- Email: [email protected]
Emergency Management of Water Distribution Systems: The Nodal Demand Control
Available online at www.sciencedirect.com
ScienceDirect Procedia Engineering 186 (2017) 428 – 435
XVIII International Conference on Water Distribution Systems Analysis, WDSA2016
Emergency Management of Water Distribution Systems: the Nodal Demand Control Fiorini Morosini A.a,*, Caruso O.a, Costanzo F.b and Savic D.c a University of Calabria, Department of Civil Engineering, Rende (CS) 87036, Italy University of Ferrara, Department of Engineering, via Saragat 1 Ferrara (FE) Italy c University of Exeter, College of Engineering Mathematics and Physical Sciences, Centre for Water Systems, Exeter EX4 4QF Devon, United Kingdom b
Abstract During emergency situations (e.g., due to pipe bursts or other network failures), appropriate management of Water Distribution Systems (WDS) is required. Critical events often cause service failures, because the pressure head in some nodes of the network become inadequate to deliver required demand. In this paper, a new methodology is developed based on the nodal demand control. with the aim to increase the pressure head, and hence the flow rate actually delivered at critical nodes (i.e., hospitals, vulnerable customers, etc.). This is done to avoid or minimize service interruptions between the failure and the repair times. Furthermore, a pipe burst can cause isolation of a portion of the network such that the flow along pipes changes and this causes the reduction of head in some nodes. The proposed methodology is manages the delivered flow rate using a Pressure Driven Analysis (PDA) approach. This is based on operating control of valves and by identifying the nodes where the pressure control should be implemented. Those control nodes are chosen by the analysis of sensitivity matrices and the Max-Sum Method (Bush and Uber, 1998; Fiorini Morosini et al., 2014). The methodology is demonstrated on a case study for a real network of Cosenza, a town in the South of Italy. © Published by Elsevier Ltd. Ltd. This is an open access article under the CC BY-NC-ND license ©2016 2016The TheAuthors. Authors. Published by Elsevier (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of the XVIII International Conference on Water Distribution Peer-review Systems. under responsibility of the organizing committee of the XVIII International Conference on Water Distribution Systems Keywords: WDS Management; Water Network Analysis; PDA Analysis; Pipe burst.
* Corresponding author. Tel.: +39-0984-496549; fax: +39-0984-496549. E-mail address: [email protected]
1877-7058 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of the XVIII International Conference on Water Distribution Systems
doi:10.1016/j.proeng.2017.03.247
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435
1. Introduction Management of WDSs has been the focus of research and industry communities in recent years. This problem becomes prominent when one or more pipe failures occur in a network. Pipe rehabilitation studies focus on preventing failures in pipe networks and ensuring the long-term system efficiency (Engelhardt et al., 2000). Optimization and decision support techniques linked to economic aspects, such as the minimization of the repair and maintenance costs (Lansey et al.,1992; Kleiner et al., 1998), or based on performance indeces of water network reliability (Gargano and Pianese, 2000; Berardi et al., 2008), can also be used. Other studies have considered multi-objective genetic algorithm methodologies, which take into account economic and performance indeces (Farmani et al., 2005), while some management models are based on the time planning of repair and maintenance operations (de Marinis et al., 2008). Furthermore, the optimal sequence of these operations could be linked to the reliability maximization and costs minimization (Alvisi and Franchini, 2006). This work presents a different approach to the problem and proposes a methodology for emergency management of water networks during the repair and recovery period. When a pipe burst occurs, it is necessary to isolate an area (district) of the network (Giustolisi and Savic, 2010) by operating a subset of valves. A critical area is a zone where the supply must be maintained (i.e., a district with a hospital or with vulnerable customers) even during service interruptions in other areas. The methodology is based on the definition of a new nodal demand distribution at sensitive nodes. The aim is to improve the head in the critical areas of the network and to increase the effective delivered flow. A PDA (Pressure Driven Analysis) model is used such that the delivered demand depends on the nodal pressure head (Calomino and Veltri, 1980) and the sensitive nodes are chosen using the sensitivity matrix approach (Fiorini Morosini et al., 2015). 