Abstract

Pressure control is an important feature for reducing leakages in water supply systems, and the use of pressure reducing valves has been well established as an efficient option for this purpose. However, several studies have demonstrated that the energy available on such sites could be used to generate electrical energy, instead of being dissipated as head loss; therefore, a more efficient and sustainable solution could be applied for pressure control. Due to the low amount of power available, the use of pumps as turbines (PATs) is highly recommended. However, manufacturers do not provide pump curves operating as turbines, making PAT selection challenging. Different empirical methods can be found in the literature for estimating PAT performance based on the pump operating conditions. Thus, this paper presents a comparative analysis of nine different methods, using real data from 14 pumps. Furthermore, the effectiveness of these methods for PAT selection is evaluated in a hypothetical network.

INTRODUCTION

In water distribution networks (WDNs), pressure is strongly related to the topography, topology, and demand pattern. In order to ensure effective operation, standards set maximum pressure limits to avoid pipe bursts and reduce leakages, and minimum limits to avoid the intrusion of pathogens and poor supply to consumers (Lee & Schwab 2005). Water leakage is one of the most extensively studied problems in WDNs, due to its environmental and economic impact. According to Perdikou et al. (2014), in certain countries, leakages account for more than 40% of the water produced, while in benchmark systems this value is approximately 3%, indicating that major investments are still necessary in order to avoid such waste of water.

At present, pressure reducing valves (PRVs) are being widely used for pressure control. Basically, PRVs operate when the downstream pressure is above a previously set value, increasing head losses through a shutter device. Saldarriaga & Salcedo (2015) developed methods for the optimal location and operation of PRVs, significantly reducing leakages. However, considering the rational use of resources, energy potential is wasted in the operation of PRVs. A microturbine may be used instead in order to produce energy, and different studies have demonstrated the feasibility of this approach (Gallagher et al. 2015). However, the power available on these sites is usually low, which may lead to economic unfeasibility. Therefore, the use of pumps as turbines (PATs) has been recommended, due to their low cost and satisfactory efficiency (Vilanova & Balestieri 2014). Furthermore, their robustness, spare part availability, and maintenance staff experience of WDNs with pumps are advantages of PATs during their life cycle, reducing repair time, that is, producing energy during a longer period.

According to Allen et al. (2008), this form of energy production, with small plants close to load sites, presents various advantages: reduced transmission losses, decreased costs, applicability to isolated communities, environmental friendliness, and increased independence for consumers. However, micro hydropower represents only a fraction of the energy produced by these local plants, with photovoltaic technology being the most extensively used. Despite its highly efficient approach and environmental friendly characteristics, as it exploits wasted energy, due to a lack of governmental support and the conservative behavior of water companies the energy production in WDNs is still incipient (Meirelles 2013). De Marchis et al. (2016) also highlight that return of investment should be carefully studied to avoid unfeasible energy recovery projects. Furthermore, certain technical issues should be solved in order to increase its use. The major point to be observed is the difficulty of selecting an optimal PAT to operate in a network, that is, one that is capable of maintaining adequate pressure, avoids water shortages and excessive leakage, and also produces energy with high efficiency. As a result of the absence of a flow control device in PATs, pressure control is compromised owing to the dynamic behavior of WDNs, as well as the energy production, as there is significant reduction in PAT efficiency outside its Best Efficiency Point (BEP). Freni et al. (2014) also noted that with low flows, the head losses inserted by PATs are insufficient for achieving satisfactory pressure control. In fact, Williams (1995) recommended that PATs should operate above 50% of the flow of their BEP in order to avoid vibration problems. To address this issue, various alternatives have been studied: Carravetta et al. (2012) proposed the use of variable speed as well as using PATs in parallel with PRVs, while Budris (2011) proposed the use of multiple PATs, selected for operation during specific consumption periods. In addition, specific microturbines are being developed to operate under this dynamic scenario of WDNs, avoiding poor efficiencies far from BEP, such as the cross-flow turbine presented by Sinagra et al. (2017).

