The individual cost contributions to the mechanical components of a small membrane bioreactor (MBR) (100–2,500 m3/d flow capacity) are itemised and collated to generate overall capital and operating costs (CAPEX and OPEX) as a function of size. The outcomes are compared to those from previously published detailed cost studies provided for both very small containerised plants (<40 m3/day capacity) and larger municipal plants (2,200–19,000 m3/d). Cost curves, as a function of flow capacity, determined for OPEX, CAPEX and net present value (NPV) based on the heuristic data used indicate a logarithmic function for OPEX and a power-based one for the CAPEX. OPEX correlations were in good quantitative agreement with those reported in the literature. Disparities in the calculated CAPEX trend compared with reported data were attributed to differences in assumptions concerning cost contributions. More reasonable agreement was obtained with the reported membrane separation component CAPEX data from published studies. The heuristic approach taken appears appropriate for small-scale MBRs with minimal costs associated with installation. An overall relationship of net present value = (a tb)Q(−c lnt+d) was determined for the net present value where a = 1.265, b = 0.44, c = 0.00385 and d = 0.868 according to the dataset employed for the analysis.

## INTRODUCTION

Immersed membrane bioreactor (iMBR) plants may either be bespoke designed or be provided as a containerised technology with only relatively minor modifications on site. All large municipal plants are designed bespoke, and the costs incurred categorised as either part of the expenditure for operation (OPEX), the cost per treated product volume, or capital (CAPEX).

There is generally little information in the peer-reviewed literature on costs, but rather more on energy demand, a key contributor to water and wastewater treatment OPEX generally and aeration energy specifically. A number of MBR energy analyses from 2010 onwards (Verrecht et al. 2010; Liu et al. 2012; Krzeminski et al. 2012; Itokawa et al. 2014; Gabarrón et al. 2014) have demonstrated the significance of operating at full capacity to maximise the flux and minimise the specific aeration demand (SADp) – the ratio of the air demanded for membrane scouring to the permeate generated. Other studies have demonstrated the benefit of operating at low solids concentrations, which reduces energy for both mixing (Fenu et al. 2012) and biological (or process) aeration (Schaller et al. 2010; Henkel et al. 2011; Rodríguez et al. 2012). Energy-based papers have tended to illustrate the classic balance between OPEX and CAPEX (operating and capital expenditure) when considering TOTEX (total expenditure), particularly with reference to equalisation (Verrecht et al. 2010; Gabarrón et al. 2014) and waste sludge production and management (Schaller et al. 2010).