2. Methodology When a failure occurs in a WDS, it should be isolated by manipulating valves to allow the necessary time to repair pipe. Therefore, a new network configuration should be analysed to assess the impact of reduced total flow in the system. Sometimes, the pipe failure produces a decrease of head in the network and in some nodes the demand flow rate (Qreq) cannot be guaranteed. In this case a PDA model, as INetPDA (Veltri et al., 2010), is more adequate than a Demand Driven Analysis (DDA) to analyses the system and to define the effective demand (Qfail) that can be delivered when the head is lower than minimum head Hmin. The aim of this paper is to describe a methodology to increase the pressure head in particular nodes of the network, defined here as critical nodes (i.e., hospitals, vulnerable customers, etc.). These nodes would otherwise not have enough pressure to deliver the required flow rate due to the failure. To increase the head value in critical nodes, the methodology proposes to limit the demand in the nodes with an adequate head and so to reduce the circulating flow in the network. This nodal demand check could be done using valves which control the head, and define a fixed flow at the node (Savic and Walters, 1995). To choose the nodes for the demand control sensitivity matrices are used. In this matrices each element represents the variation in pressure when a variation of the demand, Qi, at nodes or the roughness coefficient, İi, in pipe occurs. The model employs the Max-Sum Method (Bush and Uber, 1998) and it is shown below (Fig. 1).
429
430
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435
Fig. 1. Methodology algorithm
3. Case Study The methodology was applied to a real case, the network of Cosenza (CS), a city in the South of Italy. The network, consisting of 48 pipes, 40 nodes and 2 source/tanks with a fixed head (Tables 1 and 2), is shown in Fig. 2. The total base demand is 256.52 l/s. The minimum head Hmin varies from 261 m to 339 m.
Fig. 2. Network of the centre of Cosenza (Italy). Table 1. Characteristics of nodes Node
Demand (l/s)
Elevation (m)
Hmin (m)
Node
Demand (l/s)
Elevation (m)
Hmin (m)
1
1.72
284.50
312.62
2
3.89
256.80
297.74
21
9.55
247.00
285.96
22
11.08
260.60
315.49
3
3.99
250.00
4
5.30
240.20
283.87
23
14.16
229.00
292.90
268.87
24
4.20
246.50
284.26
431
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435 5
5.88
238.20
264.33
25
6.62
243.10
281.55
6
5.23
237.40
263.25
26
7.14
253.80
297.01
7
3.47
270.50
314.53
27
5.87
249.70
286.62
8
3.35
238.40
264.30
28
6.55
242.30
280.60
9
2.58
233.00
261.74
29
8.32
245.00
291.19
10
4.09
264.20
301.81
30
6.85
236.70
279.25
11
5.61
261.60
301.63
31
9.43
232.00
281.39
12
6.02
259.00
299.81
32
8.75
227.90
275.28
13
4.16
245.00
282.70
33
12.68
233.00
297.66
14
7.24
240.00
279.85
34
7.72
233.10
277.72
15
3.18
235.00
260.71
35
9.94
224.00
271.40
16
2.34
233.00
261.54
36
6.90
222.00
264.66
17
9.34
232.80
274.72
37
10.13
228.00
283.22
18
3.41
266.60
299.77
38
5.73
226.60
266.85
19
6.62
255.20
297.25
39
2.65
221.00
264.20
20
14.83
268.00
338.64
40
0.00
285.00
288.60
Table 2. Characteristics of pipes Pipe
Length (m)
Diameter (mm)
Roughness (mm)
Pipe
Lengt (m)
Diameter (mm)
Roughness (mm)
1
166.00
200
1
26
637.00
200
1
2
168.00
200
1
27
442.00
400
1
3
98.50
200
1
28
475.50
400
1
4
237.50
200
1
29
161.00
400
1
5
135.50
200
1
30
241.50
400
1
6
248.50
80
1
31
305.00
400
1
7
155.50
80
1
32
75.50
400
1
8
104.50
80
1
33
177.50
400
1
9
409.50
80
1
34
210.50
200
1
10
93.00
200
1
35
385.50
400
1
11
154.50
200
1
36
278.50
200
1
12
108.00
400
1
37
168.00
80
1
13
116.50
400
1
38
635.00
200
1
14
190.00
400
1
39
724.50
400
1
15
64.00
200
1
40
117.50
400
1
16
217.00
200
1
41
621.00
400
1
17
118.50
200
1
42
334.50
400
1
18
202.50
200
1
43
328.50
400
1
19
127.50
200
1
44
295.50
200
1
20
199.00
200
1
45
255.00
125
1
21
221.50
375
1
46
350.00
400
1
22
165.50
375
1
47
391.50
400
1
432
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435
23
189.00
200
1
48
163.00
200
1
24
257.50
25
427.50
200
1
49
1517.00
400
1
200
1
Considering a failure of pipe 20, an area of the network is isolated to allow the pipe repair work (Fig. 3).
