Water Treatment Technologies and Their Costs

Abstract

This chapter is a review of drinking water treatment technologies and their unit costs as a function of scale. These are classified into six classes, based on contaminants removed. We find that a UV-based treatment plant is cost-effective.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

According to the USEPA (1996) , 94 percent of 156,000 public water systems in the US are small water systems, serving a population of fewer than 3,300 people. In Canada, the proportion of small systems in one survey was over 75 percent (Environment Canada 2004). With a smaller tax base all small water systems face special challenges, unless the government aggressively supports small water treatment systems. In Canada, many continue to encounter boil water advisories and even disease outbreaks. No doubt that with appropriate public funding, many of these problems can be reduced or eliminated. However, typically in North America, each small community or rural jurisdiction must cover its own capital and operating costs of their drinking water supply, although some jurisdictions offer a subsidy for capital costs . Often a rural community has a small population, lower average income, and consequently a lower tax base. These financial constraints as well as other risk factors were highlighted at a 2004 Montana conference on small water systems (Ford et al. 2004). These constraints are even more severe in developing countries.

Threats to public health persist in rural and small water systems even in the most advanced high-income countries like the USA, Canada, and Europe . The factors accounting for some of these waterborne disease outbreaks were explored in Chap. 2. The objective of this chapter is to present statistical models of costs based on new data obtained from manufacturers of a menu of treatment technologies suitable for water systems, small and large; this information would be of use to local government officials, water engineers, and planners. Most of these technologies are concerned mainly with plants that rely on surface water as the source.

For the USA, the American Water and Wastewater Association (AWWA) has published a number of reports that include recent water utility survey data on current disinfection practices and operations compared with practices in the late 1970s (AWWA Water Quality Division Disinfection 1992 and 2008). According to the AWWA 2008 report, chlorine gas remained the predominant disinfectant, used by 63 percent of respondents, whereas those who used chloramine accounted for 30 percent; chlorine dioxide for 8 percent; ozone for 9 percent; and ultraviolet light (UV ) for 2 percent. The comparable figures for Canada are also available: according to the Environment Canada survey of Municipal Water and Wastewater Plants (2004), in Canada there were 2,402 drinking water systems in that survey, of which 1,513 reported a population of fewer than 3,000. Of these 1,513 drinking water plants, 136 gave information on the type of disinfection technology they use. Some 93 per cent (127 out of 136) used chlorine as the only disinfectant. Those using UV or ozonation accounted for only 6 percent of the total. There is a potential for improving water quality by adopting newer technologies such as UV or ozonation and reducing the probability of waterborne disease outbreaks. At the same time, there is an enormous market potential for corporations that can sell a competitive technology that is also cost-effective. The rest of the chapter presents average “first approximation” cost per cubic meter based on statistical modeling on recently collected data on costs for different flow rates from equipment manufacturers in North America. We show that there exists a menu of cost-effective technologies that might be considered for possible modernization of water treatment plants, including small water treatment plants.

We classify these technologies into six classes, depending on the contaminants removed. Our statistical results show that average costs (including capital, operating, and maintenance) of production of these technologies depend on the flow rate as well as the number of contaminants removed. The larger the flow rate is, the lower the cost will be per unit of volume treated, and the more the contaminants removed are, the higher will be the cost, for any given flow rate. One of our major findings is that for surface waters except those with high color and turbidity, UV -based treatment technologies are cost-effective. However, for any particular system, water engineers would take site-specific features into account to determine what technology is most appropriate.