Despite the numerous advantages and promising studies for improving PAT efficiency, great uncertainty remains in the selection of a machine, as the classical methods available in the literature are empirical. Basically, these methods relate pump operation (available from manufacturers) to turbine operation (achieved by the respective authors), creating coefficients of head and flow that are extrapolated to different machines. Considering the high sensitivity of the technical and economic feasibility of energy recovery projects with PAT efficiency, this study evaluates the accuracy of nine different methods used to estimate PAT performance. The complete curves of 14 pumps, with specific speeds ranging from 0.46 to 4.94, as obtained by Torley & Chaudhry (1996), are used for comparison. Furthermore, the methods are applied to a hypothetical network proposed by Gomes et al. (2011) in order to evaluate the impact of uncertainties in PAT selection on the economic and technical feasibility of the energy recovery project, considering both energy production and leakage reduction.

PAT PERFORMANCE ESTIMATION

The basic procedure for estimating PAT performance involves relating its BEP in both modes, obtaining the correction coefficients of flow, CQ, and head, CH, as shown in Equations (1) and (2): 
formula
(1)
 
formula
(2)
where Qt is the BEP flow as turbine, Qp is the BEP flow as pump, Ht is the BEP head as turbine, and Hp is the BEP head as pump.

These correction coefficients are based on experimental tests, and each PAT performance estimation method includes a procedure for calculating their values. Two major research lines exist: the first uses efficiency in the pump mode to predict its behavior as a turbine, while the second uses the specific speed as a turbine to define its BEP in pump mode. The second approach is more suitable for PAT selection, as the pump BEP is directly obtained, while the first requires an iterative process. Table 1 displays the methods investigated in this study, where ηp is the efficiency of the machine operating as pump, and nst is the specific speed considering its operation as turbine.

Table 1

Correction coefficients for different methods studied

AuthorCHCQSpecific speed range
Based on pump efficiency 
Stepanoff (1957)    40–60b 
Childs (1962)    – 
Sharma (1985)    40–60b 
Schmiedl (1988)    – 
Alatorre-Frenk (1994)    – 
Based on turbine specific speed 
Grover (1980)    10–50b 
Hergt et al. (1982)    – 
Viana (1987) a – – 0.25–1.25c 
Chapallaz et al. (1992) a – – 0.2–2.5c 
AuthorCHCQSpecific speed range
Based on pump efficiency 
Stepanoff (1957)    40–60b 
Childs (1962)    – 
Sharma (1985)    40–60b 
Schmiedl (1988)    – 
Alatorre-Frenk (1994)    – 
Based on turbine specific speed 
Grover (1980)    10–50b 
Hergt et al. (1982)    – 
Viana (1987) a – – 0.25–1.25c 
Chapallaz et al. (1992) a – – 0.2–2.5c 

aGraphical method.

bSpecific speed in turbine mode.

cSpecific speed in pump mode.

In order to evaluate the performance of each method, the complete Suter curves of 14 pumps tested by Torley & Chaudhry (1996) were used. The specific speed ranges from 0.46 and 4.94, which provides solid insight into the methods under different conditions. Certain methods have a limited range, and for these, only reliable pumps were used.

HYDRAULIC MODEL

Epanet (Rossman 2000) was used to simulate the selected PAT and evaluate its hydraulic performance. The PAT was simulated as a General Purpose Valve (GPV), and its head loss curve was created by using the dimensionless data of a similar machine (with a similar specific speed) from the set of 14 curves tested by Torley & Chaudhry (1996). Thus, the PAT curve can be expressed by means of Equation (3): 
formula
(3)
where Hi is the PAT head at time i, Qi is the PAT flow at time i, and a and b the power curve adjustment coefficients.
Leakage was also modeled using emitters on each consumption node. The emitters model a flow through an orifice, discharging to the atmosphere as a function of the pressure at the node, as shown in Equation (4): 
formula
(4)
where Ql is the leakage flow, C is the discharge coefficient, β is the pressure exponent, and pn,i is the pressure at node n and time i. The pressure exponent value was set to 0.5 for all nodes, and the discharge coefficient of each node was randomly defined in order to create a scenario with water losses of approximately 25%, which is a plausible value for most WDNs.