When considering TOTEX for the installation and operation over the entire life cycle of an installation, other factors besides energy can contribute significantly to overall cost. In such cases, it is appropriate to calculate the net present value (NPV), to account for the change in base utility and service costs, the value of the asset, and the value of money itself over the course of the plant life. The few reports in peer-reviewed literature of either TOTEX and/or NPV have tended to corroborate the importance of energy demand in determining OPEX. The energy demand appears to make up 27–34% of the total OPEX for a large (19–48 MLD, or megalitres/day, flow capacity) municipal MBR (DeCarolis et al. 2007; Brepols et al. 2010; Young et al. 2013, 2014). Contributions from individual OPEX items can vary significantly according to assumptions made (Figure 1) but the overall calculated 2015-corrected NPVs for the MBR for the analyses from these studies, based on a 19 MLD plant, are comparable at $60–80 million. Figure 1 Breakdown of percentage OPEX contributions, published papers. Figure 1 Breakdown of percentage OPEX contributions, published papers. While these heuristic studies provide representative data, they are limited to specific installations/scenarios or else are aspecific in itemisation: cost contributors are not always consistently identified, and the impact of specific items varies according to the circumstances of the site. For example, waste sludge disposal is not identified as a single cost item in the reports by some authors, yet this item contributes 21% of the OPEX at Nordkanal in Germany (Brepols et al. 2010); sludge management costs have been postulated as being comparable to aeration costs (Schaller et al. 2010). The Nordkanal analysis, conversely, excludes membrane replacement – which makes up 28% of the OPEX in the DeCarolis et al. (2007) study (Figure 1). Moreover, these costs are all based on relatively large flows (19,000–48,000 m3/d). There have been few full cost studies based on small-scale MBR systems, and those that have appeared have either been largely limited to OPEX (Verrecht et al. 2012) or have been based on small containerised systems (Fletcher et al. 2007). For small-scale MBRs, including containerised plants, the onus is on design simplicity and minimal manual intervention since the associated labour costs significantly add to TOTEX (Verrecht et al. 2012). Optimisation based on energy demand, as applied to large-scale systems, is less appropriate than system robustness in such cases. It is of interest to assess the costs associated with the construction, installation and operation of a small to medium-sized bespoke flat sheet (FS) MBR typically applied to small municipal wastewater flows, since these make up the vast majority of MBR installations globally. This paper aims to provide a detailed and consistent cost analysis for such MBR installations focusing on the CAPEX of individually itemised components and OPEX determined on a consistent basis. Three different flow capacities provide CAPEX and OPEX cost curves. ## METHODOLOGY The cost analysis proceeded by itemising each individual component (Table 1) of a standard treatment scheme (Figure 2) for an MLE (modified Ludzack-Ettinger) process. All components were individually specified, ostensibly in terms of size, flow capacity and power rating, and priced from supplier information or based on proprietary data with reference to the geographical area of Southern China. Table 1 MBR plant components and design basis Category Component(s) ID Description/purpose Note Tanks Raw water T1 Storage tank for inlet wastewater Primary sedimentation T2 Removal of gross, settleable solids Equalisation (EQ) T3 Equalisation of flow Anoxic (Ax) T4 Denitrification Aeration (Ae) T5 Nitrification and biological oxidation Membrane T6 Membrane separation Treated water T7 Storage of permeate water Sludge T8 Storage of wasted sludge Chemicals storage T9,10 Pumps Settled sludge transfer P1 Submerged, settled sludge to sludge storage tank Feed P2 EQ tank through rotary screen Permeate P3 Self-priming, membrane suction filtration Sludge return/discharge P4 Submerged, sludge recirculation and excess Sludge transfer P5 WAS to dewatering Chemicals P6,7 Cleaning chemicals transfer to membrane, x2 Blower Process B1 Biological process aeration Membrane B2 Membrane scouring Mixer EQ tank mixer X1 High speed, equalisation tank Ax tank mixer X2 Low speed, anoxic tank Screen Rotary screen S1 Fine screening of feed Membrane Membrane module M1 FS membrane plus frame with built-in aerator 10 Diffusers Fine bubble diffuser D1 Process aeration Coarse bubble diffuser D2 Membrane aeration Category Component(s) ID Description/purpose Note Tanks Raw water T1 Storage tank for inlet wastewater Primary sedimentation T2 Removal of gross, settleable solids Equalisation (EQ) T3 Equalisation of flow Anoxic (Ax) T4 Denitrification Aeration (Ae) T5 Nitrification and biological oxidation Membrane T6 Membrane separation Treated water T7 Storage of permeate water Sludge T8 Storage of wasted sludge Chemicals storage T9,10 Pumps Settled sludge transfer P1 Submerged, settled sludge to sludge storage tank Feed P2 EQ tank through rotary screen Permeate P3 Self-priming, membrane suction filtration Sludge return/discharge P4 Submerged, sludge recirculation and excess Sludge transfer P5 WAS to dewatering Chemicals P6,7 Cleaning chemicals transfer to membrane, x2 Blower Process B1 Biological process aeration Membrane B2 Membrane scouring Mixer EQ tank mixer X1 High speed, equalisation tank Ax tank mixer X2 Low speed, anoxic tank Screen Rotary screen S1 Fine screening of feed Membrane Membrane module M1 FS membrane plus frame with built-in aerator 10 Diffusers Fine bubble diffuser D1 Process aeration Coarse bubble diffuser D2 Membrane aeration 1 0.48–0.58 hours hydraulic residence time (HRT). 2 6.4–6.5 hours HRT. 3 8.1–8.2 hours HRT total. 4 Flow capacity of 125% of feed flow. 5 Flow capacity of 250% of feed flow (Recycle ratio = 2.5Q). 6 Flow capacity of 10–12% of feed flow. 7 10 L/min, applied weekly for 120 minutes total. 8 Process aeration rate SADp,bio = 11.5 Nm3/m3; aeration energy = 0.017–0.024 kWh/Nm3. 9 Membrane air scour rate SADp = 26 Nm3/m3; aeration energy = 0.015–0.022 kWh/Nm3. 10 Flux of 0.5 m/d. Figure 2 Plant schematic; see Table 1 for codes. Figure 2 Plant schematic; see Table 1 for codes. The basis of the design is included in the Table 1 footnotes, with other key cost assumptions provided in Table 2. A design value of 26 Nm3 air per m3 permeate for membrane air scouring (SADp) was used, a significantly higher value than those normally associated with large-scale wastewater treatment (SADp ∼ 10 Nm3/m3, Judd (2014)). The high SADp required for the smaller plants considered in the current analysis reflects a more conservative design required to minimise the risk of process upset and unscheduled manual intervention. The selected design process aeration rate (for sustaining biological treatment) of 11.5 Nm3/m3 assumes intermittent operation of the process blower for moderate strength municipal wastewaters, based on a dissolved oxygen concentration set-point concentration of 1–2 mg/L. Table 2 Key cost components Cost component Value (range) Electrical energy cost LE$/kWh 0.1
Membrane cost LM $/m2 79.16 Chemicals cost LC$/m3 0.0091
Membrane life tmem
Plant life t  5–20
Margin, proportion of equipment purchase costs 15%
Assembly, proportion of CAPEX 20%
Discount rate 5%
Labour
Annual salary of operator/maintenance engineer   $/y 15,167 O/h on salary (incl. employer tax & pension contrib.) 30% No. working hours per year hrs/y 2,700 Proportion of working day spent maintaining MBR 15% Flow, Q m3/d 100 500 2,500 Number of staff Labour cost LL$/m3 0.061 0.024 0.012
Contribution of labour costs to total 14.0% 5.6% 2.8%
Cost component     Value (range)
Electrical energy cost LE $/kWh 0.1 Membrane cost LM$/m2 79.16
Chemicals cost LC $/m3 0.0091 Membrane life tmem Plant life t 5–20 Margin, proportion of equipment purchase costs 15% Assembly, proportion of CAPEX 20% Discount rate 5% Labour Annual salary of operator/maintenance engineer$/y 15,167
O/h on salary (incl. employer tax & pension contrib.) 30%
No. working hours per year hrs/y 2,700
Proportion of working day spent maintaining MBR 15%
Flow, Q m3/d 100 500 2,500
Number of staff
Labour cost LL $/m3 0.061 0.024 0.012 Contribution of labour costs to total 14.0% 5.6% 2.8% Three different plant capacities were considered: 100, 500 and 2,500 m3/d. Specifications for the categorised components listed in Table 1 (tanks T1–T8, pumps P1–P7, blowers B1 and B2, mixers X1 and X2, screen S1, membrane M1, and diffuser D1) were used to obtain itemised costs provided by regional suppliers in Taiwan and Southern China. Costs within each category were then summed for further analysis of CAPEX trends. The OPEX in$ per m3 permeate was determined according to the combined cost of energy demand, critical component replacement, chemical consumption, waste management/disposal and labour (Judd 2014):
1
where LE is the cost of electrical energy in $/kWh, LM the membrane cost per m2 membrane area, J the flux in m/d and tmem the membrane life in years. LC, LW and LL represent the specific costs per m3 treated water for chemicals consumption, waste disposal and labour, with all these other than waste disposal included in the current analysis. The total specific energy demand Etot in kWh/m3 is provided by the sum of the power consumption rating W in kW of the individual component, the (estimated) fractional operational period τ (hours per day of operation divided by 24) and the permeate flow Qp in m3/d: 2 CAPEX and OPEX can be combined to estimate the averaged NPV according to the simplified relationship (Verrecht et al. 2010): 3 where i the annual discount rate and t the amortisation period (or plant life) in years. The above equation allows for inflation and assumes zero residual value of the plant at the end of its life. All costs for the literature data were converted to US$ based on annual retail price index data.

