INTERNATIONAL JOURNAL OF ENERGY RESEARCH Int. J. Energy Res.

2011; 35:436448 Published online 29 April 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/er.1706

An iterative method for modelling the air-cooled organic Rankine cycle geothermal power plant
M. Imroz Sohel1,,y, Mathieu Sellier2, Larry J. Brackney3 and Susan Krumdieck2
1 2

Scion, Te Papa Tipu Innovation Park, 49 Sala Street, Rotorua, New Zealand Department of Mechanical Engineering, University of Canterbury, Private bag 4800, Christchurch, New Zealand 3 Commercial Building Systems Electricity, Resources, and Building Systems Integration Center, National Renewable Energy Laboratory (NREL) 1617 Cole Blvd, Mailstop 5202, Golden, CO 80401, U.S.A.

SUMMARY
This work presents an iterative method for modelling the effect of ambient air temperature on the air-cooled organic Rankine cycle. The ambient temperature affects the condenser performance, and hence the performance of the whole cycle, in two ways. First, changing the equilibrium pressure inside the condenser, the turbine outlet pressure and the turbine pressure ratio vary. Since the turbine pressure ratio is a major parameter in determining the power generated by a turbine, the plant output is directly affected. Second, changing the condenser outlet temperature with ambient temperature, the pump inlet and outlet conditions are changed. Thus, the vapourizer equilibrium temperature and pressure are inuenced. The developed method iteratively seeks the equilibrium conditions for both the condenser and vapourizer. Two case studies based on a real plant performance have been carried out to demonstrate the validity of the method. The developed method demonstrates robustness and converges regardless of the initial conditions allowed by the physical properties of the working uid. This method is effective for cycles that use saturated vapour as well as superheated vapour under static or dynamic conditions with appropriate initial conditions and constraints. The developed method may be applied to any Rankine cycle with closed cycle operation. Copyright r 2010 John Wiley & Sons, Ltd.
KEY WORDS geothermal power plant; air-cooling; organic Rankine cycle; performance analysis Correspondence *M. Imroz Sohel, Scion, Te Papa Tipu Innovation Park, 49 Sala Street, Rotorua, New Zealand. y E-mail: mohammed.sohel@scionresearch.com Contract/grant sponsor: University of Canterbury Received 17 November 2009; Revised 11 March 2010; Accepted 12 March 2010

1. INTRODUCTION
Geothermal resources, concentrated solar energy and waste heat are being increasingly investigated for power generation. The Kalina Cycle has been investigated for low-temperature resources, but limited commercial application has been demonstrated [14]. The advantages of two component working uids have been theoretically demonstrated [5], but organic Rankine cycle (ORC) power plants have been found to be the most economic and proven technology [6]. The literature of major applications of the ORC include combined heat and power [7,8], waste heat recovery [916], solar thermal application [1719], biomass heat and power plants [20,21] and geothermal [6,2226].
Copyright r 2010 John Wiley & Sons, Ltd.

The effect of ambient temperature on the gas turbine power plant is well studied [2729], but it differs signicantly from the ORC. In a gas turbine power plant, the ambient air temperature dictates the density of air and hence the power output. In an ORC, the ambient air temperature affects the plant performance by affecting the heat rejection from the system. ORC geothermal power plants often use air-cooled condensers that make these power plants more susceptible to weather conditions. Generally, these plants are designed assuming a reasonable, stable, ambient temperature. As the ambient temperature increases, especially during the summer, the performance of an ORC plant is signicantly reduced [25]. Power companies make the bulk of their revenue when demand is high. Therefore, it is very important

Air-cooled ORC geothermal power plant method for modelling

M. I. Sohel et al.

for them to have high power capacity to maximize the benet during high demand periods. The sensitivity of binary cycle power plants (such as the ORC) to ambient conditions creates challenges for producers to predict power output accurately. A model that can predict hourly plant performance on a daily basis would be very useful for optimizing capacity. Both fundamental and empirical modelling methods have been used to predict binary cycle plant performance. Investigations based on the second law of thermodynamics are prevalent [6,25,30]. The ambient air acts as the heat sink temperature for an air-cooled condenser plant. As the ambient temperature increases, the Carnot efciency decreases, Z1 TL TH 1

vapourizer equilibrium temperature and pressure. The method presented here seeks the equilibrium conditions of both the condenser and the vapourizer based on initial conditions (supplied to the model as the starting point of the calculation) and external factors. Stability and rate of convergence of the developed method are important considerations and are also discussed in this paper. Two case studies are presented that demonstrate the application of the iterative modelling method. For modelling, MatLab [36] interfaced with the thermophysical property database REFPROP [37] has been used.

