ENERGY ASSESSMENT OF WATER NETWORKS, A CASE STUDY

Transcription

1 NRGY ASSSSMNT OF WATR NTWORKS, A CAS STUDY Hernández,. *, Pardo, M.A. **, Cabrera,. ** and Cobacho, R. ** *Aqualia C/ Ulises, 18 1ª lanta Madrid. Sain. **ITA, Universidad Politécnica de Valencia, C/Camino de Vera s/n, 46022, Valencia, Sain. Corresonding authors: miari@ita.uv.es, ecabrera@ita.uv.es Abstract The comlete urban water cycle requires large amounts of energy and so, there is an increasing motivation to otimize its consumtion. In addition, the eriodic energy crises (the last one, July 2008, brought the rice of the oil barrel to 150 USD, the acute worldwide commitments to reduce greenhouse gas emissions and, last but not least, the need to minimise economic losses linked to leaks (including the energy costs lace water-energy issues on the front age of the ortfolio s research. A fact highlighted by the reort California s Water-nergy Relationshi (CC, According to this study, u to 19% of the total California s energy consumtion is related to water cycle, 6.5% associated to water distribution ste, the one analysed here. Cabrera et al. (2010 develoed a methodology to erform energy audits in ressurized water distribution systems obtained from the integral energy equation and its integration in extended eriod. Inut energy (ums, reservoirs is equal to the energy consumed by users (through demanded water lus leakage and friction energy losses in ies. nergy audit requires the revious water audit as well as the mathematical model of the distribution network. From the energy audit, context and erformance indicators (Cabrera et al., 2010 are calculated in order to assess the energy erformances of the system. Furthermore, these indicators will hel to identify future actions devoted to imrove the network s energy efficiency. Cost-benefit analysis is required to decide the best strategy to imlement in ractice. The aer is organised as follows. The fundamentals are firstly outlined and then, a case study to assess a real network from an energetic oint of view is resented. The real network sulies Denia city (Alicante, Sain and the surrounding areas. The whole distribution system sulies water in a very touristic area, close to eole, which is an interesting case study because of current water scarcity and high energy consumtion. Both facts exlain the high fees aid by the final consumer, comared with those aid in the rest of Sain. Being Denia a hilly city at the east Mediterranean coast of Sain, the energy assessment is an interesting academic exercise although for the utility comany (Aqualia is much more than that imroving water-energy erformances is a key objective to be cometitive. Keywords: nergy audit, water networks assessment, erformance indicators 1 INTRODUCTION nergy assessment of water distribution networks is a key goal for utilities. In a ermanent energy crisis scenario, its imortance can be justified in different ways, as sustainable water management is becoming very energy consuming. But, in a few words, wasting significant amounts of water and energy through leaks is, simly, unaccetable. nergy is commonly wasted as a result of network leakage, and such energy loss results not only from the energy leaving the system through leaks (which can be quite relevant deending on the energy footrint of the revious stes of the urban water cycle, mainly when water comes from desalination lants but also from the extra energy needed to overcome additional friction losses created by higher flow rates in ies.

2 The audit resented in Cabrera et al. (2010 allows to identify the final uses of the energy that enters the system, and thus to erform an assessment that characterizes the network behaviour from an energy ersective. Context information and energy indicators summarise the energetic erformance of the whole system. The energy audit can be used to evaluate the GHG imacts, which deend on the sources of water and energy (Cabrera et al., In order to have a more holistic oint of view, a cost-benefit analysis including environmental costs can be erformed. As a matter of fact, these tools could easily be used from a regulatory or administrative ersective to create incentives for a more sustainable urban energy management in water distribution systems. The energy audit, like related indicators, requires a revious water audit and a calibrated model. Both audits must be alied to similar boundaries (either to the whole network or a sector. In this aer this methodology is alied to a real network, Denia (a coastal hilly city between Valencia and Alicante, Sain. For such urose an agreement between Aqualia, the water utility resonsible of the network and the Universidad Politécnica de Valencia was established. U to now, this methodology was only alied to synthetic networks. Among many candidates, Denia was selected because it is mainly sulied from desalination (which uses brackish water from Racons River and groundwater. Although this aer only focuses on the influence of the energy losses at the distribution stage, this kind of water source (really energy consuming involves high values of energy water footrint, and erforming a global assessment of the whole urban water cycle becomes essential. 2 RVISION OF TH STRUCTUR AND NRGY AUDIT CALCULATION This section describes how to evaluate the amount of energy consumed in water distribution networks. Further details can be found in Cabrera et al. ( Inut nergy Sulied By The Reservoir (Natural nergy The external energy sulied by reservoirs is: i nn tk t N ( t QNi HNi t (1 i 1 tk t1 Where is the secific weight of water, Q Ni ( t k and H Ni ( t k are, resectively, the flow rate sulied from the reservoir i (being nn the number of reservoirs and its iezometric head at time t k. Since the analysis in extended time corresonds to a given eriod k t, the k time intervals t of the analysis must be added to totalise this eriod. t 2.2 Incoming nergy to the Network Sulied by the Puming Station (Shaft Work The shaft work sulied by the um is: i n P t k t P ( t QPi HPi t (2 i1 t k t1

