- Email: [email protected]
Preliminary modeling and control studies in AQUASOL project
Desalination 222 (2008) 466–473
Preliminary modeling and control studies in AQUASOL project Lidia Rocaa, Manuel Berenguelb, Luis Yebrac*, Diego C. Alarcónc a
Convenio Universidad de Almería-Plataforma Solar de Almería, Ctra. Senés s/n, 04200 Tabernas, Almería, Spain b Universidad de Almería, Dpto. Lenguajes y Computación, Ctra. Sacramento s/n, 04120, Almería, Spain c CIEMAT, Plataforma Solar de Almería, Ctra. Senés s/n, 04200 Tabernas, Almería, Spain Tel. +34 950 387923; Fax +34 950 365015; email: [email protected] Received 20 December 2006; accepted 7 January 2007
Abstract The objective of AQUASOL project at Plataforma Solar de Almería (CIEMAT) is to develop and test a distillation system using multi-effect technique whose energy source is solar. In order to guarantee the optimization in the operation, it is important to control in an efficient and secure way different process variables, in addition to provide the operators signal information in a flexible and appropriate way. This article shows a description of the plant, a summary of the dynamic modeling works realized up to date and a scheme of the base control system. Simulation results of the control system behavior during operation of the plant are presented. Keywords: Modeling; control; SCADA system; Multi-effect desalination
1. Introduction The main objective in a commercial process is to obtain a minimum production cost, keeping the system in safe operation points. In this way, control loops in desalination systems have been developed from years. Alatiqi et al. [1] explain the basic controller variables in reverse osmosis, as the production of freshwater with acceptable *Corresponding author.
purity level, pH to avoid scale formation and inlet pressure. For the multi-stage flash point of view, Alatiqi et al. [1], Ali et al. [2] and Maniar and Deshpande [3] explain that the controller variables are inlet water temperature to avoid scale formation, pressure and vapor temperature to reach saturated conditions, brine and distillate condensate level to avoid damages in pump system or brine flow to maintain the system efficiency. Regarding to variables to be controlled in multi-effect plants, Standiford [4] explains some of them as seawater
Presented at the conference on Desalination and the Environment. Sponsored by the European Desalination Society and Center for Research and Technology Hellas (CERTH), Sani Resort, Halkidiki, Greece, April 22–25, 2007. 0011-9164/06/$– See front matter © 2006 Published by Elsevier B.V. doi:10.1016/j.desal.2007.01.159
L. Roca et al. / Desalination 222 (2008) 466–473
flow to avoid high levels of concentrated brine, seawater pretreatment or inlet water temperature to avoid scale formation. On the other hand, multieffect plants does not require so many constraints as multi-stage flash technique, so controller design to maximize efficiency can be developed. AQUASOL project (“Enhanced Zero Discharge Seawater Desalination using Hybrid Solar Technology”), financed by CIEMAT (National Lab of the Education and Science Spanish Ministry) is a technical development in desalination that combines a thermal desalination system and a solar field with a double effect absorption heat pump to reduce the cost of water production. Preliminary studies in AQUASOL system show the necessity to focus initially the control on the solar field coupled to the multi-effect plant in order to maintain the thermal power source. For this reason, a feedback linearization based technique control system has been developed to reach a desired behavior in solar field outlet temperature. This paper is organized as follows: first AQUASOL plant and its Supervisory Control And Data Acquisition (SCADA) will be briefly explained in section 2 and the control problematic in section 3. Solar field model that will be used in the main control system design will be shown in section 4, while control implementation with simulation results in section 5. Finally, conclusions will be summarized in section 6.
V1 V2 Effect1 CPC SOLAR FIELD
Gas Boiler
Secondary Water Tank
Primary Water Tank
DEAHP
Seawater MED
Brine Distillate
Fig. 1. AQUASOL scheme.
operation, the feed-water inlet temperature in the first cell must be 66.5°C. It is possible to reach this temperature with heat from a solar field and also by steam generated by an auxiliary gas boiler coupled to a double effect absorption heat pump, DEAHP (LiBr-H2O), that can work at variable loads of the steam (from 30% to 100%). Solar collector field consists of a compound parabolic concentrator (CPC) collectors (Fig. 3) coupled to two water storage tanks, 12 m3 capacity each one, used as heat storage system. Energy
2. System description The AQUASOL system of Plataforma Solar de Almería located in the southern of Spain (Fig. 1), proposes a solar distillation technology based on a multi-effect distillation (MED) plant with 3 m3/h nominal distillate production. The MED plant (Fig. 2) consists of 14 effects in vertical arrangement at decreasing pressures from cell 1 to cell 14 [5]. Seawater is preheated and pumped to the first cell where it goes down to the following cells by gravity at the same time that part of the water is evaporated. For optimal
467
Fig. 2. Multi-effect plant.
