See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/339554426

Control of Stock Consistency in Head Box Approach Flow System

Conference Paper · October 2019

DOI: 10.1109/CISCT46613.2019.9008146

CITATIONS READS
6 17

5 authors, including:

Pradeep Kumar Juneja Mayank Chaturvedi

Graphic Era University Griffith University
164 PUBLICATIONS   371 CITATIONS    60 PUBLICATIONS   192 CITATIONS   

SEE PROFILE SEE PROFILE

Amiya Kumar Ray Vikas Joshi

Indian Institute of Technology Roorkee Graphic Era University
65 PUBLICATIONS   364 CITATIONS    11 PUBLICATIONS   15 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

dec 2016 View project

Design of Smith Predictor for dead time compensation for a SOPDT process View project

All content following this page was uploaded by Mayank Chaturvedi on 18 November 2021.

The user has requested enhancement of the downloaded file.

 Control of Stock Consistency in Head Box Approach
Flow System
P K Juneja M Chaturvedi A. K. Ray
Dept. of Electronics & Comm. Enineering Dept. of Electrical Enineering Dept. of Paper Technology
Graphic Era (Deemed to be University) Graphic Era (Deemed to be University) IIT Roorkee
Dehradun, India Dehradun, India Roorkee, India
[email protected] [email protected] [email protected]

Vikas Joshi N Belwal

Dept. of Electrical Enineering Dept. of Electronics & Comm.
Graphic Era (Deemed to be University) Engineering
Dehradun, India Graphic Era (Deemed to be University)
[email protected] Dehradun, India
[email protected]

Abstract— It has been identified that head box of a wet end paper
machine is important to control to maintain the quality of feed liquor
for production of pulp for raw materials during cooking in digesters.
Controlling the head box efficiently will not only make the industry
sustainable in terms of good quality of production of paper and
paper products but also combats efficiently the solid waste disposal
problem by converting this waste to wealth. Presence of delay time
and availability of different transfer functions for the same
parameter challenge the situation further. Therefore, an attempt has
been made to solve this issue.

Keywords – consistency; headbox; FOPDT; dead time

I. INTRODUCTION
Consistency is one of the most significant variable for paper
machine, as without its information it is beyond the realm of
Fig. 2. Consistency control with white water
imagination to expect to improve efficiency with ideal quality.
Generally the dilution water from different sources is added to the Feedback provided by consistency sensor is transmitted
thick stock preceding fan pump and after that prompted stream to through a transmitter to the controller. An error signal is generated
a consistency sensor, before feeding it to the headbox. by comparing measured variable with the set point. This error
signal is then provided to the controller to regulate valve position
which is governing the flow of dilution water to the stock.
Site of the sensor, loop scheme and process nature determines
the value of the transportation lag for the consistency control.
Consistency control can be practiced subsequent to blending chest.
A graphic representation is shown in Fig. 1 and in Fig. 2.
II. LITERATURE SURVEY
David wood [1] provided an exhaustive report for the
contemporary model of headbox control, including maintaining
the consistency of paper. According to the report, consistent
properties can be achieved by arranging the slice adjacent to the
parallel by alleviating the headbox and slice lip mounting against
Fig. 1. Dilution before entering headbox for consistency control temperature deviations, through the paper breadth. He further
emphasized that the automation and adjustable slice lip can
improve headbox control.

978-1-7281-2443-8/19/$31.00 ©2019 IEEE

Authorized licensed use limited to: Griffith University. Downloaded on December 23,2020 at 23:28:21 UTC from IEEE Xplore. Restrictions apply.
 Talvio [2] executed hypothetical and experimental studies
analysis of control and stability for the headbox of paper machine. 0.5
Step Response

An extensive survey on the advancements of control of paper G1

pade 1
manufacturing headbox was conducted by Weber et al. [3]. He 0
pade 2

emphasized upon the headbox for twin-weir and fourdrinier pade 3

pade 5
machine with their equipment and control description. pade 15
-0.5 pade 20