3 Where Q Pi ( t k and H Pi ( t k are resectively the flow rate umed by the station and the um head at time t.this calculation needs to be done for the n uming stations that suly shaft work to the k system at the k different time instants. In this balance, and because ums do not belong to the system, their efficiencies (an essential arameter for the energy otimization are not considered. In any case, they can be easily included dividing, for each time interval, this shaft energy term by the corresonding um s efficiency. In this aer and since the focus is on new concets, these energy losses are not included in the analysis. 2.3 nergy Delivered to Users at Consumtion Nodes The useful energy delivered is: i n tk t U ( t qui tk Hi t ( ( k t (3 i 1 tk t1 Where n is the number of demand nodes of the network, q ui ( t k and H i ( t k are resectively the flow rate delivered to users and the iezometric head at node i and time. t k 2.4 Outgoing nergy Through Leaks Leaks reresent energy leaving the system, formally analogous to the energy delivered to users, although from the oint of view of the audit it is lost energy. This term is: i n tk t L( t qli tk Hi t ( ( k t (4 i 1 tk t1 With n the number of leaking nodes in the network, qli (tk the leaked flow rate in the ies adjacent to node i (and therefore associated to this node at time t k, while H i ( t k is the iezometric head at time t in the node where the leak q li t has been concentrated. k ( k 2.5 Friction Dissiated nergy The energy dissiated due to friction in ies is: j nl tk t F ( t q j tk h j t ( ( k t (5 j 1 tk t1 Where n is the number of lines of the network, h j t are friction losses in line j at time (this l term is in ie j the difference in iezometric heads between the initial and final nodes, a value known from the mathematical model of the system, q ( and q are, in line j, the flow rate necessary to uj tk satisfy the users demand and the flow rate that finally is lost through breaks, resectively. Therefore, the total flow rate in line j, q j t, is the sum of the two revious values. Local losses may be added to this term by calculating their equivalent iing length. ( k ( k lj ( t k t k

4 The energy dissiated due to friction in valves is: j n v t k t V ( t q j tk h j t ( ( k t (6 j 1 t k t1 Where n is the number of valves of the network, h j t and q j t are friction losses and flow rate v in valve j at time. t k ( k ( k 2.6 nergy Comensation of the Downstream Tank Many networks have a comensation tank to accumulate water during low consumtion hours while releasing it in eak ones. However, the net flow of water and energy in one of these tanks, when integrated through a long enough eriod, is zero, and so it is their contribution to the analysis as well. Short term would then be the eriod of time after which the energy stored in a tank is below a threshold, while the long term would just be the oosite case. The variation of otential energy stored in tanks of constant section for a given eriod of time is: C i n C i 1 in C C i ( t C i ( t Ai zi ( t zi ( t1 / 2 i 1 ( t (7 With A i the section of comensation tank i and z i ( t, z i ( t 1 the levels of the free surface of water of tank i at the initial and final times. The maximum variation of this energy, C max, obviously corresonds to total oscillation between emty and full tanks of the whole system. This last term will be necessary to conclude the tye of analysis (short or long term erformed. 2.7 Final Balance From the receding terms, being t the eriod of calculation of the revious exressions (as is the case of a water audit, commonly one year, the following final balance results: t ( t ( t t t t t t (8 inut ( N P U ( L( F ( v ( C( Outut ( t Dissiated ( t Comensation( t quation (8 states that the energy (natural and shaft sulied to the water coming into the network is equal to the energy delivered to the users (throughout the water sulied lus the losses (leakage and friction and the variation of energy at the comensation tank. From this balance, energy losses can be evaluated and its knowledge allows outlining efficient actions aimed to imrove system s efficiency.