Burst area and critical nodes Critical area
Fig. 3. Network of Cosenza with a failure in pipe 20 and the closure of a subset of valves
In this case, assuming the required nodal demand Qreq, under a PDA condition, the head values Hi and the real delivered flow Qfail are the following (Fig. 4).
Fig. 4. Variation between the base demand Qreq and the real delivered flow Qfail
As shown in Fig. 4, there are 10 nodes in which the delivered demand is lower than the requested demand Qreq and there are 2 nodes completely isolated from the network. Among these there is the hospital (node11). The actual flow
433
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435
in the network is 230.64 l/s instead of 256.52 l/s. In particular, the delivered nodal demand in critical nodes is 54.68 l/s instead of 62.32 l/s. The sensitivity matrices have been calculated based on the work by Fiorini et al. (2014) both for the scheme with PIPE 20 failure and the closure of the subset of valves. The methodology works by varying independently the roughness in pipes and the demand at nodes. The top sensitive nodes obtained are similar for each hypothesis considered (Table 3). Table 3. The list of sensitive nodes calculated using roughness or nodal demand variation Variation of roughness
Variation of nodal demand
5
5
4
4
3
3
8
8
13
13
6
6
9
9
17
17
16
16
15
15
14
14
39
23
The analyses of the network have been carried out considering the pipe failure and a reduction of 5% of the total demand. The new total demand in the network is Qred and the reduction is localized at the first three sensitive nodes (3, 4, 5), as shown in Fig. 5.
Fig. 5. Comparison between the request nodal demand Qreq and the new demand distribution Qred.
Assuming Qred, the network simulation in PDA conditions shows that the calculated heads HF and the effective delivered flows (Qdel) increase at critical nodes (Fig. 6). The total circulating flow in the network is equal to 219.68 l/s instead of 230.64 l/s corresponding at the failure scenario. The delivered nodal demand in critical nodes is 55.63 l/s instead of 54.68 l/s corresponding to failure pipe without reduction of nodal demand.
434
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435
Fig. 6. Comparison between the delivered flow Qfail and the actual delivered flow Qdel with the new distribution nodal demand Qred
More analyses have been carried out to compare the benefits in terms of delivered flow at critical nodes changing the demand at sensitive nodes and varying the number of sensitive nodes in which reduce the nodal demand. The nodal demand has been varied by 5%, 10% and 15%, while the number of sensitive nodes in which to perform the nodal control has been varied from 3 to 12. The results in terms of improvement of delivered flow at critical nodes are shown in Fig. 7.
Fig. 7. Improvement in critical nodes changing the demand at sensitive nodes and varying the number of sensitive nodes
The benefits at critical nodes are evident for all the analysed considered. The benefits at critical nodes decrease when the number of sensitive nodes in which the demand is reduced increases. The benefit increases when the amount of demand reduction at sensitive nodes increases. For other failures scenarios, considering different pipe failure, similar results have been obtained.