2. THE ITERATIVE METHOD

Figure 1 shows a T s presentation of a Rankine cycle. Process 4-1 represents constant pressure heat rejection by the condenser. Applying the rst law of thermodynamics (neglecting kinetic and potential energies) for an open system, we can write: _ _ mcycle h1 h4 Qcon 2

In Equation (1), TL and TH are the absolute temperatures of the heat sink and source, respectively, and Z is the Carnot efciency. The plant performance may also be expressed as an empirically derived function of various operating parameters including ambient temperature and condenser working pressure and temperature. Empirical modelling approaches are common practice in industry and are well described in the literature [3133]. The plant performance is generally expressed as a function of the turbine pressure ratio result in a set of charts or curves. Such tted curves are often plant specic and only apply for a narrow range of operations, rendering them limited for addressing the broader class of plant design and optimization problems. The effect of condenser pressure on plant performance has been widely discussed in the literature [3134]. The general consensus is that increasing condenser pressure generally results in a decrease in power output. A clear discussion on how ambient temperature affects the condenser performance and hence the overall plant performance is not readily available in the literature. Durmayaz and Sogut [35] presented an iterative method to calculate the condenser equilibrium condition of a pressurized water reactor nuclear power plant with cooling water temperature affected by weather conditions. The iterative approach allows the performance of the whole plant to be calculated using fundamental rather than empirical relationships. The present work introduces an iterative method for modelling a closed ORC power plant. The heat sink (ambient) temperature affects the condenser performance, consequently inuencing the whole cycle performance in two ways. First, changing the equilibrium pressure inside the condenser leads to changes in both the turbine outlet pressure and the related turbine pressure ratio. Second, changing the condenser outlet temperature (via the heat sink temperature) results in changes in both the pump inlet and outlet conditions. This, in turn, inuences both the
Int. J. Energy Res. 2011; 35:436448 r 2010 John Wiley & Sons, Ltd. DOI: 10.1002/er

_ where mcycle is the mass ow rate in the cycle, h1 is the enthalpy at state point 1, h4 is the enthalpy at state _ point 4 and Qcon is the condenser heat load. Since, process 4-1 is an isobaric process: p4 p1 3

_ The condenser heat load, Qcon , is a function of inlet condition (state point 4), outlet condition (state point 1), heat sink temperature, mass ow rate in the cycle and design of the condenser. During steady-state operation of a plant, the mass ow rate in the cycle is conserved. We assume that the design of a particular condenser is xed. Therefore, condenser inlet and outlet conditions and the heat sink temperature are the primary parameters inuencing the condenser performance. A comprehensive model of the condenser heat load was developed previously using the fundamentals of
T

3 2 1 4s 4

2s

s
Figure 1. Ts diagram of a Rankine cycle.

437

M. I. Sohel et al.

Air-cooled ORC geothermal power plant method for modelling

thermodynamics, heat transfer and condenser design presented elsewhere [26]. Assuming that TD is the designed heat sink tem_ perature of the condenser, the mass ow rate, mdesign , in the cycle at sink temperature, TD, may be expressed as: _ _ mdesign mcycle _ Qcon h1 h4 4

Applying the rst law of thermodynamics (neglecting kinetic and potential energies) for an open system [39] for the process 3,0 to 30 , the following equation is obtained: du u30 u3;0 h30 h3;0 p30 v30 p3;0 v3;0 7 where u is the internal energy and v is the specic volume. The subscript 3,0 represents the initial condition of state 3, which is calculated from the supplied operating pressure and temperature (typical) values of the vapourizer to start the calculation. The specic volume (v) of the uid inside the boiler/ vapourizer can change if the operating parameters (p, T) change from one steady-state operating point to another (i.e. from 3,0 to 30 or 30 to 3). However, if the volume inside the boiler/vapourizer is kept unchanged by means of any control mechanism (which is normally the case), then the specic volume (v) can be assumed constant. In the intermediate processes (i.e. from 30 to 3), no work is done by the system nor does any heat transfer take place. Rather, equilibrium is attained by the addition or departure of some mass to or from the system, which ensures constant specic volume of the system. The relative quantity of mass (compared with the holdup mass) to be added or departed to or from the vapourizer to attain equilibrium in the vapourizer is very small. If we neglect the effect of this added or departed mass, then the process from 30 to 3 can be assumed to be closed system operation. Now, applying the rst law of thermodynamics for a closed system [39] and neglecting the kinetic and the potential energies: du30 3 dQ30 3 dW30 3
30 3 30 3

Since the condenser heat load is directly related to the ambient temperature at constant vapourliquid equilibrium condition, if TD increases or decreases, the condenser heat load varies inversely resulting in a lower or higher mass ow rate in the cycle. This has an adverse effect on the cycle performance as power plants are generally optimized for a specic operating condition. The only way to maintain the operating condition is to change the vapourliquid equilibrium state in the condenser to cool the same amount of working uid. The following method is used to nd the required vapourliquid equilibrium condition in the condenser to maintain constant mass ow: _ Step 1: Calculate Qcon based on the heat sink temperature using a condenser model [26] then a value of _ mcycle is calculated from Equation (4). _ _ Step 2: If mcycle mdesign , the condenser is operating at the designed vapourliquid equilibrium condition and no further calculation is necessary. _ _ Step 3: If mcycle omdesign , the equilibrium pressure, _ _ p1, is reduced until mcycle mdesign and h1 is calculated as h1 fp1 ; x 0, where x represents quality. _ _ Step 4: If mcycle 4mdesign , the equilibrium pressure, _ _ p1, is increased until mcycle mdesign and h1 is calculated as h1 fp1 ; x 0. The back pressure of the condenser dictates the turbine outlet pressure. For the positive ow to occur, the condenser pressure must be less than the turbine outlet pressure (p1 op4 ). In practice, the turbine outlet pressure is slightly higher than the condenser equilibrium pressure. Recalling Figure 1, process 1-2 represents isentropic compression of the working uid by the cycle pump. Owing to the fact that the pump work is relatively small for an ORC, it can be assumed constant [38]. The work input to the pump may be calculated as: _ _ Wpump mcycle h2 h1 5

