4. Discussion
We wanted to study the effect of uncertainty in waste compo¬sition in a waste management system comprising the recycling of paper, plastic, glass and metals, and the incineration of residues with heat recovery. We have combined waste composition analyses and the annual amounts of source-separated waste frac¬tions reported by the relevant municipal authorities to estimate waste compositions for five cities. There can be many reasons for uncertainty in the estimated waste composition, stemming from uncertainty in both the composition analyses and in reported waste amounts, and from the combination of these two sources of information. A Norwegian study compared the composition analysis of 52 municipalities and found an uncertainty in each waste fraction of between three and six per cent (hazardous waste and textiles had higher uncertainty) (Skullerud et al., 2010). Gentil et al. (2009) found large variation in literature data on waste com¬positions when performing an assessment of the waste systems in several European countries, and Dalilen et al. (2009) discuss uncer¬tainty in public waste data. Large difference can be found if we compare the five municipalities included in the present study. Sogndalen has a paper fraction with a percentage by weight of 28.7, while the paper percentage by weight for Trondheirn is 37. For plastic the percentage by weight is 8.2 in Skien, while in Sogndalen It is 17.4. The reason for these differences could be due to the dif¬ferent sizes of the cities, their geographical location, seasonal vari¬ations influencing waste composition analysis, etc. A comparison of waste compositions in Norwegian cities found large cities to have a larger percentage by weight of paper, hazardous waste and other non-burnable waste, and less metal, food and other burnable waste (Skullerud et al., 2010). We will, however, not go further into the reasons for uncertainty in waste composition, but rather discuss the effect of such uncertainty on modelling of waste systems.