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435
4. Conclusions In this paper, a methodology for emergency management of Water Distribution Systems (WDS), based on a nodal demand control has been proposed. The scope of the method is to manage the network in emergency condition and to guarantee for critical nodes (i.e. hospitals and vulnerable customers) the nodal demand with an adequate head. The management model, when a failure pipe occurs, has shown the benefit obtainable in the areas of the network where the head decreases and the delivered nodal demand are less than the requested demand. The analysis of the network has shown that at an improvement of critical nodes functioning corresponds a contained reduction of total circulating flow in the network. The results depend on the network complexity because the procedure is based on using sensitivity matrices and on the network sensitive nodes.
References [1] C.A. Bush and J.G. Uber, “Sampling design methods for water distribution model calibration”, Journal of Water Resources Planning and Management 124 (6), 1998, pp. 334-344. [2] M.O. Engelhardt, P.J. Skipworth, D.A. Savic, A.J. Saul, G.A. Walters, Rehabilitation strategies for water distribution networks: A literature review with a UK perspective, Urban Water, pp. 153-170, 2000. [3] K.E. Lansey, N. Duan, L.W. Mays, T.K. Tung, Optimal maintenance scheduling for water distribution systems, Civ. Eng. Syst., 9(3), 211–226, 1992. [4] Y. Kleiner,, B.J. Adams, J.S. Rogers, Selection and scheduling of rehabilitation alternatives for water distribution systems, Water Resources Research, 34 (8), 2053-2061, 1998. [5] R. Gargano, and D. Pianese, Reliability as Tool for Hydraulic Network Planning, Journal of Hydraulic Engineering, ASCE , (126 5), pp 354364, 2000. [6] L. Berardi, A.F. Colombo, O. Giustolisi, Analisi di affidabilità delle reti in pressione in presenza di valvole di intercettazione, XXXI Convegno Nazionale di Idraulica e Costruzioni Idrauliche, Perugia 2008. [7] R. Farmani, G.A. Walters, D.A. Savic,, Trade-off between Total Cost and Reliability for Anytown Water Distribution Network”, Journal of Water Resources Planning and Management, 131(3), 161-171, 2005. [8] G. de Marinis, R. Gargano, Z. Kapelan, M.S. Morley, D.A. Savic, C. Tricarico, A rehabilitation approach for water distribution systems based on a pressure driven analysis, XXXI Convegno Nazionale di Idraulica e Costruzioni Idrauliche, Perugia 2008. [9] S. Alvisi, M. Franchini. Near-optimal rehabilitation scheduling of water distribution systems based on a multi-objective genetic algorithm. Civil Engineering and Environmental Systems, 2006, 23.3: 143-160. [10] Giustolisi O. and Savic D., 2010. 'Identification of segments and optimal isolation valve system design in water distribution networks. Urban Water Journal, 7: 1, 1 — 15. [11] F. Calomino, P. Veltri, Un procedimento di calcolo automatico per la verifica delle reti idrauliche in pressione a portate incognite in ingresso ed in uscita, XVII Convegno di Idraulica e Costruzioni Idrauliche Palermo, 27-29 Ottobre 1980. [12] A. Fiorini Morosini, F. Caruso O., P. Veltri and Costanzo F., Water distribution Network Management in Emergency Conditions. Procedia Engineering, 2015, 119: 908-917. [13] P. Veltri, A. Fiorini Morosini, F. Orlando, La calibrazione delle scabrezze delle reti idriche: un approccio di tipo probabilistico basato su un modello PDA di risoluzione della rete, IV Seminary on "La ricerca delle perdite e la gestione delle reti di condotte", Aversa, September 2009, L’Acqua, n°2 (March/April), 2010. [14] D.A. Savic and G.A. Walters, “An Evolution Program for Optimal Pressure Regulation in Water Distribution Networks”, Engineering Optimization, Vol. 24, No. 3, 1995, pp. 197-219. [15] A. Fiorini Morosini, F. Costanzo, P. Veltri and D.A. Savic, Identification of Measurement Points for Calibration of Water Distribution Network Models. Procedia Engineering, 2014, 89: 693-701.