Tuladhar et al [4] analyzed simulation results of mill test on an

Amplitude
operating paper machine and MD basis weight signal flow. The -1

result indicated the presence of very low frequency variation,

caused by consistency dynamics. The simulation result indicated -1.5
that the decoupled control action is feasible using MPC controller.
A case study was presented by Paattilamm J. et al. [5] on a -2

paper machine headbox control. Kumar Prasanna et al [6] has

described paper machine control and optimization for different -2.5
0 5 10 15 20 25 30 35 40 45
sub system of paper machine loop including consistency control Time (sec)

in approach flow, headbox, mainly single variable and total head

have been controlled traditionally by controlling air pad for stream Fig. 3. Step responses for G1 transfer function without approximation (blue) and
flow valve, the recirculation valve or by fan pump speed control. Padè approximation of first (green), second (red), third (cyan), fifth (magenta),
fifteenth (yellow) and twentieth (black) order
Aidun et al [7] investigated two classes of secondary streams
in headbox resulting from symmetrical properties, kinematics and TABLE I. STEP RESPONSE CHARACTERISTICS FOR DIFFERENT PADÈ
APPROXIMATIONS FOR CONSISTENCY PARAMETER IN A HEADBOX
the others, from unsettled signal of the flow field and the prototype
established is examined by means of direct arithmetic simulation.
Percentage Percent error
Nissinen, et al. [8] studied the feasibility of designing error in rise in settling
multivariable PI controller for headbox with rectifier rolls without Order of Padè Rise Settling time in time in
overflow provided with air cushion in a multi-grade paper time, time, comparison comparison to
approximation
sec sec to without without
machine. The system has been considered as MIMO system approximat approximatio
instead of SISO system. The process dynamics for such system ion n
was identified. Based on the process model and the structure of the
multivariable PI controller a process simulator was built. The without dead time
simulator was used to test different tuning methods which could 8.44 21.9 0 0
approximation
be applied to the system.
On-line real time SISO feedback control of recycled white With Ist order
11.5 23.0 36.25 5
water consistency can be a way to retention control. Thick stock approximation
wet end chemistry processes and machine direction basis weight
variation play key roles in this regard. Hence papermaker’s control With second order
9.39 21.3 11.25 2.73
strategy should be to minimize the deviation in the recycled white approximation
water consistency [9, 10].
With third order
Small errors in the headbox and tray consistency and ash data 8.83 21.6 4.62 1.36
approximation
can produce large variation in the calculated first pass retention
values [11]. With fifth order
8.58 21.7 1.65 0.913
approximation
III. DEAD TIME APPROXIMATION IN CONSISTENCY PROCESS
DYNAMICS With fifteenth
order 8.42 21.8 0.236 0.45
Generally, consistency offers a first order with time delay approximation
response [12] and consequently it is a significant candidate for
executing the dead time approximation for additional With twentieth
order 8.42 21.8 0.236 0.45
investigation. The dead time expression comprises an exponential approximation
term in the transfer function and to handle dead time in Laplace
domain, it is expected to use approximations for the dead time In this section, dead time approximation of consistency process
expression. The Padè approximation is quite advantageous in transfer function, for various orders of Padè approximations viz
control study in this respect [13, 14]. first, second, third, fourth, fifth, tenth, fifteen and twentieth order
approximation is performed for comparative analysis of step

978-1-7281-2443-8/19/$31.00 ©2019 IEEE

Authorized licensed use limited to: Griffith University. Downloaded on December 23,2020 at 23:28:21 UTC from IEEE Xplore. Restrictions apply.
responses and exhibited by Fig. 3. The consistency transfer TABLE II. FIVE TRANSFER FUNCTION MODELS FOR CONSISTENCY
AVAILABLE IN LITERATURE
function can be modeled as FOPDT model as under [15, 16]:

constant (τ)

Bump tests
Delay time

value of K
Bump test
Gain (K)

Transfer
function
System

θD / τ
Time
( )=

(θD)
(1)

%
(1 + )

−2.035 .
where, θD is the transportation. G1
-
3.84 6.84 1.78 5.0
-0.0407
2.03 Span=2% (1 + 3.84 )
Nancy formulated the following dynamics with time delay of
the order of 6.84 seconds due to transmitter location comparative 0.03 0.035
to the dilution point. The sensor measurement dynamics cased a -
G2 0.75 10 5 0.5 span = (1 + 10 )
time constant of 3.84 seconds. 16.67
4%

