, COD removal was low (up to 52 %). The non-zero background concentration for BOD varied between 4.3 and 7.4 mg/1 for the different cells while areal BOD reaction rate constants varied from 0.055 - 0.114 m/day (20 - 42 m/yr). The reaction rates are reported for pulp and paper mill wastewater for the first time. Typha cells had consistently higher TSS removal efficiency than Phragmites and papyrus in continuous flow. Besides TSS removal in the wetland bed with developed plant roots, the presence of macrophytes does not seem to enhance BOD and COD removal when compared to unplanted cells. However, the presence of plants is essential for nutrients and phenol removal.
Mean phenol removal efficiencies based on mass flows ranged from 73 % to 96 %. Good buffering was achieved even during the highest inflow phenol concentration of 1.3 mg/1 and the highest hydraulic loading rate (HLR) of 9.8 cm/day in the wetland. For batch operation, optimal removal was achieved at 5-day hydraulic retention time with a mean outflow concentration of 0.053 ± 0.004 mg/1. The major processes of phenol removal are microbial breakdown (60 %) followed by sedimentation/adsorption (up to 30 %). However, ultimate biodégradation in a mature wetland may be higher as some of the sedimented and/or adsorbed phenols may be re-cycled and microbially degraded. The removal was enhanced by wetland age and presence of aquatic macrophytes especially when they were at an exponential growth stage. During this stage plant uptake rates were
5.4 - 12.7 mg phenol/day accounting for 10 - 23 % of phenol removed in SSF cells. The study reports, for the first time, phenol reduction rates (k values) in a constructed wetland. Average k, for SSF cells was 27 m/yr and 44 m/yr respectively at hydraulic loading rates of 4.1 - 4.9 cm/day and 9.8 cm/day, respectively. The reduction rate was 38 m/yr for the free water surface flow cells at a HLR of 9.3 cm/day. Mean volumetric removal rate for batch operation was 0.57 d' for Phragmites and papyrus cells.
Water quality of Nzoia River
PANPAPER Mills's effluent discharge in a "business-as usual" scenario causes an increase in the concentration of pollutants downstream (3 km) compared to an upstream location (500 m). The increase is highest during low flow in the River, usually in the dry season (January to March). This was reflected in an increase in the concentration of various pollutants by between 20 % and 120 % at the downstream sampling point during the month of March. From the findings of this study I predict that the quality of water in River Nzoia downstream of the discharge point would improve significantly if a full-scale constructed wetland with similar performance as that in the pilot study were to be integrated with the existing treatment ponds as a tertiary stage. Such an intervention would decrease pollution by more than 90 % for phenols. TSS, COD and BOD would be reduced by 44 %, 30 % and 63 %, respectively. Nitrogen concentration downstream of discharge would remain the same as that in the upstream location (100 % pollution reduction) while phosphorus concentrations would reduce by 50 % of the current level.
An appraisal of these findings is given in Chapter 7, which also includes recommendations for design, set-up and maintenance of a full-scale wetland.