where dQ and dW are innitesimally small, so du30 3 should tend to zero for the hypothetical process 30 to 3 at equilibrium condition. Now, Equation (7) can be re-arranged to identify state 3 as: u3 u3;0 h3 h3;0 p3 p3;0 v3;0 9 Equation (9) can be solved iteratively for x 5 1 by altering p3, until the left-hand side of Equation (9) equals du of Equation (7). Knowing p3 and x 5 1, the rest of the thermodynamic properties associated with state 3 may be determined. If we assume that the work done by the turbine is done isentropically, then: s 3 s4 10 Knowing s4 and p4, the rest of the thermodynamic properties associated with state 4 may be determined. Work done by the turbine is: _ _ WT mcycle h4 h3 11 _ Equation (11) presents the ideal work done (WT ) by the system. However, owing to irreversibilities associated with the processes (i.e. heat transfer to the surroundings, mechanical losses, etc.), the actual
Int. J. Energy Res. 2011; 35:436448 r 2010 John Wiley & Sons, Ltd. DOI: 10.1002/er

The enthalpy at the boiler/vapourizer outlet is described by _ _ mcycle h30 h2 Qin 6

_ where Qin is the heat input to the system and we have dened a hypothetical intermediate state 30 , which is a trial solution for state 3. Knowing h30 and x 5 1, all other thermodynamic properties related to state 30 may be calculated (more complex problems such as super heating are discussed in the case study section).
438

Air-cooled ORC geothermal power plant method for modelling

M. I. Sohel et al.

electric power (Pel) produced by the unit is less than the ideal: _ Pel ZT mcycle h3 h4 where ZT is the turbine-generator efciency. The heat transfer to the cycle is: _ Qin UADTm 13 _ where Qin is the heat input to the cycle, U is the overall heat transfer coefcient, A is the heat transfer area and DTm is the log mean temperature difference. A is constant for a heat exchanger. The overall heat transfer coefcient is calculated as a function of geothermal uid mass ow rate using an approximate method [40,41]: _ _ U Ur m=mr 0:5 14 12

3. CONVERGENCE, STABILITY AND UNIQUENESS

The solution based on the developed method converges exponentially if the search parameter (equilibrium pressure) is updated each iteration proportionally to the relative error, similar to the Kalman lter [43, 44]. Figure 3 illustrates the typical convergence of the mass ow rate using Equation (4) to nd the vapourliquid equilibrium pressure for a typical air-cooled condenser and the typical convergence of du in Equation (9) while searching for the vapourliquid equilibrium pressure in a typical vapourizer. Here: p1 n11 p1 n k1 e1 n and p3 n11 p3 n k2 e2 n 17 where p3 and p1 are the vapourizer and condenser pressures, respectively, k1 and k2 are tuneable parameters and n represents the iteration number. For the solutions shown, k1 5 0.2 and k2 5 0.1 were used. e1 and e2 are calculated according to:   _ _ _ e1 m m =mdesign  18
n design cycle n

16

_ where Ur and mr are the reference overall heat transfer coefcient and the reference mass ow rate. These two parameters are determined via system identication methods. With typical operation of a heat exchanger involving a phase change, DTm remains almost unchanged [42]. Therefore, Equation (13) can be reduced to: _ _ _ _ Qin Qin;r m=mr 0:5 15 _ where Qin;r is the reference heat input. An initial guess of the inlet state of the working uid to the condenser is provided to the model. By working our way around the cycle, we predict a new inlet condition with an improved value. Subsequent iterations around the loop yield better results and the process converges to a unique solution within a few iterations. Figure 2 shows the effect of the ambient temperature on condenser heat load and equilibrium pressure of a superheated vapour ORC unit. The condenser equilibrium pressure is very sensitive to the ambient air temperature and that explains the strong dependence of the performance of an air-cooled condenser geothermal power plant on ambient air temperature. Details of the modelling involved are presented in later sections.
3.00E+04 heat load

  e2 n du dueqbrm n =du

19

The stability of the solution lies in the choice of values for the two constants k1 and k2. Larger values of k improve the rate of convergence, but may introduce instabilities. Lower values of k improve stability at the expense of convergence. Appropriate tradeoffs may be achieved by suitable tuning. Independent properties (i.e. p, T, v, h and s) are point functions, meaning that they are not dependent on path. Any solution obtained from the method described here must be unique. As long as the pressure values fall within the limits allowed by the thermophysical properties of the working uid, one should obtain the same solution regardless of the initial values of pressure used in Equations (16) and (18), and (17)
120 pressure 100 80 60 40 20 0

Condenser heat load [kW]

2.00E+04 1.50E+04 1.00E+04 5.00E+03 0.00E+00 -5 0

10

15

20

25

30

Ambient temperature [C]

Figure 2. Condenser heat load and condenser equilibrium pressure as a function of ambient temperature.

