A rnold Gmtinder c aSwiss Federal Institute for Environmental Science and Technology (EA WAG), Ueberlandstrasse 133,

7. 2. I n t e g r a t e d systems I n the pilot studies performed in Cornol the i ntegrated membrane system MF-NF was tested ( Fig. 3). During the experiments it was observed t hat MF was more susceptible to fouling than UF [1], therefore the suggested configuration of an i ntegrated system for karstic water is UF-NF. D ue to lack of cost data for integrated systems a r ough estimation based on values for UF and NF w as made. T he planning and engineering costs for h ousing and automation of an integrated plant a re assumed to have the largest potential for c apital costs reduction. On the one hand the p rocess operates as single unit (UF-NF) but on t he other hand the two systems have to be c ontrolled separately. This limits the economy p otential which is estimated to be about 25% c ompared to the sum of the costs of both s eparate systems. Assuming the identical operat ional conditions as discussed for the single s ystems, the whole facilities would require 50% m ore space than a single membrane process. O &M costs for pumping energy, chemicals, m embrane replacement, wastewater and sludge d isposal and plant maintenance and repairs are a ssumed to be the same as for UF and NF alone. T herefore they can be added up. Labour is the o nly parameter which may considerably change d ue to the higher efficiency. Here it is assumed t o be about 12 h a week for an integrated plant of t he size of Cornol and about 14 for one of the s ize o f Betteraz. W ith these considerations treatment costs of t he UF-NF integrated system amount to 41 c E uro/m 3 for Betteraz and 51 c Euro/m 3 for T he cost analysis of conventional and memb rane treatment schemes for the treatment of k arstic spring water in the size range of 30300 m3/h reveals that membrane filtration with M F or UF is competitive to the rather complex p rocess combinations presently applied in practice. The specific production costs for MF and U F plants are about 5% lower than the ones for c onventional w ater production (which includes f locculation, sand filtration, advanced ozonation, G AC filtration) If quality problems with m icropollutants have to be solved in addition to t urbidity, MF or UF may be combined with PAC d osing. This increases O&M costs considerably l eading to overall treatment costs o f about 20% h igher than the costs for conventional flow s chemes. However, it can be expected that the o peration of membrane systems is somewhat e asier and labour costs, so far not included in this c omparison, may be lower for MF/PAC or UF/ P AC systems. This would again bring the overall p roduction costs for membranes into the cost r ange of conventional treatment systems. The p roduction costs with a combination of NF with M F or UF are far above the other alternatives a nd is considered as not competitive as long as t he removal of nitrate, sulphate or hardness are n ot required. A c omparison of the capital costs reveals that m embrane plants are 30-50% cheaper than the c onventional treatment plants investigated in this s tudy. On the other hand, O&M costs for conv entional plants are lower which compensates f or the higher investment. A d etailed breakdown of the O&M costs s hows that simple MF and L~ dead-end opera- t ion leads to the same energy requirements as o bserved in conventional treatment schemes. An i mportant O & M cost item is the membrane r eplacement which results in specific costs of 1 5-20% of the overall treatment costs (labour e xcluded). However, the application of MF/PAC o r UF/PAC combinations requires energyc onsuming cross-flow operation and PAC dosing w hich leads to an O&M cost increase by a factor o f 2.5 compared to simple dead-end operation a nd conventional treatment. I n the comparison of the overall treatment c osts, further criteria such as the superior f inished water quality of membrane systems, t heir higher flexibility in case o f plant upgrade or e nlargement are not included. Moreover, the a ccomplishment o f more stringent water quality r egulations in future (parasites) may give good r easons to favour membrane systems over the c omplex conventional treatment schemes p resently applied for the production of potable w ater from karstic springs.