This chapter is organized as follows. Section 3.2 of this chapter presents a scheme, which classifies water treatment technologies based on the contaminants they remove; Sect. 3.3 shows projected costs of four technologies, which are Ultra Violet disinfection (UV ), Micro filtration —ultra filtration (MF-UF), High rate Clarification & Filtration (HRC) , and Ozonation ; Sect. 3.4 is an analysis of the costs of Advanced Oxidation Processes ; Sect. 3.5 makes brief reference to Reverse-Osmosis (RO ) and Nano-Filtration (NF ), but there is a whole chapter devoted to RO and other more advanced treatment technologies (Chap. 4). In Sect. 3.6, we present examples of costs of actual existing small water treatment systems in Canada. Finally, Sect. 3.7 is a general summary with concluding remarks. Our major conclusion is that for surface water sources except those sources with high color and/or high turbidity, UV is a competitive and viable treatment technology that should be considered in a menu of suitable technologies. However, as stated before, the actual adoption of a treatment technology depends on many site-specific features (such as location, distance from major cities, and topography) that are best determined by the consulting engineers.

2 Six Classes of Water Treatment Technologies

Suppose we consider a large state-of-the-art water treatment plant and use their costs of water treatment as an initial benchmark. One such treatment plant is the Seymour-Capilano Filtration Plant run by the Greater Vancouver Regional District (GVRD) in Canada. This plant, which came on stream in December 2009, will give us a perspective on costs at a large water plant. The source water for this GVRD plant is of high quality, and is possibly free of micro-pollutants , largely because of the source water quality . Table 3.1 gives some information on this system. Due to economies of scale , the plant has the potential to produce drinking water at CAN $0.40 per cubic meter. However, when the distribution costs are added, it is estimated that the consumer would pay about $1 per cubic meter. This provides a comparative benchmark of the costs at a large state-of-the-art water treatment system and shows to what extent the costs of small water systems differ from those at a large system.

Table 3.1 Description of GVRD state-of-the-art water treatment plant

Not all systems can produce at the cost and level of drinking water quality that this Vancouver plant can produce. But our survey of new technologies suggests that there are technologies for small systems with similar low average costs per cubic meter. As stated before, in general, costs depend on the number of contaminants removed, although there may also be other nonlinearities . Below we provide a scheme, which would allow us to classify a given water treatment plant by the number of contaminants removed, based on technology being utilized at the plant. We postulate six classes of water treatment technologies in Table 3.2.

Table 3.2 Proposed water treatment classes

Class 1 represents the minimum level of treatment, which is disinfection by chlorination only. We consider chlorination the minimum disinfection treatment level since all water treatment plants are required to produce water that is free of pathogens . While most groundwater based systems would rely on chlorine only (Class 1), many surface water small water systems will be Class 2, i.e. water that has suspended solids removed and is disinfected. In a Class 3 plant, protozoa will also be removed or inactivated, possibly with the aid of UV or ozonation . If, in addition, all dissolved organic matter is also removed before chlorination, then that would be water without disinfection byproducts (DBP) , and we classify such treatment technology as Class 4.

On the other hand Class 5 (i.e. Classes 5a and 5b) represents technologies that also remove chemicals, micro-pollutants , DBPs, protozoa , and suspended solids, in addition to disinfection. In the scheme proposed in Table 3.2, each progressively higher treatment class indicates a greater removal of contaminants. However, this classification scheme is fairly broad in scope, an initial attempt, although other more finely graded classifications are possible. Note that we are classifying treatment categories or classes, not final water quality . What emerges from this classification is a way of comparing final water quality indirectly, on the basis of what treatment systems are used, and also assessing any possible long-term health threats.

In North America, most drinking water comes from surface water , which needs to be treated adequately. The data presented in the introduction to this chapter shows the dominant role played by chlorine and chlorine derivatives in North America, where this Class 1 technology is concerned almost exclusively with the removal of pathogens , although we know that chlorine is not effective against protozoa and other pathogens. However, for most large cities and populations, the conventional water treatment method is coagulation, flocculation , clarification, and filtration, and is typically followed by disinfection by chlorine or chlorine derivative. But the failure of a flocculator led to an outbreak of cryptosporidiosis in Carrollton Georgia in 1987; the failure of a chlorinator led to an outbreak of giardiasis in Bradford Pennsylvania in 1979. Thus, the conventional treatment train is best described as being Class 3 if it removes all protozoa ; it cannot be classified as Class 4 as chlorination will leave DBP precursors in the water. For this reason, in Ontario and indeed in the whole of North America, the main DBPs , called Trihalomethanes (THMs), nitrosamines and Haloacetic Acids (HAAs) are regulated with maximum contamination limits. But there are also many other DBPs , called Halides, that are not regulated at all.