## RESULTS AND DISCUSSION

The outcomes of the CAPEX analysis reveal that 70–94% of the capital equipment cost is taken up by the membranes and diffusers, the proportion increasing with plant size (Figure 3(a)). A consideration of energy demand reveals the proportional energy contribution of screening and stirring to decrease with size (Figure 3(b)). This arises from the practical limitation of sizing these items for the smallest installation: the same mixer and screen were fitted for both the 100 and 500 m3/d plants, such that the equipment for the smallest plant is oversized.

Figure 3

(a) Capital cost and (b) energy demand contributions at the three different flow capacities.

Figure 3

(a) Capital cost and (b) energy demand contributions at the three different flow capacities.

Correlations for absolute costs (Figure 4) reveal the expected decrease in specific cost (cost per unit product water) with plant capacity. According to these data, based on the available cost data and conditions outlined in Tables 1 and 2, cost trends follow the relationships with flow capacity Q:
4

5
Figure 4

Cost and energy data at the three different flow capacities.

Figure 4

Cost and energy data at the three different flow capacities.

The exponent value for the CAPEX equation (Equation (4)) approaches unity, reflecting the expected low economy of scale associated with a modular treatment process and the dominant contribution of the membrane assembly to the overall equipment costs. The decrease in OPEX with capacity arises primarily from the oversizing of components at the smallest scales.