468
L. Roca et al. / Desalination 222 (2008) 466–473
2.1. SCADA system
Fig. 3. Solar field loop in AQUASOL plant.
supplied by the solar field and DEAHP is transferred to a thermal storage tank using water as the heat transfer fluid. To reduce the overall fossil energy expenditure, the low-pressure steam from the last cell in MED is used in the DEAHP, which decreases consumption from 200 kW to 90 kW [5]. Finally, V2 is a three-way regulating valve used to reach the inlet first effect nominal temperature by mixing water from primary tank with that returned from first effect. The AQUASOL plant can operate in three different modes; solar, fossil and hybrid. In the solar mode, water from secondary tank is pumped to the solar field where it is heated to more than 66.5°C returning then to primary tank. V1 is an on-off valve used to recirculate water in the solar field, through secondary tank, until nominal temperature is reached. Through the recirculation mode, cooling situations in primary tank that can occur under some boundary conditions can be avoided. In this solar mode, MED inlet temperature is reached with solar energy only. In fossil mode, the MED plant operates with the gas boiler coupled to the DEAHP and working at full load, making possible to obtain freshwater in cloudy days and night operation. The two previous modes are combined in hybrid mode that is useful when the solar field needs to make use of fossil energy to reach the desired temperature in MED plant. In this case, DEAHP works at partial load just to be a support but reducing fossil consumption as much as possible.
Due to the fact that the control system must be efficient and secure and it must also provide the operators with fast, appropriate signal information, a SCADA system has been developed. System signals are received by read and write IMP (isolated measurement pods) cards and a National Instrument PXI (PCI eXtension for Instrumentation) system. Communication with DEAHP is carried out through OPC (OLE — object linking and embedding — for process control) server and a communication processor SIMATIC (automation system brand from Siemens) CP 343-1 connected to the DEAHP PLC (programmable logic controller). All the signals are displayed on the Human Machine Interface (HMI) developed with LabView of National Instruments. As well as being a support for the daily system operation, from the control point of view the SCADA system provides facilities to design control systems increasing the technology improvements. For the solar field control purposes that will be explained in this paper, SCADA system provides basic process variable as solar field temperatures, water flows, tank temperatures and global irradiation. 3. Control problem Since AQUASOL plant was set up, different control objectives have come up. Specifically, it has been worked in basic control loops, emergency systems and efficiency maximization. 3.1. Basic control loops Initially, the basic controllers that were implemented to set up the plant were four PID (proportional integral derivative) [6] local controllers (Fig. 4) for inlet solar field water flow, inlet water temperature and flow in the first effect of MED plant and inlet water flow to DEAHP. These four PIDs are analog controllers, so manually it is possible to assign the typical parameters
L. Roca et al. / Desalination 222 (2008) 466–473
r
e
PID controller u(t ) = Kp(e(t ) + 1 ∫ e(t )dt de(t ) Ti +Td dt )
u
469 c=0
x Process
By default
Control off
Fig. 4. PID scheme.
By default
Control on
c=1 c=0
To < Tin + 3°C
c=0
(proportional Kp, integral Ti and derivative Td terms) in order to obtain an adequate response behavior. On the other hand, reference can be replaced manually or in remote way through SCADA system.
Tmax > 85°C HighT To > Tin + 8°C
Tmax > 75°C
Tmax > 85°C
Turn off pump
To < Tin + 8°C
To < Tin + 8°C Turn on pump
3.2. Emergency loops One situation that must be avoided in solar collector field is cold temperatures. During the night, collectors reach low temperatures that can produce damages in the materials. In order to avoid irreparable damages, a state machine based control system was developed so solar field pump is turned on when a minimum temperature is detected. Water flow is kept until a complete renovation of water has been produced. In the same way, another state machine acts over high temperatures to avoid evaporation situations in the water inside collectors (Figs. 5 and 6).
By default
Tmin < 6°
Turn on pump c=0
Timeout
Control on
By default
c=1
c=0
Fig. 5. LowT state machine.
Control off
Fig. 6. HighT state machine.
3.3. Efficiency loops In order to obtain an optimal production of distillation, it is necessary to reach a specific water temperature range at the inlet of the heat exchanger in first effect of the MED plant. Higher temperatures could produce scale formation in the heat exchanger, but lower temperatures result in a distillation production reduction. Although inlet MED temperature is controlled by using the three-way valve, it is necessary to keep the energy in the storage system in order to guarantee the source of inlet water temperature. For this reason, a control system for the solar field temperature is required. One possibility to guarantee enough temperature in the storage system is to develop an outlet solar field temperature control. Control system in solar field must keep the outlet solar field temperature using inlet solar field pump as the signal control in spite of disturbances. The most important disturbance involved in the system is irradiation because of the continuous variation due to the cycle day and the unpredictable cloud
470
L. Roca et al. / Desalination 222 (2008) 466–473
transitories. Constraints in inlet MED temperature are solved by valve, but there are also solar field constraints that must be considered. Due to the pump characteristics, maximum inlet flow is limited to 4.7e-3 m3/s and, in order to maintain hydraulic equilibrium in the solar collectors, minimum flow is limited to 1e-3 m3/s. This control possibility allows operators select temperature reference, but could be also possible let control system find the optimal reference temperature subject to certain constraints. In this other case the objective is to control outlet solar field temperature in the same way than in the previous situation, but reference will be defined now by an optimization process that maximizes power supplied to storage water system. In order to avoid damages in the collectors due to high-temperature differences, as well as flow, outlet temperature constraints must be considered. Specifically, the constraint in solar field is defined to keep a temperature difference, ΔT between 10 and 20.