5 3 NRGY AUDIT OVRVIW. CONTXT INFORMATION AND FFICINCY INDICATORS ach system is, from an energetic oint of view, different. The network toograhy is not modifiable and of course it can result in a high or low interest energy analysis. For instance, a hilly city with numerous intermediate uming stations, with water coming from dee wells or desalination lants would require significant amounts of energy, and consequently, a great interest study. The oosite case is a lain city, fed by surface water and without any uming stations. The difference in context between these two situations is summarised by the context information (Table 1. The first one, C 1, shows which ortion of energy delivered to the system is natural. It ranges from 0 to 1 (best value, the whole amount of energy is natural, whereas the second one, C 2, takes into account how energy demanding the network is. In articular, it is the ratio between the minimum useful energy, min,useful (defined in each node from the minimum required head, hmín, i zi PMín / and a theoretical minimum required energy (for a flat, leak free and frictionless network min, flat. Since this ideal network corresonds to a flat layout with all nodes located at the same maximum height zmax, the best ossible value of is one. C 2 Table 1. Context information C 1 nergy nature C 1 N Inut ( t ( t C 2 min, useful min, flat C 2 Network energy requirement k t n n qui hmin i t u, i ( t hmin, i k t1 i 1 i 1 Pmin P ( min U t U ( t To erform the analysis and assessment, five erformance indicators were roosed and reviewed herein (Table 2: I1 I 4 I 1 xcess of sulied energy inut ( t min, useful L( t n i1 inut u, i ( t Table 2. nergy efficiency indicators ( t h I 4 Leakage nergy Dissiated Inut min, i ( t ( t ' Dissiated I 2 Network energy efficiency ( t I 2 U Inut ( t ( t I I 3 nergy dissiated through friction I 3 Dissiated Inut I 5 Standards comliance U ( t 5 n u, i ( t hmin, i i 1 ( t ( t Further information on these indicators can be found in Cabrera et al. (2010. Their values range as follows:

6 I 1 1. It shows that the head at the nodes is close (but always above to minimum required head. The closer to one, the better. 0 I 2 1. It reresents which fraction of the total energy inut is useful. The closer to one, the better. 0 I 3 1. It reresents which fraction of the total energy inut is dissiated in ies and valves. The closer to zero, the better. 0 I 4 1. It shows energy losses due to leakage (including the additional energy required to overcome friction in ies and valves with the extra flow rate. It is desirable to reach low values of this indicator.. Better as closer to one. This is the direct ratio between the energy delivered to users and the minimum required useful energy. A value close to 1 indicates greater efficiency in meeting the ressure service above standards (condition required in our analysis. Values below one (unaccetable would show that ressure at some junctions is below standards. I CAS STUDY 4.1 Problem Descrition The case study here is based on the Denia water distribution network (Figure 1. The network contains aroximately 434 km of ies sulying water to a oulation of 100,000 inhabitants (which includes Denia and the surrounding areas, 11,500 connections (27 connections/km. The original hydraulic model was rovided by Aqualia in PANT 2 software. The model contains 6,296 nodes, 3 reservoirs, 9 tanks, 6,562 ies, 14 ums and 16 valves. Pie diameters range between 600 and 12 mm. PANT2 is a well known demand-driven water distribution network modelling software that uses temoral demand attern multiliers (DPMs to reresent a diurnal curve, i.e. the temoral variation of demand, tyically for 24 hrs (although PANT 2 simulator reeats a attern where the duration of an extended eriod simulation exceeds the duration of the attern. In the case study here, the network nodes are assumed to follow two different diurnal curves (i.e., two sets of DPMs. The first covers 28% of the average demand (57.2 l/s; while the other covers the demand left (147.5 l/s, 72%. The duration of the extended eriod may be 24 hours (short term simulation or 1 month (long term one. Pattern time ste was set to 1 hour at every simulation and the hydraulic time ste was set to 5 minutes.