RESULTS AND DISCUSSION

Comparison of methods for PAT performance estimation

Figure 1(a) and 1(b) illustrate the results for the head and flow estimation, respectively. None of the methods evaluated presented effective estimation for all 14 pumps, and in some cases, the error exceeded 100%. On average, the method proposed by Alatorre-Frenk (1994) produced the most effective results, with an average error of 14.3% for head estimation and 28.9% for flow estimation. The methods based on specific speed, which are used for PAT selection, exhibited effective results for flow estimation, but poor accuracy for head estimation. The most suitable method was found to be that proposed by Viana (1987). However, this method is restricted to low specific speeds, which reduces its applicability. The results demonstrate the high uncertainty of these empirical methods, which could significantly affect the technical and economic feasibility of PATs used for pressure control in WDNs.

Figure 1

Comparison of PAT performance estimation for different methods: (a) Head; (b) Flow.

Figure 1

Comparison of PAT performance estimation for different methods: (a) Head; (b) Flow.

Effectiveness in a WDN

The hypothetical network presented by Gomes et al. (2011) was used to evaluate the performance of the methods based on specific speed, which are widely used for PAT selection. The PAT was installed directly after the reservoir, as shown in Figure 2.

Figure 2

Network layout and demand pattern.

Figure 2

Network layout and demand pattern.

Considering the network demand pattern, the PAT was selected for maximum consumption, which results in maximum energy production. The available head is defined as the difference between the pressure at the critical node (node 10) during the maximum consumption period and the minimum pressure allowed in the network, which in this case is set to 10 m. Therefore, it is guaranteed that all nodes will experience at least the minimum pressure during the 24 h period of simulation, since the maximum head loss produced by the PAT occurs only during maximum consumption. Using the available head and flow (Ht and Qt), the four methods based on specific speed were applied to select a PAT for this condition, and a hydraulic simulation with the PAT installed was created in order to evaluate the pressure control and energy production for each case. Figure 3 shows the pressure at the critical node for each case during the 24 h period. An improvement in pressure control is observed for all cases, except for the high pressure observed during low consumption periods, which was expected as the PAT was selected for maximum consumption. Therefore, the head loss produced for low flows is insignificant. Differences can clearly be observed among the methods with an increase in consumption. The method proposed by Viana (1987) exhibited the most effective results, achieving the minimum allowed pressure during maximum consumption. Table 2 displays the energy production and leakage reduction for each method. Considering the two best results (Viana 1987 and Chapallaz et al. 1992), the difference in energy production is 23.4%, which may significantly affect the economic feasibility of the project. Even so, comparing the results presented in Table 2, the benefit obtained with a PRV, only due to leakage reduction, is greater than the combined benefit obtained using PATs, showing that its application should be carefully studied, especially in WDNs with a high leakage index. The use of PATs and PRVs in parallel can also be a good strategy, extracting the benefits of both technologies. It is also necessary to highlight the uncertainty resulting from the schematization of PAT characteristic curves in the hydraulic model and the use of information of pumps with similar specific speeds, which can also be an error source for the feasibility of the project.