To model the solar field, hydraulic equilibrium must be considered. This means, collectors are arranged in reverse feeding mode so outlet temperature of the solar field will be the same than the outlet temperature in each group of nine collectors. But also it is possible to model this group of nine collectors modeling a hypothetic equivalent tube collector with an equivalent length, Leq, and an equivalent inlet mass flow meq that is a portion of the real inlet solar field mass flow. Dynamic behavior of this solar field tube can be determined by a distributed parameter nonlinear model based on partial differential equations as Camacho et al. [7] explain. Outlet solar field temperature, T ′, varies depending on disturbances in irradiation I, ambient temperature Ta, mass flow mF and inlet solar field temperature Tin as it is showed in expression (1). ρ ⋅Cp ⋅ A ⋅
∂T ′(t ) H = η ⋅ G ⋅ I (t ) − ⋅ (T (t ) − Ta (t )) ∂t Leq
− C p ⋅ m eq (t ) ⋅
4. Solar field dynamic model Solar collector field is composed of 252 CPCs, with a total surface area of approximately 500 m2, divided into four rows of 63 collectors. Within each row, groups of three collectors are connected in parallel, and each group of three is connected to another in series. Finally, these groups of nine collectors are connected in parallel to the row header pipe (Fig. 7).
T Ti (t − dt in)
Tin To
Inlet water Outlet water
Fig. 7. AQUASOL solar field scheme.
T (t ) =
T ′ (t ) − Tin (t − dtin ) Leq
T ′ (t ) − Tin (t − dtin ) Leq
(1)
(2)
where model parameters are as follows: • r = water density (kg/m3) • Cp = thermal capacity (J/kg/°C) • A = absorber cross-section area (m2) • h = optical efficiency • G = collector aperture (m) • H = thermal losses coefficient (J/s/K) • Leq = equivalent pipe length (m) In order to simplify the model for control purposes, it is assumed a constant value for the heat losses coefficient, H, obtained experimentally calibrating the model in different operating days. Beside this, absorber temperature is modeled – as the mean temperature, T , between outlet and inlet water temperature.
L. Roca et al. / Desalination 222 (2008) 466–473
Another simplification is the use of nontemperature dependence in water properties in order to reduce as much as possible outlet temperature model. Specifically and due to the operating temperature range, fluid properties in model correspond at 65°C. Variables I, Ta, Tin and mF are measured with sensors placed in solar field. It is important to notice that between the inlet solar field temperature sensor and the inlet pipe of a group of nine collectors exists a transport delay, dtin that varies depending on the fluid velocity, v(t) and using the distance, l1, that the fluid has to cover to reach the CPC tube as Normey [8] explains:
l1 = ∫
dtin
0
v(t )dt
(3)
To obtain the outlet solar field temperature (To) it must be considered another delay in the outlet group of nine collectors temperature (T ′) that depends on the inlet mass flow in the same way as in the inlet solar field temperature, but taking into account the pipe length, l2, from the outlet of the group of nine collectors1 and the position of the outlet temperature sensor. To = T ′(t − dto)
471 Ti (t − dtin) Boundary conditions + State variables
Ta I To
To*
+–
Feedback linearization control
m ˙F
CPC solar field
To
Fig. 8. Solar field control scheme.
nonlinearities with the feedback loop as it is explained in Slotine and Li [9]. The block structure applied to reach the desired outlet solar field temperature is represented in Fig. 8. Feedback linearization control has a PI (proportional integral) implementation as the linear control and a mapping based on the solar field model explained in section 4. Disturbances are taken into account inside mapping structure. On the other hand, Fig. 9 shows the same scheme as in the temperature control but including a power optimization function in order to maximize the power delivered from the solar field to the storage system. The reference in this case is calculated by the optimization function while feedback linearization control works as in the previous case. Because of the inlet flow-outlet temperature dependence, the optimal control
5. First results maximizing thermal Power Nonlinearities present in solar field system can be faced for control purposes by linearizing the model around an operation point, so a linear control can be applied. But, due to the high range of operation points in which solar field works during the whole day, an alternative solution with a feedback linearization control has been taken. The general idea of feedback linearization technique is the use of a linear control but transforming the nonlinear system into linear one cancelling
1 ≤ m˙ F ≤ 5 kg/s Tin ≤ To ≤ Tin + 20 Ti (t − dt in) Ta I To
Feedback linearization control
To*
Boundary conditions + State variables
m ˙F
CPC solar field
To
Power optimization
1
l1 and l2 lengths depend on the group of nine collectors selected.