435
ScienceDirect Procedia Engineering 186 (2017) 428 – 435
XVIII International Conference on Water Distribution Systems Analysis, WDSA2016
Emergency Management of Water Distribution Systems: the Nodal Demand Control Fiorini Morosini A.a,*, Caruso O.a, Costanzo F.b and Savic D.c a University of Calabria, Department of Civil Engineering, Rende (CS) 87036, Italy University of Ferrara, Department of Engineering, via Saragat 1 Ferrara (FE) Italy c University of Exeter, College of Engineering Mathematics and Physical Sciences, Centre for Water Systems, Exeter EX4 4QF Devon, United Kingdom b
Abstract During emergency situations (e.g., due to pipe bursts or other network failures), appropriate management of Water Distribution Systems (WDS) is required. Critical events often cause service failures, because the pressure head in some nodes of the network become inadequate to deliver required demand. In this paper, a new methodology is developed based on the nodal demand control. with the aim to increase the pressure head, and hence the flow rate actually delivered at critical nodes (i.e., hospitals, vulnerable customers, etc.). This is done to avoid or minimize service interruptions between the failure and the repair times. Furthermore, a pipe burst can cause isolation of a portion of the network such that the flow along pipes changes and this causes the reduction of head in some nodes. The proposed methodology is manages the delivered flow rate using a Pressure Driven Analysis (PDA) approach. This is based on operating control of valves and by identifying the nodes where the pressure control should be implemented. Those control nodes are chosen by the analysis of sensitivity matrices and the Max-Sum Method (Bush and Uber, 1998; Fiorini Morosini et al., 2014). The methodology is demonstrated on a case study for a real network of Cosenza, a town in the South of Italy. © Published by Elsevier Ltd. Ltd. This is an open access article under the CC BY-NC-ND license ©2016 2016The TheAuthors. Authors. Published by Elsevier (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of the XVIII International Conference on Water Distribution Peer-review Systems. under responsibility of the organizing committee of the XVIII International Conference on Water Distribution Systems Keywords: WDS Management; Water Network Analysis; PDA Analysis; Pipe burst.
* Corresponding author. Tel.: +39-0984-496549; fax: +39-0984-496549. E-mail address: [email protected]
1877-7058 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of the XVIII International Conference on Water Distribution Systems
doi:10.1016/j.proeng.2017.03.247
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435
1. Introduction Management of WDSs has been the focus of research and industry communities in recent years. This problem becomes prominent when one or more pipe failures occur in a network. Pipe rehabilitation studies focus on preventing failures in pipe networks and ensuring the long-term system efficiency (Engelhardt et al., 2000). Optimization and decision support techniques linked to economic aspects, such as the minimization of the repair and maintenance costs (Lansey et al.,1992; Kleiner et al., 1998), or based on performance indeces of water network reliability (Gargano and Pianese, 2000; Berardi et al., 2008), can also be used. Other studies have considered multi-objective genetic algorithm methodologies, which take into account economic and performance indeces (Farmani et al., 2005), while some management models are based on the time planning of repair and maintenance operations (de Marinis et al., 2008). Furthermore, the optimal sequence of these operations could be linked to the reliability maximization and costs minimization (Alvisi and Franchini, 2006). This work presents a different approach to the problem and proposes a methodology for emergency management of water networks during the repair and recovery period. When a pipe burst occurs, it is necessary to isolate an area (district) of the network (Giustolisi and Savic, 2010) by operating a subset of valves. A critical area is a zone where the supply must be maintained (i.e., a district with a hospital or with vulnerable customers) even during service interruptions in other areas. The methodology is based on the definition of a new nodal demand distribution at sensitive nodes. The aim is to improve the head in the critical areas of the network and to increase the effective delivered flow. A PDA (Pressure Driven Analysis) model is used such that the delivered demand depends on the nodal pressure head (Calomino and Veltri, 1980) and the sensitive nodes are chosen using the sensitivity matrix approach (Fiorini Morosini et al., 2015). 