7 Figure 1. Denia network layout Leakage was not originally modelled as ressure driven demand, and so, the first ste was to model it using an orifice function as (Rossman 2000: q li C, i Hi (9 Where H j ( t k is the head difference across the leak (at our case, Hi Hi( tk HGW, and H i ( t k and H GW heads in the ie and in the surrounding groundwater at line i and time t k. It is assumed that HGW 0 and so, H j Hij( tk, C, i is the coefficient assigned to each node (named emitter coefficient, being their units, m 3- /s, while =1.1 is the emitter exonent that models the characteristics of the ie material. Another hyothesis relies on the satial distribution of leaks, which in Denia is assumed to be homogeneous with a uniform distribution of leaks, only taking into account the lengths of the lines and the time variation of ressure at nodes. So, leakage is simulated as an emitter assigned to a node that considers the weighted length of the lines connected to it (Almandoz et al., The sum of all nodal leakages rates should equal the total leakage of the system (in Denia reresents around 23% of the total injected volume.

8 5 NRGY AUDIT RSULTS Four cases are resented. They corresond to daily and monthly simulations for both ideal (leak free and real networks. These can be summarized as follow: Case A. Daily simulation, real network Case B. Monthly simulation, real network Case C. Daily simulation, ideal network Case D. Monthly simulation, ideal network The aforementioned comensation term is only relevant in short-term simulations. The threshold value, t, B, boundary between the short and the long term is calculated imosing a threshold value (i.e. 1%, from the maximum comensation energy ( c,max and the daily system energy inut ( Inut. The simulation can be considered as long term one if the maximum comensation energy is lower than this small ercentage of the system energy inut. The equation used to calculate is, then: t, T c,max ( days (10 1 Inut( daily 100 As the inut energy is inut ( t = 13, kwh/day and the maximum variation in the 9 comensation tanks is c,max = 2, kwh. The threshold value is t, T 18 days (equation 10 and so, the daily simulation can be considered as short term, but not the monthly one that qualifies as long-term. t, B 5.1 Results Prior to erform the energy audit, it is comulsory to solve the hydraulic roblem. The hydraulic model of the network and its water audit (aarent losses are considered as additional demand is required. Table 3 resents the water audit results: Table 3. Water balance in Denia network. Case A (m 3 Case B (Hm 3 Case C (m 3 Case D (Hm 3 Injected water 42, , Delivered water 31, , Real losses 7, Volume stored in tanks 3, , In case B, water losses er unit of length and time are equal to 0.91 m 3 /kmh. This indicator tends to range between 0.1 m 3 /kmh and 2 m 3 /kmh, so the attained value corresonds to the mean value of the range and to a relevant leakage level. Table 4 shows the energy audit. Theoretical energies, defined as min, useful and min, flat, are equal to 2, kwh/day (65.92 MWh/month and 1, kwh/day (31.84 MWh/month resectively.

9 Table 4. nergy balance in Denia network Real network Ideal network (no leaks Inut ( t nergy Outut ( t Case A Case B Case C Case D Short Term t t, T Long Term t t, T Short Term t t, T Long Term t t, T (kwh/day (MWh/month (kwh/day (MWh/month N ( t 2, (16.2% (18.5% 2, (19.7% (22.4% P ( t 11, (83.8% (81.5% 8, (80.3% (77.6% U ( t 3, (28.3% (33.2% 4, (43.9% (58.3% L ( t 2, (15.3% (17.7% - (0% - (0% C ( t 1, (10.0% 2.03 (0.6% 1, (11.7% 1.88 (0.7% Dissiated ( t F ( t 6, (44.9% (46.7% 4, (42.9% (39.2% V ( t (1.5 % 6.49 (1.8% (1.5% 5.31 (1.8% Additionally, results in Table 4 show that: The energy required at the distribution ste is very high. In articular, for case B, the energy intensity er unit of water injected is as high as 0.32 kwh/m 3. In fact, it is equal to the uer limit for the distribution ste rovided by CC (2005. The inut energy savings in a leak-free network are significant. At the daily simulation, it is given by the difference between 13, kwh/day and 10, kwh/day whereas at the monthly one it can be calculated from the difference between MWh/month and MWh/month. The energy delivered to users is higher in a leak-free network than a real one (resectively MWh/month and MWh/month. This increase shows imrovement at the network erformance. Accordingly to this, the artial or total recovery of these energy surluses requires the otimization of the oerating conditions of the network. So, it increases otential energy savings. The energy losses linked to leaks (outgoing energy through breaks lus additional friction losses is 2, , , = 3, kwh/day for the short term simulation. This value is MWh/month for the monthly case. This otential savings reresents 26.45% and 33.54% resectively of the total energy in use. A significant figure, indeed. The energy dissiated in valves is not a high value, which imlies a low ercentage figure comared to energy dissiated due to friction in ies. 5.2 nergy Assessment of the Denia Network The results obtained with the long term simulation are now used (Cases B and D to calculate both context and efficiency indicators. Table 5 shows them all. Context indicators remark the relevance of the analysis, while energy indicators show that, with an adequate system management, there is a huge room for imrovements. As exected, it is easy to notice that all the indicators imrove in a non leaky network.