-0.042
−0.0407 . G3 -1.4 3 3 1 -5.0 Span=
−1.4
( )= (2) (1 + 3 )
(1 + 3.84 ) 3%

- -0.0625 −2.08
G4 5 5 1 -5.0
From Table I, it is clear that in Padè approximation for first 2.08 Span=3% (1 + 5 )
order the percentage error in rise time is very large i.e. 36.25% as
compared to Padè approximation for fifteenth order, for which the - -0.0386 −1.93 .
G5 3.51 5.7 1.62 4.91
error is 0.236%. The error is same for fifteenth and twentieth order. 1.93 Span=2% (1 + 3.51 )
The error is decreasing with increase in order of approximation.
Similarly, percentage error in settling time for Padè first order
approximation is 5% and for fifteenth order is 0.236%. The error It is obvious that their input and output conditions are different
is same for twentieth order approximation. The error is decreasing for all the cases though the transmitter ranges and nominal values
with increase in order of approximation. The higher the order of are nearly the same. The step responses for these five transfer
the approximation used, the more accurate it is at higher function models are shown in Fig. 4 and their corresponding Bode
frequencies. plots in Fig. 5.
−1.93 .
IV. CONSISTENCY CONTROL LOOP ANALYSIS WITH KNOWN ( )= (3)
PROCESS DYNAMICS (1 + 3.51 )
The first step is to identify the process transfer function in the
analysis of dynamics and control. Analysis of first principles, i.e. Step Response
the conservation principles and in-plant experiments are two 0.5

approaches possible to derive the process transfer function. The G1

G2
experimental approach is used more often, which involves 0
G3
performing a “bump test”. A bump test is an online process which G4
involves testing the effect of small step change in controller output -0.5 G5
on process and manipulated variables. Many bump tests should be
Amplitude

done to ensure the results to be representative. -1

Five different transfer function models are selected for the

consistency parameter available in literature as shown in Table II. -1.5
It is interesting to note that the dead times are different; in two
cases (G1(s) and G5(s)) the value of dead time is higher than -2
process time constant. The two cases (G3(s) and G4(s)) have the
dead time value equal to time constant and in one case G2(s) the -2.5
dead time value is equal to time constant. It is obvious that their 0 10 20 30 40 50 60

input and output conditions are different for all the cases though Time (sec)

the transmitter ranges and nominal values are nearly the same. Fig. 4. step responses for five process models

978-1-7281-2443-8/19/$31.00 ©2019 IEEE

Authorized licensed use limited to: Griffith University. Downloaded on December 23,2020 at 23:28:21 UTC from IEEE Xplore. Restrictions apply.
 Table IV shows the comparison between the different models
40
Bode Diagram
of consistency parameter for stability margins. Although all the
20
G1oltf
G2oltf
models are stable but the model with G3(s) transfer function
model has a minimum gain margin and phase margin compared
Magnitude (dB)

G3oltf

to others. The models G1(s) and G5(s) have same gain margin. The
0 G4oltf
G5oltf
-20
model G5(s) has the maximum gain margin and phase margin
-40 requirements.
-60
270 V. CONCLUSION
225
G3(s) gave the best step response among these five transfer
Phase (deg)