Condenser equilibrium pressure [k Pa]

2.50E+04

Int. J. Energy Res. 2011; 35:436448 r 2010 John Wiley & Sons, Ltd. DOI: 10.1002/er

439

M. I. Sohel et al.

Air-cooled ORC geothermal power plant method for modelling

18 16 Absolute error [kg/s] 14 12 10 8 6 4 2 0 1 2 3 4 5 6 Iteration [-]

20 18 14 12 10 8 6 4 2 0 Absolute error [kJ/kg]

8

16

Figure 3. Convergence of mass ow using Equation (4) and convergence of du using Equation (9) to calculate the vapourliquid equilibrium pressure.

30 Absolute error [kJ/kg] 20 10 0 -10 -20 -30 -40 Iteration [-] 1 2 3 4 5 6 7 9

Figure 4. Convergence of du using two different initial conditions of equilibrium pressure.

and (19). Figure 4 shows this to be the case by demonstrating the convergence of du with two different initial values of equilibrium pressure. The continuous line presents the solution obtained using an initial guess of the equilibrium pressure of 1010 kPa. The dashed line shows the solution obtained when an initial guess of 2020 kPa was used. Both solutions converged to a unique operating value of 1818 kPa for the equilibrium.

case. The value of De can be calculated as [6]: De hin hout T0 Ds 22

where hin and hout are the inlet and outlet enthalpies of the geothermal uid, respectively. Ds is the difference between the inlet and outlet entropies and T0 is the equilibrium temperature (dead state temperature). For simplicity, the equilibrium temperature is assumed to be 251c.

4. EFFICIENCY
The cycle efciencies are calculated as follows: _ Wnet Z1 _ Qin Z2 _ Wnet : m De

5. CONSTRAINTS
20 The performance of system components (e.g. boiler/ vapourizer and condenser) is constrained by the system design. The maximum and minimum allowable pressures and temperatures of these devices are predened. The plant performance is dependent on these limits as well as operator interaction to maintain operating conditions for maximum output. Such constraints must be applied to the equilibrium condition obtained by the iterative method. Therefore, all iterations (Equations (4)(19)) must be terminated and assigned feasible values if extremum operating conditions are reached.
Int. J. Energy Res. 2011; 35:436448 r 2010 John Wiley & Sons, Ltd. DOI: 10.1002/er

21

where Z1 and Z2 are rst and second law (energetic and _ exergetic) efciencies, respectively. Wnet is the net work : done by the cycle, m is the geothermal uid ow rate and De is the specic exergy input to the ORC. The brine inlet and outlet conditions of the ORCs are controlled; hence, De can be assumed constant for our
440

Air-cooled ORC geothermal power plant method for modelling

M. I. Sohel et al.

GENERATOR
TURBINE
G

TURBINE 3
M

STEAM

VAPORIZER CONDENSATE TANK AIR COOLED CONDENSER

CONDENSATE

2 STEAM WORKING FLUID CYCLE PUMP

Figure 5. Schematic of the saturated vapour cycle.

6. CASE STUDY 1: SATURATED VAPOUR ORC

A geothermal steam-driven ORC with pentane as the working uid has been chosen for this case study. This unit works as a bottoming unit of a steam turbine. The geothermal resource temperature is 2051C for both of the case studies. The schematic diagram of the process is presented in Figure 5. There are four basic processes involved:  Reversible adiabatic pumping process in the pump.  Constant pressure heat transfer in the vapourizer.  Reversible adiabatic expansion in the turbine.  Constant pressure heat transfer in the condenser. There is no direct measurement of mass ow rate in the physical implementation of the cycle concerned. An approximation is obtained using Equation (12), Z 5 1 and the generated electric power (Pel). Using the iterative method, the unit is modelled for 48 h of operation and compared against observed data. The experimental initial conditions are supplied to the model as a starting point for the simulation. The efciency of the turbine is estimated to be 0.76. A constant pressure loss between the turbine outlet and condenser inlet is assumed to be the nominal observed value of 10.1 kPa. Condenser heat load, Qcon fp4 ; T4 ; Tamb ; vair , is calculated from the developed condenser model [26]. Figure 6 presents the observed and modelled condenser outlet temperatures. The average error (absolute)
Int. J. Energy Res. 2011; 35:436448 r 2010 John Wiley & Sons, Ltd. DOI: 10.1002/er

60 55 Temperature [C] 50 45 40 35 30 25 20 0 10 20 30 40 50 Time [h]

Figure 6. Condenser outlet temperatures for the saturated vapour unit over 48 h of operation.
Observed Modelled

is 3.27%. Figure 7 compares the observed and modelled vapourizer outlet pressure with an average percentage error of 2.15%. Figure 8 compares the observed and modelled vapourizer outlet temperatures with an average error of 1%. Figure 9 compares the observed and modelled electric power output of the system with an average error of 4.20%. The iteration is terminated when the tolerance limit, |e| 5 o0.1, is reached. Figure 10 presents the relative error (absolute value) of modelled electric power output of the saturated vapour unit. The relative errors largely lie within 10%. Figure 11 presents corresponding rst law and second law efciencies. The rst law efciency varies between 7 and 9% and the
441

M. I. Sohel et al.

Air-cooled ORC geothermal power plant method for modelling

second law efciency varies between 37 and 47% depending on the ambient air temperature and geothermal uid ow rate.