Fig. 9. Solar field control scheme with power optimization.
472
L. Roca et al. / Desalination 222 (2008) 466–473
reference will be subject to the equation (1) and the control constraints:
Tin ≤ To ≤ Tin + 20 To simulate and compare both schemes, disturbances obtained from an experiment day on the 7th of March 2006 will be used. Fig. 10 shows global irradiation with small variations at the beginning but strong falls at the end of the operation. Inlet solar field temperature is increasing progressively due to the temperatures reached in storage system, but at 12 h temperature falls due to multi-effect plant was turned on flowing cold water to primary tank. Finally ambient temperature varies between 12°C and 21°C. Using the first scheme with a reference varying in order to obtain a solar field temperature gain of 15°C, results in Fig. 11 were obtained. In this graph, it can be observed that outlet temperature is following the reference by modifying control signal (Fig. 12 ) although low variations in irradiation occur. On the other hand, when hard disturbances in irradiation occur, control shows an aggressive behavior to try to reach the
90 Outlet solar field temperature (°C)
1 ≤ m F ≤ 5 kg/s
100
80 70 60 Temperature reference with power optimizator Outlet temperature with power optimizator Temperature reference for 15°C of gain Outlet temperature for 15°C of gain
50 40 30 20 10 0
9
10
11
12 13 Local time (h)
15
16
Fig. 11. Solar field control simulation, 7 March 2006.
reference but, with so fast variations, it is difficult to keep the temperature to the desired value. Fig. 11 also shows output with power optimization, where reference is lower than in the previous case so inlet flow curve stays over the curve without power optimization. A simulation of the power delivered to the storage water system is shown in Fig. 13 where the desired results of greater delivered power reached with optimization can be observed.
1500
5.0
Power optimizator Temperature control
1000 4.5 9
10
11
Irradiation (W/m2) 13 14
12
15
20 15 Ambient temperature (°C) 10 9 80
10
11
12
13
14
15
4.0
16
25
16
Water flow (L/s)
500 0
14
3.5 3.0 2.5 2.0
60 1.5
40 20 9
Inlet solar field temperature (°C) 10
11
12 13 Local time (h)
Fig. 10. Disturbances, 7 March 2006.
14
15
1.0 16
9
10
11
12 13 Local time (h)
14
15
Fig. 12. Control signal simulation, 7 March 2006.
16
L. Roca et al. / Desalination 222 (2008) 466–473
Power delivered by solar field to the thermal storage system (W)
For example, a new cost function including electrical losses can be formulated in future works.
×105
3.5
473
Power optimizator Temperature control
3.0 2.5
Acknowledgements
2.0
The authors wish to thank CIEMAT, the Ministerio de Educación y Ciencia (CICYT-FEDER DPI200407444-C04-04) and to the European Commission for the funds provided for AQUASOL Project (contract np. EVK1-CT2001-00102).
1.5 1.0 0.5 0
−0.5
9
10
11
12 13 Local Time (h)
14
15
16
Fig. 13. Power delivered simulation by solar field to the thermal storage system, 7 March 2006.
References [1]
[2]
6. Conclusions Primary implementations in control field in AQUASOL project have been described. The strategy followed during the start-up of the plant was first of all to define basic control loops for the operation and emergency systems for the damages that can appear working out of defined operation points. Once these two parts were finished, the objective is the development of advanced controllers to maximize production of freshwater and reduce fossil consumption. As a first example, a solar field control has been shown in order to obtain maximum power delivered to the thermal storage system. Although power obtained is higher with an optimization function, high flows are required, so electric power consumption increases. For this reason, it is important to study in every operational day the control objectives depending on weather conditions or production demand.
[3]
[4] [5]
[6] [7]
[8]
[9]
I. Alatiqi, H. Ettouney and H. El-Dessouky, Process control in water desalination industry: an overview, Desalination, 126 (1999) 15–32. E. Ali, K. Alhumaizi and A. Ajbar, Model reduction and robust control of multi-stage flash (MSF) desalination plants, Desalination, 121 (1999) 65–85. V.M. Maniar and P.B. Deshpande, Advanced controls for multi-stage flash (MSF) desalination plant optimization, J. Process Contr., 6 (1996) 49–66. F.C. Standiford, Control in multiple effect evaporator plants, Desalination, (1986) 263–292. E. Zarza, Solar thermal desalination project phase II results and final project report, Colecci ‘on Documentos Ciemat, Madrid, 1995. K.J. Astrom, Computer-Controlled Systems Theory and Design, Prentice Hall, New Jersey, 1997. E.F. Camacho, M. Berenguel, and F.R. Rubio, Advanced control of solar plants, Springer Verlag, London, 1997. J.E. Normey, A Robust adaptive deadtime compensator with application to a solar collector field, IFAC International Workshop on Linear Time Delay Systems, Grenoble (France), 1998, pp. 105–110. J. Slotine and W. Li, Applied nonlinear control, Prentice Hall, USA, 1991.