Extrapolating these trends to the higher flows on which analyses conducted by previous authors were based permits comparison with the mean reported published data (Figure 5). Data from such studies relate to the membrane and ancillary capital equipment component specifically, this being considered most pertinent to the current study. Results show the OPEX trend to be in good agreement with the mean data from the identified studies. The sharp increase in OPEX at low flows reflects trends reported by other authors (Fletcher et al. 2007; Brepols et al. 2010; Verrecht et al. 2012). The extent of this increase is highly dependent on values assigned to key contributors such as peak loading factor, electrical and labour costs, and the degree of manual intervention.

Figure 5

Cost trends (Equations (3) and (4)) against reported data for MBR OPEX and membrane component CAPEX.

Figure 5

Cost trends (Equations (3) and (4)) against reported data for MBR OPEX and membrane component CAPEX.

Cost estimates at low flows are challenged by variations in the degree of contingency assumed. The peak loading factors and the associated oversizing demanded at low flows have led to recorded mean specific energy demand values of between 1.8 kWh/m3 (Brepols et al. 2010) to as high as 10 kWh/m3 (Verrecht et al. 2012; Itokawa et al. 2014). For the Verrecht et al. study, labour costs were calculated to contribute more than half the operating cost. Moreover, a study of 84 municipal MBR installations between 2,000 and 22,000 m3/d capacity revealed the number of FTE (full time equivalent) operators per 1,000 m3/d of treated wastewater to vary between 0.3 and 2.5 (Mathis 2011), the ratio increasing with decreasing flow capacity. The corresponding range of values adopted for the current study is 0.3 FTE per 1,000 m3/d at the highest flow considered to 1.5 FTE at the lowest.

OPEX estimates are subject to increased imprecision at low flows (<500 m3/d). However, given a consistent set of assumed values for the key OPEX components in Equation (1), the trend above 500 m3/d flow capacity appears to be reasonably representative. Trends in CAPEX, conversely, are more vagarious – primarily due to the different scope and itemisation of costs in published studies combined with errors introduced by overly simplistic normalisation (Newnan et al. 2013). The projected CAPEX of a 19,000 m3/d plant is around $6 million according to the current study. This figure is 86% of the mean membrane equipment cost determined by DeCarolis et al. (2007), and around half that of the corresponding cost calculated by Young et al. (2013). The only other cost study detailing individual components and based on low flows, in this case employing containerised MBR systems, reported equivalent CAPEX figures (in 2015$) of around 49–62k USD equipment costs at the maximum flow of 40 m3/d considered in their study (Fletcher et al. 2007). This is comparable to the CAPEX of $59k determined for the smallest plant of 100 m3/d considered in the current study. Moreover, the CAPEX figure for the membrane component of the cost quoted by Brepols et al. (2010) for a 10,000 population equivalent plant (assumed to equate to a flow of 2,500 m3/d) was slightly lower than the projected value based on Equation (3) (Figure 5). The mean NPV (Figure 6(a)) determined at each flow capacity follows the general trend with plant life t (with an R2 value of > 0.985 in all cases): 6 where a and b are empirical constants and the exponent b varies between 0.419, for the largest plant, and 0.406 for the smallest. Plotting NPV vs. Q (in m3/d) at plant life values between 1 and 20 years generates a series of cost curves (Figure 6(b)) following the same exponential form as above (i.e. NPV = mQn). Simple curve fitting of the coefficient and exponent values for these cost curves provides the following overall relationship for NPV as a function of flow capacity Q and plant life t in years: 7 Figure 6 NPV vs. (a) plant life at the three different flow capacities and (b) flow capacity at five different plant life values. Figure 6 NPV vs. (a) plant life at the three different flow capacities and (b) flow capacity at five different plant life values. Thus, according to the cost data in Table 2 and based on the generalised OPEX equation (Equation (1)): • the plant NPV follows a power law relationship with both flow and plant life; • the coefficient of the power law relationship between NPV and flow Q follows an approximate square root relationship with plant life t; and • the exponent value for the above NPV vs. Q equation decays logarithmically with plant life. There appear to have been no cost functions previously published for MBR installations. However, Equation (7) is consistent with the general expectation of decreased specific costs (cost per unit volume water treated) with both flow capacity and plant life. The quantitative NPV of$22 million USD determined at Q = 19,000 m3/d and a 20y plant life is considerably lower than the $55–80 million range calculated by previous authors for the same conditions since it is ostensibly limited to the membrane and equipment component alone. Installation and other site-specific costs outside of those directly relating to capital equipment (e.g. civil, mechanical and electrical) make up a more significant proportion of the costs for large-scale municipal installations. ## CONCLUSIONS An analysis of the cost of a small membrane bioreactor, 500–2,500 m3/h capacity and based on flat sheet membrane technology, has been conducted based on individual itemisation encompassing both purchase cost and energy demand. The analysis provides a cost curve, the cost as a function of flow and plant life, and reveals: • CAPEX to increase with flow Q according to a power law relationship with an exponent value of 0.87; • OPEX (per m3 treated water) to decrease logarithmically with flow, with an intercept of ∼ 0.45; • NPV (net present value in$) to increase with flow Q in m3/d and plant life t in years according to the equation:
where, according to the dataset employed, a = 1.265, b = 0.44, c = 0.0039 and d = 0.87.