7. CASE STUDY 2: SUPERHEATED VAPOUR ORC

Figure 12 presents the schematic of the superheated vapour ORC unit used for case study 2. This cycle is driven by separated brine from the geothermal uid (2051C). Superheating in the cycle adds some complexity. However, the same iterative approach may be used to implement the system model. From a modelling perspective, there are two basic differences between the saturated and superheated vapour ORC cycles: superheating and the addition of a recuperator for heat recovery. The maximum temperature of a superheated vapour cycle is typically much higher than that of a saturated cycle. Considering the second law of thermodynamics, the effect of ambient air (heat sink) variation will be less prominent in the superheated cycle compared with a saturated cycle [30]. The pressure loss between the vapourizer outlet and turbine inlet is assumed constant and assigned an observed value of 50.5 kPa. The pressure loss in the recuperator is assumed constant and xed at 50.5 kPa (4a-4), which is consistent with the observed value.

650 600 550 Pressure [k Pa] 500 450 400 350 300 0 10 20 30 40 Time [h]

Observed Modelled

50

Figure 7. Vapourizer outlet pressures for the saturated vapour unit over 48 h of operation.
105 100 Temperature [C] 95 90 85 80 75 70 65 60 0 10 20

Obseved Modelled

30

40

50

Time [h]

Figure 8. Vapourizer outlet temperatures for the saturated vapour unit over 48 h of operation.

5000 4500 Electric power [kW] 4000 3500 3000 2500 2000 1500 1000 0 10 20 30 40 50 Time [h]
Observed Modelled

Figure 9. Electric power output for the saturated vapour unit over 48 h of operation.

442

Int. J. Energy Res. 2011; 35:436448 r 2010 John Wiley & Sons, Ltd. DOI: 10.1002/er

Air-cooled ORC geothermal power plant method for modelling

M. I. Sohel et al.

16 14 Relative error [%] 12 10 8 6 4 2 0 1 7 13 19 25 31 Data point [-] 37 43

Figure 10. Observed relative error in modelled electric power output for the saturated vapour unit over 48 h of operation (48 data points).

0.5 0.45 Second law efficiency [-] 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 1 6 11 16 21 26 Time [h] 31 36 41 46 1 2

0.15 0.14 0.12 0.11 0.1 0.09 0.08 0.07 0.06 0.05 First law efficiency [-] 0.13

Figure 11. First and second law efciencies of the saturated vapour unit over 48 h of operation.

The condenser equilibrium condition and corresponding outlet temperature are derived in the same manner as in the case of the saturated vapour ORC. If the pump input work is assumed constant, the enthalpy of state 2 is calculated using Equation (5) and the enthalpy of state 3 is calculated using Equation (6). In the superheated cycle, heat input is given as _ _ _ Qin Qbrine 1Qrecuparator 23

_ where Qbrine is the heat input to the system from brine _ and Qrecuparator is the heat recovered from the turbine exhaust pentane vapour. Since the heat recovered in the recuperator is related to the mass ow rate of pentane (which remains relatively constant), the heat recovered is also approximately constant. Heat input to the system from geothermal brine is equal to _ _ _ Qbrine Qvaporizer 1Qseparator 24

Figure 13 presents the observed and modelled condenser outlet temperatures with an average error of 6.54%. Figure 14 shows the observed and modelled vapourizer pressures with an average error of 1.58%. Figure 15 presents the observed and modelled vapourizer outlet temperatures with an average error of 1.56%. Figure 16 presents the observed and modelled electric power output of the unit with an average error of 3.16%. Figure 17 presents the relative error (absolute value) of modelled electric power of the superheated vapour, which shows that the relative error remained within 7%. Lastly, Figure 18 presents the corresponding rst law and second law efciencies. The rst law efciency varies between 13 and 15% and the second law efciency varies between 37 and 43%.

8. DISCUSSION
Most of the modelling works available in the literature compare their results against experimental data performed under controlled environments [9,45]. In a real plant, uncertainty of a physical model increases over time due to degradation of the plant [46]. In
443

_ _ here Qvaporizer and Qseparator are calculated using Equation (15). The state at point 4 is determined as before, assuming isentropic expansion in the turbine. The value of electric power output is calculated using Equation (12) with an approximate turbine efciency of 0.9.
Int. J. Energy Res. 2011; 35:436448 r 2010 John Wiley & Sons, Ltd. DOI: 10.1002/er

M. I. Sohel et al.

Air-cooled ORC geothermal power plant method for modelling

TURBINE

3 BRINE SUPPLY VAPORIZER SEPARATOR

4a RECUPERATOR M 4

BRINE RETURN

AIR COOLED CONDENSER

Working fluid Brine 2 1

CYCLE PUMP

Figure 12. Schematic of the superheated vapour cycle.

