CoDIT16 - April 6-8, 2016, Malta 388

Continuous Timed Petri Net Model of Brayton

Cycle for a Gas Turbine
A. Nez, J. Pineda Cruz, E. E. Rodriguez
Centro de Ingeniera y Desarollo Industrial
CIDESI, Pie de la cuesta 702, Col. Desarrollo San Pablo
Quertaro, Qro
Email: cnunez@posgrado.cidesi.edu.mx

AbstractThis work developed a continuous timed is developed in Section 3. In Section 4 the continuous
Petri Net model of a gas turbine, using a rst linear Petri Net model based on the equations of section 3 is
order approximation for every state of the Brayton presented. In Section 5 the simulation of the Brayton cycle
Cycle. The Brayton cycle rules the functioning of a
gas turbine, and it is composed by four states: Com- as an hybrid system is shown. Finally, the conclusions and
pression, Combustion, Expansion and Cooling. In every further work are presented in Section 6.
state (excepting the cooling phase) temperature and
pressure are modeled as rst order linear systems, II. Fundamentals
therefore, every system is translated into a continuous A general description of the gas turbine process and an
timed Petri Net. The complete model can be used in
further work in order to design supervisory controllers introduction to continuous Petri Nets are introduced in
to ensure an ecient performance of a gas turbine. this section.
Although the model presented is extremely simplied,
it will be used as a starting point to developed more A. Gas turbine
complex models. For further reading, review (1) and (2). A generic gas
turbine is composed by three elements, each one is asso-
I. Introduction
ciated to a state of the Brayton cycle. The elements are
Gas turbines are used in transport systems as aircraft shown in Fig. 1. The general process is explained below.
and in power generating systems. The principle of the
gas turbine is developed by the Brayton cycle, a ther-
modynamic process which intervenes in the gas turbine
components. The steady-state behavior of the gas turbine
has been widely investigated in engineering area. More-
over, The dynamic behavior has been studied using non
linear models of its components, leading to complicated
mathematical representations. The gas turbine thermody-
namic and mechanic nature is considered dynamic but
the control algorithms are referred to specic punctual
measurements, making the control algorithm work as a
supervisor instead of a dynamic regulator. In this scenario,
the need of modeling the gas turbine system as a Discrete
Event System (DES) has become greater in the latest
years. However, it is impossible to ignore the dynamic
behavior of the system during the Brayton cycle, so a
standard modeling of the gas turbine as discrete state
system is not enough to capture its behavior. Therefore,
in this work the Brayton cycle is considered as an hybrid
system with four discrete states, compression, combustion, Figure 1. Gas turbine
expansion and freezing. Each state has an inherent pres-
sure and temperature dynamics, which has been reduced A contained quantity of air is in the compressor at at-
to rst order systems and modeled as continuous Petri mospheric pressure, then the compression starts increasing
nets. In Section 2, the basic concepts of gas turbines and the pressure of the air and leading it into the combustion
continuous Petri Nets are introduced. The mathematical chamber. Hence, the fuel is injected and the combustion
reduction of the gas turbine dynamics used in this work begins, increasing the temperature of the air-fuel mix and

978-1-5090-2188-8/16/$31.00 2016 IEEE

 CoDIT16 - April 6-8, 2016, Malta
directing it in a very fast speed to the turbine. The turbine
stars rotating, decompressing the air in the output and
generating a constant angular speed. Finally, the air in the
output is freeze by the environment. This closed process
is named the Brayton cycle and its introduced in the
following subsection.
B. Brayton cicle
The Brayton cycle is used to dene the behavior of a
gas turbine. It is composed by four states:
1) Compression. Ideally is an adiabatic process (no
heat transfer with the environment). However, the
temperature does increase, as has been shown in
several experiments.
2) Heat addition (combustion). An isobaric process (no
change on the uid volume).
3) Expansion. Again, an ideally adiabatic process.
4) Heat rejection. An isobaric process.
In Fig. 2 the Brayton cycle in terms of the temperature
and the pressure is shown. For the propose of this work,
the variables considered in the Brayton Cycle are pressure
and temperature. Usually in Brayton process variables
such as entropy are analyzed, but for simplicity of the
measurements, entropy is not included in the present
analysis.
Figure 2. Brayton cycle
In this description is remarkable that each state occurs
in a component of the gas turbine.
C. Continuos timed Petri Nets
The following subsection has been taken directly form
(3). The reader is refered to (4) and (5) for related subjects
in Continuos timed Petri Nets.
A continuous timed Petri net is a 3-tuple ContP N =
(N, , m0 ), where (N, m0 ) is a continous Petri Net and
|T |
: T {R+ } is a function associating a ring rate
with each transition.
389
The state equation of a ContPN is shown in Eq. 1
) = C f ( )
m( (1)
where is the time variable. C is the net incidence
matrix, C = P ost P re.
A ContPN is called innite server semantic if the ow
of a transition ti is
fi = i minppre(ti ) {m(p)/P re[p, ti ]} (2)
Eq. 1 can be expressed as a piecewise linear system given
by
m
= C(m)m (3)
The ring rate matrix is denoted by = diag(1 , ..., |T | ).
A conguration of a ContPN at m is a set of (p, t) arcs
describing the eective ow of all transitions:

1
P re[i,j] if pi is constraining tj
(m)[i, j] = (4)
0 otherwise
III. Mathematical description
In this section the model of the Brayton cycle as ah
hybrid system is developed. A high-level discrete event
system is constructed using the four states of the Brayton
cycle, explained in section II-B. In each state, the dynamic
equations of temperature and pressure are obtained, based
on the thermodynamical relationships introduced in (6).
A. High-level discrete state model
Brayton cycle is composed by 4 states
1) Compression
2) Combustion
3) Expansion
4) Cooling
In add to this states, an initial state representing no
operation of the gas turbine is included in the model. The
corresponding Finite State Machine of the Brayton cycle
is presented in Fig. 3
B. Mathematical representation of temperature and pres-
sure
In each state of the system, the temperature and pres-
sure evolve according to dynamical non linear equations.
For this work, the relationships have been reduced into
rst order linear dynamical equations in the form of
equation 5.
= m + u
m (5)
> 0 and > 0.
With u an input variable which increases the value
of f . The coecients and are determined based
in experimental results of the temperature and pressure
behavior in each state.
 CoDIT16 - April 6-8, 2016, Malta 390

2) Combustion dynamic equations: Temperature in the

output of combustion chamber T3 evolves with the applied
heat Q generated by the ignition. Q is proportional to the
amount of air mass in the combustion chamber, which is
proportional to the open percentage of angle of the stator
position Air . Eq. 13 represents the evolution of T3 with a
control input Air .

T3 = T3 [T3 T1 ] + T3 Air (13)

Pressure in the output of combustion chamber P3
evolves according to the heat transference, such as T3 ,
leading to Eq. 14

P3 = P3 [P3 P2 ] + P3 Air (14)

Figure 3. Brayton cycle modeled as a Finite State Machine
1) Compression dynamic equations: In the state of
compression, the input variable is the angular speed of
the compressor and it increases the value of the pressure
in the output of the compressor P2 , giving as result Eq. 6
3) Expansion dynamic equations: Similarity to com-
pression, pressure P4 in the output of the turbine is repre-
sented as a rst order linear system with a control input ,
as seen in Eq. 15. Making an analysis of the temperature
exchange in this process similar to compression process,
Eq. 16 represents the temperature behavior in the output
of the turbine, considering T 3 and P 3 as constants during
the expansion process.

P2 = P2 P2 + P2 (6) P4 = P4 P4 + P4 (15)

With initial condition the atmospheric pressure P1 . The T3

thermodynamic relationship between pressure and tem- T4 = P4 T4 + P4 (16)
P3
perature into the compressor is developed in (6) and is
Brayton cycle for gas turbine are not a closed process,
presented in Eqs. 7-8.
therefore the air in the output of the turbine is not reused
as input for the compresor. Thus, the cooling dynamics is
P2
Kc (t) = (7) not necessary to analyze the system behavior.
P1
IV. Continuous Timed Petri Net Model
Using the equations developed in section 3, Continuous
T 2 = Kc T 1 (8)
timed Petri net models are described in this section.
Being T2 the temperature in the output of compressor A. Compressor Petri net model
and T1 the environment temperature. From 7 and 8 Using Eqs. 12 and 6 the Petri net model shown in Fig.
4 is developed.
P2
T2 = T1 (9)
P1
Considering P1 and T1 constants (this assumption is valid
because of the slow rate change of environmental condi-
tions) and dierentiating 9

P2
T2 = T1 (10)
P1

T1
T2 = (P2 P2 + P2 ) (11)
P1

T1
T2 = P2 T2 + P2 (12) Figure 4. Compressor Petri net model
P1
 CoDIT16 - April 6-8, 2016, Malta 391

Figure 5. Combustion chamber Petri net model

M1 P2
B. Combustion chamber Petri net model M2 T2
M3
M4 P3
Using Eqs. 13 and 14 the Petri net model shown in Fig. M5 Air
5 is developed. M6 T3
M7 T0
M8 P4
M9 T4
Table I
Place numeration

C. Turbine Petri net model

marking places, as shown in Eq. 17

Using Eqs. 16 and 15 the Petri net model shown in Fig.
6 is developed. M 1 = P2 M1 + P2 M3
M 4 = P3 [M4 M1 ] + P3 M5
M 8 = P M8 + P M3
4 4

T1 (17)
P2 = P2 P2 + P2 M3
P1
M 6 = T3 [M6 M7 ] + T3 M5
T3
M 9 = P4 M9 + P4 M3
P3
Merging the models shown above, a complete gas tur-
bine Petri net model is obtained with the following inci-
dence matrix C and ring rate matrix , using the place
index as shown in Tab. I.

