Professional Documents
Culture Documents
Flow Control of Fluid in Pipelines Using PID Controller: IEEE Access February 2019
Flow Control of Fluid in Pipelines Using PID Controller: IEEE Access February 2019
net/publication/331142823
CITATIONS READS
26 5,438
4 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Cristobal Vargas on 17 June 2019.
ABSTRACT In this paper, a PID controller is utilized in order to control the flow rate of the heavy oil in
pipelines by controlling the vibration in a motor pump. A torsional actuator is placed on the motor pump
in order to control the vibration on a motor and consequently controlling the flow rates in pipelines. The
necessary conditions for the asymptotic stability of the proposed controller are validated by implementing
the Lyapunov stability theorem. The theoretical concepts are validated utilizing numerical simulations and
analysis, which proves the effectiveness of the PID controller in the control of flow rates in pipelines.
INDEX TERMS Fluid flow control, control engineering, PID control, feedback.
2169-3536
2019 IEEE. Translations and content mining are permitted for academic research only.
VOLUME 7, 2019 Personal use is also permitted, but republication/redistribution requires IEEE permission. 25673
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
S. Razvarz et al.: Flow Control of Fluid in Pipelines Using PID Controller
This paper deals with the modeling and control of flow conditions as well as external forcing, also the sensitivity to
rate in heavy-oil pipelines. For this aim, the PID control algo- spatially localized disturbances.
rithm is utilized to control the flow mechanism in pipelines. From the viewpoint of incompressible, constant-density,
A torsional actuator is placed on the motor-pump in order constant-viscosity flows associated with Newtonian fluids,
to control the vibration on motor. The stability of the PID the nonlinear Navier–Stokes equation is defined for a nondi-
controller is verified using Lyapunov stability analysis. The mensional velocity field (x, t) : Rn × R → Rn , pressure field
stability analysis of the controller results in a theorem which p (x, t) : Rn × R → Rn and Reynolds number Re > 0 as
validate that the system states are bounded. The theoretical below equation,
concepts are validated using numerical simulations and anal- ∂u ∇p
ysis, which proves the effectiveness of the PID controller in =− − u.∇u + Ff (1)
∂t ρ
the control of flow rates in pipelines. Further, it is the first
attempt to place a torsional actuator on the motor-pump in where,
order to control the vibration on motor and hence control the ρ is the density in mkg3
flow rates in pipelines. u is the flow velocity in ms ,
This paper is structured as follows. Firstly, in Section II the ∇ is the divergence,
kg
pump model system is established. The PID control method is p is the pressure in m.s 2,
where ẍp is the flow acceleration of the pump. Since τ, T , τp , The PID control is expressed as,
and n are known quantities of pump so ẍp can be estimated. Z t
The external force generated by the pump is 9 (t) = −κ p e (t) −κ i e(t)dτ − κd ė (t) (22)
0
fp = mp ẍp (20) where kp , ki , as well as kd are positive definite and ki is the
where ẍp is the acceleration of the motor and mp is volumetric integration gain. For the flow control, X d is desired reference
mass of the pump. and also X d = Ẋ d = 0. Hence, equation (22) is rewritten as
The shape force vector fb can be modeled as a linear or a below,
nonlinear model. Z t
From (12), by considering shape force vector fb as 9 (t) = −κ p X −κ i Xdτ − κd Ẋ (23)
0
non-linear model, the following analysis is illustrated:
In simple non-linear case, (12) becomes For analyzing the PID controller, equation (22) can be
stated as below,
0 Ẍ + 8Ẋ + fb = fp (21) Z t
where fb is taken to be non-linear. 9 (t) = −κ p X − κd Ẋ − ϑ ϑ = κi Xdτ , ϑ (0) = 0
0
(24)
III. THE TUNING METHOD BASED ON PID CONTROLLER
Since 1700’s the control of continuous process has been The closed-loop system of eq. (12) along with the PID
carried out by utilizing feedback loop. System with feedback control (eq. (23)) is demonstrated as below,
control contains drawback which is related to the instability 0 ẍ + 8ẋ + Fg = −κ p X − κd Ẋ − ϑ ϑ̇ = κi X (25)
of the system. In order to resolve this problem an appropriate
controller should be chosen and also it must be ideal for In matrix form, the closed-loop system is defined as,
the monitoring system. The proportional feedback control
ϑ κi X
is uncomplicated and relatively easy to implement. Never- d
X = Ẋ (26)
theless, its outcome is completely sensitive to the sensing dt Ẋ
−0 −1 (8Ẋ + Fg + kp X + kd Ẋ + ϑ)
location as well as feedback gain. It is concluded from the
control theory that these drawbacks of the P control should be Here the stability of the PID control demonstrated by
overcome by adopting I as well as D controls. Nevertheless, eq. (23) is analyzed.