40 35 Temperature [C] 30 25 20 15 10 5 0 0 10 20 30 40 50 Time [h]
Observed Modelled

Figure 13. Condenser outlet temperatures for the superheated vapour unit over 48 h of operation.
2000 1900 1800 1700 1600 1500 1400 1300 1200 1100 1000 0 10 20 30 Time [h]

Pressure [kPa]

Observed Modelled

40

50

Figure 14. Vapourizer outlet pressures for the superheated vapour unit over 48 h of operation.

444

Int. J. Energy Res. 2011; 35:436448 r 2010 John Wiley & Sons, Ltd. DOI: 10.1002/er

Air-cooled ORC geothermal power plant method for modelling

M. I. Sohel et al.

220 200 Temperature [C] 180 160 140 120 100 0 10 20 30 40 50 Time [h]
Observed Modelled

Figure 15. Vapourizer outlet temperatures for the superheated vapour unit over 48 h of operation.

5000 4500

Electric power [kW]

4000 3500 3000 2500 2000 1500 1000 500 0 0 10 20 30 40 50

Observed Modelled

Time [h]

Figure 16. Electric power output for the superheated vapour unit over 48 h of operation.

7 Relative error [%] 6 5 4 3 2 1 0 0 10 20 30 40 50 Data point [-]

Figure 17. Observed relative error in modelled electric power output for the superheated vapour unit over 48 h of operation (48 data points).

recent years, application of articial neural networks and genetic methods for modelling thermal power plant has become popular [46,47]. These models provide a high degree of accuracy without complicated physics-based models and also address the problem of increased model uncertainty with age. However, such models are very plant specic and cannot be readily used in conceptual design and developments. In contrast, the method developed here is based on fundamental thermodynamics and could be very useful in conceptual design and development.
Int. J. Energy Res. 2011; 35:436448 r 2010 John Wiley & Sons, Ltd. DOI: 10.1002/er

In the saturated vapour cycle, the error lies largely within 10% (Figure 13). Only one data point is found to lie above 14%. However, the external parameters, i.e. ambient temperature and geothermal uid ow rate, for this data point are very similar to the neighbouring data points. Therefore, this point can be assumed to be noise or an outlier [46]. Table I summarizes the average and maximum relative errors of the saturated vapour cycle and the superheated vapour cycle. Wei and co-workers [9] have reported a maximum relative error of 4% for their semi-empirical model. Quoilin and co-workers [45]
445

M. I. Sohel et al.

Air-cooled ORC geothermal power plant method for modelling

0.45 Second law efficiency [-] 0.4 0.35

0.185 First law efficiency [-] 0.175

2 1

0.165 0.155

0.3 0.145 0.25 0.2 0.135 1 6 11 16 21 26 31 36 41 46 0.125

Time [h]
Figure 18. First and second law efciencies of the superheated vapour unit over 48 h of operation. Table I. Summary of results of observed and modelled electric power output. Average error (|e|) (%) Saturated vapour cycle Superheated vapour cycle 4.20 3.16 Maximum error (|e|) (%) 9.25 6.48

have reported the maximum error of their model to lie within 10% and commented that their error was a consequence of cumulated subcomponent models inaccuracies. Smrekar and co-workers [46] have reported a maximum relative error of 7.19% of a boiler model of a real power plant. They have used articial neural networks for the modelling purpose. In our work, the maximum error remained at 10% and seems consistent with the existing literature. Moreover, the efciencies of both of the ORC units (energetic and exergetic) found to be very consistent with the literature [6]. It should be noted here that we have compared the real plant performance of a decade old plant with the model. There are a minimum number of inputs require for our model. The developed model is very generic and can be used for conceptual design, analysis and optimization. Therefore, the method presented in this paper can be considered reasonably accurate (in the context of existing methods) and very useful.

9. CONCLUSION
This work has introduced an iterative method for modelling a closed ORC. The heat sink temperature, in this case the ambient temperature, affects the modelled condenser performance. Consequently, it inuences the performance of the whole cycle. This occurs in two ways: (i) changes in the equilibrium pressure inside the condenser result in a change in turbine outlet pressure and pressure ratios and (ii) changes in the condenser outlet temperature caused by the heat sink temperature also affect the pump inlet and outlet conditions as well
446

as the vapourizer equilibrium temperaturepressure. These are competing effects. However, changes related to the turbine pressure ratio tend to dominate the power. Calculating the vapourliquid equilibrium condition of the condenser was performed by assuming that the mass ow rate in an ORC in steady-state operation remains relatively constant. The vapourliquid equilibrium condition of the vapourizer is found by assuming that the specic volume inside the vapourizer is unchanged for steadystate operation. Termination of the iterative search for unique state solutions is achieved when reaching a slack equilibrium condition within a prescribed tolerance or by meeting a constraint. As the model essentially assumes steady-state operation of the power cycle, the possible unit time where this model can be applied is bounded by the time required by a system to come into steady state. The saturated vapour cycle yielded an average error of 4.20% with a maximum error of 9.25% and the superheated vapour cycle yielded an average error of 3.16% with a maximum error of 6.48%. The main advantage of using the developed method lies on the fact that it requires a minimum number of inputs: condenser (p,T), vapourizer (p,T), condenser heat load, turbine efciency (overall), pump work and the extremum conditions of all the components. These inputs should represent typical operating conditions of a plant. The model can predict the appropriate plant performance depending on the system heat input (geothermal uid ow in this case) and the heat sink temperature. As the method is based on basic thermodynamics, rather than empirical or semi-empirical approaches, it is widely applicable. The main focus of this work is on the ORC but the developed method is applicable to any closed Rankine cycle.