2. Methodology When a failure occurs in a WDS, it should be isolated by manipulating valves to allow the necessary time to repair pipe. Therefore, a new network configuration should be analysed to assess the impact of reduced total flow in the system. Sometimes, the pipe failure produces a decrease of head in the network and in some nodes the demand flow rate (Qreq) cannot be guaranteed. In this case a PDA model, as INetPDA (Veltri et al., 2010), is more adequate than a Demand Driven Analysis (DDA) to analyses the system and to define the effective demand (Qfail) that can be delivered when the head is lower than minimum head Hmin. The aim of this paper is to describe a methodology to increase the pressure head in particular nodes of the network, defined here as critical nodes (i.e., hospitals, vulnerable customers, etc.). These nodes would otherwise not have enough pressure to deliver the required flow rate due to the failure. To increase the head value in critical nodes, the methodology proposes to limit the demand in the nodes with an adequate head and so to reduce the circulating flow in the network. This nodal demand check could be done using valves which control the head, and define a fixed flow at the node (Savic and Walters, 1995). To choose the nodes for the demand control sensitivity matrices are used. In this matrices each element represents the variation in pressure when a variation of the demand, Qi, at nodes or the roughness coefficient, İi, in pipe occurs. The model employs the Max-Sum Method (Bush and Uber, 1998) and it is shown below (Fig. 1).
429
430
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435
Fig. 1. Methodology algorithm
3. Case Study The methodology was applied to a real case, the network of Cosenza (CS), a city in the South of Italy. The network, consisting of 48 pipes, 40 nodes and 2 source/tanks with a fixed head (Tables 1 and 2), is shown in Fig. 2. The total base demand is 256.52 l/s. The minimum head Hmin varies from 261 m to 339 m.
Fig. 2. Network of the centre of Cosenza (Italy). Table 1. Characteristics of nodes Node
Demand (l/s)
Elevation (m)
Hmin (m)
Node
Demand (l/s)
Elevation (m)
Hmin (m)
1
1.72
284.50
312.62
2
3.89
256.80
297.74
21
9.55
247.00
285.96
22
11.08
260.60
315.49
3
3.99
250.00
4
5.30
240.20
283.87
23
14.16
229.00
292.90
268.87
24
4.20
246.50
284.26
431
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435 5
5.88
238.20
264.33
25
6.62
243.10
281.55
6
5.23
237.40
263.25
26
7.14
253.80
297.01
7
3.47
270.50
314.53
27
5.87
249.70
286.62
8
3.35
238.40
264.30
28
6.55
242.30
280.60
9
2.58
233.00
261.74
29
8.32
245.00
291.19
10
4.09
264.20
301.81
30
6.85
236.70
279.25
11
5.61
261.60
301.63
31
9.43
232.00
281.39
12
6.02
259.00
299.81
32
8.75
227.90
275.28
13
4.16
245.00
282.70
33
12.68
233.00
297.66
14
7.24
240.00
279.85
34
7.72
233.10
277.72
15
3.18
235.00
260.71
35
9.94
224.00
271.40
16
2.34
233.00
261.54
36
6.90
222.00
264.66
17
9.34
232.80
274.72
37
10.13
228.00
283.22
18
3.41
266.60
299.77
38
5.73
226.60
266.85
19
6.62
255.20
297.25
39
2.65
221.00
264.20
20
14.83
268.00
338.64
40
0.00
285.00
288.60
Table 2. Characteristics of pipes Pipe
Length (m)
Diameter (mm)
Roughness (mm)
Pipe
Lengt (m)
Diameter (mm)
Roughness (mm)
1
166.00
200
1
26
637.00
200
1
2
168.00
200
1
27
442.00
400
1
3
98.50
200
1
28
475.50
400
1
4
237.50
200
1
29
161.00
400
1
5
135.50
200
1
30
241.50
400
1
6
248.50
80
1
31
305.00
400
1
7
155.50
80
1
32
75.50
400
1
8
104.50
80
1
33
177.50
400
1
9
409.50
80
1
34
210.50
200
1
10
93.00
200
1
35
385.50
400
1
11
154.50
200
1
36
278.50
200
1
12
108.00
400
1
37
168.00
80
1
13
116.50
400
1
38
635.00
200
1
14
190.00
400
1
39
724.50
400
1
15
64.00
200
1
40
117.50
400
1
16
217.00
200
1
41
621.00
400
1
17
118.50
200
1
42
334.50
400
1
18
202.50
200
1
43
328.50
400
1
19
127.50
200
1
44
295.50
200
1
20
199.00
200
1
45
255.00
125
1
21
221.50
375
1
46
350.00
400
1
22
165.50
375
1
47
391.50
400
1
432
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435
23
189.00
200
1
48
163.00
200
1
24
257.50
25
427.50
200
1
49
1517.00
400
1
200
1
Considering a failure of pipe 20, an area of the network is isolated to allow the pipe repair work (Fig. 3).