The most significant drinking water outbreak of cryptosporidiosis was in Milwaukee Wisconsin from March to April of 1993, the worst waterborne disease outbreak in the US history. Two water treatment plants supplying water to Milwaukee used water from Lake Michigan. Both plants used conventional treatment of coagulation , flocculation , sedimentation , rapid sand filtration , and chlorination treatment (Solo-Gabriele and Neumeister 1996, p.81) . Again the failure to remove a protozoon indicates that these plants functioned as no more than Class 2 treatment systems.

Based on the evidence and the above classification system, we are led to the conclusion that the conventional treatment plants in North America are at best Class 3, and no more than Class 2 when they fail to remove protozoa . Note that this conclusion is based on treatment technologies and not on the quality of final drinking water, which may be quite good in some areas, depending on the characteristics of the source water; our focus here is on treatment.

It should also be noted that after a large fall in unit costs of ozonation , many water utilities are choosing ozonation¹ as the primary treatment option (Class 4). In Europe the treatment of choice is granular activated carbon, which we classify as Class 5a. Granular activated carbon (GAC) has been used extensively for the removal of dissolved organics from drinking water. In the early 1970s, it was reported that bacteria, which proliferate in GAC filters may be responsible for a fraction of the net removal of organics in the filter. Following this discovery, pre-ozonation was found to enhance significantly the biological activity on GAC. The combination of ozonation and GAC is commonly referred to as the biological activated carbon (BAC) process, or biologically enhanced activated carbon process. This was implemented in many large water treatment plants in Europe in the 1980s (Dussert and Stone 2000). The efficacy of activated carbon in removing all sorts of contaminants has been further confirmed by Rodriguez-Mozaz (2004).

Advanced oxidation processes (with ozonation or UV -based) are essentially the same as Class 5a, but experiments show a greater efficacy of removal of the same contaminants as those in Class 5a; we, therefore, classify Advanced Oxidation processes as Class 5b.

We should also note that for 90 percent of the residents of Ontario, the source water is the Great Lakes, which also receive wastewater that is not always treated to remove chemicals, particularly pesticides , pharmaceuticals and personal care products ; this topic is deferred to the chapter dealing with wastewater and its impacts on drinking water.

In Germany , roughly 74 percent of drinking water is drawn from ground and spring water, and the remainder is drawn from surface water sources, such as lakes and rivers (Althoff 2007). By 2010, 63 percent of the groundwater bodies in Germany had achieved a rating of “good chemical status” (BMU 2014). Of the total 1,000 groundwater bodies, only 4 percent have not achieved a “good quantitative status,” i.e. 4 percent of the aquifers did not have enough water. The status of surface water is such that 88 percent of water bodies achieved a “good” chemical status, while only 10 percent of all surface water bodies had obtained at least a “good” ecological status (BMU 2014). Given the quality of groundwater, practically no disinfection is needed. The 2011 Profile of the German Water Sector states:

The quality of drinking water is so good that the use of disinfectants in water treatment can even be forgone in many places without [compromising] the high hygienic drinking water standard.

Since there is no chlorine, there are no DBPs ; in areas where the source is groundwater, there are no chemical residues in the water and of course no salinity. Thus, for the groundwater sources we can conclude that German drinking water from the water treatment plants is equivalent to Class 5. In North Rhine -Westphalia, in the City of Cologne, they use groundwater as the source, which is then filtered through activated carbon, producing a very high quality of water. To quote from the City of Cologne website (RheinEnergie 2013):

Some waterworks in Cologne used disinfectant to prevent an increase in the number of germs, and thus hygienic deterioration of the drinking water quality on the way to the customer. Our water lab proved, however, that the perfect hygienic quality of drinking water can be guaranteed even without the use of chlorine dioxide or chlorine.

