J. agric. Engng Res. (1998) 69, 63 —71 Treatment of Agricultural Wastewater in Downflow Reed Beds: Experimental Trials and Mathematical Model G. Sun; K. R. Gray; A. J. Biddlestone School of Chemical Engineering, The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK (Received 28 April 1997; accepted in revised form 3 October 1997) This paper presents an examination of the perfor- mance of a full-scale downflow reed-bed system for the treatment of high-strength agricultural wastewater with BOD  in the range 400—1500 mg/l. The effects of organic loading, hydraulic loading and the frequency of intermit- tent feeding on the treatment results were identified. The removal of organic matter in terms of BOD  and COD averaged 74·3 and 53·0%, respectively. These removal rates appeared to increase with organic loading and feeding frequency but decrease with hydraulic loading. Percentage removals of suspended solids, NH  —N and PO  —P were 39·6, 23·1 and 34·7%, respectively. Organic pollutants, expressed as BOD  , were postu- lated to be retained in the downflow reed beds when the influent flow flushed through the beds; they were then aerobically decomposed by micro-organisms during the intervals between intermittent feeding when oxygen can diffuse into the beds more readily. A mathematical model was developed to describe this treatment mechanism based on an assumption that the pollutants are removed from the wastewater by the consecutive processes of adsorption and aerobic decomposition. The values of the constants for the model were derived from the experi- mental data. Comparison between the experimental and calculated results demonstrates that the mathematical model can be used for design purposes.  1998 Silsoe Research Institute Notation A  reed bed area (m) C equilibrium concentration of the pollutant in solution (mg/l) C equilibrium BOD  value of wastewater (mg/l) C  influent BOD  value (mg/l) C  effluent BOD  value (mg/l) C  BOD  value of the biofilm on bed matrix (mg/l) h reed-bed depth (m) K  , K  equilibrium constants indicative of the strength of adsorption K  adsorption constant K  reaction constant (h) m amount of adsorbent (mg) t  time interval between two flushes (h) »  water volume of each flush (l) »  volume of the biofilm on bed matrix (l) x  amount of pollutant adsorbed (mg) x amount of the adsorbed pollutants, expressed as the oxygen demand for their biological decomposition (mg) x oxygen demand of the pollutants adsorbed in the bed matrix before each flush (mg) 1/n degree of non-linearity  superficial density of reed-bed matrix (mg/m) 1. Introduction Since the 1980s, reed-bed treatment systems (RBTS) have become a popular treatment alternative in Europe and USA for a variety of wastewaters. 1–4 These systems are now well established for the tertiary treatment of sewage in small-to-medium-sized sewage works, and for its secondary treatment in small communities. In 1985, a research project was launched at the University of Birmingham to explore the use of RBTS for treating high-strength agricultural ‘‘dirty waters’’. These are wastes, generally with less than 3% dry matter, made up of water contaminated by manure, crop seepage, milk or cleaning materials. 5 They represent a significant potential pollution threat to the water environment owing to their 0021-8634/98/010063#09 $25.00/0/ag970224 63  1998 Silsoe Research Institute