Table 2

Energy production and leakage reduction for each case studied

Without controlGrover (1980) Hergt et al. (1982) Viana (1987) Chapallaz et al. (1992) PRV
Leakage [m3/dia] 719 698 665 637 658 589 
Difference [%] – 2.9 7.5 11.5 8.5 18.1 
Energy [kWh/dia] – 17.2 44.6 65.3 50.0 – 
Energy savings [$/year] – 3,765 9,774 14,307 10,948 – 
Leakage savings [$/year] – 15,330 39,420 59,860 44,530 94,900 
Total benefit [$/year] – 19,095 49,194 74,167 55,478 94,900 
Without controlGrover (1980) Hergt et al. (1982) Viana (1987) Chapallaz et al. (1992) PRV
Leakage [m3/dia] 719 698 665 637 658 589 
Difference [%] – 2.9 7.5 11.5 8.5 18.1 
Energy [kWh/dia] – 17.2 44.6 65.3 50.0 – 
Energy savings [$/year] – 3,765 9,774 14,307 10,948 – 
Leakage savings [$/year] – 15,330 39,420 59,860 44,530 94,900 
Total benefit [$/year] – 19,095 49,194 74,167 55,478 94,900 
Figure 3

Pressure at critical node for each case studied.

Figure 3

Pressure at critical node for each case studied.

CONCLUSIONS

This study demonstrates the accuracy of nine different methods for estimating pump performance operating as turbine. None of the methods presented consistent results for all 14 pumps that were used for comparison. The most effective method found, namely that proposed by Alatorre-Frenk (1994), still exhibited a high average uncertainty: 14.3% for head estimation and 28.9% for flow estimation. In addition to the uncertainty observed, the specific speed range restriction reduces the applicability of the methods based on specific speed used for PAT selection, with the exception of the method proposed by Chapallaz et al. (1992). When these methods were used to select a PAT to operate in a WDN, differences could be clearly observed. Despite the pressure control improvements, only the PAT selected using the method proposed by Viana (1987) could achieve minimum pressure. Even so, in the example studied, the available head and flow were suitable for this method, and as discussed previously, the selected PAT could be inefficient for different conditions. The differences observed are reflected in energy production, which may significantly affect the economic feasibility of the project, particularly for sites with reduced energy potential, where the cost/benefit ratio is reduced by the scale factor. However, due to the dynamic operation of a WDN, the pressure remains high during low consumption periods. This behavior require a detailed study of the benefits obtained using PATs, which can be surpassed by the leakage reduction produced with PRVs, to determine the best alternative for pressure control. The use of PATs and PRVs can be an interesting option to maximize benefits. In addition, WDNs with high leakage can lead to a specific PAT selection. During its operation, pressure drop will reduce leakage, and consequently the PAT flow, reducing its efficiency and energy production. Therefore, PAT selection in these cases should be performed taking this behavior into consideration, as proposed by Carravetta et al. (2012) and Meirelles et al. (2017). Thus, the presented methods, which are simpler, may be used only for a pre-project phase, as well as for estimating the energy production and necessary investments, while detailed selection and laboratory tests should be carried out for the final project.