Where surface water is used in North Rhine -Westphalia, they detected perfluorooctanoate (PFOA) in drinking water at concentrations up to 0.64 µ/L in Arnsberg, Sauerland, Germany . In response, the German Drinking Water Commission (TWK) assessed perfluorinated compounds (PFCs) in drinking water and in June 2006 became the first in the world to set a health-based guideline value for safe lifelong exposure at 0.3 µ/L (sum of PFOA and perfluorooctanesulfonate, PFOS). PFOA and PFOS can be effectively removed from drinking water by percolation over granular activated carbon.

For each treatment class, we also hypothesize the shape of the cost curves. Average costs per volume of water treated will vary with (a) source water quality , (b) flow rate, and (c) target water quality. We expect that for a given type of source water quality, average costs per cubic meter depend on economies of scale . For a given source water quality, Fig. 3.1 below shows the hypothesized (theoretical) average costs as a function of the flow rate for different treatment classes. This graph assumes that contaminants are additively separable and linear.

Fig. 3.1

Hypothetical costs curves and scale of treated drinking water

In reality, that assumption of linearity and additive separability would not hold as some technologies can have an overlap in their functions. For example, technologies that can remove suspended solids (Class 2) can also remove some pathogens (Class 3) and possibly some DBP precursors (Class 4), if used in conjunction with coagulation. Nevertheless, it might be useful to assess the cost differentials between some of the abovementioned treatment classes, and the extent to which nonlinearities might indicate that it would be better to aim at a higher treatment class that happens to have lower average costs per cubic meter even if water quality regulations require just disinfection and no additional removal of contaminants. There is also a further nonlinearity already implicit in Fig. 3.1, namely economies of returns to scale, which suggests that for some smaller communities it might make economic sense to consider a somewhat larger plant scale in the expectation of a future growth in water demand, or consider an amalgamation of two or more small communities to be supplied by a single but larger treatment plant.

It would be interesting to find the average costs per unit of volume of water of the broad water treatment classes and find any nonlinearity in costs where the actual average costs curves may not conform to the hypothetical graph in Fig. 3.1, but in fact exhibit discrete “jumps,” indicating the presence of nonlinearities in costs and complementarities in contaminants removed.

3 Projected Costs: Ultra Violet, Micro Filtration —Ultra Filtration (MF-UF), High Rate Treatment and Clarification (HRC) , and Ozonation

In this section, we present four technologies that may be suitable for small systems: High rate treatment and clarification, UV , MF-UF, Advanced Oxidation Processes (based on UV), Reverse Osmosis -Nano Filtration (RO-NF) and ozonation, although RO is treated in greater detail in Chap. 4 to set the stage for the econometric estimation of “breakeven” and other prices in Chap. 5.

The raw data for costs for different flow rates were obtained from the actual manufacturers (see footnotes for details). For all estimated models, we find average costs per volume of treated water, where the costs are (1) capital costs , amortized (by straight-line depreciation) over a 20 year period, and (2) O&M costs, that include labor, materials, and energy costs for given flow rates.

In the case of surface water , UV -based technologies would most likely require that source water be pretreated using a filtration or sediment removal process before being disinfected by UV . For communities that are concerned about pesticides and other micro-pollutants , advanced oxidation processes (AOPs) may be worth considering. AOPs may not be practical for small systems, but with the implementation of new regulations on drinking water quality in the future, it may be worthwhile for small systems to include UV -oxidation -based treatment technologies in their menu of possible technology options. We note that there are some small communities that are already using AOPs for surface water treatment and also for groundwater remediation, even at a small scale.²

We briefly describe each technology in Table 3.3 ³ and illustrate the statistically modeled costs associated with each of them thereafter. All the technologies considered here produce municipal standard drinking water, and most assume that the raw source water is surface water , which is easily contaminated by animals and/or human activity.

