The drainage discharge of a municipal solid waste incinerator (MSWI) bottom
ash landfill was simulated using various modelling approaches. Two functio
nal models including a neural networks approach and a hydrological linear s
torage model, and two mechanistic models requiring physical/hydrodynamic pr
operties of the waste material, HYDRUS5 and MACRO (Version 4.0) were used.
The models were calibrated using an 8-month data set from 1996 and validate
d on a 3-month data set from winter 1994/1995. The data sets comprised hour
ly values of rainfall, evaporation (estimated from the Penman-Monteith rela
tionship), drainage discharge and electrical conductivity. Predicted and me
asured discharges were compared.
The discharge predicted by the functional models more exactly followed the
discharge patterns of the measured data but, particularly the linear storag
e model, could not cope with the non-linearity of the system that was cause
d by seasonal changes in water content of the MSWI bottom ash. The lit of t
he neural networks model to the data improved with increasing prior informa
tion but was less smooth than the measured data. The mechanistic model that
included preferential discharge, MACRO, better modelled the discharge char
acteristics when inversely applied, indicating that preferential flow does
occur in this system. However, even the inverse application of HYDRUS5 coul
d not describe the system discharge as well as the linear storage model. Al
l model approaches would have benefited from a more exact knowledge of init
ial water content. (C) 2001 Elsevier Science B.V. All rights reserved.