Preliminary modeling and control studies in AQUASOL project Lidia Rocaa, Manuel Berenguelb, Luis Yebrac*, Diego C. Alarcónc a
Convenio Universidad de Almería-Plataforma Solar de Almería, Ctra. Senés s/n, 04200 Tabernas, Almería, Spain b Universidad de Almería, Dpto. Lenguajes y Computación, Ctra. Sacramento s/n, 04120, Almería, Spain c CIEMAT, Plataforma Solar de Almería, Ctra. Senés s/n, 04200 Tabernas, Almería, Spain Tel. +34 950 387923; Fax +34 950 365015; email: [email protected] Received 20 December 2006; accepted 7 January 2007
Abstract The objective of AQUASOL project at Plataforma Solar de Almería (CIEMAT) is to develop and test a distillation system using multi-effect technique whose energy source is solar. In order to guarantee the optimization in the operation, it is important to control in an efficient and secure way different process variables, in addition to provide the operators signal information in a flexible and appropriate way. This article shows a description of the plant, a summary of the dynamic modeling works realized up to date and a scheme of the base control system. Simulation results of the control system behavior during operation of the plant are presented. Keywords: Modeling; control; SCADA system; Multi-effect desalination
1. Introduction The main objective in a commercial process is to obtain a minimum production cost, keeping the system in safe operation points. In this way, control loops in desalination systems have been developed from years. Alatiqi et al. [1] explain the basic controller variables in reverse osmosis, as the production of freshwater with acceptable *Corresponding author.
purity level, pH to avoid scale formation and inlet pressure. For the multi-stage flash point of view, Alatiqi et al. [1], Ali et al. [2] and Maniar and Deshpande [3] explain that the controller variables are inlet water temperature to avoid scale formation, pressure and vapor temperature to reach saturated conditions, brine and distillate condensate level to avoid damages in pump system or brine flow to maintain the system efficiency. Regarding to variables to be controlled in multi-effect plants, Standiford [4] explains some of them as seawater
Presented at the conference on Desalination and the Environment. Sponsored by the European Desalination Society and Center for Research and Technology Hellas (CERTH), Sani Resort, Halkidiki, Greece, April 22–25, 2007. 0011-9164/06/$– See front matter © 2006 Published by Elsevier B.V. doi:10.1016/j.desal.2007.01.159
L. Roca et al. / Desalination 222 (2008) 466–473
flow to avoid high levels of concentrated brine, seawater pretreatment or inlet water temperature to avoid scale formation. On the other hand, multieffect plants does not require so many constraints as multi-stage flash technique, so controller design to maximize efficiency can be developed. AQUASOL project (“Enhanced Zero Discharge Seawater Desalination using Hybrid Solar Technology”), financed by CIEMAT (National Lab of the Education and Science Spanish Ministry) is a technical development in desalination that combines a thermal desalination system and a solar field with a double effect absorption heat pump to reduce the cost of water production. Preliminary studies in AQUASOL system show the necessity to focus initially the control on the solar field coupled to the multi-effect plant in order to maintain the thermal power source. For this reason, a feedback linearization based technique control system has been developed to reach a desired behavior in solar field outlet temperature. This paper is organized as follows: first AQUASOL plant and its Supervisory Control And Data Acquisition (SCADA) will be briefly explained in section 2 and the control problematic in section 3. Solar field model that will be used in the main control system design will be shown in section 4, while control implementation with simulation results in section 5. Finally, conclusions will be summarized in section 6.
V1 V2 Effect1 CPC SOLAR FIELD
Gas Boiler
Secondary Water Tank
Primary Water Tank
DEAHP
Seawater MED
Brine Distillate
Fig. 1. AQUASOL scheme.
operation, the feed-water inlet temperature in the first cell must be 66.5°C. It is possible to reach this temperature with heat from a solar field and also by steam generated by an auxiliary gas boiler coupled to a double effect absorption heat pump, DEAHP (LiBr-H2O), that can work at variable loads of the steam (from 30% to 100%). Solar collector field consists of a compound parabolic concentrator (CPC) collectors (Fig. 3) coupled to two water storage tanks, 12 m3 capacity each one, used as heat storage system. Energy
2. System description The AQUASOL system of Plataforma Solar de Almería located in the southern of Spain (Fig. 1), proposes a solar distillation technology based on a multi-effect distillation (MED) plant with 3 m3/h nominal distillate production. The MED plant (Fig. 2) consists of 14 effects in vertical arrangement at decreasing pressures from cell 1 to cell 14 [5]. Seawater is preheated and pumped to the first cell where it goes down to the following cells by gravity at the same time that part of the water is evaporated. For optimal
467
Fig. 2. Multi-effect plant.
