Sie sind auf Seite 1von 24

DYNAMIC MODELLING AND OPTIMIZED MODEL

PREDICTIVE CONTROL STRATEGIES FOR THE ORGANIC


RANKINE CYCLE

V. Lemort, S. Quoilin, S. Declaye and Assaad Zoughaib


Laboratoire de Thermodynamique
Universit de Lige
Ecoles de Mines de Paris

First International Seminar on ORC Power Systems

Introduction
 Organic Rankine Cycle (ORC) : Small scale, lower
temperature
 Application fields : Waste heat recovery, Solar sources,
engine exhaust gases, biomass CHP, geothermy, ...
 Selection of the working fluid and optimization of the
working conditions are key issues in waste heat recovery
applications
 Dynamic models are required to define proper control
strategies under transient conditions

First International Seminar on ORC Power Systems

Table of contents
Introduction
Steady-state models
 Description
 Experimental validation

Dynamic models
Cycle optimization and control
Model predictive control
Conclusions

First International Seminar on ORC Power Systems

Volumetric expander model








First International Seminar on ORC Power Systems

Heat transfers
Friction
Internal Leakage
Pressure drops
Built-in internal volume ratio
Under and over-expansion

ORC cycle steady-state model


Plate Heat exchangers model:

Cycle model:

Pump model:
.

(Pr,su,p ; V s,p )

(Pr,ex,p ; V s,p )
(X p ,s )

First International Seminar on ORC Power Systems

Experimental validation

First International Seminar on ORC Power Systems

Table of contents
Introduction
Steady-state models
 Description
 Experimental validation

Dynamic models
Cycle optimization and control
Model predictive control
Conclusions

First International Seminar on ORC Power Systems

HEAT EXCHANGER MODEL


Simplified heat transfer and
pressure drop laws:

Conservation of energy:
Conservation of mass:
Metal wall:
First International Seminar on ORC Power Systems

CYCLE MODEL
Performance indicators:

First International Seminar on ORC Power Systems

CYCLE DISPLAY

First International Seminar on ORC Power Systems

10

Table of contents
Introduction
Steady-state models
 Description
 Experimental validation

Dynamic models
Cycle optimization and control
Model predictive control
Conclusions

First International Seminar on ORC Power Systems

11

CONTROL STRATEGIES
Degrees of freedom:

First International Seminar on ORC Power Systems

Controlled variables:

12

Optimal working conditions


Main optimization parameter : evaporation temperature

No Recuperator!

13

First International Seminar on ORC Power Systems

OPTIMAL EVAPORATION TEMPERATURE


Static optimization of the cycle
Tev,optim depends on the heat source/sink conditions:

First International Seminar on ORC Power Systems

14

CYCLE FEEDBACK CONTROL STRATEGIES

First International Seminar on ORC Power Systems

15

CYCLE FEEDBACK CONTROL STRATEGIES

First International Seminar on ORC Power Systems

16

Table of contents
Introduction
Steady-state models
 Description
 Experimental validation

Dynamic models
Cycle optimization and control
Model predictive control
Conclusions

First International Seminar on ORC Power Systems

17

Model predictive control


 Future action on the manipulated variable is based
on a model.
 Balance between two different approaches:
model evaluation of the manipulated variable to
reach the set point
Constant rectification of the trajectory by
measurement of the control variable.

 Anticipates the effects of external perturbations


 Well-performing with constrained processes
First International Seminar on ORC Power Systems

18

Control unit
 First order model to predict
the process:
process output K e Tds
G(s) =
=
process input 1+ s

 Defined a desired
trajectory to the set point
 Define an horizon time
for which the model
predicts an intersection
between the desired
trajectory and the process.

First International Seminar on ORC Power Systems

19

Accounting for measured pertubations


 The effect of an
external pertubation is
anticipated by the
control.
 A first order model
must be defined to
account for these
external influences.

First International Seminar on ORC Power Systems

20

Modelica implementation

First International Seminar on ORC Power Systems

21

Simulation results
16
PID controller
MPC controller

14

ex,ev

[K]

12
10
8
6
4
2
0

200

400

600

800

1000

1200

1400

1600

1800

time [sec]
First International Seminar on ORC Power Systems

22

Summary
 A robust dynamic model of an ORC has been developed, based
on validated steady-state component models
 Different control strategies have been proposed an compared
 A varying set-point control strategy allows increasing the output
by more than 4%
 MPC controller follow the superheating set point in a much
better way than PI-based controllers.
 Future work will focus on the implementation and the
improvement of MPC controllers with delayed processes, during
start & stop procedures and in more severe conditions.

First International Seminar on ORC Power Systems

23

Thank you!

First International Seminar on ORC Power Systems

24