Burst area and critical nodes Critical area
Fig. 3. Network of Cosenza with a failure in pipe 20 and the closure of a subset of valves
In this case, assuming the required nodal demand Qreq, under a PDA condition, the head values Hi and the real delivered flow Qfail are the following (Fig. 4).
Fig. 4. Variation between the base demand Qreq and the real delivered flow Qfail
As shown in Fig. 4, there are 10 nodes in which the delivered demand is lower than the requested demand Qreq and there are 2 nodes completely isolated from the network. Among these there is the hospital (node11). The actual flow
433
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435
in the network is 230.64 l/s instead of 256.52 l/s. In particular, the delivered nodal demand in critical nodes is 54.68 l/s instead of 62.32 l/s. The sensitivity matrices have been calculated based on the work by Fiorini et al. (2014) both for the scheme with PIPE 20 failure and the closure of the subset of valves. The methodology works by varying independently the roughness in pipes and the demand at nodes. The top sensitive nodes obtained are similar for each hypothesis considered (Table 3). Table 3. The list of sensitive nodes calculated using roughness or nodal demand variation Variation of roughness
Variation of nodal demand
5
5
4
4
3
3
8
8
13
13
6
6
9
9
17
17
16
16
15
15
14
14
39
23
The analyses of the network have been carried out considering the pipe failure and a reduction of 5% of the total demand. The new total demand in the network is Qred and the reduction is localized at the first three sensitive nodes (3, 4, 5), as shown in Fig. 5.
Fig. 5. Comparison between the request nodal demand Qreq and the new demand distribution Qred.
Assuming Qred, the network simulation in PDA conditions shows that the calculated heads HF and the effective delivered flows (Qdel) increase at critical nodes (Fig. 6). The total circulating flow in the network is equal to 219.68 l/s instead of 230.64 l/s corresponding at the failure scenario. The delivered nodal demand in critical nodes is 55.63 l/s instead of 54.68 l/s corresponding to failure pipe without reduction of nodal demand.
434
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435
Fig. 6. Comparison between the delivered flow Qfail and the actual delivered flow Qdel with the new distribution nodal demand Qred
More analyses have been carried out to compare the benefits in terms of delivered flow at critical nodes changing the demand at sensitive nodes and varying the number of sensitive nodes in which reduce the nodal demand. The nodal demand has been varied by 5%, 10% and 15%, while the number of sensitive nodes in which to perform the nodal control has been varied from 3 to 12. The results in terms of improvement of delivered flow at critical nodes are shown in Fig. 7.
Fig. 7. Improvement in critical nodes changing the demand at sensitive nodes and varying the number of sensitive nodes
The benefits at critical nodes are evident for all the analysed considered. The benefits at critical nodes decrease when the number of sensitive nodes in which the demand is reduced increases. The benefit increases when the amount of demand reduction at sensitive nodes increases. For other failures scenarios, considering different pipe failure, similar results have been obtained.
