United Kingdom

Cranfield University


Frederic Coulon
College road
MK 230 AL, United Kingdom

T: +44 123 475 49 81



Our understanding of landfill biochemistry shows that there are number of phases in which gas production goes through, depending on the age of the landfill and the length of time that the waste has been covered for.  Generally methanogenesis will commence shortly after the acetogenesis phase, with the microbial population growing and converting acetic acids into methane.  Methane is a gas with a high greenhouse gas [GHG] effect, but is also high in energy value; as a result ‘biogas’ (gases produced from landfills, mostly methane and carbon dioxide) need to be carefully managed.  Models of biogas production can enable operators to understand how much biogas is expected to be generated from the waste within the landfill, however due to the variable nature of landfills; these models need to be validated against actual data. In a selected number of landfills the biogas production is much lower than expected, and so core samples from landfills managed by Viridor are being extracted in order to understand the reasons behind this.


To determine the effect of changes in waste composition on biogas production on UK landfills. To this end, six landfill sites will be investigated to determine how the waste composition of the landfill sites since 2005 changed and to establish an empirical relation between the waste inputs composition and the LFG production. Multi-criteria analysis (MCA) and principal component analysis (PCA) of the waste categories will be carried out in the first place to identify whether there is any temporal correlation between the waste category changes and the gas production – Analysis of variance (ANOVA) will also be carried out to identify whether there is any significant changes in waste composition between the years and within sites that can explain the changes in landfill gas (LFG) production. Then it is proposed to determine whether machine learning (ML) models such as support vector regression (SVR), M5 model Trees or Rules (M5R) can assist in predicting LFG production at landfill site when waste categories are known.

FUNDING TYPE: Industrial funding Viridor
BUDGET: £ 20.000

Contact us


Department of Materials Engineering
Kasteelpark Arenberg 44
3001 Leuven, Belgium