Theequilibrium of eq. (26) is presented
there exist very few studies investigating the application of by ϑ X Ẋ = ϑ̂ 0 0 . As at equilibrium point X = 0 as
the PID control for fluid-mechanics problems. In addition, well asẊ = 0, the equilibrium is [ f (0) , 0, 0 ]. For moving
there are a very limited number of studies dealing with the the equilibrium to the origin, the following is defined,
P control and PI controller for pipeline. Due to this lack of
investigations, this paper aims to develop a PID control for ϑ̂ = ϑ − f (0) (27)
flow rate in the pipeline. Therefore, the final closed-loop equation is defined as,
The control mechanism is demonstrated in Figure. 3 which
shows the entire control process of the flow rate of the 0 Ẍ + 8Ẋ + Fg = −κ p X − κd Ẋ − ϑ + f (0)ϑ = κi x
heavy-oil in pipelines. (28)
For analyzing the stability of equations (28), the following
properties are required,
Property 1: The positive definite matrix 0 should satisfy
in the below condition
0 < λmin (0) ≤ k0k ≤ λMax (0) ≤ γ̄ (29)
such that λmin (0) as well as λMax (0) are considered as
the minimum and maximum eigenvalues of the matrix 0,
respectively also γ̄ > 0 is taken to be the upper bound.
FIGURE 4. PID controller. Property 2: f is taken to be Lipschitz over x̃ and ỹ if
kf (x̃) − f (ỹ)k ≤ kx̃ − ỹk (30)
The PID control is considered as a control law in which the
existence of output for feedback is essential. This practical As Fg is first-order continuous functions also satisfies in
control method is widely utilized in the control society. In the Lipschitz condition, Property 2 is hereby established.
PID control, the controller is made of a simple gain (P con- The lower bound of Fg can be calculated as below,
trol), an integrator (I control), a differentiator (D control) or Z t Z t Z t
some weighted composition of these possibilities [27], see fdx = Fg dx + du dx (31)
Figure 4. 0 0 0
q
3 λmin (0)λm (κp )
Rt Rt 1
The lower bound of 0 Fg dx is stated as −F̂ g also 0 du dx
In a case that σ ≤ 1
λMax (8) , the following is
as -D̂u . Therefore, the lower bound of is defined as, obtained,
2
= −F̂ g − D̂u
2 !
(32) 1 31 λmin κp kX k2 + λmin (0)
Ẋ
V3 ≥
2 +σ λMax (0) kX k
Ẋ
The stability analysis of PID control approach is given by
the below mentioned theorem. s 2
1 λmin κp
Theorem: By taking into consideration the structural sys-
p
kX k + λMax (0)
Ẋ
≥ 0 (40)
=
tem of equation (12) controlled by the PID control approach 2 3
of equation (22), the closed-loop system of equations (28)
is Therefore,
taken to be asymptotically
T stable at the equilibriums
ϑ − f (0) , X , Ẋ = 0, if the following gains are satisfied, q
1 3 λmin (0) λm κp
1
3
1/2 ≥σ ≥
λMax (0)
1 1
ke 2 (λmin κi
−1
λmin κp )
λmin (κd ) ≥ λmin (0) λmin κp
1+
4 3 λMax (0)
(41)
− λmin (8)
" # The derivative of equation (26) is obtained as below,
1/2 λmin κp
1 1
λMax (κi ) ≤ λmin (0) λmin κp σ
6 3 λMax (0) V̇ = Ẋ T 0 Ẍ + Ẋ T κp X + ϑ T κi−1 ϑ + Ẋ T ϑ + Ẋ T ϑ + X T ϑ
2
3 σ T σ T
λmin κp ≥ [ + 4] + Ẋ 0 Ẋ + X 0 Ẍ + σ Ẋ T κd X + Ẋ T Fg (42)
(33)
2 2 2
Proof: The Lyapunov function can be stated as below, For matrix the inequality of below equation is
validated [28],
1 T 1 σ
V = Ẋ 0 Ẋ + X T κp X + ϑ T κi−1 ϑ + X T ϑ AT B + BT A ≤ AT 3A + BT 3−1 B (43)