Figure 6. Turbine Petri net model
Using the table I, Eqs 6-16 are translate in terms of
V. Simulation results
Using SIMHPN matlab toolkit (7), the dynamic models
of each state in the Brayton cycle are simulated using the
values presented in Eq. 17 . The results are shown in Figs.
7-14.
 CoDIT16 - April 6-8, 2016, Malta 392

1 1 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 1 1 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 1 1 0 0 0 0 0 0 0

0 0 1 0 1 1 0 0 0 0 0 0 0 0 0

C=
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 1 1 1 0 0 0 0 (17)

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 1 1 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 1 1

= diag 10 100 10 0.2 20 10000000 5 36000 5 1200 1.25 20 945 20 2.63

80 1200

70 1000
Temperature (C)

Temperature (s)
60 800

50 600

40 400

30 200

20 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
time (s) time (s)

Figure 7. T2 (temperature in the output of compressor) Figure 9. T3 (temperature in the output of combustor)

5
x 10
5 5
x 10
4

3.5 4
Pressure (pa abs)

3
Preesure (pa abs)

3
2.5

2
2
1.5

1 1

0.5
0
0 1 2 3 4 5 6
0 time (s)
0 1 2 3 4 5 6
time (s)

Figure 10. P3 (pressure in the output of compressor)

Figure 8. P2 (pressure in the output of compressor)

VI. Conclusions and further work

In the beginning of the process, a gear is used to start Using a linear approach for every stage of the Brayton
the compression stage. Once in stedy state, the gear is cycle allows to represent the system as a continuous
o and the system is sustain by the air ux, which is timed Petri Net. This representation is commonly used
submitted to a Brayton cycle as described in Fig. 14. for discrete event systems in order to apply supervisory
 CoDIT16 - April 6-8, 2016, Malta 393

500 4000

3500
400
3000

Angular speed (RPM)

Temperature (C)

300 2500

2000
200 1500

1000
100
500

0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
time (s) time (s)

Figure 11. T4 (temperature in the output of turbine) Figure 13. (angular speed of the rotor)

5
x 10
2

1.5
Pressure (pa abs)

0.5

0
0 1 2 3 4 5 6
time (s)

Figure 12. P4 (pressure in the output of turbine) Figure 14. Brayton cycle simulates using continous timed Petri net
models

control to constraint the system behavior. Using marking

constraints or behavioral constraints, the system can be
submit to an specication behavior. The constraints can
directly forbid certain place to reach a value, such as
temperature must be always bellow 1000 C or disable
some action in order to prevent a bad performance, such
as cut the air or the fuel supply if the angular speed is
increasing. The formal techniques of supervisory control
are widely developed and the approach presented in this
work allows the synthesis of supervisory controllers for a
gas turbine. Using the approach as a simplied model is
signicant in order to add non-linear behavior.
References
[1] A. Valera, Tecnologias emergentes en turbinas de
gas, in Congreso y Exposicion en Turbomaquinaria,
2014.
[2] E. Thirunavukarasu, Modeling and simulation
study of a dynamic gas turbine system in a
virtual test bed environment, Masters thesis,
University of South Carolina, 2013, retrieved from
http://scholarcommons.sc.edu/etd/2254.
[3] J. L. Tovany, R. Ross-Leon, J. Ruiz-Leon, A. Ramirez-
Trevio, and O. Begovich, Greenhouse modeling using
continuous timed petri nets, Mathematical problems
in engineering, 2013.
[4] H. Alla and R. David, Continuous and hybrid petri
nets, Journal of Circuits, Systems and Computers,
1998.
[5] L. Recalde, E. Teruel, and M. Silva, Autonomus con-
tinuous p/t systems, Aplication and Theory of Petri
Nets, 2005.
[6] E. E. Rodriguez, Modal model of gas turbine by
using measurements of vibration and acoustics, 2015,
submitted to ASME Turbo Expo 2016.
[7] J. Julvez and C. Mahulea, SimHPN: A MATLAB
toolbox for hybrid Petri nets user manual, 2012.