10 Table 5. nergy Indicators C 1 C 2 I 1 I 2 I 3 I 4 I 5 Real network Ideal network The first context information shows that less than 20% of the inut energy is natural. The slight variation of this value between the real and ideal networks is negligible, since it is suosed to be indeendent of the state of the network (Alegre et al., This is due to the fact that C 1 is not strictly a context indicator because it deends on the ercentage of energy sulied by the reservoirs, and, in turn, it varies according to the system behaviour. C 2, shows that Denia s network is rather hilly, as it is highlighted by the difference between the highest and lowest node (180 m. The first efficiency indicator, I 1, shows that the inut energy of the network is more than 5 times the minimum amount of energy necessary to suly the service. As a matter of fact, when leakage disaears, this indicator is brought down to 4.4. The second indicator shows the ercentage of energy delivered to users, 33% in the real network comared to the inut energy. This leaves 67% of the energy lost through either leakage or friction. In a leakage-free scenario, the value goes u to 58%, reresenting a relevant imrovement. I 3 shows how much energy is used to overcome friction in ies and valves. In this case, a value this high (47% indicates high length of the network, or tight ie diameters, or both. In a leak-free situation, this value is 39%, high enough to trigger the substitution of key mains with larger ones, although a cost-benefit analysis is required to exlore other otions (i.e., demand management olicies. The fourth indicator evaluates total energy lost due to leaks. Its high value (34% means a lot of energy wasted, 180 MWh/month. At the distribution ste, this energy can reresent economic losses of aroximately 20,000 /month (according to the Sanish electric tariff. Finally, as exected, I 5 increases in a free-leak network. So, there is more surlus of energy delivered to users. It means that, in absence of leaks, the level of ressure increases, and network s erformance can be imroved by means of regulation (valves or variable-seed ums. Anyway, erformance indicators show that it is worth to exlore different ways to imrove erformances and, throughout the corresonding cost-benefit analysis, to decide the order and the time to imlement them. And this is, indeed, the second ste of the study. To this regard, next section outlines the ways to be exlored. 6 TH WAY FORWARD Due to the fact that water sources in Denia (groundwater and desalination are very energy consuming, the oerator is interested in an overall energy assesment, that will include all the stes of the urban cycle, and not just the distribution hase u to now described. In any case, this aer will only focus on the energy assessment of the distribution stage. The strategies can be divided into two grous. Those that imrove the system oeration (i.e., variable seed ums and those that minimize the flows through the network, either reducing leaks or demands. The first ones can only be faced by the utility, whereas the second ones can be undertakaen by both, utility and users. This last otion is the ond finally adoted to outline the way forward.