2 2 4
σ σ
Z t
+ X T κd X + X T κd X + Fg dx − (34) The inequality of (42) is valid for any A, B ∈ Rn×m and
2 4 0 any 0 < 3 = 3T ∈ Rn×n , hence the scalar variable Ẋ T Fg
can be stated as,
where V (0) = 0. For representing that V ≥ 0, we divide it
into three parts, V = V1 + V2 + V3 , where, 1 T 1
Ẋ T Fg = Ẋ Fg + FgT Ẋ ≤ Ẋ T 3Fg Ẋ + FgT 3−1
Fg Fg (44)
2 2
σ
Z t
1
V1 = X T κp X + X T κd X + Fg dx − Utilizing equation (42) the following is obtained,
6 4 0
σ σ
≥ 0, κp > 0, κd > 0 (35) − X T 8Ẋ ≤ 48 X T x + Ẋ T Ẋ (45)
2 2
1 T σ 11
X κp X + ϑ T κi−1 ϑ + X T ϑ ≥ λm κp kxk2
V2 = where k8k ≤ 48 . Therefore, ϑ = ki , ϑ T ki−1 ϑ becomes x T ϑ
6 4 23
also x T ϑ becomes x T ki . Utilizing equation (45) the following
σ λmin (κi−1 ) is extracted,
+ kϑk2 − kxk kϑk (36)
4 h σ σ i
V̇ = −Ẋ T 8 + κd − 0 − 4 Ẋ
In a case that σ ≥ 3
, the following is 2 2
(λmin (κ −1
i )λmin (κ p ))
hσ σ i
obtained, − XT κp − κi − 4 X
2 2
!2 σ
− X T Fg − f (0) + Ẋ T f (0)
s
λmin (κ p ) (46)
r
1 3 2
V2 ≥ kxk − kξ k ≥0 (37)
2 3 4(λmin (κp ))
By applying the Lipschitz condition of equation (30) the
Also, following is obtained,
σ T σ
σ X f (0) − Fg ≤ κf kX k2
1 T 1 (47)
V3 ≥ X kp X + Ẋ T 0 Ẋ + X T 0 Ẋ (38) 2 2
6 2 2 σ σ
− X T [Fg − f (0)] ≤ X T X (48)
Since 2 2
From equation (42) we have,
X T AX ≥ kX k kAX k ≥ kX k kAk kX k ≥ λMax (A) kX k2
(39) Ẋ T f (0) ≥ −f T (0)3−1 f (0) (49)
By utilizing equation (38) torsional actuator with the developed controllers thus validat-
h σ σ i ing the effectiveness of the proposed control approach using
V̇ = −Ẋ T 8 + κd − 0 − 4 Ẋ PID controllers. A simulation period of 20s is considered for
hσ 2 2
σ σ i evaluation. For the simulation purposes, the weight of the
− XT κp − κi − 4 − κf X + Ẋ T f (0) (50)
2 2 2 torsional actuator is considered to be 5% of the motor and
Becomes pump weight in combination.
h σ σ i The Theorem proposed in this paper generates sufficient
V̇ ≤ −Ẋ T λmin (8) + λmin (κd ) − λMax (0) − 4 Ẋ conditions for the minimum amounts of the proportional as
2 2
T σ σ σ i
h well as the derivative gains. This Theorem validates that both
λmin κp − λmin (κ i )− 4 − X (51)
−X proportional and derivative gain must be positive as negative
2 2 2
gains can make the systems unstable. The PID gains are
Therefore, V̇ ≤ 0, kX k diminishes if the following selected within the stable range by the stability analysis in
conditions are held: order to ensure the efficiency.
σ Since the maximum flow rate of the pipeline is 13m3 s
λmin (8) + λmin (κd ) ≥ [λMax (0) + κc ]
nonlinear force associated with has to be less
2 so the other
2 than 13m3 s. Hence, we select = 13 m3 s, and
λmin kp ≥ λMax (k i ) + 4 +
σ
λmin (γ1 ) = 1.0002, λmin (γ2 ) = 1.000206, λmin (41 )
By utilizing equation (41) as well as λmin (κi−1 ) = λMax1(κ ) ,
i = 2.09, λmin (42 ) = 2.09
the following is obtained,
From Theorem 2, we use the following PID gains
1/2
1 1
λmin (κd ) ≥ λmin (0) λm κp λmin κp1 ≥ 453, λmin (κd1 ) ≥ 8, λMax (κi1 ) ≤ 928
(55)
4 3
4
× 1+ − λmin (8) (52) Also for pipe 2 we have,
λMax (0)
λmin κp2 ≥ 453, λmin (κd2 ) ≥ 8, λMax (κi2 ) ≤ 928
REFERENCES
[1] K. J. Astrom and B. Wittenmark, Computer-controlled Systems: Theory
and Design, 2nd ed. Englewood Cliffs, NJ, USA: Prentice-Hall, 1990.