NOMENCLATURE
A e h 5 area (m2) 5 relative error () 5 specic enthalpy (kJ kg1)

Int. J. Energy Res. 2011; 35:436448 r 2010 John Wiley & Sons, Ltd. DOI: 10.1002/er

Air-cooled ORC geothermal power plant method for modelling

M. I. Sohel et al.

k1 k2 : m p Pel Q _ Q s T u U v vair W : W x Subscript amb con H L in r T 14 Greek letters Z

5 proportionality constant () 5 proportionality constant () 5 mass ow rate (kg s-1) 5 pressure (kPa) 5 electric power (MW) 5 heat transfer (MJ) 5 heat transfer rate (MW) 5 specic entropy (kJ K1 kg1) 5 temperature (1C) 5 specic internal energy (kJ kg1) 5 overall heat transfer coefcient (MW m2 K1) 5 specic volume (m3 kg1) 5 velocity of air (m s1) 5 work done (MJ) 5 work rate or power (MW) 5 quality ()

5 ambient 5 condenser 5 high 5 low 5 input 5 reference 5 turbine 5 states

5 efciency ()

ACKNOWLEDGEMENTS This work was funded and supported by the University of Canterbury.

REFERENCES
1. Ogriseck S. Integration of Kalina cycle in a combined heat and power plant, a case study. Applied Thermal Engineering 2009; 29(1415):28432848. 2. Lolos PA, Rogdakis ED. A Kalina power cycle driven by renewable energy sources Energy 2009; 34(4):457464. 3. Zamrescu C, Dincer I. Thermodynamic analysis of a novel ammoniawater trilateral Rankine cycle. Thermochimica Acta 2008; 477(12):715. 4. Mlcak H, Mirolli M, Hjartarson H, Lewis B. Notes from the North: a report on the debut year of the 2 MW Kalina cycle geothermal power plant in Husavik, Iceland. Geothermal Resource Council Conference, Reno, NV, U.S.A., 2002.
Int. J. Energy Res. 2011; 35:436448 r 2010 John Wiley & Sons, Ltd. DOI: 10.1002/er

5. Mlcak HA. An introduction to the Kalina cycle. Joint Power Generation Conference, Houston, TX, U.S.A., vol. 30, 1996; 765776. 6. DiPippo R. Second law assessment of binary plants generating power from low-temperature geothermal uids. Geothermics 2004; 33(5):565586. 7. Aoun B. Micro combined heat and power operating on renewable energy for residential building. Ph.D. Thesis, CEP-Centre Energetique et Procedes, 2008. 8. Schuster A, Karellas S, Kakaras E, Spliethoff H. Energetic and economic investigation of organic Rankine cycle applications. Applied Thermal Engineering 2009; 29(89):18091817. 9. Wei D, Lu X, Lu Z, Gu J. Dynamic modeling and simulation of an organic Rankine cycle (ORC) system for waste heat recovery. Applied Thermal Engineering 2008; 28(10):12161224. 10. Gnutek Z, Bryszewska-Mazurek A. The thermodynamic analysis of multicycle ORC engine. Energy 2001; 26(12):10751082. 11. Hung TC, Shai TY, Wang SK. A review of organic Rankine cycles (ORCs) for the recovery of lowgrade waste heat. Energy 1997; 22(7):661667. 12. Larjola J. Electricity from industrial waste heat using high-speed organic Rankine cycle (ORC). International Journal of Production Economics 1995; 41(13):227235. 13. Bombarda P, Invernizzi CM, Pietra C. Heat recovery from diesel engines: a thermodynamic comparison between Kalina and ORC cycles. Applied Thermal Engineering 2010; 30(23):212219. 14. Akkaya AV, Sahin B. A study on performance of solid oxide fuel cell-organic Rankine cycle combined system. International Journal of Energy Research 2009; 33(6):553564. 15. Husband WW, Beyene A. Low-grade heat-driven Rankine cycle, a feasibility study. International Journal of Energy Research 2008; 32(15):13731382. 16. Mago PJ, Srinivasan KK, Chamra LM, Somayaji C. An examination of exergy destruction in organic Rankine cycles. International Journal of Energy Research 2008; 32(10):926938. 17. Manolakos D, Kosmadakis G, Kyritsis S, Papadakis G. Identication of behaviour and evaluation of performance of small scale, low-temperature organic Rankine cycle system coupled with a RO desalination unit. Energy 2009; 34(6):767774. 18. Bruno JC, Lopez-Villada J, Letelier E, Romera S, Coronas A. Modelling and optimisation of solar organic Rankine cycle engines for reverse osmosis desalination. Applied Thermal Engineering 2008; 28(1718):22122226.
447