The trend determined for OPEX was in good agreement with that reported from three other studies of larger-scale plants. CAPEX and NPV were both significantly lower, however, since these parameters are largely associated with the MBR technology per se rather than ancillary installation costs (such as civil, mechanical and electrical, and miscellaneous site work). It is concluded that this heuristic approach is appropriate to small-scale installations (≲5,000 m3/d) and/or the equipment component of larger MBRs, but requires modification for larger-scale installations.

## ACKNOWLEDGEMENT

The authors are grateful to Ecologix Technologies Asia Pacific, Inc., Taiwan, for providing technical and cost data for this article.

## REFERENCES

REFERENCES
Brepols
B.
Schäfer
H.
Engelhardt
N.
2010
.
Water Sci. Technol.
61
(
10
),
2461
2468
.
DeCarolis
J.
S.
Pearce
W. R.
Hirani
Z.
Lacy
S.
Stephenson
R.
2007
Cost trends of MBR systems for municipal wastewater treatment
. In:
Proceedings of the Water Environment Federation
,
13–17 October
,
San Diego
, pp.
3407
3418
.
Fenu
A.
De Wilde
W.
Gaertner
M.
Weemaes
M.
de, Gueldre
G.
Van De Steene
B.
2012
.
J. Membrane Sci
.
421–422
,
349
354
.
Fletcher
H.
Mackley
T.
Judd
S.
2007
.
Water Res
.
41
,
2627
2635
.
Gabarrón
S.
Ferrero
G.
Dalmau
M.
Comas
J.
Rodriguez-Roda
I.
2014
.
J. Environ. Management
134
,
8
14
.
Henkel
J.
Cornel
P.
Wagner
M.
2011
.
Water Sci. Technol
.
63
(
2
),
3034
3038
.
Itokawa
H.
Tsuji
K.
Yamashita
K.
Hashimoto
T.
2014
.
Water Sci. Technol
.
69
(
5
),
1088
1093
.
Judd
S.
2014
Industrial MBRs
,
Judd and Judd Ltd
,
Cranfield
,
UK
.
Krzeminski
P.
van der Graaf
J. H. J. M.
van Lier
J. B.
2012
.
Water Sci. Technol
.
65
,
380
392.
Liu
H. Y.
Freeman
B.
Sunano
S.
Munehiro
N.
Bartels
C.
Oda
Y.
2012
.
Water Practice Technol
.
7
(
4
),
1
7
.
Mathis
K.
2011
MBRs for cost-effective wastewater treatment. Water and Wastes Digest, 14 October 2011.
Newnan
D. G.
Lavelle
J. P.
Eschenbach
T. G.
2013
Engineering Economic Analysis
,
12th edn
.
Oxford University Press
,
USA
.
Rodríguez
F. A.
Reboleiro-Rivas
P.
Osorioa
F.
Martínez-Toledo
M. V.
Hontoria
E.
Poyatos
J. M.
2012
.
Biochem. Eng. J.
60
,
135
141
.
Schaller
J.
Drews
A.
Kraume
M.
2010
.
Desalination and Water Treatment
18
,
315
320
.
Verrecht
B.
Nopens
I.
Maere
T.
Brepols
C.
Judd
S.
2010
.
Water Res
.
44
,
5274
5283
.
Verrecht
B.
James
C.
Germain
E.
Birks
R.
Barugh
A.
Pearce
P.
Judd
S.
2012
J. Environ. Eng
.
138
,
594
.
Young
T.
Smoot
S.
Peeters
J.
Côté
P.
2013
.
Proc. Water Environment Federation.
8
,
6354
6365
.
Young
T.
Smoot
S.
Peeters
J. J.
Côté
P.
2014
.
Water Practice Technol
.
9
(
3
),
316
323
.