468
L. Roca et al. / Desalination 222 (2008) 466–473
2.1. SCADA system
Fig. 3. Solar field loop in AQUASOL plant.
supplied by the solar field and DEAHP is transferred to a thermal storage tank using water as the heat transfer fluid. To reduce the overall fossil energy expenditure, the low-pressure steam from the last cell in MED is used in the DEAHP, which decreases consumption from 200 kW to 90 kW [5]. Finally, V2 is a three-way regulating valve used to reach the inlet first effect nominal temperature by mixing water from primary tank with that returned from first effect. The AQUASOL plant can operate in three different modes; solar, fossil and hybrid. In the solar mode, water from secondary tank is pumped to the solar field where it is heated to more than 66.5°C returning then to primary tank. V1 is an on-off valve used to recirculate water in the solar field, through secondary tank, until nominal temperature is reached. Through the recirculation mode, cooling situations in primary tank that can occur under some boundary conditions can be avoided. In this solar mode, MED inlet temperature is reached with solar energy only. In fossil mode, the MED plant operates with the gas boiler coupled to the DEAHP and working at full load, making possible to obtain freshwater in cloudy days and night operation. The two previous modes are combined in hybrid mode that is useful when the solar field needs to make use of fossil energy to reach the desired temperature in MED plant. In this case, DEAHP works at partial load just to be a support but reducing fossil consumption as much as possible.
Due to the fact that the control system must be efficient and secure and it must also provide the operators with fast, appropriate signal information, a SCADA system has been developed. System signals are received by read and write IMP (isolated measurement pods) cards and a National Instrument PXI (PCI eXtension for Instrumentation) system. Communication with DEAHP is carried out through OPC (OLE — object linking and embedding — for process control) server and a communication processor SIMATIC (automation system brand from Siemens) CP 343-1 connected to the DEAHP PLC (programmable logic controller). All the signals are displayed on the Human Machine Interface (HMI) developed with LabView of National Instruments. As well as being a support for the daily system operation, from the control point of view the SCADA system provides facilities to design control systems increasing the technology improvements. For the solar field control purposes that will be explained in this paper, SCADA system provides basic process variable as solar field temperatures, water flows, tank temperatures and global irradiation. 3. Control problem Since AQUASOL plant was set up, different control objectives have come up. Specifically, it has been worked in basic control loops, emergency systems and efficiency maximization. 3.1. Basic control loops Initially, the basic controllers that were implemented to set up the plant were four PID (proportional integral derivative) [6] local controllers (Fig. 4) for inlet solar field water flow, inlet water temperature and flow in the first effect of MED plant and inlet water flow to DEAHP. These four PIDs are analog controllers, so manually it is possible to assign the typical parameters
L. Roca et al. / Desalination 222 (2008) 466–473
r
e
PID controller u(t ) = Kp(e(t ) + 1 ∫ e(t )dt de(t ) Ti +Td dt )
u
469 c=0
x Process
By default
Control off
Fig. 4. PID scheme.
By default
Control on
c=1 c=0
To < Tin + 3°C
c=0
(proportional Kp, integral Ti and derivative Td terms) in order to obtain an adequate response behavior. On the other hand, reference can be replaced manually or in remote way through SCADA system.
Tmax > 85°C HighT To > Tin + 8°C
Tmax > 75°C
Tmax > 85°C
Turn off pump
To < Tin + 8°C
To < Tin + 8°C Turn on pump
3.2. Emergency loops One situation that must be avoided in solar collector field is cold temperatures. During the night, collectors reach low temperatures that can produce damages in the materials. In order to avoid irreparable damages, a state machine based control system was developed so solar field pump is turned on when a minimum temperature is detected. Water flow is kept until a complete renovation of water has been produced. In the same way, another state machine acts over high temperatures to avoid evaporation situations in the water inside collectors (Figs. 5 and 6).
By default
Tmin < 6°
Turn on pump c=0
Timeout
Control on
By default
c=1
c=0
Fig. 5. LowT state machine.
Control off
Fig. 6. HighT state machine.
3.3. Efficiency loops In order to obtain an optimal production of distillation, it is necessary to reach a specific water temperature range at the inlet of the heat exchanger in first effect of the MED plant. Higher temperatures could produce scale formation in the heat exchanger, but lower temperatures result in a distillation production reduction. Although inlet MED temperature is controlled by using the three-way valve, it is necessary to keep the energy in the storage system in order to guarantee the source of inlet water temperature. For this reason, a control system for the solar field temperature is required. One possibility to guarantee enough temperature in the storage system is to develop an outlet solar field temperature control. Control system in solar field must keep the outlet solar field temperature using inlet solar field pump as the signal control in spite of disturbances. The most important disturbance involved in the system is irradiation because of the continuous variation due to the cycle day and the unpredictable cloud
470
L. Roca et al. / Desalination 222 (2008) 466–473
transitories. Constraints in inlet MED temperature are solved by valve, but there are also solar field constraints that must be considered. Due to the pump characteristics, maximum inlet flow is limited to 4.7e-3 m3/s and, in order to maintain hydraulic equilibrium in the solar collectors, minimum flow is limited to 1e-3 m3/s. This control possibility allows operators select temperature reference, but could be also possible let control system find the optimal reference temperature subject to certain constraints. In this other case the objective is to control outlet solar field temperature in the same way than in the previous situation, but reference will be defined now by an optimization process that maximizes power supplied to storage water system. In order to avoid damages in the collectors due to high-temperature differences, as well as flow, outlet temperature constraints must be considered. Specifically, the constraint in solar field is defined to keep a temperature difference, ΔT between 10 and 20.