[2] J. G. Ziegler and N. B. Nichols, ‘‘Optimum settings for automatic con-
trollers,’’ Trans. ASME, vol. 64, no. 1, pp. 759–768, 1942.
[3] J. C. Basilio and S. R. Matos, ‘‘Design of PI and PID controllers with
transient performance specification,’’ IEEE Trans. Edu., vol. 45, no. 4,
pp. 364–370, Nov. 2002.
[4] D. E. Rivera, M. Morari, and S. Skogestad, ‘‘Internal model control: PID
controller design,’’ Ind. Eng. Chem. Process Des. Develop., vol. 25, no. 4,
FIGURE 6. Comparison of motor vibration attenuation using PID pp. 252–265, 1986.
controller for pipeline 2. [5] J. C. Basilio, J. A. Silva, Jr., L. G. B. Rolim, and M. V. Moreira,
‘‘H∞ design of rotor flux-oriented current-controlled induction motor
drives: speed control, noise attenuation and stability robustness,’’ IET
Control Theory Appl., vol. 4, no. 11, pp. 2491–2505, 2010.
[6] J. E. F. Williams and B. C. Zhao, ‘‘The active control of vortex shedding,’’
J. Fluids Struct., vol. 3, no. 2, pp. 115–122, 1989.
[7] K. Roussopoulos, ‘‘Feedback control of vortex shedding at low Reynolds
numbers,’’ J. Fluid Mech., vol. 248, pp. 267–296, Mar. 1993.
[8] A. Baz and J. Ro, ‘‘Active control of flow-induced vibrations of a flexible
cylinder using direct velocity feedback,’’ J. Sound Vib., vol. 146, pp. 33–45,
Apr. 1991.
[9] E. Berger, ‘‘Suppression of vortex shedding and turbulence behind oscil-
lating cylinders,’’ Phys. Fluids, vol. 10, no. 9, pp. S191–S193, 1967.
FIGURE 7. Stability of flow rate with using of PID controller in pipeline 1. [10] S. Hiejima, T. Kumao, and T. Taniguchi, ‘‘Feedback control of vortex
shedding around a bluff body by velocity excitation.,’’ Int. J. Comput. Fluid
Dyn., vol. 19, no. 1, pp. 87–92, 2005.
[11] S. W. Sung, J. Lee, and I-B. Lee, Process Identification and PID Control.
Piscataway, NJ, USA: IEEE Press, 2009.
[12] A. Kusters and G. A. J. M. van Ditzhuijzen, ‘‘MIMO system identification
of a slab reheating furnace,’’ in Proc. 3rd IEEE Conf. Control Appl.,
Glasgow, U.K., Aug. 1994, pp. 1557–1563.
[13] A. O’Dwyer, Handbook of PI and PID Controller Tuning Rules, London,
U.K.: Imperial College Press, 2009.
[14] H. Choi, W.-P. Jeon, and J. Kim, ‘‘Control of flow over a bluff body,’’ Annu.
Rev. Fluid Mech., vol. 40, no. 1, pp. 113–139, 2008.
[15] S. S. Collis, R. D. Joslin, A. Seifert, and V. Theofilis, ‘‘Issues in active flow
FIGURE 8. Stability of flow rate with using of PID controller in pipeline 2. control: Theory, control, simulation, and experiment,’’ Prog. Aerosp. Sci.,
vol. 40, pp. 237–289, May/Jul. 2004.
[16] M. Gad-el-Hak, A. Pollard, and J.-P. Bonnet, Flow Control: Fundamentals
and Practices. Berlin, Germany: Springer, 1998.
controller are used, the flowrates initiate from zero and main- [17] E. A. Fadlun, R. Verzicco, P. Orlandi, and J. Mohd-Yusof, ‘‘Combined
taining stable flow rate, which proves the effectiveness of PID immersed-boundary finite-difference methods for three-dimensional com-
plex flow simulations,’’ J. Comput. Phys., vol. 161, pp. 35–60, Jun. 2000.
controller. [18] J. Mohd-Yusof, ‘‘Combined immersed-boundary/B-spline methods for
simulations of flow in complex geometries,’’ Annu. Res. Briefs, Center
Turbulence Res., NASA Ames Stanford Univ., vol. 1, no. 1 pp. 317–327,
V. CONCLUSIONS 1997.
In this paper, a novel active control strategy for the attenuation [19] A. Nandy, S. Mondal, P. Chakraborty, and G. C. Nandi, ‘‘Development of
of motor vibration is proposed which consequently controls a robust microcontroller based intelligent prosthetic limb,’’ in Proc. Int.