Table 3.3 Treatment Technologies

We use the nonlinear least squares (NLLS) estimation process since it can capture a wider range of functional forms than the ordinary least squares (OLS) method. Simple linear models may not describe certain data generating processes very well especially if the functional form changes over its domain. For instance, our cost data for UV (see Fig. 3.2) shows that a much better description of the data can be had if a nonlinear approach (solid line) is used instead of a strictly linear one (dashed line). In fact, since most of our data followed the same format as in Fig. 3.2, we used the NLLS method to estimate cost functions for the different classes of technology. The NLLS technique has the added advantage of yielding better estimates when the amount of data is limited.⁴

Fig. 3.2

Ultraviolet (UV) Linear Versus Nonlinear Estimation of Cost Data

Table 3.4 shows the estimated cost functions for the various technologies⁵ described above with the functional form $y_{i} = \beta_{1} X_{i}^{{\beta_{2} }} + \varepsilon_{i}$ where $y_{i}$ is the average cost per cubic meter, defined as capital plus O&M, $X_{i}$ is the flow rate in cubic meters and $\varepsilon_{i}$ is the error term, which satisfies the standard Gaussian assumptions .

Table 3.4 Estimated average cost functions for High Rate Clarification & Filtration (HRC) , UV, MF-UF and Ozonation in 2008 CDN dollars

Details of the NLLS regressions and model fit statistics are given in Appendix A. The estimations provided in Table 3.4 above are based on disinfection for the particular technology only and do not take into account the additional cost of residual chlorine for the distribution system, which is required in the US and Canada. We assume that this additional cost would be the same for all the technologies listed above in Table 3.4, and it was, therefore, left out. In any case for any actual plant, there will be many plant-specific costs that the consulting engineers will need to take into account. Therefore, the costs given by the cost models should be viewed as the first approximation to costs; costs of specific water treatment plants are likely to vary.

From Table 3.4 we can observe that both the MF-UF and High Rate Clarification and Filtration (HRC) drinking water treatment can cost on average 10 cents per cubic meter for a 100 m³ size plant. For surface water s, UV seems to be cheaper than HRC, but direct comparison could be misleading, as a lot of location-specific factors need to be taken into account. (Examples of location-specific factors would be the quality of source water, the presence of color or turbidity, etc.) For UV, some additional costs must be added for suspended solid removal, such as sand filtration , which could add up to 5 cents per cubic meter, and has been included in Table 3.4 and in Fig. 3.3. Ozonation seems to be the most expensive, but of course it can remove more contaminants and goes beyond disinfection. Perhaps this jump in the classes is a nonlinear feature, and therefore the cost per cubic meter increases by an anomalous amount from Class 3 to Class 4.

Fig. 3.3

Estimated cost curves: Class 2 for HRC , Class 3 for UV and MF-UF, Class 4 for Ozonation and Class 5 for a UV-based AOP

Ozone treatment plants⁶ have expanded rapidly in small systems across Saskatchewan and Manitoba in Canada, mostly for surface water sources. By one count, there were about 30 small ozone plants in operation (at the end of 2010). Compared to a UV-based treatment plant, ozonation is more expensive, but nevertheless it is proving to be attractive to a number of smaller communities.

4 Class 5 Treatment Technologies

UV-based advanced oxidation process (AOP) is classified as a Class 5 treatment technology in Table 3.1. Hydrogen peroxide absorbs UV light in order to form free hydroxyl radicals , which aid in breaking down contaminants. A combination of UV-photolysis and UV–Oxidation is therefore used in the treatment process. In Table 3.5 we present the estimated NLLS average cost function for such an AOP.