To model the solar field, hydraulic equilibrium must be considered. This means, collectors are arranged in reverse feeding mode so outlet temperature of the solar field will be the same than the outlet temperature in each group of nine collectors. But also it is possible to model this group of nine collectors modeling a hypothetic equivalent tube collector with an equivalent length, Leq, and an equivalent inlet mass flow meq that is a portion of the real inlet solar field mass flow. Dynamic behavior of this solar field tube can be determined by a distributed parameter nonlinear model based on partial differential equations as Camacho et al. [7] explain. Outlet solar field temperature, T ′, varies depending on disturbances in irradiation I, ambient temperature Ta, mass flow mF and inlet solar field temperature Tin as it is showed in expression (1). ρ ⋅Cp ⋅ A ⋅
∂T ′(t ) H = η ⋅ G ⋅ I (t ) − ⋅ (T (t ) − Ta (t )) ∂t Leq
− C p ⋅ m eq (t ) ⋅
4. Solar field dynamic model Solar collector field is composed of 252 CPCs, with a total surface area of approximately 500 m2, divided into four rows of 63 collectors. Within each row, groups of three collectors are connected in parallel, and each group of three is connected to another in series. Finally, these groups of nine collectors are connected in parallel to the row header pipe (Fig. 7).
T Ti (t − dt in)
Tin To
Inlet water Outlet water
Fig. 7. AQUASOL solar field scheme.
T (t ) =
T ′ (t ) − Tin (t − dtin ) Leq
T ′ (t ) − Tin (t − dtin ) Leq
(1)
(2)
where model parameters are as follows: • r = water density (kg/m3) • Cp = thermal capacity (J/kg/°C) • A = absorber cross-section area (m2) • h = optical efficiency • G = collector aperture (m) • H = thermal losses coefficient (J/s/K) • Leq = equivalent pipe length (m) In order to simplify the model for control purposes, it is assumed a constant value for the heat losses coefficient, H, obtained experimentally calibrating the model in different operating days. Beside this, absorber temperature is modeled – as the mean temperature, T , between outlet and inlet water temperature.
L. Roca et al. / Desalination 222 (2008) 466–473
Another simplification is the use of nontemperature dependence in water properties in order to reduce as much as possible outlet temperature model. Specifically and due to the operating temperature range, fluid properties in model correspond at 65°C. Variables I, Ta, Tin and mF are measured with sensors placed in solar field. It is important to notice that between the inlet solar field temperature sensor and the inlet pipe of a group of nine collectors exists a transport delay, dtin that varies depending on the fluid velocity, v(t) and using the distance, l1, that the fluid has to cover to reach the CPC tube as Normey [8] explains:
l1 = ∫
dtin
0
v(t )dt
(3)
To obtain the outlet solar field temperature (To) it must be considered another delay in the outlet group of nine collectors temperature (T ′) that depends on the inlet mass flow in the same way as in the inlet solar field temperature, but taking into account the pipe length, l2, from the outlet of the group of nine collectors1 and the position of the outlet temperature sensor. To = T ′(t − dto)
471 Ti (t − dtin) Boundary conditions + State variables
Ta I To
To*
+–
Feedback linearization control
m ˙F
CPC solar field
To
Fig. 8. Solar field control scheme.
nonlinearities with the feedback loop as it is explained in Slotine and Li [9]. The block structure applied to reach the desired outlet solar field temperature is represented in Fig. 8. Feedback linearization control has a PI (proportional integral) implementation as the linear control and a mapping based on the solar field model explained in section 4. Disturbances are taken into account inside mapping structure. On the other hand, Fig. 9 shows the same scheme as in the temperature control but including a power optimization function in order to maximize the power delivered from the solar field to the storage system. The reference in this case is calculated by the optimization function while feedback linearization control works as in the previous case. Because of the inlet flow-outlet temperature dependence, the optimal control
5. First results maximizing thermal Power Nonlinearities present in solar field system can be faced for control purposes by linearizing the model around an operation point, so a linear control can be applied. But, due to the high range of operation points in which solar field works during the whole day, an alternative solution with a feedback linearization control has been taken. The general idea of feedback linearization technique is the use of a linear control but transforming the nonlinear system into linear one cancelling
1 ≤ m˙ F ≤ 5 kg/s Tin ≤ To ≤ Tin + 20 Ti (t − dt in) Ta I To
Feedback linearization control
To*
Boundary conditions + State variables
m ˙F
CPC solar field
To
Power optimization
1
l1 and l2 lengths depend on the group of nine collectors selected.