180 functions in terms of desirable minimum values of characteristics,

135
rise time and settling time. The process model with G5(s) process
transfer function is relatively more stable than others and G3(s) is
90
10
-2
10
-1
10
0
10
1 2
10
the least stable. In relative comparison of stability for all the
Frequency (rad/sec) models the sequence is G5(s), G1(s), G2(s), G4(s), and G3(s) in
decreasing order of stability.
Fig. 5. Bode diagram for five different models indicating stability margins
REFERENCES
TABLE III. COMPARISON OF IMPORTANT CHARACTERISTICS FOR
DIFFERENT CONSISTENCY TRANSFER FUNCTION MODELS [1] D. wood, “Head box control developments,” Paper Technology, pp34-38,
June 1995.
Rise time Settling time [2] P.A. Talvio, “A study of paper machine headbox control system with linear
Model
(sec) (sec) transfer function,” IFAC,session 22, paper 22A, London, 1966
[3] R.Weber, N.Y.Gaitonde, “Non-interactive distillation tower analyzer
G1(s) 8.44 21.9 control,” Proc. Ame. Control conference, Arlington, VA, p.87, 1982.
[4] A. Tuladhar, M.S .Devies, C.Vim, G.R.Woods, “Headbox modeling and wet
G2(s) 22 44.1 end pressure pulsation analysis,” Journal of pulp and paper Canada,
98:9,III, pp.91-94,1997.
G3(s) 6.59 14.7 [5] J. Paattilammi, P.M. Makila, “Fragility and robustness : a study on paper
machine headbox control,” Control Systems, IEEE, Volume 20, Issue 1,
G4(s) 11 24.6 2000, pp. 13 - 22.
[6] V.S. P. Kumar, M.S. Rant, “Paper machine process control optimization,”
IPPTA Journal, vol.-9,no.2, pp.1-8, June 1997.
G5(s) 7.71 19.4
[7] C. K. Aidun., E.K. Agnes, “Hydrodynamics of the forming section: the
origin of non-uniform fiber orientation,” TAPPI Journal, vol.78, no. 11,
pp.97-106, Nov. 1995.
TABLE IV. COMPARISON BETWEEN THE DIFFERENT MODELS OF [8] A.I. Nissinen, H.N. Koivo, T.J. Huhtelin, “Headbox control using a
CONSISTENCY PARAMETER FOR STABILITY MARGINS multivariable PI controller in a distributed automation system,” Pulp &
paper Canada, pp.38-41, 98:5, 1997.
Closed [9] P.K. Juneja, M. Chaturvedi, S. Suman and K. Antil, “Modeling of Stock
Wg Wp
System Gm (dB) Pm (deg) loop Consistency in the Approach Flow System of the Headbox,” In 2018 3rd
(rad/sec) (rad/sec)
stable International Conference On Internet of Things: Smart Innovation and
G1 8.11 55.5 0.342 0.113 Yes Usages (IoT-SIU), IEEE, 2018, pp. 1-4.
[10] P.K. Juneja, M. Chaturvedi, A K Ray, G Yadav, “Modeling for headbox and
associated wet end systems,” In Proceedings of ICETIT 2019, Springer,
G2 7.99 54.5 0.427 0.160 Yes Cham, pp. 396-406, 2020.
[11] P.K. Juneja, A.K. Ray, R. Mitra, “Modelling deadtime in consistency
G3 6.65 42.7 0.691 0.311 Yes process dynamics in a paper mill,” Macmillan Publishers India Ltd. in
proceedings of Advances in Chemical Engineering - AchemE 2009, Patiala,
G4 6.82 44.5 0.419 0.183 Yes February, 2009, pp 201-205.
[12] P.K. Juneja, A.K. Ray, R. Mitra, “Various PID controller algorithms for
closed loop performance of consistency parameter of paper machine, IPPTA
G5 8.11 58.6 0.418 0.133 Yes , Vol. 23, No. 2, June 2011.
[13] P.K. Juneja, N. Jain, M Chaturvedi, S K Singh, “Effect of delay
As clear from Table III, G3(s) has the minimum rise time value approximation using Padè technique on controller performance designed for
a SOPDT model,” ICSNCS-2016, LNEE, Vol. 396, Springer pp 181-187,
of 6.59 seconds and G2(s) has the maximum rise time of 22 2016.
seconds. The settling time is least for G3(s) with value of 14.7 [14] M. Chaturvedi, P. K. Juneja, “Effect of variation in filter coefficients for
seconds and maximum for G2(s) with a value of 44.1 seconds. It different PID controller structures on performance of SOPDT process,”
means that G3(s) has both these parameters minimum. ICSNCS-2016, LNEE, Vol. 396, Springer pp. 249-254, 2016.

978-1-7281-2443-8/19/$31.00 ©2019 IEEE

Authorized licensed use limited to: Griffith University. Downloaded on December 23,2020 at 23:28:21 UTC from IEEE Xplore. Restrictions apply.
[15] M. Chaturvedi, P. Chauhaan, P K Juneja, “Design of time delay compensator
for a FOPDT process model,” ICSNCS-2016, LNEE, Vol. 396, Springer, pp.
205-211, 2016.
[16] D. Naithani, M Chaturvedi, P. K. Juneja, “Controller performance analysis
for a delayed process based on Integral Error performance indices,” Oriental
Journal of Chemistry, Vol. 32, No. 3, pp. 1671-1674.

978-1-7281-2443-8/19/$31.00 ©2019 IEEE

Authorized licensed use limited to: Griffith University. Downloaded on December 23,2020 at 23:28:21 UTC from IEEE Xplore. Restrictions apply.
View publication stats