Table 3.5 Estimated average cost function for UV-Based AOP in 2008 CDN dollars

Details of the NLLS estimate are shown in Appendix B. We included an additional cost for filtration for surface water s for this AOP of 5 cents per cubic meter in the predicted costs in Table 3.5. We estimate that Class 5 treatment can cost $0.21 per cubic meter for a small plant with a daily capacity of 100 m³. Note that our statistical modeling estimation, based on data supplied by manufacturers, indicates that this Advanced Oxidation Process is cheaper than ozonation and will remove a number of micro-pollutants (see description in Table 3.3). When plant-specific costs are taken into account, our information indicates that a representative plant at a scale of 3800 cubic meters per day would cost around $0.45 per cubic meter (in 2008 Canadian dollars).⁷

We hasten to add that our cost estimation models yield what we can call “first approximation costs” and what is the most appropriate technology will depend on site-specific (i.e. the particular location) factors. It is best left to the consulting engineers to do a thorough cost estimation for specific sites.

5 Reverse Osmosis and Nanofiltration (Class 6)

Reverse Osmosis and Nano Filtration , which can also remove salinity, is classified as Class 6. Dore (2005) shows that for a flow rate of 5,000 cubic meters per day, the cost of producing drinking water was US $0.50 per cubic meter per day in 2005. In a later article, Fritzmann et al. (2007) put the costs at actual desalination plants to be between US$0.48 and $0.53 cents. Finally, in a comprehensive review of the cost of desalination literature, Karagiannis and Soldatos (2008) show that for capacities between 500 and 1,000 m³, RO costs range from US$0.75 cents to $3.93 per m³ per day. For capacities less than 1000 m³, they find that the costs range from US $2.22 to as much as $19 per m³ per day. All authors mentioned here recognize the importance of economies of scale in the determination of unit costs. To some extent, RO with granular activated carbon is a “state-of-the-art” technology, mainly suitable for large systems, and so RO and other such technologies are considered in greater detail in Chap. 4.

We can also compare the above cost data with the costs of a Point-of-use (POU) Reverse osmosis system. POU costs range from 2.5 to 5 cents per liter or $25 to $50 per cubic meter. These are obviously expensive technologies and possibly not suitable for small water systems.⁸

6 Examples of Actual Costs of a Few Existing Plants

In this section, we present costs and flow rates at some existing water treatment plants in select small communities in British Columbia (BC), Canada. As before, the costs are made up as follows: (1) capital costs , amortized over a 20-year period, and (2) O&M costs, that include labor, materials, and energy costs for given flow rates. Some of these plants are managed by private corporations as operators, and therefore include their profit markup. The cost information was obtained from the managers of these water treatment plants.

Table 3.6 shows the Class and flow rate as well as its associated average operating cost per cubic meter per day. The largest flow rate plant analyzed here produces the least expensive drinking water (compared to other facilities in the same province) at $0.39 per cubic meter per day. The plant that provides the most costly drinking water also has one of the lowest flow rates.

Table 3.6 Some examples of existing small water treatment facilities in BC for 2008

Using the actual data from these select small systems in BC, we estimate various cost functions for different classes of technology. Note that for Class 1, for some of these communities, the costs reflect (a) profit markup for private sector management, (b) higher transportation costs of hazardous materials such as chlorine and (c) higher transportation costs due to remoteness. These privately managed water systems have costs that include a 100 percent markup on labor costs. We estimated the average cost functions based on the NLLS estimation procedure (see Table 3.7).

Table 3.7 Examples of estimated average cost functions for BC small systems in 2008 CDN dollars for three capacity levels

Details of the NLLS estimation are shown in Appendix C. Costs shown above for Class 1 are operating costs for treatment only. Class 1 plants with a daily flow rate of 100 m³ can produce drinking water at an average cost of $1.00, while the cost is almost quadrupled for a similar sized plant producing Class 4 drinking water on an island off the coast of British Columbia.