A. Fiorini Morosini et al. / Procedia Engineering 186 (2017) 428 – 435
4. Conclusions In this paper, a methodology for emergency management of Water Distribution Systems (WDS), based on a nodal demand control has been proposed. The scope of the method is to manage the network in emergency condition and to guarantee for critical nodes (i.e. hospitals and vulnerable customers) the nodal demand with an adequate head. The management model, when a failure pipe occurs, has shown the benefit obtainable in the areas of the network where the head decreases and the delivered nodal demand are less than the requested demand. The analysis of the network has shown that at an improvement of critical nodes functioning corresponds a contained reduction of total circulating flow in the network. The results depend on the network complexity because the procedure is based on using sensitivity matrices and on the network sensitive nodes.
References [1] C.A. Bush and J.G. Uber, “Sampling design methods for water distribution model calibration”, Journal of Water Resources Planning and Management 124 (6), 1998, pp. 334-344. [2] M.O. Engelhardt, P.J. Skipworth, D.A. Savic, A.J. Saul, G.A. Walters, Rehabilitation strategies for water distribution networks: A literature review with a UK perspective, Urban Water, pp. 153-170, 2000. [3] K.E. Lansey, N. Duan, L.W. Mays, T.K. Tung, Optimal maintenance scheduling for water distribution systems, Civ. Eng. Syst., 9(3), 211–226, 1992. [4] Y. Kleiner,, B.J. Adams, J.S. Rogers, Selection and scheduling of rehabilitation alternatives for water distribution systems, Water Resources Research, 34 (8), 2053-2061, 1998. [5] R. Gargano, and D. Pianese, Reliability as Tool for Hydraulic Network Planning, Journal of Hydraulic Engineering, ASCE , (126 5), pp 354364, 2000. [6] L. Berardi, A.F. Colombo, O. Giustolisi, Analisi di affidabilità delle reti in pressione in presenza di valvole di intercettazione, XXXI Convegno Nazionale di Idraulica e Costruzioni Idrauliche, Perugia 2008. [7] R. Farmani, G.A. Walters, D.A. Savic,, Trade-off between Total Cost and Reliability for Anytown Water Distribution Network”, Journal of Water Resources Planning and Management, 131(3), 161-171, 2005. [8] G. de Marinis, R. Gargano, Z. Kapelan, M.S. Morley, D.A. Savic, C. Tricarico, A rehabilitation approach for water distribution systems based on a pressure driven analysis, XXXI Convegno Nazionale di Idraulica e Costruzioni Idrauliche, Perugia 2008. [9] S. Alvisi, M. Franchini. Near-optimal rehabilitation scheduling of water distribution systems based on a multi-objective genetic algorithm. Civil Engineering and Environmental Systems, 2006, 23.3: 143-160. [10] Giustolisi O. and Savic D., 2010. 'Identification of segments and optimal isolation valve system design in water distribution networks. Urban Water Journal, 7: 1, 1 — 15. [11] F. Calomino, P. Veltri, Un procedimento di calcolo automatico per la verifica delle reti idrauliche in pressione a portate incognite in ingresso ed in uscita, XVII Convegno di Idraulica e Costruzioni Idrauliche Palermo, 27-29 Ottobre 1980. [12] A. Fiorini Morosini, F. Caruso O., P. Veltri and Costanzo F., Water distribution Network Management in Emergency Conditions. Procedia Engineering, 2015, 119: 908-917. [13] P. Veltri, A. Fiorini Morosini, F. Orlando, La calibrazione delle scabrezze delle reti idriche: un approccio di tipo probabilistico basato su un modello PDA di risoluzione della rete, IV Seminary on "La ricerca delle perdite e la gestione delle reti di condotte", Aversa, September 2009, L’Acqua, n°2 (March/April), 2010. [14] D.A. Savic and G.A. Walters, “An Evolution Program for Optimal Pressure Regulation in Water Distribution Networks”, Engineering Optimization, Vol. 24, No. 3, 1995, pp. 197-219. [15] A. Fiorini Morosini, F. Costanzo, P. Veltri and D.A. Savic, Identification of Measurement Points for Calibration of Water Distribution Network Models. Procedia Engineering, 2014, 89: 693-701.
435