M. I. Sohel et al.

Air-cooled ORC geothermal power plant method for modelling

19. Delgado-Torres AM, Garcia-Rodriguez L. Comparison of solar technologies for driving a desalination system by means of an organic Rankine cycle. Desalination 2007; 216(13):276291. 20. Drescher U, Bruggemann D. Fluid selection for the organic Rankine cycle (ORC) in biomass power and heat plants. Applied Thermal Engineering 2007; 27(1):223228. 21. Chinese D, Meneghetti A, Nardin G. Diffused introduction of organic Rankine cycle for biomassbased power generation in an industrial district: a systems analysis. International Journal of Energy Research 2004; 28(11):10031021. 22. Bombarda P, Gaia M. Geothermal binary plants utilising an innovative non-ammable azeotropic mixture as working uid. Proceedings 28th NZ Geothermal Workshop, vol. 6, 2006. 23. Madhawa Hettiarachchi HD, Golubovic M, Worek WM, Ikegami Y. Optimum design criteria for an organic Rankine cycle using low-temperature geothermal heat sources. Energy 2007; 32(9): 16981706. 24. DiPippo R. Ideal thermal efciency for geothermal binary plants. Geothermics 2007; 36(3):276285. 25. Sohel MI, Sellier M, Brackney LJ, Krumdieck S. Efciency improvement for geothermal power generation to meet summer peak demand. Energy Policy 2009; 37(9):33703376. 26. Sohel MI, Sellier M, Brackney LJ, Krumdieck S. Dynamic modelling and simulation of an organic Rankine cycle unit of a geothermal power plant. Proceedings World Geothermal Congress 2010, Bali, Indonesia, 2529 April 2010; accepted. 27. Al-Fahed SF, Alasfour FN, Abdulrahim HK. The effect of elevated inlet air temperature and relative humidity on cogeneration system. International Journal of Energy Research 2009; 33(15):13841394. 28. Zaki GM, Jassim RK, Alhazmy MM. Brayton refrigeration cycle for gas turbine inlet air cooling. International Journal of Energy Research 2007; 31(13):12921306. 29. Ameri M, Shahbazian HR, Nabizadeh M. Comparison of evaporative inlet air cooling systems to enhance the gas turbine generated power. International Journal of Energy Research 2007; 31(15): 14831503. 30. DiPippo R. The effect of ambient temperature on geothermal binary-plant performance. Geothermal Hot Line 1989; 19:6870. 31. Chuang CC, Sue DC. Performance effects of combined cycle power plant with variable condenser

32.

33. 34.

35.

36. 37. 38.

39. 40.

41.

42. 43. 44.

45.

46.

47.

pressure and loading. Energy 2005; 30(10): 17931801. Gay RR, Palmer CA, Erbes MR. Power Plant Performance Monitoring. R-Squared Publishing: Santa Ynez, CA, U.S.A., 2004. Li KW, Priddy AP. Power Plant System Design. Wiley: New York, 1985. Miliaras ES. Power Plants with Air-Cooled Condensing Systems, The Massachusetts Institute of Technology, 1974. Durmayaz A, Sogut OS. Inuence of cooling water temperature on the efciency of a pressurized-water reactor nuclear-power plant. International Journal of Energy Research 2006; 30(10):799810. MathWorks, 2008. Available from: www.mathworks. com. REFPROP, National Institute of Standards and Technology (NIST), 2007. Wei D, Lu X, Lu Z, Gu J. Performance analysis and optimization of organic Rankine cycle (ORC) for waste heat recovery. Energy Conversion and Management 2007; 48:11131119. Bejan A. Advanced Engineering Thermodynamics. Wiley: New York, 1997. Bai O, Nakamura M, Ikegami Y, Uehara H. A simulation model for hot spring thermal energy conversion plant with working uid of binary mixtures. Journal of Engineering for Gas Turbines and Power 2004; 126(3):445454. Nakaoka T, Uehara H. Performance test of a shelland-plate type evaporator for OTEC. Experimantal Thermal and Fluid Science 1988; 1(3):283291. Holman JP. Heat Transfer. McGraw-Hill: New York, 1992. Maybeck PS. Stochastic Models, Estimations, and Control, vol. 1. Academic press: New York, 1979. Kalman RE. A new approach to linear ltering and prediction problems. Transactions of the ASME Journal of Basic Engineering 1960; 82:3545. Quoilin S, Lemort V, Lebrun J. Experimental study and modeling of an organic Rankine cycle using scroll expander. Applied Energy 2010; 87: 12601268. Smrekar J, Assadi M, Fast M, Kus trin I, De S. Development of articial neural network model for a coal-red boiler using real plant data. Energy 2009; 34:144152. Ghaffari A, Chaibakhsh A, Lucas C. Soft computing approach for modeling power plant with a oncethrough boiler. Engineering Applications of Articial Intelligence 2007; 20(6):809819.

448

Int. J. Energy Res. 2011; 35:436448 r 2010 John Wiley & Sons, Ltd. DOI: 10.1002/er