7 Summing up and Tentative Conclusions

We can now show, in Figs. 3.3 and 3.4, that with the estimated cost functions, we can reproduce an actual set of cost functions that can then be compared to the hypothetical Fig. 3.1. Figure 3.3 shows the estimated cost curves based on manufacturers’ rated costs, while Fig. 3.4 shows the estimated cost curves based on a sample of small systems in BC.

Fig. 3.4

Estimated cost curves: Classes 1, 2 and 4 for BC Small Systems

Figure 3.3 indicates that ozone technology, a Class 4 water treatment, is more expensive than the Class 3 ( UV and MF-UF) and Class 2 (HRC) treatment types. Class 3 treatments MF-UF and UV seem to be cheaper than HRC for plants which produce less than 100 m³ of water per day and all the way up to 500 m³/day, even though HRC is a Class 2 water treatment process. But in general Fig. 3.4 suggests that the higher the Class of water treatment is, the higher will be the average costs per cubic meter for the sample of small systems in BC.

We observe that the average cost per cubic meter of the statistically estimated equations given above do not conform exactly to the hypothetical Fig. 3.1, but exhibit the nonlinearities that we expected. Another nonlinearity may be the cost of moving from one technology to another, especially when there has been a long-term commitment to a particular technology.

It is possible that older small systems continue to use higher cost older technologies as there is no incentive to modernize in the public sector. In other words, there are technologies currently available in the market that can provide higher contaminant removal at a much lower cost per cubic meter. Hence, we find that a technology, which can provide Class 3 and 4 water treatment, shows lower average cost per cubic meter than a small system, which is only providing Class 1 and 2 water treatments. Another possible reason is that there are site-specific costs that can contribute to the gap in the costs functions between technology and actual existing systems that are in the same class. For example, many of the small systems in BC mentioned above have higher transportation costs due to remoteness and the handling of hazardous materials such as chlorine. However, site-specific costs alone cannot account for this very large gap. We observe that some treatment classes at lower flow rates dominate in terms of cost-effectiveness. Class 3 MF-UF and UV provide water treatment at a much lower cost per cubic meter than BC small systems Classes 1 and 2 between output flow rates of 50 to 200 m³ per day; but at higher flow rates this gap tends to decrease. Finally, the cost per unit for these existing BC small systems is high compared to the rated costs because the systems are privately owned and costs include a markup for profit.

Before we summarize the conclusions, we need to distinguish between systems that use groundwater as the source and systems that use surface water as the source. Most of the above analysis is concerned with surface water as the source for water treatment plants.

Based on Figs. 3.3 and 3.4 and the results presented in the previous sections, we provide the following tentative conclusions:

• 1. The estimated cost curves show that small systems could achieve a higher removal of contaminants at a lower cost than their currently used technology;

• 2. A small publicly owned system could get Class 2 and 3 water treatment if they use HRC or MF-UF for about 9 to 11 cents respectively, provided the flow rate is 100 m³ per day;

• 3. For systems using surface water , UV appears to be the least expensive for small systems at only 7 cents⁹ per m³ for a plant with capacity of 100 m³ for Class 3, which shows that the competitive advantage remains even when costs of sediment removal are included. We would argue that where primary disinfection is absolutely necessary, UV would compare favorably with chlorine for primary disinfection. Of course in North America, for the distribution system the law requires chlorine residual, and perhaps that is why many small systems continue to rely on chlorine as a primary disinfection for surface water systems. The concern over disinfection byproducts (DBPs) might tip the scale in favor of UV for primary disinfection. But again site-specific considerations need to be taken into account. Furthermore, when the source water is groundwater, which is otherwise free of contaminants , the only cost is the cost of residual chlorine for the distribution system. In this case, chlorine may be cheaper than UV .

• 4. If a community is concerned with the removal of micro-pollutants , then a UV-based Advanced Oxidation Process would be cheaper than ozonation , provided the flow rate is not too small. (For example, the City of Cornwall in Canada uses AOP for 2 months of the year for taste and odor issues.)