REFERENCES

REFERENCES
Alatorre-Frenk
C.
1994
Cost Minimization in Micro Hydro Systems Using Pumps as Turbines
.
PhD Thesis
,
University of Warwick
.
Allen
S. R.
,
Hammond
G. P.
&
McManus
M. C.
2008
Prospects for and barriers to domestic micro-generation: a United Kingdom perspective
.
Applied Energy
85
(
6
),
528
544
.
Budris
A. R.
2011
Multiple ‘pump as turbine’ installations keep efficiency high over wide flow range
.
Vol. 67
,
Water World
,
Plymouth, MN
.
Carravetta
A.
,
Del Giudice
G.
,
Fecarotta
O.
&
Ramos
H. M.
2012
Energy production in water distribution networks: a PAT design strategy
.
Water Resources Management
26
(
13
),
3947
3959
.
Chapallaz
J. M.
,
Eichenberger
P.
&
Fischer
G.
1992
Manual on Pumps Used as Turbines
.
MHPG Series, Vol. 11
.
Friedr. Vieweg & Sohn Verlagsgesellschaft mbH
,
Germany
.
Childs
S. M.
1962
Convert pumps to turbines and recover HP
.
Hydrocarbon Processing and Petroleum Refiner
41
(
10
),
173
174
.
Gallagher
J.
,
Harris
I. M.
,
Packwood
A. J.
,
McNabola
A.
&
Williams
A. P.
2015
A strategic assessment of micro-hydropower in the UK and Irish water industry: identifying technical and economic constraints
.
Renewable Energy
81
,
808
815
.
Gomes
J.
,
Marques
A. S.
&
Sousa
J.
2011
Estimation of the benefits yielded by pressure management in water distribution systems
.
Urban Water Journal
8
(
2
),
65
77
.
Grover
K. M.
1980
Conversion of Pumps to Turbines
.
GSA InterCorp
,
Katonah, New York
.
Hergt
P.
,
Krieger
P.
&
Thommes
S.
1982
Die strömungstechnischen Eigenschaften von Kreiselpumpen im Turbinenbetrieb
,
Pumpentagung Karisruhe, VDMA, Section Cl
.
Lee
E. J.
&
Schwab
K. J.
2005
Deficiencies in drinking water distribution systems in developing countries
.
Journal of Water and Health
3
(
2
),
109
127
.
Meirelles
G.
2013
Microgeneration in Water Supply Systems
.
Master Thesis (in Portuguese)
,
Universidade Federal de Itajubá
,
2013
,
86
pp.
Meirelles
G. L.
,
Luvizotto
E.
Jr
&
Brentan
B. M.
2017
Selection and location of pumps as turbines substituting pressure reducing valves
.
Renewable Energy
109
,
392
405
.
Perdikou
S.
,
Themistocleous
K.
,
Agapiou
A.
&
Hadjimitsis
D. G.
2014
The problem of water leakages
. In:
Integrated Use of Space, Geophysical and Hyperspectral Technologies Intended for Monitoring Water Leakages in Water Supply Networks
.
InTech
.
Rossman
A. L.
2000
EPANET 2.0 User's Manual
.
Drinking Water Research Division, Risk Reduction Engineering Laboratory, US Environmental Protection Agency
.
Saldarriaga
J.
&
Salcedo
C. A.
2015
Determination of optimal location and settings of pressure reducing valves in water distribution networks for minimizing water losses
. In:
13th Computer Control for Water Industry Conference – CCWI
,
Procedia Engineering, 119
, pp.
973
983
.
Schmiedl
E.
1988
Serien-Kreiselpumpen im Turbinenbetrieb
.
Pumpentagung Karlsruhe, Section A6
.
Sharma
K.
1985
Small Hydroelectric Project-use of Centrifugal Pumps as Turbines
.
Technical report
.
Kirloskar Electric Co.
,
Bangalore
,
India
.
Sinagra
M.
,
Sammartano
V.
,
Morreale
G.
&
Tucciarelli
T.
2017
A new device for pressure control and energy recovery in water distribution networks
.
Water
9
(
5
),
309
.
Stepanoff
A. J.
1957
Special Operating Conditions of Centrifugal Pumps
.
John Wiley & Sons Inc.
,
New York
.
Torley
A. R. D.
&
Chaudhry
M. H.
1996
Pump characteristics for transient flow analysis
. In:
BHR Group Conference Series Publication
.
Mechanical Engineering Publications Limited
,
Bedford
,
19
, pp.
461
476
.
Viana
A. N. C.
1987
Behavior of Centrifugal Pumps Operating as Hydraulic Turbines
.
Master Thesis (in Portuguese)
,
EFEI
,
Itajubá
,
Brazil
.
Vilanova
M. R.
&
Balestieri
J. A. P.
2014
Hydropower recovery in water supply systems: models and case study
.
Energy Conversion and Management
84
,
414
426
.
Williams
A.
1995
Pumps as Turbines Users Guide
.
International Technology Publications
,
London
,
59
pp.