Conf. Contemp. Comput., 2012, pp. 452–462.
the flow rate in heavy-oil pipelines. The important theoretical [20] D. S. Park, D. M. Ladd, and E. W. Hendricks, ‘‘Feedback control of
contribution associated with the stability analysis for the PID von Kármán vortex shedding behind a circular cylinder at low Reynolds
controller is developed. The required stability conditions are numbers,’’ Phys. Fluids, vol. 6, no. 7, pp. 2390–2405, 1994.
[21] A. Herrán-González, J. M. De La Cruz, B. De Andrés-Toro, and
obtained for the purpose of tuning the PID gains. By utilizing J. L. Risco-Martín, ‘‘Modeling and simulation of a gas distribution pipeline
Lyapunov stability analysis, the sufficient conditions for the network,’’ Appl. Math. Model., vol. 33, no. 3, pp. 1584–1600, 2009.
[22] R. D. Whitaker, ‘‘An historical note on the conservation of mass,’’ J. Chem.
minimum amounts of the proportional, integrator as well as Educ., vol. 52, no. 10, p. 658, 1975.
the derivative gains are obtained. The numerical simulation [23] L. F. Moody, ‘‘Friction factors for pipe flow,’’ Trans. ASME, vol. 66, no. 8,
and analysis validates the effectiveness of PID controllers in pp. 671–684, 1944.
[24] G. O. Brown, ‘‘The history of the Darcy-Weisbach equation for pipe flow
the minimization of motor vibration to control the flow rate resistance,’’ Environ. Water Resour. Hist., vol. 4, no. 1, pp. 34–43, 2003
in pipelines. The main contributions of this paper are: [25] C. Roldán, F. J. Campa, O. Altuzarra, and E. Amezua, ‘‘Automatic identifi-
cation of the inertia and friction of an electromechanical actuator,’’ in New
1) In this work, the stability of PID controller is validated Advances in Mechanisms, Transmissions and Applications (Mechanisms
which has not been given importance in earlier researches and Machine Science), vol. 17. Dordrecht, The Netherlands: Springer,
considering the flow rate control. pp. 409–416. doi: 10.1007/978-94-007-7485-8_50.
[26] L. Wozniak, ‘‘A graphical approach to hydrogenerator governor tuning,’’
2) The technique of using torsional actuator on the motor- IEEE Trans. Energy Convers., vol. 5, no. 3, pp. 417–421, Sep. 1990.
pump arrangement is entirely a new concept. [27] K. De Cock, B. De Moor, and W. Minten, ‘‘A tutorial on PID control,’’
Future work is intended towards the development of the Katholieke Univ. Leuven, Leuven, Belgium, Rep. ESAT-SISTA/TR, 1997.
[28] A. V. den Bos, ‘‘Appendix C: Positive semidefinite and positive de finite
experimental setup for further investigation and the improve- matrices,’’ in Parameter Estimation for Scientists and Engineers. New
ment of the controller by fuzzy methods York, NY, USA: Wiley, 2007, pp. 259–263. doi: 10.1002/9780470173862.
SINA RAZVARZ received the B.S. degree in RAHELEH JAFARI received the B.S. degree in
mechanical engineering, heat and fluids from pure mathematics from Islamic Azad University,
Islamic Azad University, Iran, in 2009, and the Shabestar, Iran, in 2008, the M.S. degree in applied
M.S. degree in mechanical engineering from mathematics from Islamic Azad University, Arak,
Islamic Azad University, Science and Research in 2010, and the Ph.D. degree in automatic control
Branch, Tehran, in 2010. He is currently pursu- from the CINVESTAV-IPN (National Polytech-
ing the Ph.D. degree with the Departamento de nic Institute), Mexico City, Mexico, in 2017. She
Control Automtico, CINVESTAV-IPN (National is currently a Postdoctoral Research Fellow with
Polytechnic Institute), Mexico City, Mexico. His the Department of Information and Communica-
research interests include heat transfer, fluid tion Technology, University of Agder, Grimstad,
dynamic, fuzzy equation, neural networks, and nonlinear system modeling. Norway. Her research interests include fuzzy logic, fuzzy equation, neural
networks, artificial intelligence, fuzzy control, and nonlinear system mod-
eling. She serves as an Associate Editor for the Journal of Intelligent and
Fuzzy Systems.