• 5. Our results indicate that ozonation is competitive (2008 CDN $), and so there are number of ozonation plants in Saskatchewan and Manitoba. We estimate that at the beginning of 2011, there were 30 small systems using this technology in the two provinces (Table 3.8).

• 6. In general, manufacturers’ rated costs tend to be lower than actual plant-level average costs as they do not include some plant-specific costs, such as higher labor, energy, and transportation costs due to remoteness from large urban areas (Table 3.9).

• 7. It should be noted that some of the estimations are based on limited data. Needless to add that the costs estimates cannot be treated for predictive purposes, as all useful predicted costs must also take into account a number of location-specific costs (Table 3.10).

Our general conclusion is that while any specific water treatment facility will need to take account of raw source water quality , the actual target quality for small systems seems to be to meet only the minimum regulatory requirements. Our results show that for surface water , unless the raw water is high in color and in turbidity, a UV -based plant would be economical and cost-effective even when the additional cost of sediment removal is added. This conclusion is especially true for small plants producing less than 100 cubic meters per day. Such a plant could obtain the same or better quality water with UV for less than 8 cents per cubic meter per day. Our finding of the cost-effectiveness of UV is in agreement with USEPA (1996), Gadgil (1998) and Parrotta and Bekdash (1998) .

Appendices

Appendix A

Estimation results for Table 3.4 based on the model: $y_{i} = \beta_{1} X_{i}^{{\beta_{2} }} + \varepsilon_{i}$ where $y_{i}$ is the average cost per cubic meter, $X_{i}$ is the capacity in cubic meters and $\varepsilon_{i}$ is the error term, which satisfies the standard Gaussian assumptions

Table 3.8 Estimated regression coefficients

Appendix B

Estimation results for Table 3.5 based on the model: $y_{i} = \beta_{1} X_{i}^{{\beta_{2} }} + \varepsilon_{i}$ where $y_{i}$ is the average cost per cubic meter, $X_{i}$ is the capacity in cubic meters and $\varepsilon_{i}$ is the error term, which satisfies the standard Gaussian assumptions

Table 3.9 Estimated regression coefficients

Appendix C

Estimation results for Table 3.7 based on the model: $y_{i} = \beta_{1} X_{i}^{{\beta_{2} }} + \varepsilon_{i}$ where $y_{i}$ is the average cost per cubic meter, $X_{i}$ is the capacity in cubic meters and $\varepsilon_{i}$ is the error term, which satisfies the standard Gaussian assumptions

Table 3.10 Estimated regression coefficients

Appendix D

4.1 Sources of Data

John Meunier Inc . Water and Wastewater treatment: Actiflo^® ACP package plants. http://www.johnmeunier.com. Accessed 01 October 2011.

Koch Membrane . Municipal Water Ultrafiltration (UF) Systems for Small Communities. http://www.kochmembrane.com/mww_prepacked.html. Accessed 16 May 2011.

Kruger USA . Actiflo^® for Drinking Water Treatment. http://www.krugerusa.com/en/files/16362.htm. Accessed 16 May 2011.

Lenntech. Micro filtration System (MFS): Membrane Technology. http://www.lenntech.com/microfiltration.htm. Accessed 16 May 2011.

Mainstream Water Solutions Inc Municipal Systems. http://www.mainstreamwater.com/topmenu_municipal.html#1. Accessed 16 May 2011.

Siemens. MEMCOR^® CP - Ultrafiltration System. http://www.water.siemens.com/en/products/membrane_filtration_separation/ultrafiltration_membrane_systems/Pages/memcor_product_cp_membrane_system.aspx. Accessed 13 May 2011.

TrojanUV Technologies. TrojanUVSwiftSC. http://www.trojanuv.com/solutions/municipal/drinkingwater/products/trojanuvswiftsc. Accessed 17 May 2011.