Fig. 9. Solar field control scheme with power optimization.
472
L. Roca et al. / Desalination 222 (2008) 466–473
reference will be subject to the equation (1) and the control constraints:
Tin ≤ To ≤ Tin + 20 To simulate and compare both schemes, disturbances obtained from an experiment day on the 7th of March 2006 will be used. Fig. 10 shows global irradiation with small variations at the beginning but strong falls at the end of the operation. Inlet solar field temperature is increasing progressively due to the temperatures reached in storage system, but at 12 h temperature falls due to multi-effect plant was turned on flowing cold water to primary tank. Finally ambient temperature varies between 12°C and 21°C. Using the first scheme with a reference varying in order to obtain a solar field temperature gain of 15°C, results in Fig. 11 were obtained. In this graph, it can be observed that outlet temperature is following the reference by modifying control signal (Fig. 12 ) although low variations in irradiation occur. On the other hand, when hard disturbances in irradiation occur, control shows an aggressive behavior to try to reach the
90 Outlet solar field temperature (°C)
1 ≤ m F ≤ 5 kg/s
100
80 70 60 Temperature reference with power optimizator Outlet temperature with power optimizator Temperature reference for 15°C of gain Outlet temperature for 15°C of gain
50 40 30 20 10 0
9
10
11
12 13 Local time (h)
15
16
Fig. 11. Solar field control simulation, 7 March 2006.
reference but, with so fast variations, it is difficult to keep the temperature to the desired value. Fig. 11 also shows output with power optimization, where reference is lower than in the previous case so inlet flow curve stays over the curve without power optimization. A simulation of the power delivered to the storage water system is shown in Fig. 13 where the desired results of greater delivered power reached with optimization can be observed.
1500
5.0
Power optimizator Temperature control
1000 4.5 9
10
11
Irradiation (W/m2) 13 14
12
15
20 15 Ambient temperature (°C) 10 9 80
10
11
12
13
14
15
4.0
16
25
16
Water flow (L/s)
500 0
14
3.5 3.0 2.5 2.0
60 1.5
40 20 9
Inlet solar field temperature (°C) 10
11
12 13 Local time (h)
Fig. 10. Disturbances, 7 March 2006.
14
15
1.0 16
9
10
11
12 13 Local time (h)
14
15
Fig. 12. Control signal simulation, 7 March 2006.
16
L. Roca et al. / Desalination 222 (2008) 466–473
Power delivered by solar field to the thermal storage system (W)
For example, a new cost function including electrical losses can be formulated in future works.
×105
3.5
473
Power optimizator Temperature control
3.0 2.5
Acknowledgements
2.0
The authors wish to thank CIEMAT, the Ministerio de Educación y Ciencia (CICYT-FEDER DPI200407444-C04-04) and to the European Commission for the funds provided for AQUASOL Project (contract np. EVK1-CT2001-00102).
1.5 1.0 0.5 0
−0.5
9
10
11
12 13 Local Time (h)
14
15
16
Fig. 13. Power delivered simulation by solar field to the thermal storage system, 7 March 2006.
References [1]
[2]
6. Conclusions Primary implementations in control field in AQUASOL project have been described. The strategy followed during the start-up of the plant was first of all to define basic control loops for the operation and emergency systems for the damages that can appear working out of defined operation points. Once these two parts were finished, the objective is the development of advanced controllers to maximize production of freshwater and reduce fossil consumption. As a first example, a solar field control has been shown in order to obtain maximum power delivered to the thermal storage system. Although power obtained is higher with an optimization function, high flows are required, so electric power consumption increases. For this reason, it is important to study in every operational day the control objectives depending on weather conditions or production demand.
[3]
[4] [5]
[6] [7]
[8]
[9]
I. Alatiqi, H. Ettouney and H. El-Dessouky, Process control in water desalination industry: an overview, Desalination, 126 (1999) 15–32. E. Ali, K. Alhumaizi and A. Ajbar, Model reduction and robust control of multi-stage flash (MSF) desalination plants, Desalination, 121 (1999) 65–85. V.M. Maniar and P.B. Deshpande, Advanced controls for multi-stage flash (MSF) desalination plant optimization, J. Process Contr., 6 (1996) 49–66. F.C. Standiford, Control in multiple effect evaporator plants, Desalination, (1986) 263–292. E. Zarza, Solar thermal desalination project phase II results and final project report, Colecci ‘on Documentos Ciemat, Madrid, 1995. K.J. Astrom, Computer-Controlled Systems Theory and Design, Prentice Hall, New Jersey, 1997. E.F. Camacho, M. Berenguel, and F.R. Rubio, Advanced control of solar plants, Springer Verlag, London, 1997. J.E. Normey, A Robust adaptive deadtime compensator with application to a solar collector field, IFAC International Workshop on Linear Time Delay Systems, Grenoble (France), 1998, pp. 105–110. J. Slotine and W. Li, Applied nonlinear control, Prentice Hall, USA, 1991.