11 1. From the utility side A better system oeration, mainly through ressure management. Pressure control can reduce leakage, as well as other ressure driven demands (i.e. garden watering and the frequency of bursts. Also, it rovides a more steady service to costumers. That ressure control must requires to imlement district metering areas (DMA, ressure management areas (PMA or both. It is very convenient that both areas will coincide because, in that case, water and energy audits can be articularly alied and energy indicators identified. Deending on the imrovements achived for each articular DMAs, the subsequent actions can be scheduled. Obviously, the technique used to reduce the ressure might differ according to each PMA s articularities. Tyical otions range from the installation of ressure reducing valves to the relacement of constant seed ums by variable seed ones. A better network management, through a more active leakage control. There are two ways to reduce leaks. The first, ie renovation and the second one a more active leakage control. A ie renovation olicy will, indeed, imrove Denia s network erformance. Pardo (2010 dealt with the influence of water and energy costs in ie renovation eriods. On the other hand, areas where the number of bursts er km and year resents a reasonable figure and, hence, ie renovation is not fully justified, a more active leakage control should be romoted. 2. From the users side Water demand reduction. It is quite clear that water use reduction would lead into energy savings. CC (2005 showed that although water efficiency rograms and conservation efforts exist in that state, there are many missed oortunities to save energy, and the achievable benefit could be higher than the one obtained with energy efficiency measures. It is shown (Table 3 that in Case A, the consumed volume is 31,490 m 3 /day which results in l/ca/day, a rather high value comared to the sanish average (157 l/ca/day. So, high water demand reduction ossibilities rely behind. This roject of Denia s network energy assessment tries to imrove the sustainability by romoting a more efficient use of water and energy. Hydraulic and energy efficiency are inextrincably couled and so, every assessment decision has a double synergetic effect. Now, the roject overcame this first ste, the energy audit highlighted the current state of the network and showed huge otential savings which are now under study. Cost benefit analyses (deending on the water and energy costs should be erformed to evaluate different otential actions, and once more, the energy audit herein lays a crucial role. 7 CONCLUSIONS This aer shows the first alication of the energy audit to a real water network. Results demonstrate that it is a owerful tool to rovide key information to ease oerator s decisions. Now, the strengths and weaknesses from an energetic oint of view are well known. This case study shows huge amounts of energy losses due to leakage and dissiation. Moreover it hels to clarify the relationshi between water and energy. It is obvious that the leakage reduction leads to energy savings and the results show huge otential savings because of the comlex toograhy, the lack of secific assessment, etc. The energy assessment olicy will have to address all this info in order to evaluate roerly the benefits of each individual decision, as well as all the lans considered simultaneously (observing the

12 synergy amongst all of them. The energy audit suorts all these facts in numbers, not just words. Undoubtely, energy is a key factor in order to take decissions. As aforementioned, this aer does not consider the energy water footrint of revious stes of the urban water cycle. In Denia, water comes from very energy consuming sources i desalination, at 3.5 kwh/m 3 (NRC, 2008, and ii groundwater at 0.35 kwh/m 3 er 100 m of elevation. So, imroving the network efficiency is crucial for the water utility, mainly if environmental costs (as those derived from the GHG emission are considered. 8 ACKNOWLDGMNTS The research corresonding to the energy audit has been suorted by the Ministry of Science and Innovation of Sain, throughout Project No. CGL References Alegre H., Batista J.M., Cabrera. Jr., Cubillo F., Duarte P., Hirner H., Merkel W., Parena R., (2006. Performance Indicators for Water Sully Services, IWA Publishing. Almandoz, J., Cabrera,., Arregui, F., Cabrera Jr.,. and Cobacho, R. (2005. Leakage Assessment through Water Distribution Network Simulation. Journal of Water Resources Planning and Management. Vol 131, Nº6, November 1, Cabrera,., Pardo, M.A., Cobacho, R., Arregui, F.J. and Cabrera,. Jr.(2009. valuation of Carbon Credits Saved by Water Losses Reduction in Water Networks.Waterloss ISBN Pages Cabrera., Pardo M.A., Cobacho R. and Cabrera. Jr., (2010. nergy Audit of a water Distribution network. Journal of Water Resources Planning and Management American Society of Civil ngineering. (Acceted. To be ublished in short. CC (2005. California s Water-nergy Relationshi reort California nergy Commission. November 2005 CC SF NRC, National Research Council (2008. Desalination a national ersective. NAP Press, Washington, D.C. Pardo, M.A. (2010. Influencia de los costes del Agua y de la energía en la renovación de tuberías. Doctoral Thesis. 18th June Valencia. Available at: htt://hdl.handle.net/10251/8426 (accesed June, 2010 and in Sanish Rossman, L. A. (2000. PANT 2 Users Manual. U.S. nvironmental Protection Agency, Cincinnati.